Is mHealth a Useful Tool for Self-Assessment and Rehabilitation of People with Multiple Sclerosis? A Systematic Review

The development of mobile technology and mobile Internet offers new possibilities in rehabilitation and clinical assessment in a longitudinal perspective for multiple sclerosis management. However, because the mobile health applications (mHealth) have only been developed recently, the level of evidence supporting the use of mHealth in people with multiple sclerosis (pwMS) is currently unclear. Therefore, this review aims to list and describe the different mHealth available for rehabilitation and self-assessment of pwMS and to define the level of evidence supporting these interventions for functioning problems categorized within the International Classification of Functioning, Disability and Health (ICF). In total, 36 studies, performed with 22 different mHealth, were included in this review, 30 about rehabilitation and six for self-assessment, representing 3091 patients. For rehabilitation, most of the studies were focusing on cognitive function and fatigue. Concerning the efficacy, we found a small but significant effect of the use of mHealth for cognitive training (Standardized Mean Difference (SMD) = 0.28 [0.12; 0.45]) and moderate effect for fatigue (SMD = 0.61 [0.47; 0.76]). mHealth is a promising tool in pwMS but more studies are needed to validate these solutions in the other ICF categories. More replications studies are also needed as most of the mHealth have only been assessed in one single study.


Introduction
People with Multiple Sclerosis (pwMS) may manifest heterogeneous symptoms and functioning problems that require continuous and long-term rehabilitation programs in clinical and community settings across the disability spectrum. In high-income countries, the pressure on healthcare systems is increasing [1] and the continuity of high-level care is threatened due to lack of reimbursement, while in some countries access to the specialized MS centers has always been poor [2]. Furthermore, a vast majority of pwMS often present fatigue, emotional or cognitive dysfunction, or restricted physical mobility or a combination of those which limits access to rehabilitation centers. In this context, the WHO stated that lack of access to specialized centers or healthcare professionals is one of the most important limitations for the rehabilitation process [3]. The use of mobile technologies and electronic health (eHealth) could be an alternative to tackle the above-mentioned limitations (i.e., lack of access to centers) of rehabilitation of pwMS or complement current rehabilitation services. eHealth is also expected to facilitate the monitoring of functioning of pwMS between medical consultations, which is informative to define whether to continue or adapt medical treatment.

•
Population: pwMS performing training (rehabilitation exercises) or self-assessment in home-environment, studies with inpatient treatment or assisted-rehabilitation were not included. • Intervention: mHealth rehabilitation intervention (planned and supervised interventions), or self-assessment studies with repeated measurements over time, using any type of support (e.g., smartphones, phones, apps, web applications). Studies using non-specific games, virtual reality or active video games (e.g., Nintendo Wii, Microsoft Xbox Kinect), or computer-supported therapy were not included. A flow diagram of the study selection with the screened articles and the selection process is presented in Figure 1.

Quality Assessment
Since we included different types of articles, the critical appraisal of the methodological quality was based on the Downs and Black checklist [17], as this checklist is the best option to assess the quality and risk of bias for both RCT and non-RCT [18].

Data Extraction
The following information was extracted from the included studies: characteristics of the patients (age, sex ratio, type of MS and severity), type and duration of the mHealth intervention, study design, main outcomes, and ICF domains evaluated.

Statistical Analysis
For studies assessing the efficacy of a rehabilitation program, we performed a metaanalysis. The measure of treatment effect was the standardized mean difference effect size (standardized mean difference (SMD)), defined as the between-group difference in mean values divided by the pooled SD computed using the Hedge's g method (Hedges g = M 1 −M 2

SD pooled
). If different tests were used to assess the same ICF domains in the same study, the different results were pooled to have one unique SMD as recommended by Cochrane's group [19]. A positive SMD implies better therapeutic effects in the intervention group compared to the control. We assessed the heterogeneity in stratified analyses by type of ICF domains. We calculated the variance estimate tau 2 as a measure of between-trial heterogeneity. We prespecified a tau 2 of 0.0 to represent no heterogeneity, 0.0-0.2 to represent low heterogeneity, 0.2-0.4 to represent moderate heterogeneity, and above 0.4 to represent high heterogeneity between trials [20]. To deal with high or moderate heterogeneity we used random-effect models and presented forest plots for the different ICF domains. We checked for publication bias using funnel plot [21] and Egger's test for the intercept was applied to check the asymmetry [22].

Ethical Approval
This systematic review was reported following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) recommendations [23]. For the present study, no ethics committee approval was necessary.

Results
For the sake of clarity, this section has been divided into three different parts; first, we will present the characteristics of the included studies and the patients; then we will describe the different mHealth used in these studies and finally, we will present the clinical efficacy for the different domains in the ICF.

Search Results
In total, 1346 articles were found with the systematic review. A total of 112 full-text articles were assessed and 36 papers were included in the analysis. The PRISMA flowchart on the study selection is presented in Figure 1.

Characteristics of the Included Studies
Thirty studies about the use of mHealth for rehabilitation interventions of pwMS were included in this review, representing a total of 1962 patients . The majority of these studies (n = 25; 3%) were RCTs. Concerning the patients, the majority of the patients were female (76 ± 10%); concerning the type of MS the majority of the included patients (79%) have relapsing-remitting multiple sclerosis (RRMS), 16% have secondary progressive multiple sclerosis (SPMS) and 5% primary progressive multiple sclerosis (PPMS), and the average EDSS is 3.5 ± 1.1. The median duration of the intervention was 9 weeks [p25 = 8 weeks, p75 = 12 weeks] for a median time of 18 h [p25 = 13.25 h, p75 = 27 h]. Finally, for the ICF, 16 (53%) of the studies reported outcomes related to cognition, 11 (37%) to fatigue, 10 (33%) to quality of life, 7 (23%) on motor function, and 6 (20%) on activity level; we observed that most of the studies are assessing different primary outcomes (ICF domains). The complete description of the included studies is presented in Table 1. Amongst the 30 studies, 16 different mHealth apps have been tested.
The characteristics of the studies and participants are summarized in Table 2.

Quality Assessment
The quality of the included papers was checked using the Downs and Black checklist. Overall, the average score for the included studies is 21.9 out of 28 (22.2 for studies on rehabilitation, 20.3 for studies assessing self-assessment). The average results for the different questions and sub-analysis of the Downs and Black checklist are presented in Figure 2. When analyzing individual items, we observed that, due to the nature of the training, the blinding of the participants was not possible, on the other hand, the blinding of the investigators was not guaranteed either. Another potential source of bias is the uncertainty about the randomization assignment until the complete recruitment. The last important point is that a high number of studies do not take into consideration the patients that did not complete the intervention (loss in follow-up) so leading to uncertainty on reasons of non-adherence. Only a few studies used intention-to-treat analysis. On average 90.6% of the included patients completed the entire protocol and were included in the final analysis.

Description of the Available mHealth Solutions
First, we present the mHealth solutions that are mainly used for rehabilitation purposes. Most of the proposed mHealth solutions have been studied for cognition, QoL, and fatigue and were limited to one single ICF domain. We later discuss the applications for the respective domains, although some overlap occurred.
RehaCom [24,30,33,48] is a comprehensive and sophisticated system of software for computer-assisted cognitive rehabilitation. It proposes different solutions for screening and training cognitive functions and offers apps for home training.
BrainHQ [32,41,42] is a platform providing a set of more than 30 mini brain training exercises designed to challenge different cognitive functions (processing speed, attention, working memory, and executive function through visual and auditory domains). The initial level of challenge is low and the difficulty is adapted on an individual basis as learning and abilities improve over time. The company was previously known as Posit Sciences [27].
Luminosity [25] is a platform providing cognitive training exercises embedded in games. As for BrainHQ different cognitive functions can be challenged in a set of different mini-games.
The Memory, Attention, and Problem Solving Skills for Persons with Multiple Sclerosis (MAPSS-MS) intervention [35] aims to help pwMS acquire the highest level of cognitive functioning and functional independence. It includes problems solving and lifestyle adjustments (sleep, stress management, physical activity) that support cognitive functioning and will support persons with MS in the use of compensatory cognitive strategies and cognitive skills. The cognitive training is done with Luminosity app.
BrainStim [28] is a computerized training tool based on the working memory (WM) model of Baddeley [60]. It consists of three different modules targeting both, verbal and visual-spatial aspects of WM.
COGNI-TRAcK [31] implements three different types of exercises which were shown to be effective in improving the cognitive status of healthy subjects. The exercises consisted of (i) a visuospatial WM task; (ii) an "operation" N-back task; (iii) a "dual" N-back task [61].
The Attention Processing training (APT) program [25] is a cognitive rehabilitation intervention that targets focused, sustained, selective, alternating, and divided attention. The aim is to increase the ability to respond to specific stimuli [62].
ELEVEDIA [37] content is based on cognitive behavioral therapy strategies and is conveyed chiefly via the technique of a 'simulated dialogue'. Program modules are composed of an introduction and a summary and include homework tasks. Patients are advised to access the program once to twice per week.
The MS Home Automated Telehealth (MS HAT) system [34,43] is supporting patientcentered care, self-management and allows easier patient-provider communication. Three interfaces are available: patient unit, server, and clinical unit. The patient unit had interactive options for data collection, educational content, exercise information, and therapistpatient communication, access to exercises, response to exercise-specific assessments, and documenting exercise data from home. Exercises consist of task-oriented training such as digitized writing tracking or manipulating light bulbs or keys. Exercise adherence feedback was via diary entries, calendars, and graphs [63].
AKL-TO3 [51] is engaging the patients in simultaneous sensory and motor tasks and is designed to engage the frontal neural network. It enabled real-time monitoring of progress and continuously challenges patients so that the training is never too easy or difficult encouraging patients to improve performance.
RELAXaHEAD [49] is designed for pain management and in particular migraine and neck pain. It contained a headache diary, which includes features for tracking headache characteristics, headache medications and sleep, and tracking medication side effects and menstrual cycles. The app also contains a serious game module to ease muscle relaxation.
WalkWithMe [50] has been developed to motivate and stimulate patients to walk more and farther. It allows to track the walking activities and follow up on progress. The app detects the walking speed and gives feedback during walking with verbal feedback (i.e., pace) by the virtual coach.
Webbasedphysio.com [47] is an internet-delivered therapeutic exercise program. The web-based physio allows people the flexibility to do their own, individualized exercise program at a time and location which is convenient to them, thus enhancing the individual's ability to self-manage their condition on a long-term basis.
Deprexis [29] is an online program based on principles of cognitive-behavioral therapy. It consists of 10 sequential modules-psycho-education, behavioral activation, cognitive modification, mindfulness and acceptance, interpersonal skills, relaxation, physical exercise and lifestyle modification, problem-solving, expressive writing and forgiveness, positive psychology, and emotion-focus interventions.
The Mindfulness Based Stress Reduction (MBSR) [38] deals with stress management, relaxation training, sleep hygiene, fatigue, and social relationships. The course materials were developed using existing informative MS videos, created by the Italian MS Association, recording new interviews and generating new exercises.
Concerning the mHealth apps that are mainly used for self-assessment: MSdialog [55] is a web and mobile (i.e., cell phone and tablet) based software application that combines information from RebiSmart (with health information recorded by patients via their personal computer or smartphone to collect and store real-time data regarding administration of Rebif (interferon β-1a), clinical outcomes, and patient reported outcomes). MSdialog offers a practical means by which patients record and exchange information with their healthcare specialists intending to support the patient-physician relationship and offering patients a method of engaging in the pharmaceutical management of their MS and patients' self-reported outcomes [64].
MS Telecoach [56] provides a combination of monitoring, self-management, and motivational messages, focusing on energy management of physical activity to improve fatigue levels. It has two components: telemonitoring (physical activity through accelerometers and self-reported fatigue impact levels) and tele-coaching (motivational messages and advice).
Floodlight [57] is a combination of continuous sensor data capture and standard clinical outcome measures. It involves performing a set of daily active tests to evaluate cognition, upper extremity function, gait, and balance domains and contribute sensor data via passive monitoring, also including self-reported patient outcomes.
The Mellen Center Care Online (MCCO-enhanced) [54] is an electronic messaging system between clinician and patient. It contains a self-monitoring and self-management system to assess MS symptoms and the pwMS receives graphical feedback to evaluate symptom changes FatigueApp.com [58] is collecting data to correlate fatigue measures with other symptoms and quality of life. Fatigue questionnaires are completed every morning for 6 consecutive days and again 4 weeks later.
ElevateMS [59] allows collecting different data in the real-world environment of the patients such as self-reported measures of symptoms and health via 'check-in'-surveys Independent assessments of motor function occur via sensor-based active functional tests, participants are encouraged to complete surveys daily, and notifications to perform more comprehensive functional tests are provided once a week.
The overall effect of mHealth intervention in pwMS is moderate (SMD = 0.50 [0.35; 0.66]) and statistically significant (p < 0.0001). Since different studies evaluated human functioning at different aspects according to the ICF, we then performed subgroup analysis to assess the efficacy across the different ICF. The forest plot is presented in Figure 3. Note that for activity level these results must be interpreted carefully due to the small number of included studies (n = 3). For the domains of motor function and quality of life the results were not significant but only included two and three studies, respectively. Using a fixed-effect model to summarize the overall ICF functioning we found an overall moderate effect (SMD 0.47 [0.37; 0.56], p < 0.0001).
We then summarized the main results and conclusions of the studies that were not included in the meta-analysis.
Concerning cognitive function, Fuchs et al., 2019 investigated the clinical characteristics predicting response to a home-based restorative cognitive training. Significant improvements were observed after training [41]. Villou et al., 2020 is an explorative study that reported statistical improvement of various cognitive functions after training such as visuospatial memory, visual attention, task-switching, reading speed and response inhibition, and verbal learning [42].
For fatigue, Stuifenbergen et al., 2018 analyzed the acceptability and effect of MAPSS-MS on cognitive function and fatigue. The authors find similar results as with usual care; interestingly, the improvements were maintained during the follow-up at 3 and 6 months and were superior in the intervention group [35].
For the quality of life, Cavalera et al., 2018 showed an improvement of QoL after 8 weeks of intervention using a mindfulness program but the progress was not maintained over time (6-month follow-up after the end of the intervention) [38]. Tarakci et al., 2021 compared an in-person rehabilitation program with a telerehabilitation program. After 12 weeks of training, the results were similar in the two groups for fatigue and activity level [52]. Manns et al., 2020 demonstrated a reduction of fatigue after a combined intervention (SitLess and MoveMore) but the difference was not significant compared to usual care [46]. Interestingly the total sedentary time decreased in the intervention group and these results are maintained over time.
Van Geel et al. reported that using the WalkWithMe app induced a significant improvement in quality of life, walking, and leisure, 36-Item Short-Form Health Survey (SF-36) quality of life, cognition, cognitive fatigability, lower limb strength, and dominant hand function. However, it was an observational study without a control group [50].

Self-Assessment
Concerning the efficacy of self-assessment and monitoring, only six studies were included in this review.
Miller et al. highlighted group differences between the MCCO-original and MCCOenhanced groups. MCCO-original had a higher European Quality of Life level after 12 months of regularly self-monitoring their quality of life [54].
Greiner et al. performed a 6-week longitudinal observation and showed that MSdialog was adapted to monitor patient-reported outcomes. Amongst the different functions evaluated by the pwMS, fatigue (99%), physical health (96%), cognitive deficits (93%), pain (91%) and sleep quality (91%) were the most important. These numbers represent the weight given by the patients for these different functions that scored the MS quality-of-life questionnaire using a visual analogical scale [55].
D'Hooghe et al. showed that it is feasible to use the MS TeleCoach at home without supervision. The authors observed a significant decrease in fatigue and an increase in cognitive function after 12 weeks of use [56].
Midaglia et al. assessed the usability and acceptability of the Floodlight for active monitoring and passive monitoring intervention. After 24 weeks of intervention, mHealth had an acceptable impact on daily activities including cognition and physical activity for 80% of the pwMS [57].
Newland et al. reported that the FatigueApp could collect self-reported symptoms including fatigue, self-reported EDSS (EDSS-SR), pain, and cognition [58]. Participants were asked to complete the questionnaires for 7 consecutive days and then again 4 weeks later.
Pratap et al. in a large study including more than 500 pwMS described that ElevateMS can be used to longitudinally (12 weeks period) to collect information about the most common symptoms of MS. During this follow-up, they observed that the most frequent complaints are fatigue (63%), memory issues (42%) and difficulty with walking (41%). After the intervention, there were significantly increased functional performances and QoL [59].

Summary
To summarize the findings of this study we listed the different mHealth according to the targeted ICF domain for both rehabilitation and self-assessment in Table 3. Table 3. Overview of the different mHealth solutions for rehabilitation and self-assessment according to the mean ICF targeted.

Discussion
The main result of this review is the high number of solutions (applications) currently being tested with pwMS for rehabilitation (n = 16), despite the relatively recent development and use of these new apps in rehabilitation. On another side, the development of mHealth for self-assessment and home-monitoring is still limited (six apps found). Consequently, one of the downsides is that there are only very few studies performed with the same mHealth which makes it more difficult to compare the studies and thus to determine the level of evidence. Therefore, rather than comparing the efficacy of each particular mHealth, we performed the analysis at the ICF domain level. The most studied ICF domain is cognition: we found a small but significant effect of the training using mHealth (SMD = 0. 28 [68]. For the motor function and quality of life, there are, currently, not enough studies available, but the few studies available also seem to indicate a favorable effect. The paucity of studies investigating the effects of mHealth applications to train motor functions is somehow surprising. However, we excluded studies including wearables and thus the number of interventions done to increase physical activity based on step count (i.e., [69]). The low numbers of studies investigating the effects of mHealth interventions on quality of life may be expected as the quality of life is often thought to be the result of improving specific ICF domains.
Another major finding of this systematic review regarding self-assessment is the fact that mHealth can be used directly by the patients to continuously monitor several different functions in their living environment. The solutions are not only well accepted by the patients, but several studies also show that using this type of mHealth is directly beneficial for the patients. This positive effect may be mediated by a better knowledge of the diseases and symptoms (education) [70] but also by the more active participation of the patient in his treatment (patients' engagement) [71].
There are several limitations to this review. The first one is the lack of standardization in the nomenclature used to describe the different mHealth currently tested in research. Therefore, due to the small numbers of studies published, we ended up including studies assessing different types of applications and intervention modalities or duration. The heterogeneity between the studies, and the patients, makes it more difficult to compare studies and especially to generalize the results. There is also a huge heterogeneity in the duration of the intervention for both the duration of the intervention (ranging from minimum 4 to maximal 26 weeks) and the total amount of training (ranging from 4 to 65 h). Unfortunately, due to the small number of studies included in the different ICF levels, we could not perform meta-regression to determine if there is a dose-response relationship between the amount of training using the mHealth and the clinical outcome.
A third important limitation is that most of the included studies on the rehabilitation aspects (except [32,35,38,40,45]) have relatively small sample sizes and the results are likely to be underpowered [72]. Furthermore, the percentage of participants that were included in the final analysis is 90% and information about the adherence to the intervention was missing (usually a threshold of 80% is applied to determine if the participants do a sufficient amount of exercises [73]). Concerning the meta-analysis, due to the relatively small number of included studies, the results must also be interpreted carefully, especially for the ICF motor function and quality of life. Concerning motor functions, most of the current solutions are focusing on walking while patients may also experience severe disability in upper limbs functions and dexterity, efforts must be made to develop solutions that focus on these problems. Concerning the external validity of this review and the translation to the clinic, it is important to note that the vast majority of the applications were tested in pwMS with mild disability (EDSS = 3.5 ± 1.1) with RRMS (79%), and thus not guaranteed to be applicable to the same extent in more disabled patients with restricted mobility. Further studies must therefore focus on more disabled patients to determine the feasibility of mHealth with these patients if the efficacy is similar.
Finally, most of presented solutions are still at the research project stage and applications are not yet widely available to patients or their treating clinicians.
Despite these limitations, this review highlights interesting and promising results for patients. However, there are still a few points that should be addressed before these solutions can be used in daily practice. The first, and probably most important is the recognition of the m-and eHealth apps as medical devices. In June 2020, the US Food and Drug Administration (FDA) permitted the marketing of the first game-based digital therapeutic device to improve attention function in children with attention deficit hyperactivity disorder (ADHD). The mHeath, EndeavorRx, is indicated to improve attention function as measured by computer-based testing and is the first digital therapeutic intended to improve symptoms associated with ADHD, as well as the first game-based therapeutic granted marketing authorization by the FDA for any type of condition. The device is intended for use as part of a therapeutic program that may include clinician-directed therapy, medication, and/or educational programs, which further address symptoms of the disorder [74]. Interestingly this solution is developed by Akili, the company that has developed AKL-T03 which also shows significant results in pwMS [51]. The COVID-19 pandemic has not only disrupted healthcare systems but has also allowed for a very significant acceleration in the development, implementation, and recognition of mHealth in the clinics [75]. It is important to note, however, that most of the measures taken during the crisis may be temporary and it is hoped that efforts will continue in this direction once the crisis is over. For example, it will be important to adapt the nomenclature of interventions, as mobile solutions are currently placed in the same categories as drugs, which poses problems for the validation and reimbursement of these interventions [76]. Another limitation is that, for the moment, the majority of the analyzed mHealth is being developed during research projects and is therefore not easily accessible for patients, except for BrainHQ and Luminosity that are two commercial (gaming) companies. As an indication, the price of an annual subscription to these companies is less than USD 100 per year for a full premium account. RehaCom is also already widely used by clinical centers but mostly for research purposes.
This brings us to the second biggest current limitation which is the lack of reimbursement by the social security system. The organization and involvement of healthcare systems in the revalidation process is country-specific and we will not discuss reimbursement specifically here. However, we know that two of the most important barriers to the implementation of telemedicine and telehealth for the patients, regardless of the pathologies or the specialties, are the financial issues and the lack of knowledge and familiarity with the use of (new) technology [77,78]. The pwMS being relatively young, most of them are familiar with smartphones, apps, and mobile technology, therefore the familiarity with the technology should not be an issue for most of the patients [79], but this can be a real barrier for other diseases or patient groups (e.g., older adults with dementia) [80]. Efforts must also be directed to the education of healthcare professionals as they need to be perfectly aware of the technology and its limitations to motivate the patients to use it.
As a result of all the above limitations, in practice, the solutions described in this article are only used by a small fraction of the pwMS. A recent survey performed in the US found that only 3.1% of the pwMS who took part in the survey (n = 786) are using mHealth solutions regularly [81].
We will now discuss some ideas for consideration to facilitate the implementation of these solutions in the rehabilitation process.
The first point would be to integrate the mHealth solutions into the healthcare system, with reimbursement for the patients, providing education of the mHealth solutions for healthcare professionals, and the integration of the data collected with the apps (see [54][55][56][57][58][59]) in patients' medical records. This could speed up and ease the implementation of mHealth in the daily management and rehabilitation of pwMS. This would not only save time but also allow for a more accurate and regular assessment of patients [65]. Furthermore, these assessments could be carried out in the patients' homes. This fits in perfectly with telemonitoring [82] and the use of real-world data [83]. This would therefore allow the development (by increasing the number of potential users, companies may be more inclined to invest in such solutions) and wider use of these solutions.
A last important point is the sustainability of the studied solutions [84]. The speed of the development of mobile technology (hardware and software) is one of the most important considerations, and the technology becomes quickly obsolete (for example there is a new version of the operating systems [Andoid© or iOS©] on average every 6 months). Thus, the apps that have been developed with the previous version are not supported anymore with the more recent one. This is not much of an issue with the commercial solutions, but it is more problematic with the solutions being developed during research projects.

Conclusions
This review highlights an important potential of mHealth for pwMS with evidence of a small but significant effect on fatigue and cognition. Although we have seen that current mHealth is, at the moment, not a perfect solution, given the high prevalence of fatigue and cognitive impairment in pwMS and the lack of low-cost tools to assist and stimulate the patients at home, the use of apps could be greatly beneficial.
To develop innovative, effective solutions adapted for pwMS whose cognition, quality of life, functionality, and wellbeing are impaired, researchers, clinicians, policy makers, and app developers will need to further collaborate.

Supplementary Materials:
The following are available online at https://www.mdpi.com/article/ 10.3390/brainsci11091187/s1, Table S1: Search strategy for PubMed Figure S1: Funnel plot of the included studies in the meta-analysis; Figure S2: Sensitivity analysis of the included studies in the meta-analysis.