Advances and Challenges in Low-Resource-Environment Software Systems: A Survey

Abstract

A low-resource environment has limitations in terms of resources, such as limited network availability and low-powered computing devices. In such environments, it is arguably more difficult to set up new software systems, maintain existing software, and migrate between software systems. This paper presents a survey of software systems for low-resource environments to highlight the challenges (social and technical) and concepts. A qualitative methodology is employed, consisting of an extensive literature review and comparative analysis of selected software systems. The literature covers academic and non-academic sources, focusing on identifying software solutions that address specific challenges in low-resource environments. The selected software systems are categorized based on their ability to overcome challenges such as limited technical skills, device constraints, and socio-cultural issues. The study reveals that despite noteworthy progress, unresolved challenges persist, necessitating further attention to enable the optimal performance of software systems in low-resource environments.

1. Introduction

The goal of software development is to solve problems by building, designing, and deploying software for use in diverse environments [1,2]. Software applications simplify daily tasks and have become essential components in virtually every organisation and community. In an increasingly digital globalised society, economic, social, and political activities rely on technology, and those without access to relevant ICT resources risk being left behind [3]. This disparity is particularly evident in low-income countries, often referred to as low-resource environments or low-resource settings, which lack resources that are readily available in high-income countries [4]. The availability of ICT resources, including software systems, is crucial for socioeconomic growth in any country [5].

Significant lessons have been learned from the successes and challenges in Information and Communication Technologies for Development (ICTD or ICT4D) research [6], yet key challenges persist, especially regarding the low-income levels and limited technological innovations that typify low-resource environments. Development challenges identified by ICTD researchers in these settings include cost constraints, limited connectivity, low literacy levels, linguistic diversity, limited power availability, and accessibility issues [7,8,9,10]. It is also important to recognise that software systems require varying resources depending on the environment, such as computing devices, Internet access [11], technical skills, funding, and data. For instance, as reported by Asudeh et al. [12], networks connecting remote locations are often slow and unreliable, meaning that tasks involving critical data transmission and multisite collaboration must be carefully planned and executed. This affirms the notion that software resources cannot be generalised across all contexts due to the uneven distribution of essential resources. Low-resource environments frequently lack technical expertise, high-capacity computing devices, and high-speed Internet access—all of which impact the feasibility of implementing and maintaining software systems in ways comparable to high-resource settings [13].

In line with these arguments and to gain a deeper understanding of the challenges in low-resource environments, we formulate an open-ended research question (RQ) as follows:

RQ: “What are the issues affecting the use of software in low-resource environments, and how do software systems in these settings overcome these challenges?”

The main contribution of this paper is a comprehensive survey of software systems designed for low-resource environments, offering valuable insights into both challenges and solutions. By identifying the challenges impacting software development, usage, and maintenance in low-resource settings and categorising existing software systems based on their effectiveness in addressing these issues, this paper provides a robust framework for understanding and evaluating software development in this domain. Additionally, it emphasises the persistent challenges in low-resource environments, underscoring the need for continued research and innovation. Ultimately, this work not only enhances our understanding of software systems in low-resource settings but also offers a roadmap for future research and development efforts in this critical area. It serves as a valuable resource for various stakeholders, including academic researchers, software developers, non-governmental organisations, policymakers, humanitarian and relief organisations, educational institutions, and technology consultants, by informing research, guiding policy decisions, and supporting effective software development in resource-constrained contexts.

We recognise that these challenges encompass both social and technical dimen- sions [7,14,15]. In the context of software development in low-resource environments, social issues are primarily related to human factors, culture, and societal context, while technical challenges pertain to infrastructure, technology, and expertise. Socially, challenges include limited access to technology and low digital literacy, which result in a scarcity of adequate devices, such as computers and smartphones, thus impeding the adoption of software solutions. Low literacy levels also contribute to a shortage of skilled professionals needed to develop and maintain software systems, leading to expertise gaps. Technically, one major challenge is the use of low-end devices with limited processing power, memory, and storage. Additionally, intermittent connectivity, often found in remote or rural areas, and power constraints, including unreliable or intermittent power supplies, further limit the effective use of software systems in these environments. A holistic understanding of these interconnected social and technical challenges is essential for successful software development in low-resource settings.

By categorising challenges into social and technical issues, our study provides a practical framework for developers and researchers working in low-resource settings. This classification allows practitioners to prioritise design choices that address specific constraints encountered in these environments, serving as a guide for tailored software solutions that better meet the needs of underserved communities. The categorisation also informs actionable design strategies: for example, technical constraints such as limited device capacity suggest the use of lightweight and energy-efficient solutions, while social constraints, such as low literacy levels, highlight the need for intuitive, culturally sensitive interfaces. These insights are intended to help practitioners adapt their development processes to low-resource settings effectively.

In the remainder of the paper, we first present the methodology used for this survey. We then provide an overview of low-resource environments and examine the challenges that affect software deployment, usage, and maintenance within these settings. Following this, we discuss how these challenges have historically posed—and continue to pose—obstacles to the effective use of software in low-resource environments. Next, we explore specific software systems designed for low-resource contexts, detailing how they address these challenges and categorising them based on their application domains. Finally, we offer insights into future directions and trends in the development and use of software for low-resource environments.

2. Methods

The methodology adopted for this study is primarily qualitative, focusing on a detailed review and comparative analysis of software systems designed for low-resource environments. A systematic approach was employed to explore the existing literature, academic papers, case studies, and technical reports related to software challenges in low-resource environments. The research identifies the key obstacles faced in software deployment and usage in low-resource settings and examines how some selected software systems overcome these challenges. This approach allows for a comprehensive understanding of the social and technical challenges that characterise low-resource environments.

Data collection was conducted through secondary sources, with extensive searches performed on academic databases such as Google Scholar, ACM Digital Library, and ScienceDirect. Key search terms, including “low-resource software”, “software systems in low-resource settings”, and “Software for low-income environments”, were utilised to gather relevant studies from both academic and non-academic sources. This broad search strategy ensured the inclusion of various perspectives, capturing both theoretical and practical insights. Non-academic searches through platforms such as Google also provided valuable information on real-world software applications in low-resource environments.

The selection of the software systems for analysis followed specific inclusion criteria. Only systems explicitly designed for low-resource environments and capable of addressing one or more of the identified challenges, such as low network connectivity, limited device capability, or socio-cultural issues, were included. These systems were further evaluated based on their cost-effectiveness, technical feasibility, and deployment in regions such as Sub-Saharan Africa, Latin America, and parts of Asia. The comparative analysis primarily mapped these software systems to the specific challenges they addressed, offering insights into their strengths and limitations in real-world applications. The analysis further included a comparative framework to classify the selected software systems based on their ability to tackle the challenges in low-resource settings. Social challenges such as low literacy and limited technical skills were examined alongside technical barriers such as unreliable networks and lack of device capability.

The analysis provided a holistic view of how different software solutions were tailored to meet the unique needs of low-resource environments, highlighting their practical applications and areas for future development. Despite the broad scope of the review, the study acknowledges the limitations posed by the availability of detailed implementation data, particularly regarding the long-term effectiveness of software systems. Additionally, while the survey covers a wide range of software systems, not all possible software systems could be included based on the scope of the searched literature.

3. Low-Resource Environments

This section presents low-resource environments, their associated challenges, and the classification of the challenges affecting software development, usage, and maintenance.

Low-resource environments have been considered from various perspectives, including settings that lack infrastructure, have limited human capacity, and exhibit low levels of basic and digital literacy [7,14,15]. As emphasised by Akhigbe et al. [16], a low-resource environment can also refer to resource-poor settings, limited-resource contexts, low- and middle-income countries, and under-resourced areas. These terms collectively describe settings or regions where resources are scarce, such as those found in Sub-Saharan Africa, parts of Asia, and Latin America and the Caribbean (https://www.worldbank.org/en/region/lac/overview (accessed on 20 March 2024)). The Human Development Index (HDI) (https://hdr.undp.org/data-center/human-development-index#/indicies/HDI (accessed on 20 March 2024)), a statistical tool used to assess a nation’s success across its social and economic dimensions, evaluates a country’s social and economic characteristics based on citizens’ health, educational attainment, and standard of living. In most low-resource environments, these indicators reveal significant resource limitations [15].

3.1. Challenges in Low-Resource Environments

Low-resource environments encounter a variety of challenges, including organisational, infrastructural [14], and human resource issues [17]. Ramanujapuram and Malemarpuram [18] recognise four major challenges in low-resource environments that primarily limit the adoption and use of technology: resource unavailability, limited technical capabilities, lack of a culture of data recording and use, and limited managerial capacity. Similar to these challenges, Ramanujapuram and Akkihal [17] recognise human, technical, and environmental challenges. These three categories represent significant barriers to optimal software adoption in low-resource settings, making it difficult to implement any form of data-driven decision support system.

Van-Zyl et al. [15] recognise nine major issues that characterise low-resource environments: geographical and environmental factors, financial shortages, lack of knowledge, underdeveloped infrastructure, restricted social resources, influence of beliefs and practices, human resource limitations, suboptimal healthcare service delivery, and research challenges.

Moreover, Bon et al. [7] highlight the need for an adequate framework to develop ICT services, including software development and use, in low-resource contexts due to challenges such as unidentified local needs; areas with significant low literacy; lack of essential technology and infrastructure; low purchasing power; limited basic ICT skills; misaligned goals between donors and beneficiaries; and lack of sustainability planning. These challenges, as identified across various studies, are presented in Table 1, with some overlap among them.

From these challenges, those affecting software deployment, maintenance, and usage were considered in accordance with the scope of this study. They were classified into eleven categories: network issues, power issues, expertise limitations, low literacy, economic factors, device constraints, socio-cultural issues, collaboration difficulties, data management issues, managerial capacity, and language barriers.

Table 2. Software usage challenges categorisation in low-resource environments.

Software Usage Challenges	Development Constraints	Other Related Challenges
Network issues	Technical	Technology and infrastructure unavailability [7], lack of facilities [14], underdeveloped infrastructure [15], unreliable mobile or wired networks [17]
Low literacy	Social	Paucity of knowledge [15], low literacy rate [7], accessibility and technological illiteracy [19]
Electricity issues	Technical	Lack of facilities [14], underdeveloped infrastructure [15]
Economic factors	Social	Lack of funding [14], low end-user purchasing power [7], financial shortages [15], lack of funding [16], affordability [19]
Device issues	Technical	Technology and infrastructure unavailability [7], lack of facilities [14], underdeveloped infrastructure [15], limited technology capability [18]
Economic factors	Economic	Lack of funding [14], low end-user purchasing power [7], financial shortages [15], lack of funding [16], affordability [19]
Socio-cultural issues	Social	Influence of beliefs and practices [15], unknown local needs [7], restricted social resources [15]
Collaboration issues	Social	Mismatch between donor-sponsor and end-user goals [7], research challenges and considerations [15], unknown context [7], underutilisation of ICTs [19]
Data issues	Technical	No culture of data recording and use [18], restricted information network [17]
Expertise issues	Technical	Lack of technical skills [14], paucity of knowledge [15], lack of critical skills [16], human resource limitations [15], low human capacity [18]
Managerial capacity issues	Social	Sustainability issues [7], human resource limitations [15], limited management capacity [18]
Language barrier issues	Social	Unknown local needs [7], unknown context [7]
Environmental challenges	Environmental	Geographical and environmental constraints [15], climate-related factors impacting technology access and sustainability [16]

lists these software-related challenges alongside other closely related challenges in low-resource environments. For instance, economic factors—a core challenge for software deployment in low-income countries—encompass related issues such as lack of funding [14,16], low end-user purchasing power [7], financial shortages [15], and affordability constraints [19] as identified in the literature.

Table 1. Software usage challenges from the literature.

Study	Study Area	Study Focus	Challenges
Bon et al. [7]	Burkina Faso	A framework for the development of ICT services in low resource settings.	Unknown local needs, low literacy rate, unknown context, technology and infrastructure unavailability, low end-user purchasing power, lack of ICT knowledge, mismatch between donor-sponsor and end-user goals, sustainability issues.
Suleman [14]	South Africa	Examining the design principles behind digital repositories and how well they work in low-resource environments.	Lack of technical skills, lack of facilities, lack of funding.
Pascoe et al. [20]	Tanzania	Improving the routine reporting of health data from primary health facilities to district levels by utilizing mobile phone applications to enhance timeliness and completeness of Integrated Disease Surveillance and Response (IDSR)	Poor infrastructure, the remoteness of health facilities from district offices, high transport costs for health workers, and difficulties in timely reporting of routine health data.
Morawczynski [21]	Kenya	Exploring the usage and impact of M-PESA as a mobile financial service in transforming financial practices and enhancing economic accessibility for users in both urban and rural settings	Limited access to formal financial institutions, the risks and costs associated with informal money transfer channels, and the unintended negative consequences of mobile financial services
Awokola et al. [22]	Nigeria	Practical challenges encountered in implementing an electronic medical record (EMR).	Resistance to change among healthcare workers, lack of computer literacy, inadequate technological infrastructure, high costs of implementation, and difficulties in accessing reliable Internet for system support and maintenance.
Ramanujapuram and Malemarpuram [18]	India	Developing supply chain management software for public health supply chains	Resource unavailability, limited management capacity, limited technology capability, no culture of data recording and use
Kamanga et al. [23]	Zambia	Utilizing mobile telephones for active case detection of malaria in rural health centres to improve timely reporting and management of malaria cases in unstable transmission conditions.	Inconsistent data collection, lack of diagnostic tests, and limited reliable communication systems.

Table 1. Software usage challenges from the literature.

Study	Study Area	Study Focus	Challenges
Bon et al. [7]	Burkina Faso	A framework for the development of ICT services in low resource settings.	Unknown local needs, low literacy rate, unknown context, technology and infrastructure unavailability, low end-user purchasing power, lack of ICT knowledge, mismatch between donor-sponsor and end-user goals, sustainability issues.
Suleman [14]	South Africa	Examining the design principles behind digital repositories and how well they work in low-resource environments.	Lack of technical skills, lack of facilities, lack of funding.
Pascoe et al. [20]	Tanzania	Improving the routine reporting of health data from primary health facilities to district levels by utilizing mobile phone applications to enhance timeliness and completeness of Integrated Disease Surveillance and Response (IDSR)	Poor infrastructure, the remoteness of health facilities from district offices, high transport costs for health workers, and difficulties in timely reporting of routine health data.
Morawczynski [21]	Kenya	Exploring the usage and impact of M-PESA as a mobile financial service in transforming financial practices and enhancing economic accessibility for users in both urban and rural settings	Limited access to formal financial institutions, the risks and costs associated with informal money transfer channels, and the unintended negative consequences of mobile financial services
Awokola et al. [22]	Nigeria	Practical challenges encountered in implementing an electronic medical record (EMR).	Resistance to change among healthcare workers, lack of computer literacy, inadequate technological infrastructure, high costs of implementation, and difficulties in accessing reliable Internet for system support and maintenance.
Ramanujapuram and Malemarpuram [18]	India	Developing supply chain management software for public health supply chains	Resource unavailability, limited management capacity, limited technology capability, no culture of data recording and use
Kamanga et al. [23]	Zambia	Utilizing mobile telephones for active case detection of malaria in rural health centres to improve timely reporting and management of malaria cases in unstable transmission conditions.	Inconsistent data collection, lack of diagnostic tests, and limited reliable communication systems.

All these challenges (as contained in Table 1 and Table 2) are a mixture of both social and technological challenges. Worthy of note is that these challenges often arise from one another. That is, one of these challenges can sometimes lead to or cause another. For example, a low literacy level can result in a shortage of skilled professionals, which leads to a lack of expertise. These interactions between these challenges are illustrated in Figure 1. The eleven categories of challenges are classified into social and technical challenges as follows:

• Social issues

  -

  Low literacy

  -

  Economic factors

  -

  Socio-cultural issues

  -

  Managerial capacity issues

  -

  Language barrier

  -

  Collaboration issues

• Technical issues

  -

  Network issues

  -

  Electricity or power issues

  -

  Device issues

  -

  Data issues

  -

  Expertise issues

Figure 1. Low-resource software challenges and their interactions.

Figure 1. Low-resource software challenges and their interactions.

The division into social and technical dimensions presents an interesting view of these challenges, which indicates that most of the issues revolve around societal problems. Highlighting the persistent challenges faced by low-resource environments such as African nations in terms of poverty, disease [24], ignorance, inequality, and lack of opportunity, Osei-Hwedie [25] recognises the need for better and more effective ways to tackle these problems in order to improve the welfare and wellbeing of people, for which technology innovation is key.

These eleven categorised challenges are discussed to understand how they impact software system development in low-resource environments.

3.1.1. Low Literary Issues

In low-resource environments, low literacy has been a longstanding issue [7,26,27]. Despite recent efforts to address this, a significant literacy gap remains compared to high-income countries with abundant resources. Most products, services, usage models, and research on ICT use in education originate from high-income contexts. This presents a recurring challenge for education policymakers and planners aiming to implement ICT in rural, low-income areas globally, as the literacy levels of local populations often limit access to ICT facilities [7,28]. Consequently, technology-enabled “solutions” are frequently imported and adapted to fit conditions that are substantially more challenging in low-resource contexts.

For instance, some software solutions developed for remote areas of Africa or isolated settlements in the Andes are often designed elsewhere, with limited understanding of the real-world circumstances in which these technologies will be used [7]. Many individuals who have lived and worked in low-resource environments are familiar with well-intentioned but relatively costly interventions. Combined with low literacy levels among the local population, this has often rendered technological innovations unsustainable or led to flawed user-centred design approaches.

In addition to established views on literacy in resource-poor environments, Kowalski et al. [27] highlight language barriers as a significant challenge in Haiti, a particularly resource-constrained setting. As of 2016, 95% of Haitians spoke Haitian Creole as their mother tongue [29], yet education is predominantly conducted in French, the government’s official language, which fewer than 5% of Haitians speak. With only a small percentage of pupils completing primary school and less than 10% completing secondary education, early disadvantages in literacy are difficult to overcome. This issue is compounded by underqualified and insufficiently trained teachers, resulting in less than ideal educational outcomes for those who remain in school [26,27,30].

3.1.2. Economic Factors

The intended end users have relatively little purchasing power in low-resource environments. As a result, despite being technically possible, many services may not be financially viable for various reasons. According to Jere et al. [19], the high cost of ICT equipment is a concern in low-resource environments, as their study found out that limited income is one of the issues in low-resource environments. Van-Zyl et al. [15] pointed out that financial uncertainties appear to be a vital component when defining low-resource environments, as evidenced by the existence of content categories related to financial pressure in every layer of the socio-ecological model. The underlying content categories shed light on crucial variables that contribute to financial resource uncertainty, such as insufficient income, a lack of healthcare insurance, reliance on subsidised healthcare, unemployment, subsistence employment, and undernutrition [31].

As a result of limited financial resources and revenue, which necessitates choosing between multiple investment options, persuading governments or other stakeholders to invest in technology and creative applications is typically a difficult undertaking in many low-resource environments [32,33,34]. It was suggested by Godfrey et al. [32] that technologies developed by multinational firms with headquarters in resource-rich nations do not always only find utility in environments with limited resources. Therefore, when using technology at scale in a geographically diversified, low-resource/middle-income country, the economic/financial and other social issues must be taken into consideration. It is important to note that there was a dearth of knowledge supporting the application of any new technologies outside of resource-rich countries for usage in low-income regions due to many economic factors [15].

3.1.3. Socio-Cultural Issues

Socio-cultural issues refer to customs, practices, behaviours, and ideas that are shared by a demographic group. It is a phrase associated with social and cultural variables and factors influencing how individuals make judgments in a society. It has been noted that taking into account the context, needs, vulnerabilities, and priorities of nations, there is a critical need to build resource capacity for the proper deployment of information systems. In this regard, among other issues, special focus has been placed on socio-cultural issues in low- and middle-income nations [35]. Resistance to change is one of the socio-cultural issues that has been highlighted and needs to be dealt with in low-resource settings.

Godfrey et al. [32] note the difficulties posed by socio-cultural issues and conclude that when utilizing technology at scale in a geographically diverse, low-resource/middle-income region, cultural challenges—such as differences between urban and rural life, economic/financial factors, and social influences—must be considered. Thus, the adoption of technology, even when described as low-cost, may have substantial functional repercussions, some of which could limit its usefulness if socio-cultural concerns are disregarded [36,37].

3.1.4. Managerial Capacity Issues

The ability to effectively manage capacity is likely the most important aspect for the long-term success of most IT-related projects and organisations. The need for managerial skills is critical even in developed economies since the quick pace of technological innovation results in short product lifecycles, low production yields, and frequently, extensive lead periods for production. Even though organisations may develop inventories or keep excess capacity to protect against demand fluctuation in a sustained demand environment, most are hesitant or unwilling to take on such financial risks in a down market [38]. In support of this, Ssennyonjo et al. [39] find that even though various innovations improve providers’ performance, limited managerial capacity is one of the issues affecting system performance in many low-income nations.

For instance, a report conducted by Flink and Chen [40] argues that improving infrastructure requires not only monetary inflows but also an effective management system that can make good use of the money. Taking a cue from this, it is clear that managerial capability is essential for developmental advancement in any setting, whether it be a developed one or one with limited resources. It also applies to the success of projects, whether they involve software systems or other types of development.

3.1.5. Language Issues

Developing software systems to suit the requirements of low-resource environments presents additional obstacles in the existence of language barriers. The vast majority of people native to the community have a fundamental understanding gap when it comes to the English language, which is employed in the majority of software. Because words in other languages do not necessarily have the same meaning or function in English, these language barriers are not always resolved by simple translation into the local language [41]. Mwotil [42] observed that when looking for a customisable solution in low-resource contexts, software users frequently assess the convenience of use, level of language compatibility, overhead creation needs, and level of security offered by the solution. When it comes to data-driven applications, particularly machine translation applications and other applications such as text-to-speech and other machine learning applications, low-income countries frequently experience difficulties due to a lack of open-access language datasets. This problem affects the majority of data-driven applications in low-resource environments [43]. According to Meyer et al. [44], the vast majority of languages, the majority of which are spoken in countries with low incomes, do not have open speech datasets, and even fewer languages have high-quality data with aligned text and speech.

In their study on the sustainability of multiple languages in Africa, Litre et al. [45] suggest that the success of natural language processing (NLP) is grounded on global languages that have a text corpus of hundreds of millions of words, such as English and a few other languages. However, only a small portion of the world’s languages are represented by these global languages that have extensive corpora. They went on to say that most human languages are low-resource languages, and that low-resource languages spoken in low-resource environments lack large monolingual or parallel corpora. As a result, these languages need additional tools and resources to overcome the digital gap and the resource barrier, which will enable natural language processing to provide more inclusive and widespread benefits on a global scale. A grassroots NLP research group called Masakhane [46] has pushed for the necessity for African languages to break over the native language data barrier in NLP. By iterating more quickly and efficiently, the group, which is founded on African values, enables researchers to enhance translation, terminology development, and other NLP procedures [45]. The ability to discuss science in regional indigenous languages allows for the interpretation of available research into the language of the people who speak it [46]. Thus, efficient communication and the use of software systems both depend on language.

3.1.6. Collaboration Issues

The strong need for technological innovation in environments with limited resources has been recognised as an opportunity for collaboration. It was suggested by Jawhari et al. [47] that software implementers be urged to employ already-existing systems and software, collaborate with other organisations by making use of their resources, invest in backup ability, audit user activities, offer on-site training, and track performance. Given the increased likelihood of investment and deployment in multi-institutional collaborations, it is quite likely that this will aid in the implementation of software in any environment with limited resources. This is because software development involves a variety of factors, including systems, people, processes, and products. The level of collaboration was identified as a significant difference between high-resource and low-resource environments in the study by Kolko et al. [48] and it was stated that in high-resource environments, there existed a robust enough technological infrastructure to facilitate collaboration.

Through collaboration, especially with the open-source software engineer who created the software application in the first place, many restrictions on the creation and use of software applications can be removed. Since open-source software is by its very nature conducive to sharing and collaboration, there is a sizable community of software engineers prepared to enhance and perfect open-source solutions based on insightful feedback [49,50,51,52]. This can reduce the degree of collaboration issues that are typical of many low-resource environments.

3.1.7. Network Issues

Low-income countries have the lowest number of Internet connections, with only 22% of Africa’s population having access, implying that over 1 billion Africans are without good Internet connectivity (https://www.statista.com/statistiques/273018/number-of-internet-users-worldwide/ (accessed on 24 April 2024)). Lack of infrastructure, low income and affordability, a lack of basic literacy and digital literacy, and many other factors contribute to this low level of access to networks and, as a result, to relevant software usage [36].

Even in the deployment of eHealth tools, the impact of information and communication technologies on clinical practise and the quality of health services is far more pronounced in high-income countries [18,53,54]. But countries with limited resources face numerous economic and social issues that impede the use of digital health tools as most eHealth software is cloud-based and require good networks to operate maximally. Ramanujapuram and Malemarpuram [18] posed that the unavailability of an essential resource, such as a good network, is a serious barrier in low-resource environments in order to achieve long-term gains in data quality and performance.

Due to improvements in technological capability at both the device and application levels, the high-income countries now enjoy commercial ICT services that are heavily Internet-based, such as Netflix, Spotify, and YouTube [7,10,55]. But in a large portion of Africa, where the mobile 2G (GSM) network is frequently the digital infrastructure accessible in most rural places, and 3G or 4G networks are accessible at most urban areas, this is not available [15]. A roll-out of traditional and Internet-based ICT services (software applications) in low-resource environments will not be possible due to technological limitations. To support the development of ICT services for the poor in rural areas, innovative solutions are required.

When it comes to the issue of software availability, people frequently focus on the operational stage, which is no longer a valid assumption given that software operation and maintenance occur concurrently [56]. Operation and maintenance are alternated, and simultaneous operation and maintenance requires adequate and functional networks.

3.1.8. Electricity Issues

ICT usage is still relatively low in some countries, notably in Africa, especially in rural areas with underdeveloped infrastructure and where poor access to electricity and an Internet connection are all present [7].

Bon et al. [7] note that power outages frequently occur in most urban regions and some rural areas have no electricity network in many low-resource environments. As a result, most rural areas lack access to the ICT facilities, including software, since there is not enough electricity to run the equipment used by Internet Service Providers. Despite the fact that some forms of communication technology, particularly mobile phone services, are available in some remote regions, affordability remains a major obstacle. Recently, there has been significant progress made toward electrifying rural areas, particularly in nations of Sub-Saharan Africa [11,57]. Many people believed that thorough methods in policy and planning were responsible for the great results in the electrification rate [11]. However, it should be mentioned that in 2030, 650 million people, or 8% of the world’s population, would still be without access to electricity [58]. Ninety percent of persons without access to electricity will reside in sub-Saharan Africa (SSA), in places where the national and regional grids may not always be economically viable for a variety of reasons. If these power issues are not adequately addressed, the effectiveness of ICT infrastructure, particularly software systems, will continue to be hampered.

3.1.9. Device Issues

The computing power of a device refers to how quickly it is able to carry out a specific computation [59,60]. The processing speed, the amount of memory that is needed, and the bandwidth that is utilised are the three characteristics that determine how much computing power a machine has in relation to the amount of time it takes [61]. According to Singh et al. [60], the amount of computational power that each device possesses is a crucial factor in determining how secure a communication can be when it involves multiple different types of devices. Because of this, it is of the utmost importance to perform an analysis of the computing power of the devices that are used for communication in order to devise and put into action the most appropriate security solutions [62]. This is also true for software, as it is essential to check that a device that is capable of running the software and meeting its necessary requirements is available before installing it. In low-resource settings, where most people lack the ability to acquire the appropriate hardware for software use, it is still difficult to guarantee that people will have access to devices that are capable of running modern software applications.

The operation of software systems in today’s world requires a significant amount of computing power. According to Salman et al. [61], a large number of the industrial software tools that have been created over the course of several decades is currently undergoing significant revisions as a result of a variety of factors, including but not limited to shifts in technology, shifting customer requirements, and improved development processes. The evolution of software architectures and implementations that in the past focused on single-core computing platforms operating with new data-intensive and computationally demanding applications within the system has necessitated these changes. These changes were anchored on the realization that the system now requires additional computational capacity.

The findings of the study on “integrating computational thinking and science in a low-resource setting” by Gautam et al. [63] recognise several key ingredients that contribute to the achievement of this success. These key ingredients include usable software as well as a match between the affordances of the software, the instructional purposes of the unit, and other supporting materials and student activities. They theorized that the prospects for achieving success in low-resource settings are directly proportional to the manner in which these fundamental infrastructural resources can be utilised. Therefore, they suggest that if software cannot be used optimally, for example, due to the computing device that is available, adequate success will be unattainable.

It has been noted that one of the most significant challenges posed by environments with limited resources is the absence of adequate computing devices. This challenge is not limited to the computing power of the devices but encompasses the devices themselves. For instance, in the study by Gautam et al. [63], the researchers used a school in Nepal that had been established in 2013 and had a total enrolment of 125 students with a mission to provide students with an education that is based on their areas of interest; however, the school only had three computers that were operational, and access to them was limited to students in the ninth and tenth grades. Due to insufficient funds being allocated to the education sector, this scenario can frequently be found in the majority of educational institutions located in low-income countries. In high-income countries, on the other hand, there are sufficient computing devices that support the efficient utilization of software. Even though there have been several attempts made by researchers in some of the low-income countries to develop lightweight devices [64,65] and software [62,66,67] that can work on low-resource devices, the general notion is that the availability of efficient computing devices with adequate computing power, as is available in high-income countries, is still a major challenge in low-resource environments.

3.1.10. Data Issues

Digitized data and their numerous benefits are still not realised in low-resource environments [15]. In most low-resource environments, data are still being stored on paper, thus making data availability difficult. According to Ramanujapuram and Malemarpuram [18], in most low-resource environments, the procedures for collecting and using data are not enforced, even in cases when regulations and standards are present. This ultimately results in erroneous and inadequate data.

Modern data-driven applications need accurate data in order to function. This was emphasised by Chimalamarri and Sitaram [68], who said that machine translation systems were extremely dependent on availability of relevant data. This machine translation task is difficult in low-resource environments since obtaining reliable parallel human translation corpora for this purpose is quite expensive [69]. In line with this argument, creating such applications for software developers in environments with limited resources is stressful and time-consuming. Zhang et al. [70] add to our understanding of the significance of data in real-time applications today by pointing out that the development of ICT tools for clinical use almost always necessitates vast quantities of clinical data with specific labels. The lack of clinical data prevents them from supporting pertinent research, especially for deep learning-based medical-assisted diagnosis, where the effectiveness is heavily dependent on adequate samples. According to the study’s findings, “challenges caused by insufficient volume or imbalanced medical time-series data, in particular in low-resource settings, can limit the capability of clinical medical diagnosis algorithms”.

There are methods for adjusting the data when there are few data available to train a model [71,72]. However, because these software systems need a lot of textual resources to enhance accuracy [12,73], research on areas requiring large datasets has been concentrated on high-resource environments because some low-resource domains and languages make it impossible to create training datasets from scratch due to a lack of data.

Furthermore, data issues pose a significant challenge to the development and effectiveness of deep learning software systems [74,75], especially in low-resource environments [76,77]. The lack of reliable and comprehensive data can hinder the training and optimisation of deep learning models [78,79], leading to subpar performance and limited capabilities. For instance, in a low-resource healthcare system, predicting disease outbreaks based on patient data may be compromised without sufficient and reliable data on patient symptoms and demographics, resulting in ineffective disease control measures.

3.1.11. Expertise Issues

In environments where software is widely utilised, having access to experts is essential for the efficient operation of modern applications and websites. The importance of ensuring that people from all sociodemographic divisions have equal access to expertise through personal networks was emphasised by Cornwell and Cornwell [80]. It was also stated that for a variety of reasons, “upper-class” people and members of historically powerful social groupings, particularly in advanced economies, likely have greater access to experts in various fields than others. Some of the identified reasons were as follows:

• Formal professional consultation is expensive; thus, it is more accessible to individuals with more money.
• Minorities and people from lower socioeconomic levels may have less informal connections with experts since professionals frequently come from higher status and historically dominant groups.

Toolis [81] states that in an equitable society where access to experts is prioritised in order to advance equity, effective placemaking must take into account the knowledge and input of local stakeholders, increase the availability of resources for marginalised communities, promote redistributive policies, and put environmental sustainability at the forefront. As stated by Wootton and Bonnardot [82], creating a system from scratch is entirely feasible even in low-resource environments, but it requires the necessary technical expertise. Despite the fact that the software and IT infrastructure are both made available as a single integrated package, an underpinning IT infrastructure is still necessary. To develop an effective and long-lasting service in low-resource environments, more than only the technological infrastructure is needed. In other words, while having the infrastructure is important, having engagement experts who can offer routine responses is also required for successful service. Due to the absence of suitable expertise in the low-resource environments, some software applications for low-resource settings always require the formation of project teams made up of a research team, an international expert committee, software development professionals, and foreign language translators [83,84,85,86].

It is important to take note of the work being done to create software tools in low-resource environments that do not entirely require the support of experts [87]. This is assumed to be the best approach to expertise issues in low-resource environments.

4. Low-Resource Environment Software

In today’s software systems, application properties such as scalability, availability, portability, security, and management are increasingly becoming essential even for cloud-based systems [42]. This is particularly true for scalability. Because of this, there is now a pressing requirement to make investments in new tools to facilitate the creation and distribution of modern software.

Mobile-based software applications have seen a significant growth in recent years, despite the limited resources available in low-resource environments. According to Sudar and Anderson [88], the use of mobile devices in data-focused workflows is becoming increasingly common. These deployments are often managed by organisations with limited technical capacity, making decisions about fundamental architectural elements challenging. In their study, DeRenzi et al. [89] highlight the work undertaken by researchers in the fields of Human–Computer Interaction for Development (HCI4D) and Information and Communication Technologies for Development (ICTD) to explore how mobile devices could support and enhance software systems in low-resource environments.

A key driver for mobile-based applications in these settings is the presence of resource constraints, such as limited Internet connectivity and low-end devices [90,91]. These factors significantly affect the feasibility and sustainability of implementing software systems. For instance, in a remote healthcare facility with limited or no Internet access, a mobile-based software system with offline storage capabilities could be used to collect patient data during appointments and then transmit them to a centralised database once Internet connectivity is available. This enables healthcare providers to access real-time patient information and make informed treatment decisions, even in rural or remote areas.

As observed, mobile health (mHealth) applications and e-health tools [92,93], along with other similar applications such as educational software systems, have gained significant penetration in low- and middle-income countries [94,95,96,97,98]. It is now common practice to use mobile health (mHealth) technologies to assist medical personnel [99,100]. Care coordination [101], data capture [85], the retrieval of patient data [102], and decision support are just some of the tasks that can be accomplished with the help of smartphone applications in these environments. Electronic health record (EHR) systems are currently undergoing active implementation in a variety of low- and middle-income countries around the world. EHR systems are now being used in public facilities that are run by the government in countries such as Kenya, Uganda, Nigeria, and Mozambique [103]. These EHR systems are being deployed in hundreds of thousands of public health facilities to either replace or supplement the use of paper-based records that are already in existence. This is being done with the intention of improving the quality of care that is provided to patients and providing support for the monitoring and evaluation of various programmes. When looking at the impact of mHealth tools in low- and middle-income countries, McCool et al. [100] state that the growth of digital health care marketplaces, AI-based self-diagnosis apps, and a plethora of new mHealth apps that replace or transform traditional health care services raises important questions about their sustainability. This was found in their investigation of the impact of mHealth tools in low- and middle-income countries.

Based on our findings, we realised that in various low-resource environments, more application software systems were developed in the areas of health, education, and agriculture, and that the majority of these software applications were compatible with mobile devices. This is primarily because mobile devices have become increasingly affordable and accessible, even in low-resource settings. As a result of this, there has been significant growth in the development and adoption of mobile-based software solutions in low-resource environments, driving positive change and empowering communities to overcome challenges [15].

4.1. The Selected Software Systems

The following software systems, selected from the searched lists, are presented as exemplars to demonstrate how software developed in and for low-resource environments can address some of the challenges associated with developing and using software in low-resource environments. These software systems are presented and categorised based on the challenges they address and their application domains.

4.1.1. Avaaj Otalo

An interactive speech application known as Avaaj Otalo was developed for use by small-scale farmers in the Indian state of Gujarat [104]. As a part of a collaborative effort between the Development Support Center (DSC), an organisation based in Ahmedabad, Gujarat, India, and the IBM India Research Laboratory, which is located in New Delhi, Avaaj Otalo was built as a VoiceSite by utilising the Spoken Web platform developed by IBM Research. The most well-liked component of Avaaj Otalo according to Patel et al. [104] was its voice forum, which allowed users to pose and answer questions on a variety of agricultural issues, as well as explore the questions and answers posed by other users.

4.1.2. BantuWeb

BantuWeb [105] is a digital library tool that is based on the Web and serves as a central repository for content that is written in South African resource-scarce languages (such as isiNdebele, isiXhosa, isiZulu, Sepedi, Sesotho, Setswana, siSwati, Tshivenda, and Xitsonga). These languages contents are crawled from the Internet and are generated or contributed by a community of users. The contents are culled from various locations on the Web. Gamification elements were incorporated into the platform in order to encourage users to contribute content for the purpose of enhancing the collection of resources and expanding the amount of community involvement. In addition to that, it offers a multilingual search function, which enables users to conduct informational searches in any of the languages.

4.1.3. CommCare

CommCare (https://www.commcarehq.org/ (accessed on 25 April 2024)) is a customizable, open-source mobile platform that enables non-programmers to build mobile applications for data collection, counselling, behaviour change, and a variety of other functions. To date, hundreds of organisations have used CommCare to build mobile applications that are designed to support frontline workers (FLWs) across a variety of sectors in low-resource settings. FLWs use CommCare to track and support their clients with registration forms, checklists, SMS reminders, and multimedia—all on simple Java-enabled phones or Android smart phones and tablets. CommCare users submit over 1 million forms per month from 50+ countries around the world [106]. The application was first developed by the software company Dimagi (https://www.dimagi.com/ (accessed on 25 April 2024)) in 2007 to reduce gaps that frontline programmes face in delivering services to populations in low-resource settings.

4.1.4. M-Pesa

M-Pesa [107] is a mobile application that enables customers to conduct branchless banking through the use of their mobile phones. Safaricom, the leading mobile service provider in Kenya, was the first company to officially launch the product on the Kenyan market in March of 2007, aiming to attract the unbanked and prepaid portion of the country’s population [108]. The first generation of the M-Pesa technology stack was implemented utilising .NET on SQL Server, with MSMQ serving as the messaging layer, and some Java being run on the SIM cards. It uses mobile phones to move money quickly, securely, and directly to another user of a mobile phone, regardless of where they are located or how far apart they are. On the M-Pesa application, users have the ability to check the balance of their accounts, make deposits and withdrawals, pay bills, buy mobile phone credit, and send money to other users [21]. It has now spread to South Africa, Tanzania, Mozambique, the Democratic Republic of the Congo, Lesotho, Ghana, and Afghanistan.

4.1.5. Easy-To-Use Arithmetic Learning Tool

Chakraborty et al. [109] developed a low-cost and easy-to-use arithmetic learning framework to help physically challenged people in low-resource settings. A learning framework for Braille-based arithmetic was proposed in the study. The framework was based on the design goals that were developed through comprehensive studies in two different low-income regions of the world. It was an arithmetic learning framework that was both inexpensive and simple to use, and it drew a contrast between the traditional methods of solving arithmetic problems and an easy-to-use system.

4.1.6. CALJAX

CALJAX is a repository tool developed for low-resource environments [110]. It is made up of separate components that work together to manage an online repository, as well as provide search and browse services. Initially, it was developed as a prototype in order to demonstrate the practicability of utilising recent developments in Web technology in order to improve accessibility to information on a global scale. It is arguable that CALJAX is both lightweight and distributable. It has a small amount of software that needs to be installed, but it does not sacrifice usability or functionality. A variety of objects and associated metadata can be found within the CALJAX central repository. These are saved in the form of files within directories organised in a tree structure, and each file has its own metadata file that is associated with it. When the repository is distributed offline, the contents of the central repository are merely copied to one or more forms of removable media. This offline repository then integrates large offline data collection and few recent online additions to provide users with an up-to-date experience. If the users choose to view digital objects, the local copies of the files are shown to them whenever that is possible, and the online versions are displayed otherwise. It was discovered that the CALJAX online–offline integration worked to a satisfactory degree for relatively modest collections, but that it had a linear scalability problem. During the online–offline integration, it was suggested that this limitation could be lessened by only returning selected results and not all updates in response to requests. It was believed that the strategy would be successful in an environment with a limited capacity for Internet connectivity, which is typical of most low-resource environments.

4.1.7. e-IMCI

The e-IMCI software application [111], which stands for “electronic Integrated Management of Childhood Illness”, is designed for use on a PDA to administer the IMCI protocol. It was developed to enhance the quality of medical care in Tanzania’s health facilities, with a focus on usability, flexibility, and reducing deviations from standard IMCI protocols. Initially, the system checks if the patient consents to the clinician using the PDA. It then raises alerts for four danger signs—vomiting, convulsions, difficulty drinking, and lethargy or unconsciousness. If any danger sign is selected, e-IMCI prompts the clinician to refer to the chart booklet to ensure patient safety while testing the system. By minimising unintentional deviations from the IMCI protocol, e-IMCI contributes to improved care quality in low-resource settings.

4.1.8. Krishi Kontho

The agricultural information service system known as Krishi Kontho [112] sends pre-recorded voice messages and text messages (SMS) to the mobile phones of smallholder farmers in rural Bangladesh at intervals that are meticulously timed to correspond with the growth cycles of their crops. It was made available through the partnership among several organisations, including the international NGO Christian Aid, the local NGO Gana Unnayan Kendra (GUK), the technical partner mPower, and local farmers. The main goal of the system was to disseminate agricultural guidance to low-literacy rural communities in a manner that was in sync with the cycles of their crop production techniques.

4.1.9. mYoga

The goal of the mYoga application (https://indianexpress.com/article/technology/social/whos-myoga-app-what-is-it-about-and-how-does-it-work-7369046/ (accessed on 25 April 2024)), which was developed by the Ministry of Ayush, India and the World Health Organisation in collaboration, is to make assisted yoga training available for free to anyone who possesses a smartphone. The mYoga application is designed to be used on a consistent basis by members of the general public. It offers yoga classes and practice sessions of varying lengths. A review of the scientific literature and extensive consultation with experts from around the world were two of the processes that went into the development of the application. The application is primarily composed of two tabs: one for learning, and the other for practising yoga poses. The “learning” tab is designed specifically for users who have no prior experience with yoga. It offers a series of videos that can be viewed in any order to assist viewers in learning the various yoga asanas and performing them correctly [113].

4.1.10. Mobile InterVA (MIVA)

InterVA (https://www.who.int/initiatives/behealthy (accessed on 25 April 2024)) is a well-known software in the health sector for its ability to automatically and quickly assign probable causes of death to verbal autopsy (VA) data that have been converted into a digital format [114]. The development of InterVA began in 2003, and the first version, known as InterVA-1, was released in 2005. Using the responses to a series of yes/no questions, Bayesian principles are used by InterVA to determine the most likely causes of death in a given case. The software is considerably faster than physician-coded verbal autopsy (PCVA) at identifying probable causes of death [115]. MIVA [114] is a version of InterVA that has been adapted for use on mobile devices. It is comprised of an HTML page that is kept locally on the mobile device, and JavaScript is utilised for the implementation of the data gathering and storage features. It was developed for the purpose of facilitating the direct data capture, as well as the rapid interpretation, presentation, and storage of information regarding the cause of death in low-income countries. The WHO VA standard questions were taken into consideration during the development of the software, which was then integrated into the InterVA-4 system. It was designed to promote fieldworkers’ sense of ownership of the system and, as a result, encourage them to provide helpful feedback on the design, which can be difficult to elicit in low-income countries.

4.1.11. mUzima

mUzima is an open-source, highly configurable Android application that was developed by Savai et al. [116] for patient data management. Some of the application’s features include offline management, deduplication, cohort management, relationship management, security, and error resolution. Providers who work remotely and have lower-end Android smartphones can use mUzima to access historical patient data, collect new data, view media, leverage decision support, conduct store-and-forward teleconsultations, and geolocate clients. An engaged community of software developers and end users provides support for the application. The community also helps determine which features should be prioritised. mUzima was first implemented in Kenya, and it is currently being used to a significant extent in Rwanda. Additionally, it is gaining popularity in Uganda and Mozambique. It is not specific to any one disease and is already being used to manage a variety of conditions, including HIV, cancer, chronic disease, and COVID-19. Because it satisfies all of the WHO’s Principles of Digital Development, mUzima has been acknowledged as a digital global public good and included in the WHO Digital Health Atlas as a result of its successful scaling implementation. Because of its powerful offline functionality, mUzima is able to compensate for areas with poor Internet connectivity. As a result, medical professionals who use mUzima can continue to provide care for patients even when they are not connected to the server. Non-developers can manage and fix server errors with mUzima owing to its built-in error resolution mechanism. In remote facilities, which frequently lack staff to provide advanced technical support, mUzima is extremely usable and scalable due to its capacity to fix errors on submitted data without the need for knowledge or access to the server’s back-end.

4.1.12. mira

mira [42] is an extension platform to the CI/CD (Continuous Integration/Continuous Deployment) pipeline for containerised applications for small developer teams in low-resource settings. It loosely automates the build, test, release and deploy operations of CI/CD for different programming frameworks. These operations enable containerisation of an application and its subsequent deployment to a container orchestration platform for production use. The NodeJS framework was used to implement the mira back-end service, while the ReactJS framework was used to write the front-end code. Crane Cloud’s application programming interface (API) endpoint must be used for application deployment, and user and project tokens serve as the authentication mechanism. This denotes a location of the application within the hosting platform that is part of a subset.

4.1.13. Open Data Kit (ODK)

Open Data Kit (ODK) [117] is an open-source, extensible toolkit that was developed to facilitate the creation of information services in low-resource environments. It was built with a primary emphasis on the creation of compact, composable modules and, as a result, it is potentially simpler to extend and alter. It was developed to support diverse devices and to endure and evolve over the long term. Additionally, it was designed to provide a diversity of available form-factors and to adapt to new capabilities made available by the high rate of innovation in software design. Collect, Aggregate, Voice, and Build are the four tools that are offered by ODK. Collect is a mobile platform that facilitates the manipulation of data and the rendering of application logic. Aggregate offers a server that can be deployed with a single click and is capable of storing and transferring data either within the “cloud” or on local servers. Voice translates the logic of an application into phone prompts that consumers reply to by pressing the appropriate keypad buttons. Finally, Build is the designer that is responsible for generating the logic that the tools employ. ODK core tools are based on previously established open standards and are supported by an open-source community that has contributed to its expansion. These tools can be used together or independently, depending on the user’s preference.

4.1.14. ODK Scan

ODK Scan [118] is a mobile application that is capable of digitizing data from paper forms by employing computer vision techniques. The application provides users with assistance while manually entering handwritten text and numbers, in addition to automatically classifying machine-readable data types such as bubbles and checkboxes. Community Health Workers (CHW) consumption of health commodities in Mozambique was monitored through the use of ODK Scan, and Dell et al. [118] identified it as a standardized data collection and reporting system. The design focuses on building a process that is inexpensive, familiar, and easy to use for CHWs and that also provides a mechanism for district supervisors to scan, analyse, and disseminate CHW logistical data.

4.1.15. Pendragon

Pendragon [119] is a commercial, all-purpose form design and administration system that is compatible with the palm operating system and Windows OS. It is an advanced form designer as well as software for mobile data collection. It is helpful for designing forms such as surveys, inspections, or work orders, which are then distributed to mobile devices. Because of its intuitive design and user-friendliness, Pendragon Forms can be utilised easily. On the other hand, the depth of the platform makes it possible to create more complex applications that make use of barcoding, geo-location, scripting, and integration with enterprise software. In the health sector, it has been put to use in a wide variety of applications, particularly in low- and middle-income countries [120,121,122]. Pendragon requires a moderately small one-time license fee in addition to a relatively small extra fee for each subsequent user. When compared to the operating costs of a system that relies on paper, this is significantly cheaper [120].

4.1.16. Printr

Printr [123] is an interactive system that enables low-resource organisations to quickly generate individualized paper tools. It was designed with this purpose in mind. It gives publishers the ability to specify various tasks and presents them with a set of potential paper tools at each stage of the specification process. These tools could be printed and used by the end-user that is being envisioned. It includes information about the required materials, as well as instructions for assembly and use, alongside each suggestion that it makes. It was developed as a web-based system using HTML/CSS and JavaScript. It was built with jQuery, Twitter Bootstrap, and D3 frameworks.

4.1.17. Shreddr

Shreddr [124] was developed to be a web-based service that is hosted in the cloud. Its primary function is to convert images of paper forms into structured data on demand. It is an application that combines various types of processing, such as batch processing and compression methods from database systems, automatic document processing through the use of computer vision, and value verification through the use of crowd-sourced workers. It utilises computer vision to align images before shredding them into constituent parts, and it provides the user with a web-based interface that assists in the extraction of a paper form’s schema as well as the data locations on the form. After that, the system orders and groups the shredded data according to an entropy-based metric, and it places the shredded data groups into batched worker interfaces. It enables field workers to collect information using pre-existing and familiar paper forms. These forms are then uploaded as images and iteratively digitized using a combination of automated and human-assisted techniques.

4.1.18. Voice- and Web-Based Feedback System

A mobile voice- and web-based application that operated within the browser and displayed versions of the visualizations in high-quality PNG format was designed by DeRenzi et al. [89] and distributed to ASHAs (Accredited Social Health Activists) in India with the purpose of providing feedback on their work. The system assisted low-literate ASHAs in their work processes by supplying them with information and motivation that was beneficial to their work. The application was put to the test by female ASHA participants participating in the ReMiND (e-Reducing Maternal and Newborn Deaths) project. There were 71 people who took part in the application testing that took place over the course of a year. Everyone who took part in the activity had received training in the past that prepared them to offer counselling and referrals to members of their communities’ primary health services. The interpretation of how the participants utilised the system, together with qualitative data, illustrated the importance of providing health workers in low-resource contexts with a personalized, on-demand information feedback system.

4.1.19. MISTOWA

A software application known as MISTOWA (Market Information Systems and Trader’s Organisations of West Africa) [125] was developed with the intention of expanding regional agricultural markets throughout the countries of West Africa. MISTOWA can be used on mobile devices in addition to desktop computers. Benin, Burkina Faso, Ghana, Mali, Nigeria, and Senegal were some of the countries that participated in MISTOWA at the time when it was being sponsored by USAID. Maize, rice, cassava, cattle, tomatoes, onions, and fertilisers were some of the commodities that were included in either the domestic or international market information systems of MISTOWA. The United States Agency for International Development (USAID) provided funding for the MISTOWA programme beginning in 2004, and the International Finance Corporation (IFCD) was responsible for its implementation. MISTOWA collaborated with national affiliates of Market Information Systems (MIS), Traders’ Organisations (TOs), and Producers’ Organisations (POs), as well as regional public MIS networks such as RESIMAO/WAMIS-NET and private ones such as TradeNet. These networks provide current and accurate information on 400 rural and urban agricultural commodity markets to all stakeholders via the Internet, radio, print, email, and SMS.

4.1.20. Esoko

Esoko (https://esoko.com/ (accessed on 6 May 2024)) is a messaging and profiling service for the agricultural industry that is managed on the Web and delivered via mobile devices. The application can collect and send out market data through the use of straightforward text messaging, and it is utilised by individuals, agri-businesses, and the government. It was determined through the use of the Esoko application how successful the crowdsourcing model was in facilitating the efficient exchange of information among farmers. It was developed in order to meet a need—in this case to serve as a connection point or a bridge between rural farmers and commercial enterprises that do market the farmers’ wares. The company that is now known as Esoko was originally called TradeNet when it was established in 2005 by Mark Davies, who went on to become the company’s acting CEO [126]. The name was changed to Esoko in 2009; the “E” stands for electronic, and “soko” is the word for market in the Kiswahili language. It has recently expanded its reach to include additional countries south of the Sahara [127].

4.1.21. Ushahidi

The Ushahidi software (https://www.ushahidi.com/ (accessed on 6 May 2024)) is a crisis-mapping application that can collect and visualise data, information, and actions taken by citizens through the use of interactive mapping. During the Kenyan elections in 2008, its creators saw a need for a platform that would enable ordinary Kenyans to discuss issues that were being ignored by the country’s official press [128]. It does this by utilising digital cartography and, quite frequently, data gathered from the general public in order to provide alternative narratives as well as spaces for communication and action. Ushahidi has remained an open-source project that is actively being developed, and it currently has thousands of deployments across the world [129]. In the aftermath of numerous disasters, including the earthquake in Haiti and Snowmaggedon, as well as in Syria and other countries related to the “Arab Spring”, it was utilised to assist local communities in reporting acts of violence [130].

4.1.22. OpenMRS

The OpenMRS [131] application is a collaborative open-source initiative that has served as a foundation for electronic medical record (EMR) software development in low- and middle-income countries. It was created in 2004 in response to a shortage of dependable and accessible medical record systems in this region. Since then, OpenMRS has grown to become a widely used platform, offering healthcare providers a flexible and configurable option for efficiently managing patient data [132]. Because of its open-source nature, OpenMRS has generated a global community of developers, health professionals, and implementers who are always working to improve and expand it. This collective effort has resulted in a powerful and innovative platform that can be customised to meet the unique needs of various medical facilities. It enables healthcare professionals to reliably gather, store, and manage patient health information. OpenMRS is highly adaptable and may be tailored to the specific requirements of various healthcare facilities and scenarios [133]. Furthermore, the modular nature of OpenMRS enables simple connections with other health information systems, allowing for smooth data interchange and interoperability. As a result, OpenMRS has been instrumental in improving healthcare delivery and patient outcomes in low-resource environments around the world [132].

4.1.23. Kolibri

The Kolibri [134] is an online/offline learning platform created by Learning Equality (https://learningequality.org/kolibri/ (accessed on 10 May 2024)) in 2020 to provide educators worldwide with access to verified and openly licenced educational content libraries. It is centred on an offline-first learning platform that can be accessed via a range of low-cost and legacy devices. Kolibri OER are available in a variety of languages, and the Kolibri libraries are designed to support an all-around curriculum, including both formal educational materials, such as lessons and assessments, and exploratory materials for science, technology, engineering, and mathematics. It also promotes collaboration among educators by enabling them to share their own resources and best practices, fostering a global community of educators dedicated to improving access to quality education for all.

4.1.24. EpiSurveyor

EpiSurveyor (currently known as Magpi) (https://www.magpi.com/ (accessed on 10 May 2024)) is a comprehensive data-gathering software application for public health and epidemiology [135]. One of its key benefits is its capacity to collect data on the go. It is designed to be used on mobile devices such as smartphones and tablets, making it ideal for field data collection. This feature eliminates the need for traditional paper forms and time-consuming manual data entry, simplifying data collection. The EpiSurveyor, which eventually became Magpi, was founded in 2003 (by DataDyne) as EpiSurveyor (https://www.magpi.com/about (accessed on 10 May 2024)). The project’s name was changed to “Magpi” in 2013. This change marked the beginning of a new phase in the software’s development and its greater applicability in data collection across a variety of areas, including public health and epidemiology [136,137]. One of EpiSurveyor’s distinguishing qualities is its capacity to operate in low-resource areas with minimal or no Internet connectivity. When an Internet connection becomes available, data acquired in the field can be kept on mobile devices and uploaded to the server [135]. Its other outstanding features include real-time data monitoring and quality control, survey logic and branching, GPS integration, and post-data collection.

5. Software Categorisation

The software systems discussed were categorised according to the stated challenges of low-resource environments and according to their domains of application. Rather than detail their descriptions, we focused on how specific low-resource issues affected each software system. This approach emphasises the practical impacts of resource constraints on performance and usability, providing a clearer picture of the real-world challenges and adaptations required for software in low-resource environments.

5.1. Categorisation on Challenges

The ability of this software to meet any of the eleven (11) challenges is the criterion we used for the categorisation based on the challenges. Some software systems fit into multiple categories, as multiple approaches were used to address different challenges in some particular environments. They explicitly indicate this in their features. This categorisation allowed us to gain a comprehensive understanding of how different software systems tackled specific challenges in low-resource environments. By identifying the strengths and weaknesses of each software system, we were able to provide insights on the most effective solutions for various issues. A summary of this is described in Table 3, showing key features, cost, technology, and area of first deployment of these software systems.

5.1.1. Network

The software systems in this category specifically address the lack of networks, or the lack of a high-speed connection, by incorporating offline capabilities into their features. This offline capability allows users to access and use the software even in rural or disconnected areas where Internet connectivity is limited or nonexistent. It ensures uninterrupted productivity and access to important features, making it highly suitable for users in remote locations.

The following are the software systems in this category:

• mUzima
• CALJAX
• printr
• Episurveyor

Table 3. Summary of key features of the low-resource environments software.

Software	Key Features	Challenges Category	Deployment	Cost	Technology	First Deployed
Open Data Kit (ODK)	GPS integration, language interchange, Internet-free development, and Internet support	Device	Mobile and cloud	Low	Java, JavaScript, and Python	Kenya
mUzima	Offline management, deduplication, relationship management, security, cohort management, and error resolution	Device, collaboration, and network	Mobile	Medium	Java, Gradle, and maven	Kenya
CALJAX	Online–offline integration, lightweight, and distributable	Network	Cloud	Medium	AJAX	South Africa
Voice-and web-based feedback system	Simple and minimalist	Low literacy, device	Mobile and cloud	medium	-	India
Krishi Kontho	Voice application	Device, collaboration, low literacy	Mobile	medium	-	Bangladesh
Pendragon	Mobile form designer, data collection, and database synchronization	Economic, device	Mobile	Medium	-	Africa and Asia
Avaaj Otalo	Voice integration	Collaboration	Mobile	High	VoiceSite on IBM’s Spoken Web platform	India
BantuWeb	Repository management, gamification, and language crawling	Language	Mobile and Web	Medium	-	South Africa
CommCare	Data collection, offline capabilities, and multilingual support	Device, Expertise	Mobile	Low	-	India
M-Peza	Mobile banking, biometrics authentication, and offline capabilities	Device	Mobile	Low	.NET, SQL Server, MSMQ, and Java	Kenya
e-IMCI	Simplicity, flexibility, and IMCI compliance check	Device	Mobile	Low	-	Tanzania
mYoga	Health and fitness management	Device	Mobile	-	-	India
MIVA	Verbal autopsy check	Device	Mobile	Low	-	South Africa
Esoko	Agribusiness market information and data collection	Device	Cross-platform	Low	-	Ghana
mira	CI/CD pipeline compliance	Collaboration	Cloud	Medium	NodeJS framework and ReactJS framework	Uganda
ODK Scan	Document scanning, digitizing data	Economic, device	Mobile	Low	-	Mozambique
printr	Paper tools’ management	Network, device	Web	Low	HTML/CSS and Javascript	Ghana
Shreddr	Image converter, on-demand data entry	Device, economic	Cloud	Low	-	Mali
MISTOWA	Agribusiness market information	Device, economic	Cross-platform	Medium	-	Ghana and Tanzania
Esoko	Market information sharing	Economic, data, network	Web/Cloud	Medium	Java/Python	Ghana
Ushahidi	Crowdsourced data collection	Network, Language barriers, and collaboration	Web/Cloud	Low	PHP/Python	Kenya
OpenMRS	Electronic health records’ management	Language barriers, data, and expertise	Web/Cloud	Low	Java	Kenya
Kolibri	Education content delivery	Low Literacy, Language barriers, and socio-cultural	Offline, web	Low	Python	Uganda
EpiSurveyor	Mobile data collection and surveys	Data, language barriers, and collaboration	Mobile	Low	Java/Python	Kenya

Table 3. Summary of key features of the low-resource environments software.

Software	Key Features	Challenges Category	Deployment	Cost	Technology	First Deployed
Open Data Kit (ODK)	GPS integration, language interchange, Internet-free development, and Internet support	Device	Mobile and cloud	Low	Java, JavaScript, and Python	Kenya
mUzima	Offline management, deduplication, relationship management, security, cohort management, and error resolution	Device, collaboration, and network	Mobile	Medium	Java, Gradle, and maven	Kenya
CALJAX	Online–offline integration, lightweight, and distributable	Network	Cloud	Medium	AJAX	South Africa
Voice-and web-based feedback system	Simple and minimalist	Low literacy, device	Mobile and cloud	medium	-	India
Krishi Kontho	Voice application	Device, collaboration, low literacy	Mobile	medium	-	Bangladesh
Pendragon	Mobile form designer, data collection, and database synchronization	Economic, device	Mobile	Medium	-	Africa and Asia
Avaaj Otalo	Voice integration	Collaboration	Mobile	High	VoiceSite on IBM’s Spoken Web platform	India
BantuWeb	Repository management, gamification, and language crawling	Language	Mobile and Web	Medium	-	South Africa
CommCare	Data collection, offline capabilities, and multilingual support	Device, Expertise	Mobile	Low	-	India
M-Peza	Mobile banking, biometrics authentication, and offline capabilities	Device	Mobile	Low	.NET, SQL Server, MSMQ, and Java	Kenya
e-IMCI	Simplicity, flexibility, and IMCI compliance check	Device	Mobile	Low	-	Tanzania
mYoga	Health and fitness management	Device	Mobile	-	-	India
MIVA	Verbal autopsy check	Device	Mobile	Low	-	South Africa
Esoko	Agribusiness market information and data collection	Device	Cross-platform	Low	-	Ghana
mira	CI/CD pipeline compliance	Collaboration	Cloud	Medium	NodeJS framework and ReactJS framework	Uganda
ODK Scan	Document scanning, digitizing data	Economic, device	Mobile	Low	-	Mozambique
printr	Paper tools’ management	Network, device	Web	Low	HTML/CSS and Javascript	Ghana
Shreddr	Image converter, on-demand data entry	Device, economic	Cloud	Low	-	Mali
MISTOWA	Agribusiness market information	Device, economic	Cross-platform	Medium	-	Ghana and Tanzania
Esoko	Market information sharing	Economic, data, network	Web/Cloud	Medium	Java/Python	Ghana
Ushahidi	Crowdsourced data collection	Network, Language barriers, and collaboration	Web/Cloud	Low	PHP/Python	Kenya
OpenMRS	Electronic health records’ management	Language barriers, data, and expertise	Web/Cloud	Low	Java	Kenya
Kolibri	Education content delivery	Low Literacy, Language barriers, and socio-cultural	Offline, web	Low	Python	Uganda
EpiSurveyor	Mobile data collection and surveys	Data, language barriers, and collaboration	Mobile	Low	Java/Python	Kenya

5.1.2. Low Literacy

This category includes low-resource environment software that was developed for users with low and varying levels of literacy. These software solutions are designed to be accessible and user-friendly, often incorporating visual aids and simplified interfaces. They aim to bridge the digital divide and empower individuals in underserved communities by providing them with tools to enhance their digital literacy skills.

The following are the software systems in this category:

• Voice- and Web-Based Feedback System
• Krishi Kontho

5.1.3. Economic

The cost to develop the software systems in this group is typically very low, making them most economical for a low-resource environment. They have comparatively minimal costs, both for the initial development and ongoing operation. This affordability allows organisations with limited budgets to adopt and utilise these software solutions effectively. Moreover, the low cost of development also enables frequent updates and improvements, ensuring that the software remains efficient and up-to-date in the long run.

The following are the software systems in this category:

• Pendragon
• mYoga
• Easy-to-use arithmetic learning tool
• ODK Scan
• Shreddr
• printr

5.1.4. Device

The low-resource environment software systems in this category can support a wide range of devices, as well as being lightweight and requiring fewer processing resources. This makes them ideal for devices with limited storage capacity or slower processors. They are designed to optimise performance and minimise power consumption, ensuring a seamless user experience even on low-end devices.

The following are the software systems in this category:

• Open Data Kit (ODK)
• CommCare
• M-Pesa
• MISTOWA
• Esoko
• Mobile InterVA (MIVA)
• mUzima
• e-IMCI
• Pendragon
• ODK Scan
• mYoga

5.1.5. Collaboration

The low-resource environment software systems in this category facilitate community cooperation for the purpose of further development and advancement. They provide a platform for individuals to collaborate and share resources, ideas, and knowledge. By leveraging the collective efforts of the community, they enable faster progress and innovation in a cost-effective manner.

The following are the software systems in this category:

• mUzima
• mira
• Avaaj Otalo
• Krishi Kontho
• MISTOWA

5.1.6. Data

Software systems for low-resource environments in this category support the storage and management of data. They also do not require vast quantities of data to function. These software solutions are designed to be lightweight and efficient, allowing them to run smoothly on devices with limited processing power and memory. They often have built-in data compression techniques to optimise storage space and minimise the impact on system resources.

The following are the software systems in this category:

• Open Data Kit (ODK)
• mira

5.1.7. Expertise

The low-resource environment software systems that fall into this category are capable of functioning to their maximum capability even without the assistance of experts in the relevant technology. These software systems are designed to be user-friendly, allowing individuals with limited technical knowledge to easily navigate and utilise their features.

The following are the software systems in this category:

• CommCare
• mUzima

5.1.8. Language Barrier

The low-resource environment software systems in this category are developed with the goal of eliminating the local language barrier. Low-literacy users particularly find them useful due to language localization features. They often include features such as visual interfaces and audio instructions to facilitate comprehension in local languages.

The following is the software system in this category:

• BantuWeb

5.2. Categorisation on Application Domains

Software systems have evolved to cater to various application domains, each with its own unique demands and requirements. In this subsection, we categorise the selected software systems by application domain, specifically focusing on their use in areas such as healthcare, education, and agriculture. This categorisation not only organises the software systems according to their primary fields of application but also illustrates key technical adaptations that allow these systems to function effectively in low-resource settings. For example, software applications such as CommCare demonstrate how applications can operate on older hardware, reduce computing resource consumption, and integrate with legacy systems. These use cases highlight the practical strategies developers utilise to enhance software compatibility and efficiency, addressing the unique constraints found in low-resource environments. We describe how these software systems are specifically designed to work within specific fields or areas of use. They can also feature in more than one area. For example, a healthcare application can also be used in crisis response or community engagement.

These software systems are tailored to meet the specific requirements and usage in this application domains. For example, in healthcare, software systems need to adhere to privacy and security standards to protect patient information [138].

5.2.1. Healthcare

One of the most critical areas of software system application is healthcare. Software systems in this category are designed to simplify data collection, be accessible to health workers and researchers, and facilitate efficient management of patient records in low-resource environments.

The following are the software systems in this category:

• CommCare
• OpenMRS
• e-IMCI
• mUzima
• Pendragon

These software applications demonstrate specific technical adaptations to function in low-resource environments. For instance, CommCare and OpenMRS reduce computing resource consumption by limiting background processes, using optimised code, and leveraging minimalistic interfaces that prioritise core functionality over heavy graphical elements. CommCare’s modular design enables it to allocate resources as needed, conserving memory and processing power, which is especially beneficial on basic mobile devices [106]. OpenMRS, on the other hand, integrates with legacy systems by adopting a lightweight framework and scalable architecture, making it compatible with older desktop operating systems and allowing easy upgrades as new hardware becomes available [132]. Additionally, they are built to support compatibility with other commonly used legacy health information systems, enabling seamless data transfer and integration across platforms. This integration is essential for low-resource healthcare facilities that may use a mix of digital and paper-based records. These essential features and capabilities are the same for e-IMCI, mUzima, and Pendaragon applications.

5.2.2. Education

The need for a platform that fosters learning by providing access to educational resources and bridging educational inequities by providing a wealth of knowledge even in locations with poor Internet connectivity is critical in the domain of education. This category comprises software solutions that aid in learning.

The following are the software systems in this category:

• mYoga
• Easy-to-Use Arithmetic Learning Tool
• Kolibri

These educational software systems utilise various technical adaptations to ensure compatibility with older hardware and conserve computing resources. Kolibri, for instance, uses local caching and preloading of content, allowing students to access materials offline without needing real-time data processing [134]. Its design optimises content delivery by prioritising text-based resources over multimedia where possible, thus reducing memory and bandwidth requirements. Similarly, the Easy-to-Use Arithmetic Learning Tool incorporates a minimalist design that operates on low-resource devices by focusing solely on essential arithmetic functions without added visual elements [109]. This simplicity reduces memory and processing demands, ensuring that the tool is compatible with basic mobile devices that lack advanced computing capabilities. These examples highlight how educational software can be tailored to low-resource environments, ensuring equitable access to learning tools and resources regardless of technological limitations.

5.2.3. Finance

The software solutions in this category manage financial issues in low-resource environments. They play an important role in facilitating financial transactions and assisting with financial record-keeping, particularly for small enterprises. These technologies improve financial management, promote economic stability, and increase financial access.

The following is the software system in this category:

• M-Pesa

M-Peza demonstrates how financial software can be adapted to meet the needs of low-resource settings, providing essential services while optimising for minimal resource consumption and high accessibility [107]. It exemplifies technical adaptations tailored for low-resource settings by using SMS and USSD protocols instead of relying on data-heavy applications. This approach minimises data consumption and enables transactions over basic GSM networks, ensuring that the service is accessible in regions with limited or unreliable Internet access. Additionally, its interface is streamlined for ease of use on older, text-based mobile phones, avoiding graphics-intensive features that could slow performance or require modern hardware [139]. These design choices allow M-Pesa to operate effectively on low-powered devices, reducing the barriers to financial services for users in resource-constrained environments.

5.2.4. Crisis Response and Data Collection

The crisis response includes software solutions that support gathering crucial information and involving communities during crises and public health tragedies. In low-resource environments, these software systems are crucial for effective data collection, decision-making, and community participation, which supports efficient crisis response and evidence-based treatments.

The following are the software systems in this category:

• CommCare
• Ushahidi
• EpiSurveyor
• Voice-and Web-Based Feedback System

These crisis response software applications incorporate technical features that enable them to function effectively in low-resource environments. Ushahidi, for instance, uses SMS and email as primary data input channels, reducing dependency on high-powered devices and enabling participation through basic mobile phones [140,141]. This approach minimises data requirements and allows communities with low Internet availability to contribute valuable information during crises. Also, CommCare’s offline functionality and modular design allow field workers to operate in remote areas, storing data locally on the device and syncing only when a connection is available, which conserves device resources and bandwidth. All these software systems demonstrate how crisis response and data collection systems can be adapted for low-resource environments by prioritising minimal resource consumption, offline functionality, and compatibility with basic devices. These adaptations enhance the ability of communities and organisations to respond effectively during crises, ensuring timely access to critical data and empowering local participants.

5.2.5. Agriculture

Agriculture software systems provide information, resources, and market-related assistance to farmers and agribusinesses. They are important in low-resource environments because they improve agricultural techniques, increase market access, and promote pricing transparency, thereby improving food security and economic livelihoods.

The following are the software systems in this category:

• Avaaj Otalo
• Esoko
• MISTOWA

The agriculture-focused software systems in this category incorporate specific technical features to function effectively in low-resource settings. For example, Avaaj Otalo uses a voice-based platform that operates through low-bandwidth voice channels, enabling farmers to access agricultural advice and interact with experts without needing Internet access or advanced devices [102]. This voice-based approach reduces dependency on literacy skills and works seamlessly on basic mobile phones, ensuring accessibility for a broader range of users. Esoko, on the other hand, uses SMS as its primary communication method, delivering real-time crop prices, weather forecasts, and agricultural tips directly to farmers. The SMS-based design minimises data usage, conserves device resources, and ensures compatibility with basic mobile phones, making it a sustainable choice for regions with limited connectivity [126]. Similarly, MISTOWA focuses on improving market linkages by providing market information and trading networks via simple, resource-efficient digital interfaces that can function on low-powered devices [125]. By prioritising offline functionality and lightweight data processing, MISTOWA enables rural farmers to connect with markets and gain visibility even in areas with intermittent Internet access. Together, these technical adaptations demonstrate how agricultural software systems are optimised for low-resource settings, making critical information and resources available to support farming communities regardless of infrastructure constraints.

5.2.6. Community Engagement and Feedback

The community engagement and feedback category comprises software tools that allow for two-way connections with communities and gather insightful feedback. These resources are essential for empowering and involving communities in low-resource environments. They encourage openness, respond to community concerns, and streamline information flow, ultimately enhancing community engagement and decision-making.

The following are the software systems in this category:

• Ushahidi
• Avaaj Otalo
• Voice-and Web-Based Feedback System

This category’s community engagement software systems utilise specific technical features to ensure they can operate effectively in low-resource environments. Ushahidi, for instance, leverages SMS and email as primary data input channels, allowing community members to submit feedback or report incidents through basic mobile phones without relying on Internet access [129]. This SMS-based approach minimises data usage, conserves device resources, and extends accessibility to users with limited connectivity, making it a practical tool for crisis mapping and public engagement in remote areas. Similarly, Avaaj Otalo employs a voice-based platform allowing community members to leave messages or receive information through pre-recorded responses [102]. This system operates on low-bandwidth voice channels, enabling it to work on basic mobile devices, bypassing literacy barriers, and functioning effectively without Internet access.

5.2.7. General Purpose

This category of general-purpose software systems offers versatility and adaptability for a wide range of uses and domains. They are useful in low-resource environments because they provide adaptable solutions that may be tailored to meet unique requirements. These tools enable organisations and people to develop unique and flexible solutions to a wide range of challenges in low-resource environments.

The following are the software systems in this category:

• mira
• ODK
• Shreddr
• printr
• CALJAX

The general-purpose software systems incorporate various technical features to function effectively in low-resource environments. mira, for example, is a data management tool designed to run on basic devices with limited processing power [42]. It uses lightweight coding and a streamlined interface to conserve computing resources, making it ideal for data tracking and analysis in areas where high-powered devices are not available. ODK (Open Data Kit) is another flexible tool that enables users to collect and manage data offline, syncing with a central database when connectivity is restored [118]. This offline capability makes it useful for surveys, assessments, and monitoring projects in remote locations, as it minimises data usage and conserves device resources. Shreddr and printr are general-purpose document management and printing tools optimised for minimal resource consumption. Shreddr, for instance, focuses on efficient document processing with low memory usage, allowing users to manage and store large volumes of documents, even on older or low-capacity devices [124]. These technical adaptations, from offline functionality and resource-efficient coding to compatibility with basic devices, allow these general-purpose tools to be used effectively in low-resource environments. By offering adaptable solutions that cater to various needs, these systems empower users to overcome infrastructural limitations and create customised solutions for their unique challenges.

In general, the various use cases demonstrate strategies for the software systems to function on older, non-state-of-the-art hardware, among other features. To maximise accessibility, many of these applications are optimised for minimal computing resource consumption, ensuring compatibility with basic mobile phones, legacy desktop systems, and low-powered devices. For instance, applications such as Ushahidi and Avaaj Otalo utilise SMS and voice-based channels, eliminating the need for advanced processing power and Internet connectivity and making them accessible on basic mobile devices commonly found in low-resource settings. Similarly, Open Data Kit (ODK) incorporates offline functionality, allowing data collection and storage on low-powered devices, syncing data only when network access becomes available. These adaptations enable these applications to operate effectively, even on hardware lacking the specifications of more modern applications. To harmonise with older hardware, these applications are often developed with resource-efficient design principles, such as lightweight coding, modular architecture, and the prioritisation of core functionality over graphics-heavy interfaces. These practices not only conserve device memory and power but also enhance the overall usability and reliability of software systems in and for low-resource environments.

6. Strategies and Technical Adaptations for Overcoming Challenges in Low-Resource Environments

This section analyses the strategies and technical adaptations software systems can use to address specific challenges in low-resource environments. By examining the approaches taken in areas such as healthcare, education, agriculture, and crisis response, we highlight the methods that allow these systems to operate effectively despite limitations in infrastructure, technical expertise, and available resources.

6.1. Device Compatibility and Resource Optimisation

One of the primary challenges in low-resource environments is the lack of access to modern computing devices. Many systems are designed to be compatible with older hardware or basic mobile devices to overcome this limitation. For instance, CommCare and e-IMCI in the healthcare domain use lightweight architectures that enable them to function on devices with minimal processing power and memory. By prioritising core functionality over graphics-heavy interfaces, these applications conserve computing resources and extend battery life, allowing them to run effectively on older or low-powered devices.

In the education sector, Kolibri employs local caching and offline content access, enabling students to access learning materials without constant Internet connectivity. This resource-efficient design ensures that Kolibri can run on older computers or tablets, making it accessible to schools with outdated hardware. Similarly, Open Data Kit (ODK) in data collection is designed for low-resource settings by incorporating offline capabilities, allowing data entry and storage on low-end devices, with data syncing only when a network connection is available.

6.2. Offline Functionality for Limited Connectivity

Limited or unreliable Internet access is a significant obstacle in low-resource environments. To address this, many software systems incorporate offline functionality, allowing users to operate without continuous Internet connectivity. For instance, Ushahidi and ODK rely on SMS and USSD capabilities, enabling data collection and reporting even in areas with intermittent or no Internet access. These systems store data locally and synchronize with central servers when a network is available, which not only conserves bandwidth but also ensures data availability during outages.

In healthcare, CommCare supports offline data entry, allowing health workers in remote areas to record patient information in the field and sync it to a central database when they have Internet access. This offline functionality is critical for health initiatives in remote or rural areas, where consistent connectivity cannot be guaranteed, and allows for continuous data collection and management.

6.3. Adaptation to Low Literacy Levels and User Training

Low literacy rates and limited technical expertise present significant barriers to software adoption in low-resource environments. To overcome these challenges, many systems use simplified interfaces and provide training resources for users. Avaaj Otalo, in agriculture, for example, uses a voice-based interface that enables farmers to access information without needing to read or navigate a complex interface. This approach reduces the need for high literacy levels and allows for effective user engagement.

CommCare also addresses literacy and technical skills challenges by offering a straightforward, intuitive interface for healthcare workers. In addition, the system provides extensive training resources to ensure that users can operate it effectively with minimal technical background. This approach has been widely adopted in low-resource healthcare settings to improve usability and reduce the need for on-site technical support.

6.4. Financial Constraints and Cost-Effective Design

Financial constraints are a prevalent issue in low-resource environments, where software systems must be cost-effective and affordable. Systems such as M-Pesa in finance have been designed to work on basic mobile phones, using SMS for transactions rather than requiring costly data plans or advanced smartphones. This approach allows even users with limited financial resources to access banking services, supporting financial inclusion in underserved communities.

In the education sector, Kolibri is freely available as an open-source platform, which eliminates software costs for schools with limited budgets. This cost-effective design makes Kolibri an accessible resource for educational institutions in low-income regions, where budget constraints are a significant barrier to adopting new technology.

6.5. Cultural and Contextual Adaptations

Cultural factors, such as differences in language, practices, and social norms, can impact the effectiveness of software in low-resource environments. To address this, many systems include culturally and contextually relevant adaptations. For instance, Ushahidi allows for the localisation of language in its crisis mapping platform, enabling users to report incidents in their native language, which increases usability and accessibility across diverse communities.

Additionally, Voice- and Web-Based Feedback Systems are tailored for local communities, providing a voice-based interface that accommodates users who may not be literate or familiar with digital systems. This design fosters greater community participation by allowing users to contribute feedback without needing high levels of digital literacy.

6.6. Integration with Legacy Systems and Modular Design

In environments where legacy systems are already in use, software must integrate smoothly with existing infrastructure to avoid costly overhauls. Modular design is often used to enhance compatibility and scalability. OpenMRS, a healthcare record-keeping system, integrates with existing hospital data systems and can function on legacy desktop computers. This modular approach enables gradual adoption and easy updates, allowing facilities to improve their record-keeping processes incrementally without requiring extensive new hardware.

ODK also uses a modular structure, enabling it to integrate with other data collection and analysis tools commonly used in low-resource environments. This compatibility with existing tools and infrastructure reduces the learning curve for users and makes it easier for organisations to deploy ODK within their current setup.

7. Discussion

In this section, we discuss the implications of our findings for the future design of software in low-resource environments and highlight some challenges encountered during this study.

Low-resource environments present a wide variety of challenges that impact the deployment and use of software systems. We identified several of these challenges and grouped them into eleven (11) categories.

In Table 2, network issues refer to the availability and reliability of connectivity, which can significantly hinder the use of software systems, particularly those that are web-based or cloud-based [17]. Network availability is one of the most significant challenges in low-resource environments, and, as a result, many developers have integrated online/offline capabilities into recent software systems tailored for these settings. This is a recurring design pattern across different application domains, reflecting the need for flexible connectivity solutions. It is also apparent that being “online” or “offline” is not a binary state; partially online functionality as a design goal may benefit a wide range of users worldwide, not just those traditionally considered to be in low-resource environments.

Low-resource languages [142] are also becoming increasingly important, aligning closely with low-resource software systems. Machine learning for natural language processing (NLP) does not perform well when models lack sufficient training data, and collecting large datasets in low-resource environments remains a challenge. This has led to multiple projects focused on efficiently gathering data to support low-resource languages. Additionally, low-code development [143,144] has become pivotal in creating resource-efficient software systems by streamlining the development process and optimising various aspects of design and functionality.

Device limitations are another significant obstacle in low-resource environments, as access to appropriate computing devices is often constrained [7]. While mobile devices are proposed as solutions in many projects, resource availability varies widely. For instance, users in high-income countries often have high-speed Internet and advanced mobile applications, while users in low-resource settings may rely on basic feature phones with limited functionality. This disparity reflects the digital divide not only in computer and Internet access but also in mobile technology availability and usage [145,146].

Another major challenge in low-resource environments is the need for a reliable electricity supply. In many low- and middle-income countries, consistent electricity is necessary to power devices; without it, any installed software becomes inaccessible. Providing adequate electricity is a persistent obstacle, and it is estimated that by 2030, 650 million people—8% of the world’s population—will still lack electricity access, with low-income countries accounting for around 80% of this population [58].

To address the shortage of technical expertise, some developers have focused on creating toolkits that allow users to assemble applications for specific use cases without needing extensive local software development skills. This toolkit approach can benefit all software systems but is particularly valuable in low-resource environments, where additional resources may not be readily available.

In this survey, we examined 24 software systems designed for low-resource environments and categorised them based on how they addressed specific challenges. One ongoing challenge in the literature is defining what makes software “low-resource”. Researchers and practitioners take various approaches, often focusing on hardware resources such as processing power, memory, and storage capacity. Christi et al. [147], for example, demonstrate in their study on resource usage adaptations using delta-debugging that software can be considered low-resource if it operates with minimal resources while maintaining basic functionality. Identifying the factors that contribute to a software system being classified as low-resource is essential for addressing specific needs and constraints effectively. Optimising data structures, algorithms, compression techniques, and efficient data handling practices is critical for designing resource-efficient software that performs reliably in constrained environments [75]. These approaches not only improve system performance but also ensure that software remains responsive and dependable in challenging conditions.

Specific design recommendations include the following: (1) software systems intended for areas with unreliable connectivity should prioritize offline functionality; (2) software systems for low-powered devices should employ lightweight architectures; and (3) scalable, modular designs should be employed to improve adaptability across various low-resource environments. These recommendations are tailored towards achieving reliable and efficient software deployment and usage and to guide developers with concrete strategies for creating reliable software systems in low-resource environments.

8. Conclusions and Future Direction

This survey presented a collection of studies examining software systems designed for low-resource environments and the approaches taken to address common challenges in these settings. Rather than explicitly defining a low-resource environment, we highlighted key characteristics to illustrate the breadth of challenges encountered, including network limitations, device constraints, and the need for offline functionality. A total of 24 software systems developed specifically for low-resource environments were analysed based on their design strategies for overcoming these issues. Common themes, such as the use of mobile and offline systems, emerged, suggesting that certain design principles may be universally applicable across low-resource settings.

The findings indicate that software development in low-resource environments often responds directly to the unique challenges of each project. However, recurring themes across various domains imply that a more structured approach to designing software for low-resource settings could be formalised. Such an approach could be developed into best practices or design frameworks, which could be taught to developers and applied broadly, potentially enhancing the usability and functionality of software across both low- and high-resource environments.

Future directions in this field should focus on improving the scalability and modularity of software systems, making them adaptable to varying levels of resource availability. The integration of low-code and no-code platforms is especially promising, as these enable rapid customisation of applications by users with limited technical expertise. Additionally, exploring the use of artificial intelligence and machine learning to optimise software performance in resource-constrained settings could help overcome barriers such as limited network connectivity and device capacity. This would also support the localisation of software systems for low-resource languages, thereby enhancing inclusivity and accessibility in low-income regions.

As energy access remains a critical challenge in low-resource environments, future research should investigate the development of energy-efficient software systems. This could involve optimising algorithms and processes to require minimal computational resources, thus reducing the energy consumption of devices. Furthermore, new approaches to harnessing renewable energy sources to power devices and networks in low-resource settings could help address power constraints. These innovations, along with ongoing efforts to create adaptable and modular software, are poised to drive the next generation of sustainable software systems for low-resource environments.