Correlations between Social Isolation and Functional Decline in Older Adults after Lower Limb Fractures Using Multimodal Sensors: A Pilot Study
Dragoş Arotǎriţei
Round 1
Reviewer 1 Report
Comments and Suggestions for Authors1. If the paper address to a journal with the subject "Algorithms", an important prat or at less a part mut be related to new proposed algorithm of something about
improvements or modification of some algorithms that improve the performance for a problem. In this paper, a section that deal with algorithms is almost missing so the authors must solve this somewhat.
2. The paper is rich in statistics, but a linear correlation of 0.52 cannot be a good performance (maybe a value closer to 1.0, it will be more convincing).
The result "it is possible" is a result, but it is as in flipping coin probability, 0.5. So, the authors must prove a higher certitude that their results are validates by repetition of experiments.
3. All the references must be cited in the paper. Missing citation are: 35, 35.
4.The paper is highly narrative, but for a paper that deal with record data from sensors there must be few figures with signals and few statistic parameters related to each signal.
The paper must be more argumentative in this direction.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsThe authors described a multimodal sensor platform to investigate the correlations between social isolation and functional decline by using clinical and sensor data from older adults after lower limb fractures. However, the authors used questionnaires to assess health status of patients, which can be subjective when the sample size is small (n=8), since the answers could vary a lot from person to person. A larger sample size is necessary to validate the platform.
In addition, the manuscript should be organized and written in a better way. For example, there are too many abbreviations which makes the manuscript hard to read.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
Comments and Suggestions for AuthorsReview for Algorithms-3117684
All the sensors used in this study collect personal/individual data. Why not use sensors that provide interaction/socialization information objectively? Communication and phone usage has been used to measure social isolation in young adults. Do you see a use for this in evaluating social loneliness/isolation in elderly?
What time of day the biweekly assessments took place? Morning vs afternoon? Should we expect a circadian influence on the outcome of physical tests or the patients’ responses to the questionnaires? For example, patients might be more sleepy and tired in early afternoons compared to mornings. Please discuss.
Why did the authors choose 8 weeks for the duration of follow up for each patient?
How long after the procedure did the enrollment start?
Did the patients have follow up doctor appointment, physical therapy, or at home nursing care during these 8 weeks and if yes what kind of influence is expected from these appointments on the patients’ responses to the questionnaires and their sense of isolation.
Three of the patients were employed. Did they work during the 8-week period of this study? Should authors have controlled for job status?
What was the psychological state of these patients, any issues with depression or anxiety disorders?
Have the author considered potential confounding factors such as patients’ psychological state, pain, medication, or local icing that could cause maybe less willingness to socialize and potentially affect patients’ physiological parameters such as heart rate.
For those patients who have pets how does the motion sensor or the bed sensor work assuming the pet sleeps in the same bed as the patient and moves freely around the house? Same question for that one patient who was married.
What is the accuracy of the Sleep Sensor in detecting sleep parameters, and sleep HR?
Roddick et al. in “Effects of Chronic and State Loneliness on Heart Rate Variability in Women” state “chronic loneliness is associated with altered parasympathetic function (both resting HRV and HRV reactivity) in women, and that the immediate experience of state loneliness is linked to a proximate increase in HRV among chronically lonely women.” Did authors consider heart rate variability in this study?
Was daytime napping a consideration? Goonawardene et al. in “Sensor-Driven Detection of Social Isolation in Community-Dwelling Elderly” states “elderly who perceived themselves as socially lonely tend to take more naps during the day time.”
For position, did it matter where the patient went, friend’s house vs a therapy/doctor visit?
When looking at correlation between sleep parameters and other variables, did the author use the sleep from the night before or the night after? I think this could be important to know as it could potentially points to the effect of sleep parameters on other variables or vice versa.
In Table 10 how do you interpret the negative correlation between TUG and wakeupcount or durationtosleep
What role does time play in the assessment of all these correlations? All the extracted features are time series, but authors have not considered time in their analysis. Is it possible that some of the reported correlations strengthen or weaken as time passes and patients recover.
American Heart Association as part of Life’s Essential 8 recommends 7-9 hours of sleep per night for adult population. Authors report a negative correlation between SIS and total sleep duration. What was the average total sleep duration for these patients? It would be helpful to know if on average the total sleep duration in this group falls outside of the 7-9h range or not, especially if it is above the maximum recommended duration of 9 hours per night. Also, is this correlation with the sleep from the night before or the night after SIS questionnaire was filled.
In section 5.3.2 part 1, why was there any missing clinical data?
In 5.3.2 part 3 & 5.3.1 part 3, Authors state ‘every two features
with an absolute correlation value of higher than 0.90, were deemed as “redundant”’. Could you please provide some examples of redundant features for each case?
In line# 437-438 authors indicate the lower physical health the greater the distance traveled. I was wondering if authors have any explanation for that.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsThe paper is improved and the paper satisfy the requirements for publication.
Reviewer 2 Report
Comments and Suggestions for AuthorsCurrent version looks better.
