Systematic Review and Scientific Rating of Commercial Apps Available in India 2 for Diabetes Prevention

medRxiv preprint doi: https://doi.org/10.1101/2021.02.15.21251723; this version posted February 18, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

1 Systematic Review and Scientific Rating of Commercial Apps Available in India 2 for Diabetes Prevention

3 H Ranjani1, S Nitika1, R Hariharan1, H Charumeena1, N Oliver2, R Pradeepa1, J C 4 Chambers3,4, R Unnikrishnan1, V Mohan1, P Avari 2, RM Anjana1

5 1Dr. Mohan’s Diabetes Specialities Centre and Madras Diabetes Research Foundation, 6 Chennai

7 2Department of Metabolism, Diabetes and Reproduction, Imperial College London, 8 London, UK

9 3Department of Epidemiology and Biostatistics, Imperial College London, London W2 10 1PG, UK

11 4Lee Kong Chian School of Medicine, Nanyang Technological University, Singapore 12 308232, Singapore

13 Key words: Systematic review; mhealth; apps; quality; scientific rating; diabetes prevention 14 15 Word Count: Abstract –283 Manuscript –3254 16 17 ADDRESS FOR CORRESPONDENCE: 18 RANJANI HARISH, CDE, PhD 19 Sr. Scientist and Head- Translational Research 20 Madras Diabetes Research Foundation and Dr.Mohan's Diabetes Specialities Centre 21 WHO Collaborating Centre for Non-Communicable Diseases Prevention and Control 22 IDF Centre for Education 23 No:6B, Conran Smith Road, Gopalapuram 24 Chennai, India Pin:600086 25 Ph:91-44-43968888, Fax:91-44-28350935 26 Email : [email protected] 27 Website: www.drmohansdiabetes.com, www.mdrf.in

28 *The abstract of this manuscript has been presented as part of a poster discussion at 29 the Advanced Technologies and Treatment for Diabetes (ATTD), 2020, which was held 30 from 19th – 22nd February 2020 in Madrid, Spain.

NOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice. medRxiv preprint doi: https://doi.org/10.1101/2021.02.15.21251723; this version posted February 18, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. All rights reserved. No reuse allowed without permission.

33 ABSTRACT

34 BACKGROUND

35 Scientific evidence for digital health applications (apps) which claim to help in the 36 prevention and management of Type 2 diabetes (T2D) is limited.

37 OBJECTIVES

38 We aimed to evaluate the quality of currently available health apps for prevention of T2D 39 amongst Asian Indians.

40 METHODS

41 Using the keywords, ‘diabetes prevention’, ‘healthy lifestyle’ and ‘fitness’, a total of 1486 42 apps available in India via Google Play were assessed for eligibility by two independent 43 reviewers. After initial screening using specific inclusion and exclusion criteria, 50 apps 44 underwent a pre-specified rating based on user reviews, number of downloads and app 45 size. Sixteen apps that scored ≥ 9 were shortlisted for further review using the Mobile App 46 Rating Scale (MARS). MARS contains 2 main categories: Category I (Application Quality) 47 and Category II (Application Subjective Quality). The mean MARS scores were used to 48 identify the top ranked apps.

49 RESULTS

50 The mean score for Category I of MARS rating was highest for ‘Google Fit: Health and 51 Activity Tracking’ (4.55/5). This was followed by ‘Healthifyme - Diet Plan, Health and 52 Weight Loss’ (4.45/5). For Category II of MARS, ‘Diabetes M’, ‘Google Fit: Health and 53 Activity Tracking’ and Calorie Counter- My fitness pal’, ‘Healthifyme - Diet plan, Health 54 and Weight Loss’ all scored equally well. On comparing the advantages and 55 disadvantages of each of these applications, ‘Google Fit: Health and Activity Tracking’ 56 and ‘Healthifyme - Diet plan, Health and Weight Loss’ again ranked the best.

59 CONCLUSIONS

60 Our review identifies two commercially available apps ‘Google Fit: Health and Activity 61 Tracking’ and ‘Healthifyme - Diet plan, Health and Weight Loss’ as being user friendly 62 and good quality. Although encouraging, further research is needed to evaluate the 63 efficacy of these apps for prevention of diabetes.

81 Introduction

82 Mobile technology has changed the life of millions of people around the world [1]. Smart 83 phones provide an ideal combination of user-accessibility with high functionality of 84 customized content. These have already achieved high penetration across Asia. India 85 accounts for 20% of the global smart phone market with approximately 370 million users 86 [2]. It is estimated that the number of smartphone users in India will double to reach 87 between 650 million and 700 million by 2023 [2] rendering mobile health (mHealth) an 88 indispensable component of health care [3].

90 The term mHealth refers to clinical and public health activities, made available through 91 smartphone devices, which offer health related information and services to people 92 anywhere, anytime [4]. mHealth encourages users to be part of their own health 93 management plan, particularly when it relates to prevention and/or self-management for 94 chronic conditions [5]. The increase in usage of smartphones has led to the rapid 95 development of many mHealth applications. The number of mHealth apps available at 96 Google Play in the first quarter of 2020 was 43285, compared to 23955 in the first quarter 97 of 2015 [6]. These mHealth apps allow people to be connected with their health care 98 providers like never before [7]. International guidelines support the use of mHealth apps 99 for prevention of type 2 diabetes (T2D) and cardiovascular disease (CVD) in high-risk 100 individuals [8 ,9]. However, little information on the quality of apps is available, and merely 101 selecting health apps on the basis of popularity does not provide any meaningful 102 information on app quality [10,11].

103

104 Hence, in this study we aimed to collate data on currently available commercial health 105 apps and evaluate their potential quality as tools for prevention of T2D, with particular 106 reference to Asian Indians.

107

108

109 Methodology

110 The systematic review process comprised of three stages:

111 Stage 1: Searching the Android platform 112 Two independent primary reviewers searched for health applications commercially 113 available in India on the android platform (Google Play store) on 3rd June 2020 using 114 three key phrases: ‘diabetes prevention’, ‘healthy lifestyle’, and ‘fitness’. Other key words 115 such as “exercise” and “physical activity” were not included, as these searches yielded 116 apps focusing on exercise routines, strategies related to athletic training and had no link 117 to diabetes prevention. Using the key word “diabetes” alone, yielded apps which helped 118 to diagnose and/or manage diabetes. Thus, we focused primarily on the key words 119 ‘diabetes prevention, healthy lifestyle and fitness’ as they yielded the apps that were 120 relevant to diabetes prevention. In total, 1486 apps in the Google Play store matched with 121 the keywords; 500 apps using the keyword “diabetes prevention”, 498 apps for “healthy 122 lifestyle” and 488 apps for “fitness” (Figure 1). These were assessed for eligibility by the 123 primary reviewers, with each reviewer assessing approximately 750 applications. 124 125 Inclusion criteria were an average user rating ≥ 3, more than 10,000 downloads, minimum 126 50 user reviews, app updated in the last 6 months and available in the English language. 127 As we were looking for apps which could be used across the country, we preferred to 128 select apps available in English which is the link language across urban India. Exclusion 129 criteria were medical apps providing diagnostic or clinician-led healthcare (for example, 130 ‘DIABNEXT- Make your diabetes management easy’ and ‘Diabetes Diagnostics’), apps 131 with consultation provided (‘Gadge Diabetes Care’), pharmacy apps, non-Asian food 132 charts (apps that focused on American/ western diets and not relevant to Indian cultural 133 preferences or habits), apps with quotes on lifestyle (‘Best Life tips’), apps on 134 unconventional diets for weight loss (‘Total Keto Diet’), apps promoting a specific 135 organization or products (‘Medworks Management of Diabetes’, ‘Lefum Health’, ‘Lenovo 136 life’) and promotional apps (e.g.: Fitness bands, Fitness centres).Premium features 137 requiring payment from the user were excluded, however apps providing a basic version 138 without additional cost were included. Based on these exclusion criteria, reviewer 1 and

139 reviewer 2 excluded 680 and 693 apps respectively. Thus, reviewer 1 shortlisted 50 apps 140 and reviewer 2, 47 apps (Figure 1). 141 142 Subsequently, detailed reviews posted by users were explored to study the app features 143 and specifications. A rating scale (Appendix 1) was applied based on the following 4 144 parameters that reflect their availability in the Play store: - a) overall rating of the app, b) 145 number of downloads, c) size of the app and d) number of reviews. Each primary reviewer 146 rated their list of apps on a scale of 1 to 4 [1 being the lowest and 4 being the highest, 147 total rating out of 16 (4*4)]. There were16 apps that received a total rating score ≥9 from 148 both reviewers, and these were shortlisted for further evaluation (Appendix 2). 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168

169 170 Figure 1: Flow chart to depict the systematic search of the apps on an android 171 platform 172

173 Stage 2: Call for commercial apps 174 Developers of commercial health apps (apps which already have been published in the 175 Google Play store) were invited to participate in our Digital Health Intervention study. This 176 digital intervention study is the next stage of the project, wherein the two commercial 177 mHealth apps selected through this systematic review process and two in-house apps 178 will be tested for usability, preferences, engagement, and weight loss in 1000 participants. 179 This was done “simultaneously” along with stage 1. There could have been relevant apps 180 which were missed in our search due to the descriptions given in the play store. Hence, 181 an official invite to participate was sent to the developers. An advertisement was 182 published on our official website and also on social platforms, including Facebook, 183 Instagram and LinkedIn. We received 106 responses but only one response lead was 184 relevant to our study (‘Fittr’). To increase the number of responses, we did another 185 process of Search Engine Optimization. This process added 20 more apps, out of which 186 11 had already been shortlisted by the two primary reviewers in their initial search. The 187 remaining 9 apps were reviewed and rated (as performed in Stage 1) but none scored 188 above 9. 189 190 A secondary reviewer reviewed the apps shortlisted by the two primary reviewers to verify 191 and solve any disputes regarding the final 16 apps selected from stages 1 and 2. Stages 192 1 and 2 were performed simultaneously. 193 194 Stage 3: Applying MARS 195 The Mobile App Rating scale (MARS) is a medical app quality rating tool that provides a 196 multidimensional assessment of app quality indicators like engagement, functionality, 197 aesthetics, and information quality, as well as app subjective quality (Figure 2) [11,12]. 198 MARS contains 2 main categories. The developers of this scale have not divided MARS 199 into categories but have mentioned divisions such as ‘Application Quality’ (Sections A to 200 D) and ‘Application Subjective Quality’ (Section E). For ease of understanding, we have 201 renamed ‘Application Quality’ as Category I and ‘Application Subjective Quality as 202 Category II’. Category I assesses quality across various dimensions. It contains 4 203 sections (A-D) and all items are rated on a 5-point scale from “1. Inadequate” to “5.

204 Excellent”. A mean score is given at the end of each section (A-D). These mean scores 205 are averaged to get an overall mean score for these categories (out of 5) as the app 206 quality mean score. 207 208 Category II is the Application Subjective Quality that has questions (Section E) on 209 recommendation, frequency of use, willingness to buy the app and the overall rating given 210 for the app from the user point of view. The apps are scored and presented separately 211 category wise.

212 213 Figure 2: Categories of MARS

214 The two primary reviewers independently downloaded the 15 shortlisted apps (1 app 215 could not be downloaded due to technical issues) on their android phones and used them 216 extensively to conduct in-depth evaluations of each app before rating them using MARS. 217 The reviewers were trained to use the MARS instrument by watching an online tutorial 218 [13], to ensure rating consistency amongst the reviewers. The mean MARS scores were 219 used to identify the best quality apps. Kappa statistic was used to estimate inter-reviewer 220 (rater) reliability between the two primary reviewers [14]. The reliability estimate was 0.77 221 (Cohen Kappa calculated using the SPSS version 15). As the domains in this scale are 222 subjective in nature, this reliability estimate was considered to be very good.

223 Results 224 The 15 apps which were downloaded are listed below: 225 1. Google Fit: Health and Activity Tracking 226 2. Healthifyme - Diet Plan, Health and Weight Loss 227 3. Calorie Counter - My fitness pal 228 4. Habits Diabetes coach 229 5. Diabetic Recipes: Healthy Food 230 6. Calorie Counter - MyNetDiary, Food diary tracker 231 7. 8fit Workouts and Meal Planner 232 8. Beat O SMART Diabetes Management 233 9. Diabetic Diet Recipes : Control Diabetes and Sugar 234 10. Diabetes:M 235 11. Noom: Health and Weight 236 12. mySugr-Diabetes App and Blood Sugar Tracker 237 13. Diabetes Forum 238 14. Diabetes Diary - Blood Glucose Tracker 239 15. Beat Diabetes 240 241 Figure 3 shows the results of MARS for App Quality (sections A-D). In section A, 242 Engagement, ‘Google Fit: Health and Activity Tracking’ had the highest engagement 243 mean score of 4.50/5 for including the notification features for activity tips, goal progress 244 tips, goals adjustment and completed goals. Meanwhile, ‘Beat Diabetes’ had the lowest 245 Engagement mean score of 2.30/5. This app had limited information and was less 246 interactive. 247 248 In section B; Functionality, ‘Diabetes Diary: Blood Glucose Tracker’ scored the highest 249 on functionality, with a mean score of 5.00/5. ‘mySugr: Diabetes App and Blood Sugar 250 Tracking’ and ‘Diabetes Forum’ had the lowest Functionality mean score of 3.90/5. 251 ‘mySugr: Diabetes App and Blood Sugar Tracking’ required an external glucometer to be 252 connected and ‘Diabetes Forum’ had glitches in the registration and user log in 253 processes.

254 For section C; Aesthetics, the mean score for all the apps was above 4.00. ‘Calorie 255 Counter: MyNetDiary, Food diary tracker’ had the highest score of 5.00/5. The app had a 256 clear and orderly organized layout with high quality graphics and visual design. The colour 257 scheme further enhanced the app features. On the other hand, ‘Beat Diabetes’ had the 258 lowest Aesthetics mean score of 3.65/5. 259 260 In section D; Information, ‘Healthifyme’ and ‘Google Fit: Health and Activity Tracking’ had 261 the highest score of 4.40/5. Both the apps provided a thorough description of the app 262 components and had specific and achievable goals. The quality of information was well 263 written and relevant to the goals of the app. Visual information was clear and logical. 264 ‘Google Fit: Health and Activity Tracking’ App also was evidence-based [15]. The 265 ‘Diabetes Diary: Blood Glucose Tracker’ app had the lowest Information mean score of 266 3.30/5. This app did not appear to have specific, measurable goals.

267 268 Figure 3: Mean score of MARS Sections A- App Engagement, B- Functionality, C- 269 Aesthetics and D- Information (all 15 apps) 270 The mean MARS scores for each of the 15 apps downloaded by the two primary 271 reviewers are shown in Table 1. According to the mean scores of both reviewers 272 for Category I, ‘Google Fit: Health and Activity Tracking’ was ranked at number 1 with the 273 average app quality mean score of 4.55. This was followed by ‘Healthifyme - Diet Plan, 274 Health and Weight Loss’ with a mean score of 4.45. Two apps, namely ‘Calorie Counter

275 - My fitness pal’ and ‘Habits Diabetes coach’ were tied at number 3 (mean score 4.30). 276 These were followed by ‘Calorie Counter - MyNetDiary, Food Diary Tracker’ and ‘Diabetic 277 Recipes: Healthy Food’ (mean score 4.25) at number 4, and ‘Diabetic Diet Recipes: 278 Control Diabetes and Sugar’, ‘Beat O SMART Diabetes Management’ and ‘8fit Workouts 279 and Meal Planner’ (mean score 4.20) at number 5. 280 Table 1: Overall mean MARS scores under each category for the 15 apps

Category II - App Subjective Quality Mean Category I - App Quality Mean Score Score Mean Mean Score Score of Application Name Application Name of Primary Primary Reviewers Reviewers Google Fit: Health and Activity 4.55 Diabetes:M 4.80 Tracking Healthifyme - Diet Plan, Health Google Fit: Health and Activity 4.45 4.75 and Weight Loss Tracking Calorie Counter - My fitness pal 4.30 Calorie Counter - My fitness pal 4.75 Healthifyme - Diet Plan, Health Habits Diabetes coach 4.30 4.55 and Weight Loss Diabetic Recipes: Healthy Food 4.25 8fit Workouts and Meal Planner 4.50 Calorie Counter - MyNetDiary, 4.25 Habits Diabetes coach 4.50 Food diary tracker 8fit Workouts and Meal Planner 4.20 Noom: Health and Weight 4.40 Beat O SMART Diabetes 4.20 Diabetic Recipes: Healthy Food 4.30 Management Diabetic Diet Recipes: Control Beat O SMART Diabetes 4.20 4.30 Diabetes and Sugar Management Diabetic Diet Recipes: Control Diabetes:M 4.10 4.25 Diabetes and Sugar Calorie Counter - MyNetDiary, Noom: Health and Weight 4.00 4.00 Food diary tracker mySugr-Diabetes App and mySugr-Diabetes App and 3.90 3.80 Blood Sugar Tracker Blood Sugar Tracker Diabetes Forum 3.85 Diabetes Forum 3.65 Diabetes Diary - Blood Glucose 3.75 Beat Diabetes 2.75 Tracker Diabetes Diary - Blood Glucose Beat Diabetes 3.40 2.75 Tracker 281 282 From Category II of the MARS ‘Diabetes M’ had the highest App Subjective Quality Mean 283 Score of 4.80/5, scoring highly on all the parameters, including app recommendation,

284 frequency of use, willingness to buy and the overall rating. Conversely, the two apps 285 ‘Diabetes Diary: Blood Glucose Tracker’ and ‘Beat Diabetes’ had the lowest App 286 Subjective Quality Mean Score of 2.75/5. ‘Beat Diabetes’ scored lowest for willingness to 287 buy. Thus, ‘Diabetes M’, ‘Google Fit: Health and Activity Tracking’ and Calorie Counter- 288 My Fitness Pal’ and ‘Healthifyme’ were the top ranked apps as per their App subjective 289 quality mean scores. This was followed by ‘8fit Workouts and Meal Planner’, and ‘Habits 290 Diabetes coach’ tied at number 4 and ‘Noom: Health and Weight’ at number 5. However, 291 an in-depth evaluation of the apps revealed that ‘Diabetes M’ may not be suitable for the 292 Indian cuisine (needs cultural adaptation) and was difficult to access. ‘Healthifyme’ had 293 more features (tracked weight, activity, food and water and provides a personal coach) 294 when compared to ‘Calorie Counter- My Fitness Pal’. Hence, our top ranked apps 295 considering both the ‘App quality mean score’ (Category I) and ‘App subjective quality 296 mean score’ (Category II) were Google fit and Heathifyme. 297 298 Discussion

299 To our knowledge, this is the first study, to conduct a systematic review and compare 300 quality of mHealth apps available in the Play Store for diabetes prevention in Asian 301 Indians, using a standardized rating tool. We found two commercially available apps 302 ‘Google Fit: Health and Activity Tracking’ and ‘Healthifyme - Diet plan, Health and Weight 303 Loss’ ranked highly amongst the apps available in the Google play store for prevention of 304 diabetes in Asian Indians. 305 306 Lifestyle factors such as diet, physical inactivity, excess alcohol consumption, cigarette 307 smoking, drug abuse, stress affecting mental health and lack of sleep are widely 308 recognized as major determinants of metabolic diseases including obesity, T2D and CVD 309 in most populations [16, 17]. Most of these lifestyle behaviours form the base for 310 developing an mHealth app [7]. After analysing the 15 shortlisted apps with respect to 311 these health behaviours, we found that majority dealt with diet and/or physical activity and 312 alluded very briefly to the topics of mental health, stress, sleep and other lifestyle 313 behaviours. 314

315 The ‘Google Fit: Health and Activity Tracking’ app ranked well because although it is 316 focussed on physical activity, it scored highly on Engagement and Functionality. The app 317 is intuitive to use and records physical fitness activities, such as walking, running and 318 cycling. It uses this information to estimate calories burned. Other features offered by the 319 app include weight history, duration of exercise, heart rate monitor and sleep tracker (with 320 the help of a third-party applications), as well as enabling personalized goal settings, with 321 customized tips and actionable coaching. Reports on the Play store suggest there were 322 over five million downloads within 6 months of its release. ‘Google Fit: Health and Activity 323 Tracking’ has previously been reported to be an “effortless and affordable activity tracker 324 with the potential to significantly extend the number of activity tracker users compared to 325 other devices” [15]. The main disadvantage of this app was that it does not track food, 326 however it can connect to many other apps and devices providing a common platform to 327 bring data from various sources together. 328

329 Next were ‘Healthifyme and Calorie Counter - My fitness pal’. ‘Healthifyme’ ranked better 330 than ‘Calorie Counter- My Fitness Pal’ in its Functionality and Information. ‘Healthifyme’ 331 focuses on weight loss, fitness and diabetes prevention through behavioural lifestyle 332 changes. The app sends the user reminders to track their weight, activity, food and water 333 to meet fitness goals. The app also provides personalized reminders for walking and 334 workout and has an artificial intelligence (AI) powered smart nutritionist. The premium 335 version of the app offers personalized coaching by diet, fitness and yoga experts. The 336 app also has success stories to motivate users, as well as more than 500 recipes and 337 nutritional information. In a recent publication [18] by an economist reviewing information 338 technology for primary healthcare in India, ‘Healthifyme’ has been called a promising 339 behavioural application; however, research backed evidence of its impact is not yet 340 available.

341 342 Other apps that scored highly were ‘Habits Diabetes coach’, ‘8fit Workouts and Meal 343 Planner’, ‘Noom: Health and Weight’, ‘Diabetes:M’ and ‘Calorie Counter – MyNetDiary, 344 Food Diary Tracker’. ‘Habits Diabetes coach’ can track glucose, activity, diet and weight,

345 give medication reminders, has videos and a lifestyle coach for better engagement. ‘8fit 346 Workouts and Meal Planner’ has an attractive user interface, good quality video content 347 and categorized workouts, including yoga, strength and high intensity exercise. On the 348 other hand, ‘Noom: Health and Weight’ has an in-built steps counter, food log, graphs to 349 track weight and customized course plans. ‘Diabetes:M’ has lot of features with detailed 350 report generation, which include weight, pulse, blood pressure, cholesterol, Hb1Ac and 351 sites of injection. This app allows the user to maintain multiple profiles, has a smart 352 assistant, advises on insulin bolus dose, can add reminders, give options for data 353 management and data sharing, discussion groups and track blood glucose levels. ‘Calorie 354 Counter –MyNetDiary, Food Diary Tracker’ has meals and activity log, barcode reader, 355 weight chart, basic weight loss plan, vitamin and pulse rate log, daily calorie budget and 356 feature to upload recipes. However, there are limited workout plans, complicated graphs, 357 charts and food tracker, lack of Indian food data and limited features in the basic version 358 of the app affected overall App Quality, Engagement and Information scores of these 359 apps.

360

361 All the apps reviewed in this study had a mean MARS score above 3.75 except for one. 362 This suggests the apps had an overall acceptable level of quality. On evaluating the 363 individual aspects of the app quality mean score, we found the mean score for 364 Engagement to be the lowest (3.6) for all the 15 apps; followed by Information (4.0). The 365 mean score for Aesthetics and Functionality was much higher at 4.4. This indicates that 366 app engagement and information provided by the app are potential target areas for 367 improvement. Similar outcomes were reported in another systematic review on 368 mindfulness apps by Mani et al, where mean engagement scores were low, with 369 aesthetics and information scales found to be moderate [19]. A review of commercial apps 370 available for weight loss, also showed that Information scored the lowest amongst the 371 MARS domains, indicating an overall lack of evidence-based content [20]. This suggests 372 further investment by developers in evidence-based, data-driven content, combined with 373 change techniques known to be effective in changing relevant behaviour patterns, is

374 required. This may improve the overall app quality, regardless of the perceived aesthetic 375 qualities of the app.

376

377 Published literature has shown that most apps available in the market more or less 378 resemble each other in features [21]. But most of these apps lack authentic information 379 on diabetes and/or obesity prevention, the link between lifestyle behaviours and metabolic 380 disease prevention, as well as suffer from lack of culturally relevant tips and suggestions. 381 A recent mHealth study [22] reported that personalization seems to be the key element 382 in engaging a diverse group of participants. In another study [23], the authors concluded 383 that user experience plays an important role in improving their engagement with the app. 384 Furthermore, user interfaces with real-time data interpretation and smart algorithms may 385 be useful in enhancing user experience, thereby improving engagement. The need to 386 develop new mhealth apps arise from the lacunae that exists with regards to lack of 387 scientific evidence, measurable quality standards and a value proposition for existing 388 apps [24,25]. While we were able to identify two high-quality commercial apps, there is, 389 at present, no evidence available for their efficacy in prevention of T2D. Hence, it is 390 recommended that app developers follow a step wise approach to developing any app: 391 (1) define the problem and end user, (2) review the literature extensively, (3) transform 392 information to knowledge, (4) have a proper security and privacy policy, and (5) evaluate 393 usability and efficacy through scientific means [26, 27]. Another important 394 recommendation to mhealth app developers is that while culturally/linguistically adapting 395 an app, it would be best to have a central base application in which app users could just 396 select language etc. and use. Creating multiple applications to handle language and 397 cultural variations could lead to several development and maintenance issues.

398

399 Our study has some limitations. The search terms used in the study were restricted to 400 mhealth apps commercially available in India. Also, the search conducted in Google Play 401 store (for Android phones) might have restricted the results of this review as some of the 402 apps are only found in the App Store (for Apple phones). This study evaluated the basic

403 version available of the selected apps, which would be widely accessible to general 404 population. However, we acknowledge the limitation that paid premium versions may be 405 available for download. In addition, considering the dynamic development of apps, 406 popularity and ratings may change very quickly over time.

407 Conclusions 408 Whilst an increasing number of lifestyle and behaviour change apps are being developed 409 for diabetes prevention and management, the scientific evidence base underlying their 410 use is still limited. Our review identifies two commercially available apps ‘Google Fit: 411 Health and Activity Tracking’ and ‘Healthifyme - Diet plan, Health and Weight Loss’ as 412 being highly rated for their usability. Further research through randomised controlled trials 413 is required to evaluate the usability of these apps in diabetes prevention in Asian Indians. 414 415 Author Contributions: Conceptualization of paper and writing the initial draft: HR and 416 RMA; Methodology: SN, RH, HC, HR; Data Curation: RH, SN, RMA, PA; Review & 417 Editing: All authors. All authors have read and approved the final version of the 418 manuscript, and agree with the order of presentation of the authors.

419

420 Funding: The research was supported by the UK National Institute for Health Research 421 (NIHR) Official Development Assistance (ODA, award 16/136/68). The views expressed 422 are those of the author(s) and not necessarily those of the NIHR.

423

424 Acknowledgments: At Imperial College London, the infrastructure support was provided 425 by the NIHR Imperial Biomedical Research Centre and the NIHR Imperial Clinical 426 Research Facility. The views expressed are those of the author(s) and not necessarily 427 those of the NHS, the NIHR or the Department of Health and Social Care.

428

429 Disclosures: N.O has received honoraria for speaking and advisory board participation 430 from Abbott Diabetes, Dexcom, Medtronic Diabetes, and Roche Diabetes. V.M has

431 received research or educational grants or honoraria for speaking engagements or 432 serving on advisory boards from Novo Nordisk, Servier, MSD, Novartis, Eli Lilly, M/s. 433 USV, Lifescan J & J, Sanofi Aventis, Merck, Astra Zeneca, Boehringer 434 Ingelheim, Abbott and from several Indian pharmaceutical companies.

435 Conflicts of Interest: None declared.

436 Abbreviations:

437 AI- artificial intelligence

438 Apps- applications

439 CVD- cardiovascular disease

440 MARS- Mobile App Rating Scale

441 mHealth- mobile health

442 T2D- type 2 diabetes

443 Appendix 1: Initial rating/scoring scale developed internally based on available 444 information in the google play store

445 Initial Rating Scale 446 Overall app rating in the play store Score out of 4 ≤3.4 1 447 3.5-3.9 2 4-4.4 3 448 4.5-5 4

449 No. of Downloads 10,000-49,999 1 450 50,000-99,999 2 1,00,000-9,99,999 3 451 ≥10,00,000 4

452 Size of the app >75MB 1 453 51-75MB 2 26-50MB 3 454 ≤25MB 4

455 User Reviews (shows popularity) ≤1000 1 456 1001-4999 2 5000-13999 3 457 ≥14000 4

458

459

460

461

462

463

464 Appendix 2: Basic description and review of the final 15 apps using the play store 465 data (or Initial rating scale given in Appendix 1)

Score Score Score for for Score for for number of size number Total Application Application overall download of the of Score S.No. name Description rating s app reviews out of 16 A diabetes app providing coaching along with a tracking program for people with Habits prediabetes, T2D Diabetes and starting insulin 1 coach therapy. 3 2 3 1 9 This app is compiled by clinicians specifically for people with diabetes explaining complications, diets and latest Beat treatment 2 Diabetes strategies. 4 2 4 2 12 A diabetes forum where the user can ask questions, find Diabetes support and share 3 Forum texperiences. 3 2 4 1 10

Calorie Counter - This app is a MyNetDiary, personal weight- Food Diary loss, diet, and 4 Tracker nutrition assistant. 4 4 1 4 13 This app provides healthy recipes for people with diabetes along with Diabetic step by step Recipes : instructions, videos Healthy and nutritional 5 Foods information. 2 3 4 1 10 This app assists in weight loss and fitness with the Healthifyme help of personal - Diet Plan, coach, diet plan Health and and calorie 6 Weight Loss counter. 4 4 3 4 15

Diabetes Diary - This app tracks Blood blood glucose, Glucose weight, blood 7 Tracker pressure and A1C. 4 1 4 1 10 This app helps mySugr- sync with blood Diabetes glucose meter, 8 App and import values and 4 4 3 4 15

Blood Sugar manage diabetes. Tracker One can also log in meals and medication. This app tracks almost all aspects of the diabetes treatment and provides the user with detailed reports, charts and 9 Diabetes: M statistics 4 3 4 4 15 This app recommends an activity goal to improve health. It Google Fit: can track workouts, Health and monitor goals can Activity connect to various 10 Tracking apps and devices. 2 4 4 4 14 This diabetes tracker is a monitoring app through which you can get advice from certified doctors BeatO and nutritionists, SMART helps in scheduling Diabetes your medication Managemen deliveries and 11 t diagnostic services 4 3 4 3 14

It is a fitness tracker, helps in Calorie weight loss and Counter - has a calorie My fitness counter and macro 12 pal tracker. 4 4 3 4 15 Provides a psychology based approach, and builds a custom game plan to help Noom: you form healthy Health and habits, faster and 13 Weight lose weight. 3 4 2 4 13

This app has an Diabetic Diet offline set of Healthy Sugar Recipes : Free Recipes Control along with step Diabetes wise simple 14 and Sugar instructions. 3 2 4 1 10 A fitness app providing quick easy workout 8fit routines combined Workouts with a and Meal simple healthy 15 Planner meal planner. 4 4 1 4 13 Life in This app helps Control manage diabetes Diabetes with a lifestyle Coach coach, blood sugar 3 1 4 1 9

tracker and health log 466 467 468 References 469 1. Jusoh, S. A survey on trend, opportunities and challenges of mHealth apps. 470 International Journal of Interactive Mobile Technologies (iJIM). 2017; 11: 73-85. 471 2. IAMAI & KPMG Report “India On The Go – Mobile Internet Vision Report 2017”. 472 URL:http://eeind.in/mobile-internet-use-in-india-2015-16/. Published 2015. 473 Assessed September 22, 2020. 474 3. Farrington C, Aristidou A, Ruggeri K. mHealth and global mental health: still waiting 475 for the mH2 wedding?.Global Health. 2014;10:17. 476 4. Klasnja P, Pratt W. Managing health with mobile technology. Interactions. 2014; 477 21 (1). DOI:10.1145/2540992. 478 5. Sittig S, Wang J, Iyengar S, Myneni S, Franklin A. Incorporating Behavioral Trigger 479 Messages Into a Mobile Health App for Chronic Disease Management: 480 Randomized Clinical Feasibility Trial in Diabetes. JMIR mHealth and uHealth. 481 2020; 8: e15927. 482 6. Matej Mikulic. Google Play: number of available medical apps as of Q4 483 2020.Published May 13, 2020. 484 URL:https://www.statista.com/statistics/779919/health-apps-available-google- 485 play-worldwide/. Assessed June 26, 2020. 486 7. Milne-Ives M, Lam C, De Cock C, Van Velthoven MH, Meinert E. Mobile Apps for 487 Health Behavior Change in Physical Activity, Diet, Drug and Alcohol Use, and 488 Mental Health: Systematic Review. JMIR MhealthUhealth. 2020;8(3):e17046. 489 8. WHO. Global action plan for the prevention and control of NCDs 2013-2020. 490 http://www.who.int/nmh/publications/ncd-action-plan/en/. Assessed August 8, 491 2020. 492 9. Cummings E, Borycki E, Roehrer E. Issues and considerations for healthcare 493 consumers using mobile applications. Stud Health Technol Inform. 2013;183:227– 494 231.

495 10. Girardello A, Michahelles F. AppAware: Which mobile applications are hot?. 12th 496 international conference on Human computer interaction with mobile devices and 497 services. Association for Computing Machinery, New York (NY), USA. 2010: 431– 498 434. 499 11. Stoyanov SR, Hides L, Kavanagh DJ, Zelenko O, Tjondronegoro D, Mani M. 500 Mobile app rating scale: a new tool for assessing the quality of health mobile apps. 501 JMIR MhealthUhealth. 2015;3(1):e27. 502 12. Stoyanov SR, Hides L, Kavanagh DJ, Wilson H. Development and Validation of 503 the User Version of the Mobile Application Rating Scale (uMARS). JMIR 504 MhealthUhealth. 2016;4(2):e72. 505 13. The Mobile App Rating Scale (MARS) and its Application to the Development of a 506 Youth. URL: https://www.c4tbh.org/seminar/title-tba-5/. Assessed September 9, 507 2020. 508 14. Gwet KL. Handbook of inter-rater reliability: the definitive guide to measuring the 509 extent of agreement among multiple raters, 3rd edition. 3rd ed. Gaithersburg: 510 Advanced Analytics, LLC; 2012. 511 15. Menaspà Paolo. Effortless activity tracking with Google Fit. British journal of sports 512 medicine. 2015;49(24):1598. 513 16. Weber MB, Ranjani H, Staimez LR, et al. The Stepwise Approach to Diabetes 514 Prevention: Results From the D-CLIP Randomized Controlled Trial. Diabetes 515 Care. 2016;39(10):1760-1767. 516 17. Diabetes Prevention Program (DPP) Research Group. The Diabetes Prevention 517 Program (DPP): description of lifestyle intervention. Diabetes Care. 2002;25:2165- 518 2171. 519 18. Mor Nachiket. Information Technology for Primary Healthcare in India. 2020. 520 DOI:10.13140/RG.2.2.25513.13925. 521 19. Mani M, Kavanagh DJ, Hides L, Stoyanov SR. Review and Evaluation of 522 Mindfulness-Based iPhone Apps. JMIR MhealthUhealth. 2015;3(3):e82. 523 20. Bardus M, van Beurden SB, Smith JR, Abraham C. A review and content analysis 524 of engagement, functionality, aesthetics, information quality, and change

525 techniques in the most popular commercial apps for weight management. Int J 526 BehavNutr Phys Act. 2016;13:35. 527 21. Arnhold M, Quade M, Kirch W. Mobile applications for diabetics: a systematic 528 review and expert-based usability evaluation considering the special requirements 529 of diabetes patients age 50 years or older. J Med Internet Res. 2014;16(4):e104. 530 22. Ranjani H, Nitika S, Ranjana RM, Sandhya R, Mohan V, Nalini S. Impact of NCD 531 text messages delivered via an app in preventing and managing lifestyle diseases 532 - results of the ‘myarogya’ worksite-based effectiveness study from India. Journal 533 of Diabetology. 2020;11. 534 23. Muralidharan S, Ranjani H, Anjana RM, Allender S, Mohan V. Mobile Health 535 Technology in the Prevention and Management of Type 2 Diabetes. Indian J 536 Endocrinol Metab.2017;21:334-340. 537 24. Shuhan He. mHealth App Development and Standardization of Innovative 538 Technologies. URL: https://conductscience.com/mhealth-app-development- 539 current-practices-and-future-perspectives/. Assessed December 2, 2020. 540 25. Kazemi A, Salmani H, Shakibafard A, Fatehi F. New and Emerging Mobile Health 541 (mHealth) Technologies and Solutions: A Horizon Scanning Study. Front Health 542 Inform. 2019; 8(1): e17. 543 26. Stephan LS, Dytz Almeida E, Guimaraes RB, et al. Processes and 544 Recommendations for Creating mHealth Apps for Low-Income Populations. JMIR 545 Mhealth Uhealth. 2017;5(4):e41. 546 27. L. Nurgalieva, D. O'Callaghan, Gavin Doherty . Security and Privacy of mHealth 547 Applications: A Scoping Review. IEEE Access 2020. PP. 1-1. 548 10.1109/ACCESS.2020.2999934.