Can Semantics Uncover Hidden Relations between Neurodegenerative Diseases and Artistic Behaviors?

Abstract

Semantics play a crucial role in organizing domain knowledge, schematizing it, and modeling it into classes of objects and relationships between them. Knowledge graphs (KGs) use semantic models to integrate and represent different types of data. This study aimed to systematically review related work on the topics of ontologies for neurodegenerative diseases (NDs), ontology-based expert systems for NDs, and the artistic behavior of ND patients. The utilization of ontologies allows for a more comprehensive understanding of the progression and etiology of NDs, the structure and function of the brain, and the artistic expression associated with these diseases. The data collected from ND patients highlights the presence of cases where artistic expression can be linked to the disease. By developing fuzzy ontologies for NDs and incorporating them into expert systems, early detection and monitoring can be supported. Through our systematic review, we identify and discuss open issues and challenges in understanding the relationship between ND patients and their artistic behavior. We also conclude that ontology-based expert systems hold immense potential in uncovering hidden correlations between these two. Further research in this area has the potential to address key research questions and provide deeper insights.

1. Introduction

Neurodegenerative diseases (NDs) are characterized by a progressive loss of brain neuron structure or function, affecting a large number of people without a clear understanding of their etiology, prevention, or genetic connections. These diseases are incurable and result in ataxia and dementia, including Alzheimer’s disease (AD), Parkinson’s disease (PD), Huntington’s disease (HD), multiple sclerosis (MS), and amyotrophic lateral sclerosis (ALS). Neurons, being the functional and structural units of the nervous system, play a crucial role in our interaction with the internal and external environment.

Artistic brains are thought to be supported by the functional interconnectivity between the cerebral hemispheres, particularly the right hemisphere, which is connected to creativity, and other parts of the brain such as the sensory projection areas in the occipital, temporal, and parietal lobes and cerebellum [1,2]. The artistic brain is extensively studied in the field of neuroscience, with research suggesting that certain regions of the brain are particularly active during creative tasks, such as the prefrontal cortex, which is responsible for decision-making and attentional control [3,4] and was found to be more active in artists compared to non-artists during tasks that require creativity. The limbic system, which processes emotions, was also found to be more active in artists, potentially allowing for better expression of emotions through their art and the creation of art that resonates with others on an emotional level.

To be precise, the artistic brain is characterized by increased activity in several key regions, including the prefrontal cortex, parietal lobes, temporal lobes, default mode network, and limbic system, which likely contributes to artists’ ability to create new and innovative works of art [1,2,3]. As described by Koletis et al. 2022 [5], these brain regions collaborate in the expression of artistic activities such as painting, dance, creative writing, and more.

A literature review revealed a connection between NDs and individuals who participate in performing arts. The symptomology and brain regions involved in art production were previously identified in the literature [6,7]. Studies [5,6,7] discovered a link between NDs and participation in performing arts, with a higher prevalence of musical and artistic abilities among individuals with autism spectrum disorder (ASD) compared to the general population [8]. This is believed to stem from enhanced abilities in perception and attention to detail in individuals with ASD, which are advantageous for tasks related to music and art. Participation in performing arts was found to have therapeutic benefits for individuals with NDs, such as improving social skills and reducing anxiety [9]. However, it is important to note that not all individuals with NDs will have enhanced artistic abilities or benefit from participation in performing arts, as the relationship between NDs and artistic abilities is complex and multifactorial. Advancements in ontology technology provide the capacity to manage and combine various forms of heterogenous big data, but a scale of aesthetic expression or style identification cannot be established through arts. Hence, research into the connection between NDs and artistic abilities requires the adoption of fuzzy concept research, as the mechanisms involved in NDs are complex and often uncertain. Studies examined artistic production in individuals before and after the manifestation of NDs and discovered alterations in style as well as new artistic talents in specific cases.

Our hypothesis is that there is a structural and functional overlap between NDs and aesthetic expression. This hypothesis is supported by studies in neuroscience, which found similarities in the brain regions and cognitive processes activated during both NDs and artistic tasks [2,3,4,5]. Structural overlap refers to the presence of similar brain regions that are activated during both NDs and artistic tasks, such as the amygdala, which is involved in processing emotions and has increased gray matter volume in individuals with ASD [7,8,9]. Functional overlap refers to the presence of similar cognitive processes used during both NDs and artistic tasks, such as enhanced perception and attention to detail and increased activity in the Default Mode Network (DMN), which is associated with creative thinking. Hence, the hypothesis of a structural and functional overlap between NDs and aesthetic expression is supported by research in neuroscience. Studies identified similarities in brain regions and cognitive processes activated during both NDs and artistic tasks, providing a biological basis for the connection between NDs and artistic abilities.

The rapid growth of data in recent years created numerous opportunities for research in various fields. In the context of dementia diagnosis and the detection of delayed symptoms, a large volume of data can be derived from various sources, including aesthetic observations, medical evaluations, knowledge data, and biological testing. To gain meaningful insights from these sources of big data, it is necessary to combine, harmonize, and interpret these data sources in a systematic manner. Aesthetic observations involve the collection and analysis of data related to art and design, such as user ratings and reviews of works of art or analyses of patterns in visual art. This type of data can provide valuable insights into how the human brain processes and interprets artistic stimuli. Medical evaluations, on the other hand, involve the collection and analysis of data related to health and well-being, such as electronic health records and medical imaging data. This data can help to provide a comprehensive view of an individual’s health status and history. Knowledge data refers to information that is derived from various sources, including books, articles, and online resources. This type of data provides a wealth of information that can be leveraged to improve our understanding of complex diseases, including dementia. Biological testing involves the collection and analysis of data related to genetics, biochemistry, and physiology, such as DNA sequencing and brain imaging. This type of data can help to provide a deeper understanding of the underlying biological mechanisms involved in the development of dementia. The integration of these diverse sources of data provides a rich and comprehensive view of complex phenomena, such as dementia. Artificial intelligence (AI) models can be trained using these data sources, resulting in more accurate predictions and a better understanding of the underlying mechanisms of dementia. In particular, ontology-based AI can be used to discover patterns in these data sources, providing valuable insights into the complex relationships between the various factors involved in dementia.

The aim of this research direction is to leverage the already existing ontologies that contribute to the organized classes and their relationship to our research topic. Ontologies provide a formal description of knowledge and are particularly useful for textualizing different formats of knowledge [10]. Furthermore, they do not have data volume limitations and can be updated and expanded with new extensions, such as fuzzy ontologies.

The objective of this paper was to survey related works that could potentially answer the following research question: can semantics uncover hidden relationships between NDs and artistic behaviors?

The paper is structured as follows. In Section 2, background knowledge and the main concepts used in this paper are presented. Section 3 presents the research methodology of this survey paper. Section 4 presents a review of the contemporary state of the art literature. Section 5 critically discusses related work based on a set of criteria highlighting the limited amount of research on NDs and artistic behavior. In Section 6, a proposed approach toward uncovering hidden relationships between NDs and artistic behaviors is presented. Finally, Section 7 concludes the paper.

2. Preliminaries and Background Knowledge

This section will provide a comprehensive overview of the essential concepts required for comprehending the in-depth review that will be presented in the following sections. The central themes to be analyzed include the definition of an ontology in general, the Semantic Web, knowledge graphs, and several ontology design tools. Additionally, we will briefly touch upon the fundamental principles that underlie the relationship between NDs and the arts.

NDs refer to a group of progressive and debilitating conditions that result in a gradual decline of the central nervous system. The hallmark of NDs is the degeneration of brain structure and function, leading to heterogeneous symptoms such as cognitive impairment, motor dysfunction, and behavioral changes. NDs are notoriously challenging to diagnose and treat due to the wide variability of their symptoms. There is currently no cure for these diseases, which result in permanent neurological disability until death. This leads to significant psychological, emotional, and pathological impacts on both the patients and their families, resulting in a reduced quality of life. Studies have demonstrated that patients with NDs often suffer from depression, anxiety, and social isolation, which can exacerbate their symptoms and further impair their quality of life [11]. Furthermore, NDs result in the formation of pathological proteins, the loss of neurons, brain shrinkage, and the disruption of neural networks. Overall, NDs are complex and multifactorial conditions with a major impact on the lives of those affected.

The Semantic Web (SW) represents the vision of the World Wide Web Consortium (W3C) [12] to create a more interconnected and intelligent web of data. This is achieved by adding a layer of meaning, or semantics, to the data available online. The use of standardized languages, such as Resource Description Framework (RDF) [12] and Web Ontology Language (OWL) [12], allows for the representation of data in a machine-readable format. The benefits of this approach are numerous, including improved search capabilities, enhanced data integration, and the automation of tasks that previously required human intervention. Furthermore, the use of semantics in web data was shown to result in more efficient and accurate data sharing, which can benefit a wide range of fields, such as healthcare, finance, and e-commerce [13]. The SW also enables the creation of knowledge graphs, which are networks of interconnected data that can be used to represent complex relationships between entities. The SW represents a major advancement in the field of web technologies and provides a means of making the web more interconnected and intelligent. Semantics can be broadly categorized into three subcategories: formal semantics, conceptual semantics, and lexical semantics [14]. Ontologies play a crucial role in facilitating collaboration and information sharing within the scientific community. They can be reused, extended, and provide explicit assumptions, and they consist of Classes, Subclasses, Instances, and Relations in a specific domain. They provide a standardized way of representing knowledge in a given field and serve as a common language for communication between researchers [15].

Moreover, fuzzy ontologies are a type of knowledge representation that allows for the modeling of real-world cases and events, which often involve uncertainty. These ontologies are based on the concept of fuzzy logic, which enables the representation of imprecise or uncertain information [16]. Research in the field of fuzzy ontologies revealed their potential benefits, such as the ability to capture the vagueness and uncertainty inherent in natural language and to manage incomplete and inconsistent information. Additionally, studies show that fuzzy ontologies can effectively model complex systems found in fields such as medicine and engineering, in which the relationships between entities are complex and uncertain. Furthermore, fuzzy ontologies can represent context-dependent information, providing a more accurate representation of real-world scenarios. As a result, fuzzy ontologies are a versatile tool for knowledge representation and are widely applicable to fields such as natural language processing, decision-making, and artificial intelligence [17,18]. An important aspect of fuzzy ontologies is the use of fuzzy membership to describe and define key concepts within a given domain [13,14,15,16,17].

OWL is a semantic web language designed for the representation of rich and complex knowledge about things, groups of things, and their relationships. OWL documents, also known as ontologies, can be referenced or referred to by other OWL ontologies, and they form part of the W3C’s Semantic Web technology stack alongside technologies such as RDF [12], RDFS, and SPARQL [12]. The current version of OWL, referred to as OWL 2, has many ontology development tools available, with Protégé [19] being one of the most widely used. Protégé is a free and open-source ontology editor and framework designed for building intelligent systems, and its plug-in architecture allows for the development of simple or complex ontology-based applications. Another open-source framework for building SW applications that also supports RDF, RDFS, and OWL is Jena [20].

After the creation of ontologies, it becomes necessary to interconnect the various objects within the ontologies. This is achieved through the creation and development of a knowledge graph (KG). A KG is a graph-based representation of knowledge that schematizes the relationships between entities and their properties [20]. The KG serves as a tool to organize, understand, and leverage large and complex data sets. The creation of KGs is the subject of extensive research in the field of artificial intelligence, and many research papers highlight the potential benefits of using a KG [21]. Some of these benefits include improved search and recommendation capabilities, better data integration, and the ability to automate tasks that currently require human intervention. The knowledge within a KG is organized and schematized using a multidomain graph. The nodes of the graph represent the entities of interest, and the connectors between the nodes represent the types of relationships between the entities [20,22]. By representing knowledge in this manner, it becomes possible to understand and leverage the data within the KG more effectively.

The specific pillar of this literature review focuses on the arts and artists. “The arts” is an abstract notion that refers to human expression and creativity through the use of fine arts, and those who perform the arts are known as artists. The arts can be seen as a form of communication and human expression, and the ability to perform the arts is dependent on the interaction between the psychosomatic partnership and the motor and sensory nerves. Both types of nerves start and end in the brain via a complex neurological network [17,22]. All of the knowledge represented in the ontologies and KGs should be integrated into smart systems. An expert system is a type of computer system or software that can emulate human decision-making abilities. It is a subcategory of AI, and its scope includes resolving complex problems through the use of reasoning knowledge. Expert systems are divided into four categories, with the fuzzy logic-based expert system being the category of interest in this literature review [17].

3. Survey Methodology

The methodology used in this review was based on the PRISMA approach, which is a systematic and rigorous method for reviewing and synthesizing available literature [23]. To conduct our research, we thoroughly searched four of the most commonly used web repositories, including Google Scholar, PubMed, Research Project, and Human Brain Project. Our search involved the use of keywords that were categorized into three conceptual groups. These keywords were selected to ensure that we captured the most relevant and up-to-date studies related to our research question. By following the PRISMA methodology, we were able to carefully screen and select studies that met our inclusion criteria and ultimately arrive at a comprehensive understanding of the existing literature on the topic of interest.

The first category focused on the relationship between NDs and artistic behavior, with keywords such as “neurodegenerative diseases”, “artists”, “arts”, and “painting”. The second category aimed to identify papers related to ND and expert systems, including those based on fuzzy logic, to diagnose the disease in its early stages through symptoms. Keywords used in this category were “fuzzy ontologies”, “expert systems”, and “neurodegenerative diseases”. The third category included ontology-based systems that capture the concept of ND and its related diseases, such as Parkinson’s and Alzheimer’s. Keywords used in this category were “ND ontologies”, “Parkinson disease Ontologies”, “Fuzzy Ontologies for ND”, and “Alzheimer’s disease Ontologies”.

Our review aims to provide a comprehensive and up-to-date synthesis of the literature in the field of NDs and expert systems. The initial search on Google Scholar for the first category yielded approximately 21,300 results, whereas PubMed provided about 112 findings. The Research Project database contained 195 records, and PubMed contained 112 records. After a thorough screening process, including matching and eliminating duplicates across the different repositories, we narrowed down the results to 953 works. In the second screening, we selected 351 records, focusing on works published after 2012. During the third screening, only 179 of the remaining papers were considered relevant to the review. In the fourth screening, we excluded works with unclear hypotheses, a lack of supporting evidence, poor analysis, or language problems. Ultimately, we arrived at 27 records for our final analysis (as shown in Figure 1). The keywords used in the first category included “artistic creativity”, “art therapy”, “Parkinson’s disease”, “brain damage emotion perception”, “art”, “creativity”, “dementia”, “emergent creativity”, “visual arts”, and “neurodegeneration”. These keywords were used in combination with “and” or “or” operators to yield the most relevant results.

To conduct our search for the second category of our review, we utilized Google Scholar, which yielded an initial result of 20,600 findings. Following the first screening process, which involved eliminating duplicate and matching records, we were left with 711 works. The second screening process was focused on finding works published after 2012 and resulted in narrowing down the pool of works to 125. During the third screening process, we assessed the eligibility of 84 works and further narrowed down the list to 11 records during the fourth screening process. The final list of works included those that had a clear hypothesis, adequate supporting evidence, and high-quality analysis, whereas works that had problems with language, poor analysis, or lacked sufficient evidence were excluded (Figure 2).

In combination with “and” or “or”, the keywords used for the second category were “Dementia”, “Diagnosis”, “Clinical Decision Support System”, “Fuzzy logic”, “fuzzy ontology”, “Expert System”, “Ontology”, “Fuzzy logic Alzheimer disease”, “Ontology-based modelling”, and “Brain disorder”. The inclusion of these keywords allowed us to identify works that focused on the use of expert systems based on fuzzy logic to diagnose NDs in their early stages by assessing symptoms.

In the third category of our review, we conducted a search using Google Scholar, which initially provided us with a total of 20,800 findings, and 688 records were identified from PubMed. After the first screening process, which involved filtering out duplicates and matching records, we concluded with 1458 works. The second screening process narrowed down the number of works to 656 and focused on finding works published after the year 2012 related to the first category. During the third screening process, we assessed the eligibility of 282 works and further narrowed down the list to 17 records during the fourth screening process. This final list consisted of works that had a clear hypothesis, adequate supporting evidence, and high-quality analysis. We excluded works that had problems with language, poor analysis, or lacked sufficient evidence (Figure 3).

The keywords used in combination with “and” or “or” for the third category included “Alzheimer’s disease ontology”, “Neurodegeneration”, “Ontology”, “Semantic Web”, “Ontology Alignment”, “Ontology integration”, “SNOMED CT”, “AlzPathway”, “ADMO”, “ADO”, “ADIO”, “Biomedical domain”, “Ontology”, “Data”, “OWL”, “RDF ontologies”, “ontology reuse”, and “neuroscience ontology”.

4. State-of-the-Art

This section contains a comprehensive review of the literature, which is divided into three subcategories for thorough examination. The first subcategory focuses on the examination of the findings that pertain to the relationship between NDs and artistic behavior, as the objective was to investigate the modifications that occur in one’s artistic style as a result of NDs. The second category is dedicated to the examination of the literature that encompasses knowledge regarding the development of fuzzy logic-based expert systems or decision support systems for the early diagnosis or effective therapy of ND patients. The core focus of this review and future research was ontologies, and thus, the third category provides an in-depth review of the most prominent ontologies that were established in relation to NDs.

4.1. Neurodegenerative Diseases and Artistic Behavior

Recent studies explored the relationship between various neurodegenerative disorders and changes in patients’ artistic behavior [24,25,26,27,28,29,30]. A number of different approaches were adopted, varying in methods used and participant mix. Many of these studies highlight changes in painting styles among ND patients [24,25,26,27,28,29,30], whereas others focus on the increased creativity of dementia patients [31,32]. Several authors compared the artistic behavior of ND patients to that of healthy controls of the same age [4,24,28,29,33,34,35]. In one study [24], the creative potential of dementia patients was evaluated by comparing the artistic products of 15 professional artists and 15 dementia patients with no prior artistic experience. After participating in a custom artistic education program, the results confirmed that dementia patients can exhibit a great deal of creative potential, even despite progressive brain damage.

Another study by Ruggiero et al. [4] aimed to examine the link between creativity and pathology by using the Diverse Thinking Test. 17 idiopathic PD patients, 11 frontotemporal dementia (FTD) patients, and 15 healthy subjects participated. The results showed that PD patients and healthy controls performed better than the FTD group, highlighting the connection between creativity and NDs.

Harrison et al. [35] aimed to gain a deeper understanding of the relationship between NDs and personal artistic styles. They analyzed the artwork produced by four individuals suffering from various forms of dementia and four healthy individuals who participated in an art exercise. The authors concluded that different forms of dementia can be distinguished by the unique artistic styles produced. Lauring et al. [29] compared the aesthetic and formal evaluations of paintings created by PD patients and healthy individuals. The sample consisted of 21 PD patients and 23 age and gender-matched controls who evaluated a series of paintings using a modified version of the assessment of formal art attributes (AAA). The authors included demographic data and data related to art interests, and after analyzing the data, they found a significant difference in emotional expressivity between PD patients and healthy controls, which was probably due to dopamine pathways. However, most of these studies have limitations, such as a lack of information on the severity of the disease and a limited sample size as well as not specifying if the patients had prior artistic experience.

Few researchers [27,28,36,37,38] have attempted to compare changes in artistic production among professional and even renowned artists who were diagnosed with a ND to the alterations in artistic behavior among artists who experienced normal aging. Forsythe et al. [27] utilized fractal analysis to determine the presence of alterations in artistic expression and NDs. Fractal analysis is used as a verification method to establish the authenticity of an artistic work. The fractal dimension measures the extent to which a pattern fills a space. The results indicate that the artistic production of artists who suffer from NDs is significantly different from that of artists who experience normal aging. This distinction is particularly important, as it lays the foundation for creating a system for early detection of the disease.

The authors in [28] conducted an analysis of the paintings of famous and unknown artists before and during different stages of progressive neurodegeneration caused by various types of dementia. The analysis included the examination of paintings along with medical information such as the type of ND, design, color, abstraction, simplicity, and overall modification of the art form. The findings show that each type of dementia damages different regions in the brain, leading to modifications in artistic expression. For example, visual perception may be impaired in certain regions of the brain, leading to color deficit reproduction in dementia with Lewy bodies, which causes recurrent visual hallucinations. In [38], the changes in painting style among artists with NDs were presented and compared to the painting style of normally aging healthy controls. In [39], the changes that occurred in the painting style of PD patients were particularly emphasized.

Studies such as [33,40] focused on examining the impact of different tasks on the progression of NDs in homogenous samples of patients, such as PD patients or dementia patients (DPs). Johnson et al. [40] studied the effects of an 8-week visual arts training program on cognition in people with dementia but found that the program did not yield significant benefits in overall cognition, working memory, or delayed recall. In their research [33], the authors investigated whether artistic production in PD patients was triggered by dopaminergic treatment (DT) or pre-existing artistic skills but found no evidence to support the idea that DT is the cause of an emergence of artistic creativity.

Other studies [25,30,34,41,42] explored the potential therapeutic benefits of art therapy for patients with NDs. Mirabella et al. [25] provided a striking example of how theatre training can help patients manage their motor and emotional control and provide new strategies for balancing mind and body. Windle et al. [41] examined art-based activities as therapy for DPs, whereas [42] explored the potential benefits of visual art training for these patients. Studies such as [30,43] investigated dance and music as ways to assess the severity of NDs, with [43] focusing on the impact of dance on the functional mobility of PD patients and [30] presenting music and art therapy as interventions aimed at improving the quality of life of patients and potentially slowing down or even halting disease progression.

It was observed [34] that patients with NDs may start to show artistic inspiration and creation after the onset of the disease, even if they were not previously affected by the arts. Music therapy, for example, was found to be protective against age-related cognitive decline and to improve quality of life in AD patients [34]. By practicing the arts during their illness, patients may reduce stress, apathy, and anxiety and potentially have a positive impact on neuroplasticity.

Geser and their colleagues [30] conducted a review that analyzed the effects of FTD on artistic activities and manifestation, paying close attention to the chronological development of each patient. Their work supports the idea that creative stimuli can strengthen a person’s self-healing ability and self-confidence as NDs progress. The authors recommended further analysis of the progression of NDs to gain a better understanding of the FTD process and physical impairment. In [44], the short-term clinical benefits of a dancing lesson were evaluated by comparing the functional mobility of a group of PD patients who took a dance class to another group who did not. The integration of creativity tasks into the rehabilitation process of ND patients was explored in studies [44,45,46,47]. Although the use of creativity tasks was proposed for ND rehabilitation, the authors stressed the need for further quantitative study to examine the results.

Studies relating to artistic activities as therapy have several limitations. Despite the potential for therapeutic implications to be a powerful intervention, rigorous testing of their effectiveness is necessary to establish a solid evidence base. Clinical scales, neuropsychological tests, petrophysical tests, and brain imaging techniques are some of the evaluation methods that can provide robust results and offer insights into the connection between art and brain damage [25]. Some researchers compared artistic behavior to the progressive neural damage caused by NDs [26,48,49]. In [25], articles and book chapters from older papers (published before 2017) were analyzed in terms of disease severity, aging, and the type of ND(s) involved (FTD, Dementia with Lewy Bodies, AD). A common pattern of artistic production among ND patients was found, consisting of changes in spatial construction quality, visual perceptual characteristics such as color and contrast, and a general simplification of planning. These observations are linked to brain regions damaged by NDs. The abstract representation of images reflects the distortion of global visuospatial aspects and is caused by the degeneration of nerves in the right parietal region. In [48,49], the use of electroencephalogram (EEG) and magnetic resonance imaging (MRI) was proposed to link artistic behavior to specific brain damage caused by NDs.

4.2. Expert Systems and Decision Support Systems for NDs

Alexiou et al. presented a new multi-scale ontology-based modeling technology for the accurate and progressive diagnosis of PD in their study [50]. The authors aimed to exploit heterogeneous patient data and match new associated physical and biological biomarkers through a multi-layer neural network. The input for their expert system was the patient’s health record (PHR), including various data such as neuropsychological test scores, brain tissue samples, genetic material and mutations, molecular data, MRIs, electrophysiological data, EEGs, biomarkers detected from blood, and other measurement data of mitochondrial dysfunctions. The knowledge base used by the expert system was an ontology-based model that combined structured vocabularies with a small set of relationships between the members, which described gene product characteristics, gene product annotation, and diseases. This vocabulary was specified with definitions of classes, relations, functions, and other objects. The authors’ ultimate goal was to identify the patient’s status and appropriate treatment, which was made possible through the use of a multi-layer neural network. Although the results of the proposed model were promising, the authors acknowledged that the validation of their model was a crucial aspect that needed to be addressed in future research.

Hosseini et al. [51] developed a clinical decision support system (CDSS) for the diagnosis of MS to help physicians diagnose MS with a relapsing-remitting phenotype. The CDSS was developed in four stages: requirement analysis, system design, system development, and system evaluation. To evaluate the efficiency and applicability of the software, the data of 130 medical records of patients aged 20 to 40 between 2017 and 2019 were used. The results of the evaluations of the efficiency and applicability of the software demonstrate the high acceptability of the system and the satisfaction of all of the operational expectations of users. The CDSS provided guidance in decision-making and helped physicians in the timely and accurate diagnosis of MS.

Zekri et al. tackled the challenge of differentiating and accurately diagnosing AD [52,53] by proposing a fuzzy ontology called AlzFuzzyOnto [54]. This fuzzy ontology is related to AD concepts and enables the semantic representation of medical data for the diagnosis and continuous support of AD patients. The ontology incorporated the uncertainties and inaccuracies associated with AD through the concept of fuzziness. The authors used the mind ontology [55] as an initial ontology and extended it to develop AlzFuzzyOnto. They also utilized SNOMED (Systematized Nomenclature of Medicine-Clinical Terms) CT [56], a structured clinical vocabulary for use in an electronic health record, and the Unified Medical Language System (UMLS) Metathesaurus [57] as standards for enabling the sharing and reuse of this domain ontology. UMLS is a large biomedical thesaurus that is organized by concept or meaning that links synonymous names from over 200 different source vocabularies. The initial classes used were “Doctor”, “Patient”, “Diagnosis”, “Enrolment”, “Follow-Up”, “Test”, and “TestValue”. The authors added further classes related to AD diagnosis, questioning, symptoms, and entourage as well as classes representing general concepts regarding AD support such as “Therapies”, “Phase”, and “Current State”. They fuzzified various classes such as “Patient”, “Pharmacotherapy”, “Non-Pharmacotherapy”, “Current State”, and “Diagnosis”, and introduced the fuzzy relation “In”. All of the classes were related to other classes through properties such as HasSymptoms, HasTherapies, HasCurrentState, Corresponding Patient, Corresponding Follow-Up, HasDiagnosis, and HasEnrollment. Although the proposed ontology was innovative and showed potential, the authors noted that it still needed to be validated in the future.

One example of an ontology-based expert system for neurodegenerative diseases is the Parkinson’s Disease Ontology-Driven Expert System (PD-ODES) [58]. PD-ODES is designed to assist clinicians in the diagnosis and management of PD using an ontology-based approach. The system employs a rule-based inference engine to provide recommendations for treatment and management based on patient data. PD-ODES showed promising results in preliminary studies and has the potential to improve the accuracy and consistency of diagnosis and management of PD. PD-ODES was evaluated in a pilot study with 14 patients and showed promising results in terms of the accuracy and consistency of diagnosis and management recommendations.

In a study conducted by Amanzadeh et al. [59], the authors explored the challenge of diagnosing AD. They suggested that a CDSS could provide a solution for this problem by offering an accurate and early diagnosis of AD. The authors noted that CDSSs can be designed using either non-knowledge-based technologies, such as machine learning techniques, or knowledge-based technologies that incorporate medical guidelines. Although they emphasized the important role of physicians, they also aimed to help health providers reach a more accurate diagnosis by eliminating the difficulties associated with diagnosing AD. However, the authors did not address how to integrate heterogeneous data types into the CDSS or how to update the system as new knowledge becomes available.

Stavropoulos et al. [60] developed a knowledge-based proof-of-concept (PoC) application for monitoring MS patients. The platform consists of a mobile application and a web portal that allows patients to remotely monitor their symptoms and communicate with healthcare professionals. The platform also includes CDSSs for healthcare professionals, based on clinical guidelines and patient data, to assist in the management of MS patients. The goal of the project is to improve the quality of life for MS patients, reduce the burden on healthcare systems, and increase access to specialized care for patients in remote areas.

Sherimon [61] conducted a review of existing CDSSs dedicated to the AD domain. The review was motivated by the growing interest in early detection of AD, which is key to delaying the progression of its symptoms. Sherimon [61] found that ontology-based systems were the most used, with AlzFuzzyOnto being one of the more prominent examples. The PredictND tool, [62] which was found to increase accuracy by up to 3%, was also highlighted in the review. Other ontologies mentioned in the review include Gene-Ontology (GO) in AD, Ontology Design Patterns (ODP), Neurodegenerative Disease Data Ontology (NDDO), Alzheimer’s Disease Ontology (ADO), MIND, SWAN, and SNOMED CT. These ontologies provide a semantic foundation for the diagnosis of AD and standardize the data related to this domain. Overall, the authors of both studies [59,61] highlight the potential of CDSSs to improve the accuracy of AD diagnosis and support the work of healthcare providers.

The authors of the study [59,61] concluded that the use of an CDSS in combination with semantic knowledge and big data analysis holds significant potential for more accurate and differential diagnoses of NDs. This is due to the complex and heterogeneous symptoms of NDs, which can be challenging to diagnose without a comprehensive approach. One of the interesting findings in the authors’ work is the EU neuGRID4You (N4U) project [63] that was carried out in 2019. This project was based on the use of big data and fuzzy logic and aimed to calculate a patient’s Alzheimer’s Disease Identification Number (ADIN) using a big data repository and fuzzy processing. This calculation helped to identify patients with varying degrees of AD. The authors [64] also implemented and evaluated the Alzheimer’s Disease Diagnosis Ontology (ADDO), which acts as a comprehensive semantic knowledge base. The authors [64] improved the previous version of ADDO by using fuzzy concepts and extending it into a fuzzy ontology. The definition of the fuzzy sets in this study was done using the Mini-Mental State Examination (MMSE) and the Alzheimer’s Disease Assessment Scale-Cognitive Subscale (ADAS-Cog) [65]. The ADDO was validated using the Hermit reasoner [64].

Another significant study related to ontologies and NDs was conducted by Dramé et al. [66]. This study presented an ontology-based model for building a robust ontology, taking AD diagnosis as a use case. The authors aimed to provide a structure for a proposed ontology-based fuzzy CDSS model and to compare the ADDO with existing ontologies in terms of completeness and reusability. The authors support that the use of CDSS, semantic knowledge, and big data analysis has the potential to revolutionize the way NDs are diagnosed, providing more accurate and differential diagnoses while also ensuring semantic interoperability.

In another study, the authors proposed a robust architecture for the construction of an expert system for the early diagnosis of NDs [67]. The architecture consists of five modules: the content management module for the knowledge base (KB), the ontology import/export module (OWL/Fuzzy OWL), the screen forms generation module, the user interaction module, and the administration module. The system was implemented using the Java programming language and the Spring Boot framework, and the communication between the modules was facilitated through representational state transfer (REST) [67].

One of the main advantages of using a graph database management system (GDMS) such as Neo4j [68] is that it can resolve some of the problems faced in the process of Knowledge Base engineering, such as limitations in multi-user mode, support for transaction mechanisms, and the low speed of inference mechanisms [67].

Another expert system for the early diagnosis of NDs was presented in a separate study [69]. The system consists of a user interface and a user designer, which interact with each other and with the inference engine and the knowledge base. The domain expert is connected to the knowledge database and provides it with rules and facts, making the system more comprehensive and effective [69]. The user interface interacts with the inference engine and the knowledge base while the domain expert provides input to the knowledge database. The study aimed to implement symptoms into the medical expert system in order to diagnose one of five disorders: AD, Creutzfeldt–Jakob disease, HD, MS, and PD. The system was implemented using the C# and .NET programming languages, and Microsoft SQL Server 2012 served as the relational database management system (RDBMS) [69].

4.3. Ontologies Related to NDs

Martinez Romero et al. [70] investigated the issue of selecting an appropriate biomedical ontology for a given research topic. They found that the sheer amount, complexity, and diversity of available biomedical ontologies made it difficult to choose the right one for a specific research project, as the choice would depend on the study’s characteristics. In response to this challenge, the National Centre for Biomedical Ontology (NCBO) [71] released the Ontology Recommender in 2010 [70]. This service provides assistance in choosing the right ontology by suggesting ontologies that are best suited to a biomedical text corpus or a list of keywords. The Ontology Recommender version 2.0 uses several criteria to evaluate the suitability of ontologies for biomedical text data, including the extent to which the ontology covers the input data, the level of detail of the ontology classes that cover the input data, the acceptance of the ontology in the biomedical community, and the domain-specific specialization of the ontology. The service is available via a web service API and a web-based user interface, which was developed using the Ruby-on-Rails web framework [72] and the JavaScript language. The evaluation of Ontology Recommender 2.0 showed that it provides higher quality suggestions, better coverage of input data, and more detailed information about concepts compared to previous implementations.

An ontology that is intended to support the integration and sharing of data and knowledge related to diseases across multiple domains, including clinical practice, research, and public health, is the Disease Ontology (DO) [73]. It is a formal ontology that provides a standardized vocabulary and hierarchy of concepts for representing knowledge related to human diseases. The DO includes a set of classes and relationships that define concepts related to diseases, including “disease”, “symptom”, “diagnosis”, and “treatment”. These classes are organized into a hierarchical structure, with more specific concepts being subclasses of more general concepts. For example, “lung cancer” is a subclass of “cancer”. The DO includes a set of relationships that define the connections between different concepts and is used in a variety of applications and systems, including electronic health records, CDSSs, and biomedical research databases. The DO is freely available [74] and can be accessed through several online resources, including the National Centre for Biomedical Ontology (NCBO) BioPortal [71] and the UMLS Metathesaurus [57].

Another ontology concerning diseases that are related to motor symptoms is the Geriatric Motion Assessment (GMA) ontology, which is related to the assessment of motion and function in older adults [75]. The ontology is intended to support the integration and sharing of data and knowledge related to geriatric motion assessment across multiple domains, including clinical practice, research, and education. The GMA ontology includes a set of classes and relationships that define concepts related to geriatric motion assessment, including “assessment instrument”, “test”, “task”, and “outcome measure”. These classes are organized into a hierarchical structure, with more specific concepts being subclasses of more general concepts. For example, “balance test” is a subclass of “assessment instrument”. The ontology is used in a variety of applications and systems, including electronic health records, CDSSs, and biomedical research databases.

Gomez-Valades et al. [76] conducted an analysis and comparison of ontologies used in the area of AD as part of an expert system. Ontologies play an important role in homogenizing information, allowing for the integration of different types and structures of information as well as the inference of additional information based on stored data. The authors analyzed several ontologies and evaluated them based on criteria such as standardization of terminology, information storage and retrieval, and support for diagnosis. The authors found that some of the ontologies had high reusability, class hierarchy, and metadata, whereas others were lacking in these areas. Additionally, some of the ontologies focused on term standardization, whereas others were part of an expert system.

Gibaud et al. [77] presented the NeuroLOG ref middleware data management layer in their paper, which provides a platform for sharing heterogeneous neuroimaging data using a federated approach. The data management system captures the semantics of shared information through a multi-layer application ontology and a federated schema, which aligns the heterogeneous database schemas from different legacy repositories. The NeuroLOG system ref can translate the relational data into a semantic representation and support semantic search using a semantic search engine. The authors highlight the importance of the distributed approach for neuroscience data management. The NeuroLOG repository [77] integrates data from multiple federated sites and encompasses images stored as regular files as well as metadata stored in relational databases. This metadata could include information on the content of images, image acquisition conditions, subjects involved, context of image acquisitions, and scores obtained from neurological and cognitive assessments. The OntoNeuroLOG ontology, which was built as a modular multi-layer application ontology, was based on the preliminary work done in the NeuroBase project [78] and reuses the OntoNeuroBase ontology.

The Biomedical Informatics Research Network (BIRN) [79,80] program was a pioneering effort in the field of data integration, utilizing a federated method that incorporated ontology-based mediation in the realm of neuroscience research. Launched in 2001, BIRN aimed to build an infrastructure that could facilitate data exchange and computation resources, with the ultimate goal of advancing our understanding of brain anatomy and function, from the cellular level to behavior. During its first phase (2001–2008), BIRN placed significant emphasis on the creation of ontologies, which resulted in the development of the BIRNLex ontology [79,80,81,82]. This resource, more of a lexicon than a full-fledged ontology, is now widely accessible through the NCBO BioPortal [71]. In parallel with the development of the ontologies, BIRN also introduced a technologically advanced architecture built around the storage resource broker (SRB), a mechanism that enables the virtualization of files in large federated systems [81]. The BIRN federated system currently comprises 11 HIDs. This PoC demonstrated by BIRN showed that ontology-based mediation could be an effective way of integrating data in the field of neuroscience and has had a lasting impact on the field. The BIRN program serves as a shining example of the power of collaboration and the potential for progress that can be achieved through collective efforts.

Gomez-Valades et al. presented a comprehensive work in their paper [83], outlining the development of the Neurocognitive Integrated Ontology (NIO), which is available through NCBO BioPortal [71]. The NIO ontology represents a significant contribution to the field of neuroscience research by integrating the main domains of NDs, diagnostic tests, cognitive functions, and brain areas. The authors aimed to address common limitations in existing ontologies by developing an ontology that is capable of being adapted to new projects and expanding to incorporate new knowledge domains. To evaluate the stability and usefulness of the NIO ontology, the authors performed two case studies: adapting the ontology to a new project and updating the ontology using a new ontology. The inner reasoner of Protégé [19] was used to assess the stability of the ontology. The results showed that the NIO ontology was successful in overcoming the difficulties of adapting and expanding existing ontologies, making it a valuable tool for researchers.

In the field of AD research, the early diagnosis of moderate cognitive impairment (MCI) has become a crucial goal, as it is a transitional phase between normal aging and dementia. However, information related to MCI and AD is scattered across various forms and standards created by different systems, making manual interaction problematic. To address this challenge, ontologies emerged as a solution due to their ability to provide homogeneity and agreement in data representation and reuse. The NIO ontology represents a major effort to address this challenge by integrating knowledge about neuropsychological testing (NT), AD, cognitive processes, and brain regions. The ontology promotes interoperability and data access by merging disparate information from several fields, making it relevant for other study groups as well. The authors followed the ontology-building life cycle to ensure the stability and reusability of the NIO ontology, combining and extending terms from four separate reference ontologies [83]. The NIO ontology is a well-structured and comprehensive ontology that offers a valuable solution to the challenge of integrating disparate information in the field of NDs and cognitive functions. The ontology is not only useful for its intended domain but also relevant for other study groups due to its ability to promote interoperability and data access.

Younesi et al. [84] presented the Parkinson’s Disease Ontology (PDON), which was built using the standard life cycle of ontology development and provides a standardized vocabulary and set of definitions for describing and representing information about PD and its associated symptoms, treatments, and clinical studies. The PDON is designed to support interoperability and data sharing in the PD research community and can be used to enhance the accuracy and consistency of data annotation and analysis. The authors went to great lengths to evaluate the ontology on several levels, including functional, structural, and expert evaluations, to ensure its quality and usability. They also introduced a new metric to measure the gain of new knowledge from using the ontology. Additionally, a cause-and-effect model was created and two gene expression studies from the Gene Expression Omnibus database [85], a public repository of high-throughput gene expression and other functional genomics data sets, were re-annotated to demonstrate the usefulness of the ontology. The ontology features a subclass-based taxonomic hierarchy that encompasses a broad range of biomedical concepts, from the molecular level to the clinical level, and includes 632 concepts organized under 9 different views [84]. The evaluations showed that the ontology has the ability to answer specific questions related to PD.

However, more detailed information about the patient cohort is often contained in the original article that describes the data set. In this case, the relevant data collection was disclosed in an open access paper, and the journal’s rules allowed for automated analysis of the complete text. As a result, the entire text of the paper reporting was subjected to automated analysis using PDON keywords [86,87]. This was achieved through a UIMA-based annotation procedure using words from the PDON nomenclature. The Unstructured Information Management Architecture (UIMA) [88] is a type of content analytics software that is used for natural language processing in large amounts of unstructured material, such as biomedical papers [87].

The Parkinson Movement Disorder Ontology (PMDO) [89] is a valuable ontology for PD that was developed through an extensive review of literature led by movement disorder specialists and a systematic analysis of movement disorder instruments. It was developed using an iterative ontology engineering process, and it focuses on Parkinsonian disorders. PMDO contains classes representing three broad categories—neurological findings, treatment plans, and instruments—used to evaluate various traits of PD. This ontology is an important resource for researchers and clinicians to annotate, share, and integrate data related to PD and other movement disorders, as it provides a standardized vocabulary for these conditions.

Another well-known ontology related to PD is the Human Phenotype Ontology (HPO), which is a standardized vocabulary of human phenotypic abnormalities and related medical terms organized into a hierarchical structure [90]. The HPO is used to annotate clinical and genetic data, and it enables the integration of diverse types of data in biomedical research, including genetic, clinical, and model organism data. HPO contains over 13,000 standardized terms that describe phenotypic abnormalities, such as abnormal morphology or function of organs, tissues, or cells, and the signs and symptoms of diseases [90]. The ontology also includes information about the inheritance patterns of phenotypic abnormalities and links to other biomedical ontologies and databases. The HPO is used in various applications, including clinical diagnosis, genetic counseling, and research. It is also used in the analysis of genome-wide association studies (GWAS) and in the development of computational algorithms for the identification of disease-causing genetic variants. The HPO is maintained and developed by the HPO Consortium [90], an international group of researchers and clinicians who collaborate to improve the ontology and its applications encouraging community participation and contributions to improve the quality and coverage of the ontology.

The Parkinson’s Disease Knowledge Ontology (PDKO) is another formal representation of knowledge related to PD and PD-related disorders [91]. It aims to provide a standardized vocabulary and hierarchy of concepts for representing PD-related knowledge, making it easier to share, integrate, and reason about data from different sources. The PDKO is based on the OWL and includes a set of classes, properties, and axioms that define the relationships between different concepts [91]. Some of the key classes in the PDKO include “Parkinson’s disease”, “symptom”, “neurotransmitter”, “gene”, “drug”, and “clinical trial”. These classes are organized into a hierarchical structure, with more specific concepts (e.g., “tremor” and “dopamine”) being subclasses of more general concepts (e.g., “symptom” and “neurotransmitter”). The PDKO also includes a set of properties that define relationships between concepts. There is a “causes” property that links a disease to the factors that contribute to its development and a “treats” property that links a drug to the conditions it is used to treat. It was developed as part of the National Institute of Neurological Disorders and Stroke (NINDS) Parkinson’s Disease Biomarkers Program [92], which is a multi-institutional initiative aimed at improving the diagnosis and treatment of PD. The PDKO is also a component of the larger Neuroscience Information Framework (NIF) project [93], which is a collaborative effort to develop a comprehensive ontology for neuroscience research.

Malhotra et al. [94] recognized the significant barrier to effective information retrieval and integration in the field of AD. To overcome this, they proposed the use of formal ontologies, which can help construct specific knowledge domains. Gene ontology and SNOMED already demonstrated the benefits of using ontologies for annotating and recording clinical data. The Fraunhofer Institute SCAI [95], as a bioinformatics partner in the Neuroallianz Consortium [96], saw the need for a more efficient system to represent the knowledge structure in the AD domain. This prompted them to draft an ontology that would represent various aspects of current knowledge in AD, including clinical features, treatment, risk factors, and more. The ontology, known as the Alzheimer’s Disease Ontology (ADO) [97], was designed to conform to typical higher ontologies while still maintaining concept descriptions that are similar to normal language terms. The hierarchical structure of the ADO serves as a strong navigation tree for terminology integration and text-mining applications.

Malhotra et al. [94] constructed the ADO, which contains information focused on the main biological components of AD such as preclinical, clinical, etiological, molecular, and cellular mechanisms. The ontology was enriched by adding synonyms and references, and it was validated with impressive results in real-life scenarios. The ADO has the capability to reveal both established and scattered knowledge in scientific text, and it supports semantic searches. The ontology was built following the ontology-building life cycle and adheres to the criteria of top-level ontologies. Knowledge was acquired from various sources, including review articles, online books, standard knowledge bases, encyclopedias, glossaries, and informative websites. The important feature of the ontologies is that knowledge can be acquired through web-based searches, and the NCBO [71] services were used to retrieve synonym information for automatic and manual mapping to external ontologies.

Refolo et al. [98] published the CADRO as part of a comparative analysis by the National Institute on Aging. It has three categorization systems: clinical, translational, and basic. Although ADO provides valuable insights into the mechanisms of AD, it does not fully encompass the biological domain of the disease. Another study will be conducted to investigate the representation of fuzzy features of knowledge related to AD. Henry et al. created AD Map Ontology (ADMO) [99] to improve the depiction of AD pathophysiology by integrating the AlzPathway in a consistent ontology. Shoaip et al. [100] proposed the Alzheimer’s Disease Integrated Ontology (ADIO), which integrates two biomedical ontologies related to AD: ADO and the ADMO. ADO covers the clinical, preclinical, experimental, and molecular mechanisms of the disease, whereas ADMO represents the complex pathophysiology of AD. LogMap [101], a matching and alignment system for ontologies that is used to find correspondences between concepts from different ontologies, was used to align the overlapping modules of the integrated ontologies. The use of Ontograf, a Protégé [19] plugin, allowed the authors to visualize the classes graph.

On the other hand, Salvadores et al. [102] introduced the Bio Portal, a repository of over 300 biomedical ontologies that were developed in various formats, including OBO format, Rich Release Format (RFF), OWL, and more. The repository [71] contains 190 million triples, including 9.8 million cross-ontology mappings of diverse types that were generated both manually and automatically, along with their own metadata. The quality of these ontologies was assessed using metrics, peer reviews, and categorization into different domains. All of the stored data can be accessed via REST APIs and a SPARQL endpoint, and mappings between terms in different ontologies are also available. Each ontology in BioPortal is assigned a stable identifier and is indexed with version numbers.

The services of BioPortal Web [71] include retrieving ontology metadata, ontology content, downloading an ontology, and searching ontologies. Users can request ontology material based on the ontology identifier, in which case the most recent version of the ontology is returned, or by specifying a particular version number. The search service allows for searching the entire repository or a selected group of ontologies, with options for exact or approximate matching and searching in concept names and synonyms or property values [103].

The ontologies identified by our systematic review are summarized in

Table 1. Comparative table of reviewed ontologies.

Ontology	Domain	Language	Fuzzy Logic	Availability	Status
DO	Human Disease	OWL, OBO, RDF	Yes	Open Source	Up to date
HPO	Human Disease	OWL, OBO, 1n2RDF, JSON	No	Open Source	Up to date
GMA	Elderly Motor Symptoms	OWL	Yes	Free	Last update 2018
OntoNeuroLOG	NDs	OWL, RDF, JSON	No	Free	Last update 2017
BIRNLex	Neuroscience	OWL, OBO, RDF	Yes	Open Source	Last update 2021
NIO	Neuroscience	OWL, RDF	Yes	Free	Last update 2018
PDON	PD	OWL	No	Free	Up to date
PMDO	PD	OWL	No	Free	Up to date
PDKO	PD	OWL	No	-	Dead
ADO	AD	OWL	No	Free	Last update 2021
ADMO	AD	OWL, RDF	No	Free	Up to date
ADIO	AD	OWL	No	Open Source	Last update 2020

under specific criteria. The domain for which the ontology is developed and the encoded language are presented. Fuzzy logic compatibility is also mentioned. Finally, the availability of every ontology and its status are also included in the criteria.

5. Discussing Open Issues and Challenges

This review presents state-of-the-art research on technologies in the intersection of fields related to CDSSs, NDs and artistic behavior. What makes it a challenging research topic is the research question of how to combine and analyze the personal health data and artistic behavior of patients to predict and diagnose the early onset of NDs. From this study, it can be concluded that there is limited data and technology available in this research area. We conducted our review in three categories with a small degree of overlap between them. The first category reviews work that examine the symptomatology of NDs and creative expression. The second category reviews work related to the development of fuzzy logic-based expert systems and decision support systems for medical purposes. The third category reviews the ontologies for NDs, such as ontologies for PD or AD. As presented in

Table 2. Comparative table of reviewed ES/DSS supporting specific criteria.

	Artistic Behavior	Ontology-Based	Fuzzy Logic	NDs	Sensor Data	Symbolic Reasoning	ML	Status
[48]		✓		PD	✓			PoC
[49]		✓		MS		✓		PoC
[51]		✓	✓	AD		✓		PoC
[56]		✓		PD		✓		PoC
[57]				Dementia			✓	Certified CDSS
[58]		✓		MS	✓	✓		PoC
[64]				AD, MS, PD, HD				PoC

, currently, an expert system that combines technology from the three categories does not exist.

Based on this survey, it was found that NDs can damage brain cells, neural connections, and the ability to perform mental tasks, indicating that further study is necessary to understand their mechanisms more accurately. Although changes in a person’s creative style may not always indicate a decline in health, neuroplasticity can occur. Unfortunately, there is a lack of artistic behavior ontologies, which creates a significant gap that needs to be filled. On the other hand, medical ontologies have not been updated with new knowledge gained from recent years, such as the increasing use of neuroimaging technology, specific biological and cognitive tests, and the high progression of the AI sector.

Moreover, there is a vast amount of biological and non-biological data in the medical sector that needs to be expanded to provide relevant information to both physicians and patients. However, there is no available data in open-source distribution about brain scans of individuals performing creative activities. What are easily discovered are descriptive observations of the art created, such as color intensity, lines, curves, shape delineation, mind capture, movement precision, and abstract awareness of the surrounding world.

Another data stream that is more easily collected and stored is personal health and daily activity data collected from the ND patients through wearable sensors. These data can be integrated in an expert system that takes advantage of sensor-related ontologies. The Semantic Sensor Network ontology (SSN) [104] and the Sensor, Observation, Sample, and Actuator (SOSA) ontology [104] were introduced mainly for the representation of sensors and observations they facilitate. The Smart Applications Reference (SAREF) ontology [105] was introduced as a result of ETSI standardization, and its extension, SAREF4WEAR [106], was developed for modeling using different wearables available in the market. Finally, the DAHCC ontology [107], which reuses the SAREF (core) and SAREF4WEAR ontologies, was introduced with the aim to describe the required semantics for linking wearables and sensors to sensing abilities. As presented in Table 2, few expert systems related to NDs have the ability to integrate sensor data. However, there is no expert system that integrates data related to PD, artistic behavior, and patient movement at the same time.

One positive aspect of ontologies is their ability to conceptualize historic knowledge and integrate diverse types of data, such as neuroimages or gene expression/inhibition graph networks. In recent years, there has been a rise in expert systems and decision support systems, which are AI-based software supplied with massive data linked to their expertise to “learn” the data, create patterns, and infer essential knowledge that would not be feasible in other ways, given the vast number of resources that must be constructed. Fuzzy ontologies, which can conceive the actual world with its uncertainty and produce more realistic, accurate, and specific outcomes, could be of tremendous assistance for this purpose.

In this review, the need for further research in utilizing art to predict and diagnose early-onset NDs, as well as the importance of updating existing medical ontologies, is highlighted. The development of new ontologies, specifically for artistic behavior and for monitoring patients’ movement/living, could also fill a critical gap in this field. Additionally, the development of expert systems and decision support systems utilizing fuzzy ontologies can provide more accurate outcomes to assist in monitoring and alerting ND patients.

6. Proposed Approach

The previous section highlighted the lack of an ES/DSS that uses products of artistic expression, such as paintings, combined with other personal health data to enable the diagnosis and progression of PDs. Therefore, in this section, a framework for the detection of common mechanisms between patients with PD and those who possess artistic talents is proposed (Figure 4). The proposed framework utilizes semantic models based on fuzzy ontologies with the aim of creating an integrated representation of biological and artistic knowledge. To realize this framework, an ontology-based interlinking and decision support system must be developed and evaluated, emphasizing fuzzy semantics and biomedical/bioengineering data. The use of fuzzy semantics accommodates the uncertainty and imprecision often associated with PD and artistic expression, and it can help reveal patterns and connections that may not be initially apparent. Semantics (ontologies) play a critical role in this framework, providing a structured formal and explicit representation of the data and facilitating the identification of relationships between different concepts. Moreover, ontologies are used to link various data sources, leading to a more complete understanding of the relationship between PD and artistic behaviors. The proposed system takes interconnected data and fuzzy semantics as inputs and determines (a) if PD and artistic behavior interrelations exist and (b) if preliminary PD biomarkers are present.

As previously discussed, few expert systems utilize personal health sensor-based data in order to facilitate the diagnosis and monitoring of PD. Those related works, and the expanding of the ontologies that can annotate and conceptualize sensor data, have influenced the design of the proposed framework. Taking a step further from other frameworks, an additional input for the proposed framework is the data collected through wearable devices during artistic tasks performed by the PD patients. Specifically, an application for smartwatches has already been designed and tested for the purpose of our research. The application enables the collection of movement data from ND patients during artistic activities such as drawing. The artistic outputs, drawings, paintings, etc., will be collected via scanning techniques. Therefore, multiple scanning techniques can and will be assessed for suitability with our proposed framework. Finally, PHR data will be injected. All three types of data sources will be semantically annotated using the corresponding semantics of a unified ontological model (which is under development) and RDF-encoded in a personal health knowledge graph (PHKG) for each patient. By combining biomedical/bioengineering data, sensor data (e.g., movement), artistic expression data, and fuzzy semantics, the proposed framework aims to improve the diagnosis and treatment of PDs by providing a more comprehensive understanding of their underlying mechanisms. A CDSS implemented based on the proposed framework is expected to assist in making more complex and accurate clinical decisions, guiding doctors in the diagnosis and in patients’ monitoring processes. Ultimately, the proposed framework and the CDSS implementations based on it have the potential to contribute to the development of new therapies and interventions for NDs in general.

7. Conclusions

In conclusion, the combination of existing and new technology in a novel framework that will utilize data and semantics from different sources has the potential to uncover hidden relationships between NDs and artistic behaviors. Fuzzy semantics can account for the uncertainty or imprecision often present in NDs and artistic expression data. The integration of biomedical data and ontologies provides a structured representation of the information, allowing for the identification of patterns and connections between NDs and artistic talent. This can lead to a better understanding of the underlying mechanisms connecting NDs with alterations in artistic behaviors, potentially aiding in the diagnosis and treatment of NDs. In this paper, based on the literature review and going beyond the state-of-the-art related works, a framework that can identify preliminary ND biomarkers based on artistic behavior, movement, and personal health data of PD patients is proposed. The proposed framework is expected to contribute to the understanding and treatment of NDs in general. The proposed framework is a first approach towards the development of new therapies and interventions for NDs.

Future plans include the implementation of a proof-of-concept CDSS based on the proposed framework to evaluate its value in real-world scenarios. For this purpose, we have already designed and implemented the specific components, processes, and tasks of such a system (i.e., an application for collecting patients’ movement data from a smart watch and an ontology for semantically annotating this data and the personal health data retrieved from PHRs), and we have started experimenting with others (i.e., a service for the semantic data integration and reasoning of streaming/sensor data and PHR data, running on an edge device using an Arduino kit and the RDFox high performance knowledge graph and semantic reasoning engine).