LvL UP 1.0, a holistic mHealth lifestyle coaching intervention for the prevention of non-communicable diseases and common mental disorders: a mixed methods feasibility study

LvL UP 1.0, a holistic mHealth lifestyle coaching intervention for the prevention of non-communicable diseases and common mental disorders: a mixed methods feasibility study | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article LvL UP 1.0, a holistic mHealth lifestyle coaching intervention for the prevention of non-communicable diseases and common mental disorders: a mixed methods feasibility study Jacqueline Louise Mair, Ahmad I. Jabir, Alicia Salamanca-Sanabria, and 15 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6484027/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 22 Dec, 2025 Read the published version in Scientific Reports → Version 1 posted 8 You are reading this latest preprint version Abstract LvL UP is a smartphone-based holistic lifestyle coaching intervention aimed at improving health behaviours, mental well-being, and preventing noncommunicable diseases and common mental disorders. It features a ‘talk and tools’ approach, combining automated health literacy coaching via conversational agent with digital tools such as journaling, life hacks, and slow-paced breathing exercises. An ‘in-the-wild' mixed-methods study was conducted in Singapore to evaluate LvL UP’s feasibility and acceptability to inform a future definitive trial. The app was available on iOS and Android from March to August 2023 and was promoted through online and offline strategies. Data collection included in-app surveys, usage metrics, and interviews, summarised using descriptive statistics and template analysis. The app was downloaded 307 times. Data from 99 active users were analysed. Most users were female and aged 21–35 years with mild to moderate vulnerabilities in physical activity, diet, and depressive symptoms. Engagement was highest during the first eight days, with 9% remaining engaged for up to 50 days. Users rated technology acceptance highly, finding the app enjoyable, easy to use, and informative. Suggested improvements included streamlined onboarding, fixing bugs, shortening dialogues, and adding rewards. The findings support LvL UP’s feasibility and have informed enhancements for future trials. Biological sciences/Psychology/Human behaviour Health sciences/Health care/Disease prevention/Lifestyle modification Digital health chatbot behaviour change lifestyle medicine conversational agent preventive health Figures Figure 1 Figure 2 Figure 3 Figure 4 Background The growing burden of non-communicable diseases (NCDs) and common mental disorders (CMDs), along with their associated risk factors, contribute significantly to escalating healthcare and socio-economic costs worldwide [ 1 – 3 ]. Addressing the prevention, management, and treatment of NCDs and CMDs requires a paradigm shift toward holistic interventions that prioritise health-promoting behaviours, mental well-being, and overall wellness [ 4 , 5 ]. However, the initiation and long-term maintenance of health-promoting behaviour change remains challenging and is often adopted by only a fraction of those in need [ 4 ]. Lifestyle coaching, which focuses on enhancing behaviours related to physical activity, nutrition, stress management, and sleep, has demonstrated its potential to improve health outcomes and reduce chronic disease risks [ 6 ]. Yet, traditional delivery models, typically involving in-person or group-based sessions with a single practitioner, are time- and resource-intensive, limiting accessibility and scalability [ 7 ]. To address these challenges, innovative delivery methods are required, particularly those that can overcome barriers related to cost, accessibility, and sustained engagement. Advances in mobile health (mHealth) technologies and artificial intelligence present new opportunities for scalable and user-friendly interventions. Conversational agents (CAs), or chatbots, have emerged as promising tools for delivering tailored health behaviour support [ 8 ]. With their ability to engage in interactive, real-time dialogue, CAs can mimic human coaching interactions, delivering tailored advice, reminders, and motivation while being available 24/7. Despite their potential, the use of CAs in holistic lifestyle coaching, that is, addressing multiple interconnected domains related to physical and mental health, is still relatively underexplored [ 5 ]. Moreover, few CA interventions are culturally tailored to Asian population [ 8 ]. In response to this need, we have undertaken a rigorous theory-informed and user-centred intervention development process to create “LvL UP” [ 9 ], an app-based holistic lifestyle coaching intervention designed specifically for the Singapore context. The intervention employs a ‘talk and tools’ approach [ 10 ] to deliver health literacy coaching sessions via an automated CA focused on three key areas—physical activity, diet, and mental well-being—which are complemented by behavioural tools, including a journal, life hacks, and a slow-paced breathing training game. The LvL UP intervention is set in Singapore, a highly urbanised and densely populated city-state in Southeast Asia, home to a culturally diverse population comprising mainly Chinese, Malay, and Indian ethnic groups. As a nation facing rising rates of NCDs [ 11 ] and CMDs [ 12 ], Singapore presents a compelling case for preventive health interventions, particularly among young adults. The country’s advanced digital infrastructure, high smartphone penetration, and strong government support for digital adoption make it an ideal setting for deploying mHealth interventions. This study aimed to assess the technical feasibility, user satisfaction, and recruitment capability of the LvL UP intervention in a real-world setting, with the goal of informing future trials and implementation strategies. Methods Intervention Development The LvL UP intervention has been developed as part of a 5-year programme of research under the Future Health Technologies Programme in Singapore. An iterative and user-centred development process has been carried out in line with the multiphase optimization strategy [13] and the UK Medical Research Council guidelines for developing and evaluating complex interventions [14]. Based on findings from initial market analyses [15,16], evidence syntheses [5,17,18] and qualitative studies with potential stakeholders [19,20], a behavioural diagnosis was conducted using the Capability, Opportunity, Motivation, Behaviour (COM-B) model of behaviour change [21] to develop a conceptual model underpinning the intervention [9]. We then engaged in a user-centred prototyping process to develop LvL UP version 1.0; a smartphone app containing a motivational interviewing-inspired CA to act as a ‘digital lifestyle coach’, ‘life hacks’ to encourage healthy habit formation, and ‘tools’ supported by a transdiagnostic cognitive behavioural therapy approach including a slow-paced breathing training game and a journaling tool to promote emotional literacy. Four different CA personas were developed to represent typical Singaporean lifestyles, and users could choose one to communicate with throughout the intervention (Figure 1). Additionally, gamification and storytelling features were built into the intervention design to encourage sustained engagement with the app. This included rewards for task completion and unlocking new features as the intervention progresses. Each of these components is described in detail elsewhere [9]. In brief, the CA was built using an open-source platform called MobileCoach (www.mobile-coach.eu) [22] that operates rule-based with scripted sessions and pre-defined answer options developed by domain experts of the research team. This ensures the fidelity and safety of the intervention content in the health context and has been used in similar interventions previously [23-26]. Coaching sessions deliver evidence-based psychoeducational content with a motivational interviewing-inspired communication strategy focused on three lifestyle behaviour pillars: Move More: Increasing physical activity (e.g., daily step count), exercise (home-based training), reducing sitting time/screen-based activity. Eat Well: Encouraging a nutrient-rich diet, reduced snacking, food purchasing and meal preparation suggestions, links between diet and well-being. Stress Less: Activities based on Cognitive Behavioural Therapy (CBT) evidence, including emotional regulation, behavioural activation, cognitive flexibility, and problem-solving therapy. Life hacks are actionable health and well-being-related tips, delivered either in the morning, at lunchtime or in the evening, that can be easily implemented into the daily routine. They serve to (i) ensure a holistic approach to improving health-related behaviours regardless of the coaching pillar chosen by the participant; (ii) target improvements in self-efficacy; and (iii) facilitate routine during everyday life to encourage positive habit formation [27]. Breeze is an intervention component that uses the smartphone’s microphone to continuously detect breathing phases in semi-real time (i.e., inhale, exhale, and pause between inhale and exhale), which are then used to trigger a gamified biofeedback-guided breathing training [28,29]. Gamified biofeedback is shown as a sailing boat moving down a river which speeds up when the user performs slow-paced breathing. This targets experiential outcomes in addition to instrumental outcomes of psychological well-being and heart rate variability [30]. The free-text Journal tool allows users to reflect on their thoughts, feelings, behaviours, and learnings throughout the intervention and report moments of gratitude to help them reduce stress and solve problems [31]. All journal entries were stored on the device and, therefore, remained strictly confidential to the user, with only the time stamp and mood rating assigned to each journal entry being recorded on the server. Users could view and edit their journal entries at any time. In addition to these therapeutic components, gamification, storytelling, and just-in-time intervention notifications were included as engagement strategies. LvL UP 1.0 is intended to be used daily for 21 days, with users completing one coaching session per day, after which continued use is optional. Progress was monitored through the LvL UP Shield on the app’s home screen. Each time users completed a task in the app they received a puzzle piece to fill the LvL UP Shield. Tasks were split across four categories: Coaching Sessions, Life Hacks, Tools (Breeze or Journal), and Basics (intermediary dialogues and in-app surveys to gather evaluation data). When the shield was filled with the required puzzle pieces, users would unlock an episode from the LvL UP Storytelling approach and progress to the next level with a new shield to fill. In total, there were three levels (See Supplementary File). Study Design We tested the feasibility and acceptability of LvL UP 1.0 in a one-group, in-the-wild study using a mixed-methods approach. The study was conducted entirely remotely in Singapore between March and August 2023. As this was a feasibility study, a formal sample size calculation was not appropriate [32]. To ensure saturation in identifying usability issues, a minimum of 20 participants is typically required based on heuristics in usability engineering [33], but to account for the retention challenges common in remote digital health studies [34], we aimed to recruit 200 participants. Ethical approval The Institutional Review Board of the National University of Singapore (NUS-IRB-2021-623) and the Ethics Commission of ETH Zurich (EK-2021-N-109) approved the study. The study was conducted in accordance with the Declaration of Helsinki and all participants provided written informed consent. Participant Recruitment In an attempt to address diversity issues in digital health research [35,36], the recruitment strategy aimed to reach a demographically diverse sample of adults. Both offline and online approaches were employed to maximise outreach. Offline efforts included door-to-door and on-street flyer distribution in residential areas, complemented by six student interns distributing flyers in high-footfall locations around the university campus over a 10-day period. Online recruitment leveraged various channels, including posts on Telegram groups dedicated to research participation, electronic direct mail (EDM) marketing and social media posts via the research centre’s mailing list and social platforms, and ad campaigns through a dedicated LvL UP social media page. Social media ad campaigns on Meta and Google Ads began in March 2023 for iOS and expanded to both iOS and Android from May to July 2023. Campaigns targeted English-speaking adults aged 21 and older living in Singapore. Initial ad variations were tested using Meta’s A/B testing feature, comparing a video post and a static image with positive framing (“Do you want to improve your well-being?”) against negative framing (“Are you struggling with your mental and physical well-being?”) (see Supplementary File). Over seven days, the positively framed ads outperformed the negatively framed ones, with a lower cost per click (S$0.34 vs. S$0.41), prompting their use for the remainder of the campaign. Additionally, a project website provided detailed information about the LvL UP app, the research program and study team, and a link to request an offline version of the LvL UP life hacks and journal template. To further extend reach, a referral link within the app enabled participants to invite friends and colleagues to download and engage with the app. Procedure App Onboarding On opening the app, participants were prompted to register with an email address and complete an e-consent to participate in the study. They then proceeded to choose a digital coach (CA); one of four coaching personas designed to represent the target group of young and middle-aged adults (see Supplementary File). Participants completed a ‘Welcome Day’ dialogue which provided an overview of the goals and intended use of the intervention, a vulnerability assessment, and a shared decision-making process on which pillar (Move More, Eat Well, or Stress Less) to begin with. When the vulnerability assessment identified someone as having severe depression or anxiety (PHQ-4 score >9) they were advised to seek mental health resources in Singapore (e.g., SOS Samaritans) and face-to-face counselling from a health professional. They were not prohibited from using the app or continuing with the study. Still, they were advised that the LvL UP intervention is designed to prevent disease and not treat a pre-existing condition. After completing the ‘Welcome Day’ dialogue, participants were free to use the app as they wished, although they were prompted to complete a coaching session and a life hack daily through the app’s home screen and in-app notifications. Outcome measures Socio-demographic characteristics, including age group, gender, ethnicity, educational level, marital status, and household income were obtained via in-app questionnaires. Health status was also measured by the single-item Self-Rated Health Index [37] and the World Health Organization-Five Well-Being Index (WHO-5) [38]. The health vulnerability assessment included the Patient Health Questionnaire-4 (PHQ-4) for symptoms of depression and anxiety [39], the International Physical Activity Questionnaire (short form) [40] and a dietary guideline adherence questionnaire developed for this study based on Singapore’s My Healthy Plate (see Supplementary File). The efficacy of each recruitment channel was assessed using dynamic link tracking and QR codes in all promotional materials. The feasibility of the LvL UP app was assessed by (i) user engagement in terms of app usage (i.e., response to notifications, timestamps and duration of app component use, and app usage patterns), (ii) session alliance with the CA, self-reported on a six-point Likert scale anchored from 0 (‘not at all’) to 5 (‘completely) [41], and (iii) technology acceptance, including perceived enjoyment, ease of use, usefulness, and control, self-reported on a six-point Likert scale anchored from 0 (‘strongly disagree’) to 5 (‘strongly agree’) [42] (see Supplementary File). User satisfaction with LvL UP was assessed using a short feedback survey containing two open-response questions that asked participants what they liked about the app and what needed to be improved. Additionally, the cultural relevance of the app content was assessed across three dimensions of equivalence: functional (whether it contains familiar behaviours, relevant cultural context, and culturally tailored goals and examples), conceptual (whether the underlying ideas and expressions are meaningful to the target. group), and linguistic (whether the language, including regional variations, idioms, and slang, is appropriate) [43]. These were rated on a six-point Likert scale ranging from ‘not at all’ (0) to ‘completely’ (5). For those completing the short feedback survey, an option to participate in an interview was offered. Depending on participant preference, interviews were conducted virtually using Zoom or WhatsApp chat by two members of the research team (AIJ and BF). The topic guide for the interview focused on motivation to use the app, experience using the app’s features, and perceived impact of the app. It was piloted with members of the study team and consisted of semi-structured, open-ended questions, with more specific probes used to solicit additional information (funnelling technique). After completing at least one coaching session, all self-report surveys were available in the app and tied to the LvL UP Shield progress to encourage completion. Participants were reimbursed with a S$5 shopping e-voucher upon completing the short feedback survey and a S$20 shopping e-voucher for interviews. Data Analysis We used Firebase Analytics to analyse dynamic links and summarise installs from each recruitment channel. Data from the Google and Meta Ads Manager were used to summarise ad performance including impressions, click through rate, cost per click, link clicks, and ad spend. Vulnerability assessment cut-off scores were established as follows: PHQ-4 score = 3 (indicating symptoms of anxiety and depression); IPAQ-short form total physical activity in mins per week = < 120 (below physical activity guidelines); dietary guideline adherence score = 18 (not consuming recommended portions of key food groups). App usage data were analysed using R (version 4.4.2) and Jupyter Notebook. Quantitative survey data were summarised using descriptive statistics and open-ended responses were coded and analysed together with the interview data by one researcher (JLM). Data are reported as mean ± standard deviation throughout unless otherwise specified. Interviews were automatically transcribed verbatim by the Zoom platform or downloaded directly from WhatsApp. Two members of the research team (AIJ and XL) reviewed the transcripts for errors and de-identified the files. Transcripts were then uploaded to Atlas.ti (V.9) for analysis. Template analysis, a form of thematic analysis, was used to analyse the data [44]. The initial template (codebook) was defined a priori and covered four key domains based on our research aims (i.e. to improve the LvL UP app): (i) participants’ perceptions of the app, including coaching sessions, life hacks, the journal, Breeze, storytelling (animations), gamification (Shield), in-app notifications, and overall look and feel; (ii) suggestions for improvement; (iii) barriers and facilitators to app use; and (iv) perceived impact on health and behaviour. To ensure rigor in the coding process, two researchers (AIJ and SZ) independently coded an equal subset of transcripts using a deductive approach. Each researcher maintained a reflective notebook to document emerging codes and analytical insights. Following independent coding, they reviewed each other’s transcripts, compared codes, and discussed any discrepancies or additional codes that emerged beyond the initial template. Disagreements in coding or code definitions were resolved through discussion, with a third reviewer (JLM) acting as an arbiter when consensus could not be reached. The codebook was iteratively refined until all coders reached an agreement. The template was then applied to the open-ended survey responses by JLM. The final template covered three overarching topics: (i) app components, including coaching sessions, life hacks, Breeze, and the journal; (ii) engagement strategies; and (iii) overall user experience. Results Reach The average click through rate across all social media ad campaigns was approximately 1%, which is considered reasonable for social media news feed ads [45]. While EDM achieved the highest click through rate, friend referrals proved most effective in converting clicks into app installs (Table 1). A technical issue with link tracking on Meta prevented the accurate tracking of app installs through that platform. Recruitment In total, 307 individuals downloaded the app, 249 opened the app and registered an email address, 215 verified their email address, 204 started the welcome dialogue, 105 completed the vulnerability assessment, and 99 (herein defined as ‘active users’) completed the ‘Welcome Day’ dialogue to unlock the coaching sessions. The recruitment cost per install was S$15.92 and per active user was approximately S$47.29. The main reasons people download the app were (1) to see what the app was about (n=52; 41.6%), (2) to improve overall well-being (n=36, 28.8%), (3) to lose weight (n=13; 10.4%), (4) to learn about healthy living (n=7; 5.6%), (5) to get more active (n=5; 4.0%), (6) to improve diet (n=3; 2.4%) and (7) to manage emotions (n=3; 2,4%). The active users were mostly female, aged 21-35 years, of Chinese ethnicity, single, educated to university degree level, and in employment (Table 2), with mild to moderate health vulnerabilities in terms of physical activity, diet, and mental well-being (Table 3). Of the 19 active users completing the health status survey, average mental well-being, (measured by the WHO-5) was rated 16.76 out of 25 (67.05%) and the majority self-rated their health as good (Table 2). Technical Feasibility Technology acceptance was rated good overall, with most people agreeing the app was enjoyable, easy to use, and provided useful information (n=31; mean ± SD; 3.73 ± 1.07). The cultural relevance of the app content was rated moderate (n=10; functional equivalence 3.6 ± 0.97; conceptual equivalence 3.6 ± 0.97; and linguistic equivalence 3.5 ± 1.18) and users' session alliance ratings with the CA were good (n=19; 3.9 ± 1.22). Daily engagement with the LvL UP app features (coaching sessions, life hacks, journal, and Breeze) since installation is summarised in Figure 2. Usage was highest in the first eight days . By week two, 77.8% (77/99) of the initial users were lost to attrition. Of those who spent more than one day on the app, they completed an app activity approximately every three days (3.21 ± 7.23 days). Users were most active in the afternoons (Figure 3) and on Tuesdays (Figure 4). On average, active users completed three coaching sessions (mean ± SD; 3.35 ± 4.53), five life hacks (4.92 ± 6.52), two journal entries (2.34 ± 2.30) and four slow-paced breathing training sessions (4.29 ± 4.09). Active users spent between 1 and 28 days (4.07 ± 5.02 days) completing at least one task on the app and 35.4% (35/99) were active for less than one day. Among the active users, 9.1% (9/99) completed Level 1, 4.0% (4/99) progressed to Level 2, and 3.0% (3/99) completed the intervention. Only one user managed to complete the intervention within 21 days. Five users (5%) were considered super users, engaging in the intervention regularly for 21 days or longer. A total of 405 coaching session notifications were successfully delivered. Of those, 77.8% (315/405) were responded to, and 28.9% (117/405) were within 10 minutes (1.53 ± 2.41 min). The median response time between notification delivery and observed coaching session engagement was 57 minutes. However, the notification response rate could be higher as not all activities on the app were tracked (i.e. the time user opened the app) due to system limitations. Acceptability Fifty participants completed the short feedback survey, and 14 interviews were conducted (n=11 on Zoom and n=3 on WhatsApp chat). The interviews took, on average, 33 minutes (15.8 – 50.0 min). Feedback was centred around the positive and negative aspects of (i) the app components, including the coaching sessions, life hacks, Breeze, and journal, (ii) engagement strategies, and (iii) overall user experience. LvL UP intervention and app components Many participants had positive experiences with the CA and the coaching sessions, describing the content as helpful, comprehensive, and interactive. Some mentioned that the intervention provided support and served as a reminder to make healthier choices in everyday life: “It had a lot of tips on how to eat well, and not just like cooking wise. I was quite impressed with the social situation one, because yeah, I did really think about how the social situation would affect our eating habits a lot. And then in particular…communicating how to say no in a way that doesn't affect the other party.” [Interviewer: like, how to reject the food?] “Yeah…that stood out for me .” (ID09, female, 21-35 years old) “ I think LvL UP serves as an accountability partner in terms of your well-being journey because you are required to check in every day. And if you don't then [it will] give you a notification. So, I think there is a good reminder that you have to be mindful of these new habits to incorporate and learn new things every day.” (ID08, female, 21-35 years old) ”It helps that I don't feel alone when I don't have a friend to work out with. And I don't get judged by others. I felt I looked forward to communicating with Kai [CA] again and he is very personal despite it being a chatbot only.” (ID12, female, 36-50 years old) Some felt the content was personalised which made it more motivating, while others felt the content could be more relevant to their needs. ”I like the chatbot, it makes it personalised and creates more motivation with the regular check ins and all. The exercises and concepts discussed via chatbot are comprehensive and second only to live discussions with coaches, I feel.“ (ID13, female, 21-35 years old) On the other hand, many participants mentioned that the coaching sessions were too long and responses from the CA were too slow, and that daily coaching may be too demanding. Additionally, elements of the motivational interviewing style of coaching did not seem to work well in a chat-based format, for example, asking for permission before progressing with advice. “….they [CA] would pose a question and then I'll have to click ‘yes, I want to hear more’, before they put the next topic in. Well, personally, I feel that if you just put the topic itself and carry on talking, without putting in the random interjecting statements like, ‘yes, I want to hear more’, it's a bit easier for me to process the information because it comes to me directly. I do not need to keep on clicking the button .” (ID07, male, 21-35 years old) “I think it's really cool that there's the badges that you get to earn. The icon that I was talking about when you complete an activity. So, you get to see your progress. That is something that resonates with me because I'm somebody that really likes to see gamification integrated into whatever I'm doing, because it just makes the overall experience much more enjoyable, much more fun.” (ID02, male, 21-35 years old) Engagement strategies There was some appreciation for the storytelling approach which participants found interesting and potentially effective in keeping people engaged, but it was also perceived as time consuming. “…the storyline thing…the characters that you can choose, and they have the daily check-in with you. So, it feels like a game but not really a game where you can follow the character’s development …” (ID09, female, 21-35 years old) “LvL UP tries to give you a story. I mean, it's not bad lah, the story, because maybe you identify with a certain character like the lady, Dana, who drinks bubble tea and whatever, right? But when I'm pressed for time, then it becomes very irritating.” (ID06, female, 36-50 years old) Many participants talked about the need for incentives and rewards, tied to the gamification features of the app, to encourage its continued use as well as motivate health behaviour change. “I think incentive-based health care provision is actually quite effective. Like anecdotally, I've seen people going to lengths walking their 10,000 steps a day to clock their Healthy365 [step counting app] steps, you know, so that they can get rewards. Yeah, so I do think that the use of incentives might motivate people to pursue a healthier lifestyle .” (ID07, male, 21-35 years old) “Perhaps once you manage to accumulate all these icons, then you get some sort of monetary incentives or something like that. I think that would really incentivize people to want to do the tasks.” (ID02, male, 21-35 years old) “Maybe you can connect with other partners like merchants to give discount vouchers, or like you can collaborate with I don't know, like speaker or gym source to provide more services or something. ( ID08, female, 21-35 years old) Overall user experience The LvL UP app was generally described as easy to use although some participants mentioned that there were too many components to the app and the levelling up criteria was not always clear. There was a preference for straightforward functionality, the option to tailor settings in the app (such as timing of coaching sessions), and an app that required minimal time investment. “The ability to schedule our coaching sessions was a good move to help integrate it into my lifestyle.” (ID14, male, 21-35 years old) “…because Lumihealth is kind of like touch and go once you see the quest, then you pretty much know what to do. As for LvL UP, for the most part I would have to stay on the app for maybe let's say, 10 to 15 min before I get the task done.” (ID02, male, 21-35 years old) Regarding interactions with the CA, the predefined answer options were generally appreciated from an ease of use perspective, however participants would prefer more autonomy over their responses, through either more diverse answer options or the ability to type their own response. “Offer more options to disagree with the coach. I'm going through the Move More pillar, and there have been various instances where I had to select an option equivalent to ‘Yes’ of varying degrees to agree with the chatbot because there is no negative option, even though I do not agree with the statement put forth. The conversation then ends up feeling shallow, like not true to self or lying.” (WA03, female, 36-50 years old) “But sometimes I think I am just very lazy to type, lah. That's good to have an option. But I think it can be another option to let you type yourself”. [Interviewer: So, having a variety of options to respond to the coach is something that you would prefer. Am I right to say that?] “Yeah, correct.” (ID05, female, 21-35 years old) Feedback on the user interface was mixed; some liked the colourful look and feel while others said a cleaner interface would be preferable. The animated design of the digital coaches and the story were sometimes described as “childish”, with one participant suggesting that videos of real people would be preferred. There was an expectation that the app would better integrate with existing apps and devices for a more personalised experience. “Also, consider syncing with wearable devices, allow data sharing or integrate with health-related apps to have a comprehensive experience. (ID15, female, 36-50 years old) Finally, suggested improvements were related to usability (improving the registration process and the speed of the app, fixing bugs and app crashes). Discussion This study provides insights into the feasibility and acceptability of the LvL UP 1.0 intervention delivered to adults in Singapore. The findings align with broader trends in mHealth research, highlighting both the potential and challenges of digital behavioural interventions, particularly those delivered via CAs. LvL UP 1.0 was technically feasible; all intervention components were successfully delivered, technology acceptance and session alliance were rated positively, and notifications were delivered as planned. However, active users were not representative of the broader target population, and engagement beyond eight days was low, possibly due to shortcomings in the overall user experience. Recruitment and engagement strategies play a crucial role in the success of remotely delivered mHealth interventions. In line with previous studies, we found that positively framed health messages were more effective in promoting engagement than fear-based messaging [45]. The overall click through rate of approximately 1% was consistent with industry standards for social media advertising [46], but conversion from advertisement clicks to active app users remained a challenge. Friend referrals were the most effective in converting interest into active participation, supporting existing evidence that peer recommendations enhance trust and motivation in digital health adoption [36]. The demographic profile of active users in this study suggests that LvL UP appealed primarily to young, well-educated, employed women with mild to moderate health vulnerabilities. This is consistent with broader digital health trends and underscores the ongoing challenge of reaching individuals from lower socioeconomic backgrounds or ethnic minority groups [47,48], who may stand to benefit the most from such interventions. Expanding recruitment strategies to include community-based outreach and partnerships with trusted local organizations could enhance engagement among those at higher risk of developing NCDs or CMDs but who may not (yet) be actively seeking support [36]. Additionally, while LvL UP attracted initial interest, the attrition rate—from app downloads to active participation—suggests potential barriers during early engagement, including digital literacy, perceived relevance, and trust [36]. The application of user-centered design principles in LvL UP provides a strong foundation, and future iterations can build on this by incorporating additional accessibility and culturally adapted features to ensure inclusivity and sustained engagement. Challenges with uptake and attrition may have been exacerbated by the study’s fully remote, ‘in-the-wild’ design, where users received no researcher-led guidance beyond the app’s welcome dialogue. While this approach enhanced external validity by simulating a real-world app launch, sustained usage was low, despite the reported positive experiences with the LvL UP content. Although the usage patterns align with trends seen in other app-based [49] and CA interventions [50], the drop in adherence presents challenges for evaluation the therapeutic effectiveness of remotely delivered interventions. User experience, gamification, progress tracking, and rewards are key features that can enhance user retention in mHealth interventions [17,36]. Therefore, future iterations of LvL UP could explore simplifying navigation, and offering clearer onboarding guidance, feedback on activities, and achievement-based rewards, to help improve usability and retention. Solving the problem of attrition in mHealth is an ongoing effort. However, findings from this study point to three potential solutions: (i) access to human support for those who need it, (ii) more personalised support to improve therapeutic alliance, and (iii) delivering relevant intervention support at the most optimal moments. Firstly, adherence to mHealth interventions improves when usage expectations are clearly communicated, and human support is available [17]. Participants in this study emphasized the importance of a straightforward user experience in creating positive first impressions and the need for clear guidance on the intended use of the intervention. Offering a human connection during onboarding and throughout the intervention may foster supportive accountability [51] and sustained engagement. Embedding tools like LvL UP within existing healthcare or community support systems, such as primary care, mental health services, or peer-led community programmes, could strengthen this accountability while offering additional opportunities for engagement. For example, a health professional or community worker introducing the app, guiding the initial setup, and providing periodic encouragement may bridge the gap between digital and human support. This blended model could enhance retention, particularly for users with greater needs or lower digital confidence. Nevertheless, relying on human involvement at scale may be impractical, and further research into optimising when and how to provide such support within mHealth interventions is needed [52]. Secondly, as shown elsewhere, there was a clear expectation and appreciation for personalised health behaviour support [17,19]. The CA in the LvL UP app is based on a rule-based system whereby coaching sessions are predefined based on evidence-based content, and personalization is achieved through branching. While this approach ensures fidelity of the content and offers a more interactive delivery of health literacy support, branching capability is limited and it is impossible to adapt content to an individual’s changing needs. Furthermore, delivering motivational interviewing in a chat-based format with pre-defined answer options is challenging because it relies on reflecting one’s own words. In this study, attempts to “ask, offer, ask” when providing advice or guidance were also perceived as annoying. Moving forward, strategies employing large language models to personalise the interaction further using more natural language while adhering to the spirit and principles of MI may offer a solution [53,54]. Thirdly, the engagement data suggest that LvL UP was most frequently used in the afternoons and on Tuesdays, which could inform future ‘just-in-time’ strategies by tailoring intervention delivery to peak usage periods when people are most receptive [55]. Recent research has shown the potential for machine learning approaches to predict these moments based on features such as device usage, phone battery level, location, and physical activity [56]. However, more research is needed in the context of complex health behaviour interventions such as LvL UP. Recommendations for future work The findings from this study highlight several key considerations for future iterations of LvL UP: Expand and diversify recruitment strategies by complementing social media advertising with peer referral systems, community-based outreach, and partnerships with trusted local organisations. Clearer value propositions tailored to diverse user groups may also improve conversion from interest to active use. Embed LvL UP within existing care and community ecosystems to enhance supportive accountability and engagement. Future iterations could explore blended models where LvL UP is introduced and supported by healthcare providers, community health workers, or peer mentors. Leverage machine learning and just-in-time adaptive interventions to deliver support during periods of peak user receptivity, informed by real-world usage data and contextual cues such as time of day, activity level, and device interaction. Enhance the CA experience by improving dialogue flow and incorporating hybrid interaction models—combining structured choices with optional free-text input—to allow more natural, personalised, and engaging conversations. Refine onboarding and retention strategies to create good first impressions by streamlining the initial user journey and introducing elements such as progress tracking, social or peer-based accountability, and achievement-based rewards to promote sustained engagement. Conclusion While LvL UP demonstrated feasibility and acceptability, engagement challenges highlight the need for further optimization. The findings align with broader trends in mHealth research, reinforcing the importance of personalization, human coaching support and accountability, gamification, and rewards in sustaining long-term engagement. Future work should explore how to balance scalability with tailored support, leveraging emerging technologies such as large language models to enhance personalisation and engagement, and subsequently, intervention effectiveness. Declarations Author Contributions JLM – conceptualisation, methodology, formal analysis, investigation, writing-original draft, project administration AS - conceptualisation, methodology, writing-reviewing & editing OC – methodology, investigation, writing-reviewing & editing AJ – methodology, formal analysis, investigation, writing-reviewing & editing SZ - methodology, formal analysis, investigation, writing-reviewing & editing RK – methodology, formal analysis, investigation, writing-reviewing & editing BF – methodology, investigation, writing-reviewing & editing CSL – software AA - methodology, investigation SN – data curation, writing-reviewing & editing AS - data curation, writing-reviewing & editing RVD - methodology, writing-reviewing & editing, funding acquisition EST - supervision, funding acquisition EF - funding acquisition FW - supervision, funding acquisition LTC - conceptualisation, writing-reviewing & editing, supervision, funding acquisition FMR - conceptualisation, writing-reviewing & editing, supervision, funding acquisition TK – conceptualisation, methodology, resources, writing-reviewing & editing, supervision, funding acquisition All authors read and approved the final manuscript. Data Availability Statement The datasets collected and analysed in the current study are available in anonymised form from the corresponding author on reasonable request. Additional Information The authors acknowledge the important contribution of Mr Prabhakaran Santhanam in the technical development of LvL UP 1.0 and Ms Xiaowen Lin for her assistance in processing interview transcripts. This study was funded by the National Research Foundation, Prime Minister’s Office, Singapore under its Campus for Research Excellence and Technological Enterprise (CREATE) programme. The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; or decision to submit the manuscript for publication. JLM, TK, OC, and FW are affiliated with the Centre for Digital Health Interventions, a joint initiative of the Institute for Implementation Science in Health Care, University of Zurich; the Department of Management, Technology, and Economics at ETH Zurich; and the Institute of Technology Management and School of Medicine at the University of St Gallen. CDHI is funded in part by CSS, a Swiss health insurer; Mavie Next, an Austrian health insurer; and MTIP, a Swiss digital health investor. However, none of these companies were involved in this study. All other authors declare no potential conflict of interest. References Murphy, A. et al. The household economic burden of non-communicable diseases in 18 countries. BMJ Glob. Health 5 , e002040 (2020). Murray, C. J. L. et al. Global burden of 87 risk factors in 204 countries and territories, 1990–2019: a systematic analysis for the Global Burden of Disease Study 2019. The Lancet 396 , 1223–1249 (2020). GBD 2019 Mental Disorders Collaborators. Global, regional, and national burden of 12 mental disorders in 204 countries and territories, 1990–2019: a systematic analysis for the Global Burden of Disease Study 2019. Lancet Psychiatry 9 , 137–150 (2022). Katz, D. L. et al. Lifestyle as Medicine: The Case for a True Health Initiative. Am. J. Health Promot. 32 , 1452–1458 (2018). Zheng, S. et al. Effectiveness of holistic mobile health interventions on diet, and physical, and mental health outcomes: a systematic review and meta-analysis. eClinicalMedicine 66 , (2023). Olsen, J. M. & Nesbitt, B. J. Health Coaching to Improve Healthy Lifestyle Behaviors: An Integrative Review. Am. J. Health Promot. 25 , e1–e12 (2010). Werntz, A., Amado, S., Jasman, M., Ervin, A. & Rhodes, J. E. Providing Human Support for the Use of Digital Mental Health Interventions: Systematic Meta-review. J. Med. Internet Res. 25 , e42864 (2023). Schachner, T., Keller, R. & Wangenheim, F. von. Artificial Intelligence-Based Conversational Agents for Chronic Conditions: Systematic Literature Review. J. Med. Internet Res. 22 , e20701 (2020). Castro, O. et al. Development of “LvL UP 1.0”: a smartphone-based, conversational agent-delivered holistic lifestyle intervention for the prevention of non-communicable diseases and common mental disorders. Front. Digit. Health 5 , (2023). Buen, R. et al. Talk and Tools: the best of both worlds in mobile user interfaces for E-coaching | Personal and Ubiquitous Computing. Pers. Ubiquitous Comput. 21 , 661–674 (2017). Phan, T. P. et al. Forecasting the burden of type 2 diabetes in Singapore using a demographic epidemiological model of Singapore. BMJ Open Diabetes Res. Care 2 , (2014). Subramaniam, M. et al. Tracking the mental health of a nation: prevalence and correlates of mental disorders in the second Singapore mental health study. Epidemiol. Psychiatr. Sci. 29 , (2020). Collins, L. M. Optimization of Behavioral, Biobehavioral, and Biomedical Interventions: The Multiphase Optimization Strategy (MOST) . (Springer International Publishing, 2018). doi:10.1007/978-3-319-72206-1. Skivington, K. et al. A new framework for developing and evaluating complex interventions: update of Medical Research Council guidance. BMJ 374 , n2061 (2021). Keller, R. et al. Digital Behavior Change Interventions for the Prevention and Management of Type 2 Diabetes: Systematic Market Analysis. J. Med. Internet Res. 24 , e33348 (2022). Castro, O. et al. Top-funded companies offering digital health interventions for the prevention and treatment of depression: a systematic market analysis. Arch. Public Health 82 , 200 (2024). Jakob, R. et al. Factors Influencing Adherence to mHealth Apps for Prevention or Management of Noncommunicable Diseases: Systematic Review. J. Med. INTERNET Res. 24 , (2022). Mair, J. L. et al. Effective Behavior Change Techniques in Digital Health Interventions for the Prevention or Management of Noncommunicable Diseases: An Umbrella Review. Ann. Behav. Med. 57 , 817–835 (2023). Mair, J. L. et al. Exploring the potential of mobile health interventions to address behavioural risk factors for the prevention of non-communicable diseases in Asian populations: a qualitative study. BMC Public Health 23 , 753 (2023). Salamanca-Sanabria, A. et al. Exploring the Perceptions of mHealth Interventions for the Prevention of Common Mental Disorders in University Students in Singapore: Qualitative Study. J. Med. Internet Res. 25 , e44542 (2023). Michie, S., Atkins, L. & West, R. The Behaviour Change Wheel: A Guide to Designing Interventions . (Silverback Publishing, 2014). Kowatsch, T. et al. Design and Evaluation of a Mobile Chat App for the Open Source Behavioral Health Intervention Platform Mobilecoach. in Designing the Digital Transformation. Desrist 2017. Lecture Notes in Computer Scienc vol. 10243 485–489 (Springer, Berlin, 2017). Ollier, J. et al. Can digital health researchers make a difference during the pandemic? Results of the single-arm, chatbot-led Elena+: Care for COVID-19 interventional study. Front. Public Health 11 , (2023). Ulrich, S., Lienhard, N., Künzli, H. & Kowatsch, T. A Chatbot-Delivered Stress Management Coaching for Students (MISHA App): Pilot Randomized Controlled Trial. JMIR MHealth UHealth 12 , e54945 (2024). Stasinaki, A. et al. Effects of a novel mobile health intervention compared to a multi-component behaviour changing program on body mass index, physical capacities and stress parameters in adolescents with obesity: a randomized controlled trial. BMC Pediatr. 21 , 308 (2021). Stieger, M. et al. Changing personality traits with the help of a digital personality change intervention. Proc. Natl. Acad. Sci. 118 , e2017548118 (2021). Arlinghaus, K. R. & Johnston, C. A. The Importance of Creating Habits and Routine. Am. J. Lifestyle Med. 13 , 142–144 (2018). Shih, C.-H. (Iris) et al. Breeze: Smartphone-based Acoustic Real-time Detection of Breathing Phases for a Gamified Biofeedback Breathing Training. Proc. ACM Interact. Mob. Wearable Ubiquitous Technol. 3 , 152:1-152:30 (2019). Lukic, Y. X., Teepe, G. W., Fleisch, E. & Kowatsch, T. Breathing as an Input Modality in a Gameful Breathing Training App (Breeze 2): Development and Evaluation Study. JMIR Serious Games 10 , e39186 (2022). Lukic, Y. X. et al. Physiological Responses and User Feedback on a Gameful Breathing Training App: Within-Subject Experiment. JMIR Serious Games 9 , e22802 (2021). Parks, A. & Schueller, S. The Wiley Blackwell Handbook of Positive Psychological Interventions . (John Wiley & Sons, Ltd, 2014). Billingham, S. A., Whitehead, A. L. & Julious, S. A. An audit of sample sizes for pilot and feasibility trials being undertaken in the United Kingdom registered in the United Kingdom Clinical Research Network database. BMC Med. Res. Methodol. 13 , 104 (2013). Nielsen, J. Designing Web Usability . (New Riders Publishing, Indianapolis, USA, 2000). Pratap, A. et al. Indicators of retention in remote digital health studies: a cross-study evaluation of 100,000 participants. Npj Digit. Med. 3 , 1–10 (2020). Clark, L. T. et al. Increasing Diversity in Clinical Trials: Overcoming Critical Barriers. Curr. Probl. Cardiol. 44 , 148–172 (2019). Mair, J. L. et al. Understanding and overcoming barriers to digital health adoption: a patient and public involvement study. Transl. Behav. Med. 15 , ibaf010 (2025). Ware, J. E. & Sherbourne, C. D. The MOS 36-item short-form health survey (SF-36). I. Conceptual framework and item selection. Med. Care 30 , 473–483 (1992). World Health Organization. The World Health Organization-Five Well-Being Index (WHO-5) . https://cdn.who.int/media/docs/default-source/mental-health/who-5_english-original4da539d6ed4b49389e3afe47cda2326a.pdf?sfvrsn=ed43f352_11&download=true (2024). Kroenke, K., Spitzer, R. L., Williams, J. B. W. & Löwe, B. An Ultra-Brief Screening Scale for Anxiety and Depression: The PHQ–4. Psychosomatics 50 , 613–621 (2009). Tran, V. D. et al. Validity of the International Physical Activity Questionnaire–Short Form for Application in Asian Countries: A Study in Vietnam. Eval. Health Prof. 43 , 105–109 (2020). Falkenström, F., Hatcher, R. L., Skjulsvik, T., Larsson, M. H. & Holmqvist, R. Development and validation of a 6-item working alliance questionnaire for repeated administrations during psychotherapy. Psychol. Assess. 27 , 169–183 (2015). Kowatsch, T. et al. Conversational Agents as Mediating Social Actors in Chronic Disease Management Involving Health Care Professionals, Patients, and Family Members: Multisite Single-Arm Feasibility Study. J. Med. Internet Res. 23 , e25060 (2021). Salamanca-Sanabria, A., Richards, D. & Timulak, L. Adapting an internet-delivered intervention for depression for a Colombian college student population: An illustration of an integrative empirical approach. Internet Interv. 15 , 76–86 (2019). Brooks, J., McCluskey ,Serena, Turley ,Emma & and King, N. The Utility of Template Analysis in Qualitative Psychology Research. Qual. Res. Psychol. 12 , 202–222 (2015). Gallagher, K. M. & Updegraff, J. A. Health message framing effects on attitudes, intentions, and behavior: a meta-analytic review. Ann. Behav. Med. Publ. Soc. Behav. Med. 43 , 101–116 (2012). Chaffey, D. 2024 average ad click through rates (CTRs) for paid search, display and social media. Smart Insights https://www.smartinsights.com/internet-advertising/internet-advertising-analytics/display-advertising-clickthrough-rates/ (2024). Yao, R. et al. Inequities in Health Care Services Caused by the Adoption of Digital Health Technologies: Scoping Review. J. Med. Internet Res. 24 , e34144 (2022). Woolley, K. E. et al. Mapping Inequities in Digital Health Technology Within the World Health Organization’s European Region Using PROGRESS PLUS: Scoping Review. J. Med. Internet Res. 25 , e44181 (2023). Baumel, A., Muench, F., Edan, S. & Kane, J. M. Objective User Engagement With Mental Health Apps: Systematic Search and Panel-Based Usage Analysis. J. Med. INTERNET Res. 21 , (2019). Jabir, A. I. et al. Attrition in Conversational Agent–Delivered Mental Health Interventions: Systematic Review and Meta-Analysis. J. Med. Internet Res. 26 , (2024). Mohr, D., Cuijpers, P. & Lehman, K. Supportive Accountability: A Model for Providing Human Support to Enhance Adherence to eHealth Interventions. J. Med. Internet Res. 13 , e1602 (2011). Castro, O. et al. The LvL UP trial: Protocol for a sequential, multiple assignment, randomised controlled trial to assess the effectiveness of a blended mobile lifestyle intervention. Contemp. Clin. Trials 150 , 107833 (2025). Shenoi, A. et al. MICoach: LLM-driven motivational interviewing for physical activity behaviour change. Under review, in Proceedings of the ACM International Conference on Intelligent Virtual Agents (ACM, BERLIN Germany, 2025) Steenstra, I., Nouraei, F., Arjmand, M. & Bickmore, T. Virtual Agents for Alcohol Use Counseling: Exploring LLM-Powered Motivational Interviewing. in Proceedings of the ACM International Conference on Intelligent Virtual Agents 1–10 (ACM, GLASGOW United Kingdom, 2024). doi:10.1145/3652988.3673932. Keller, R., von Wangenheim, F., Mair, J. L. & Kowatsch, T. Receptivity to mobile health interventions. in Digital Therapeutics for Mental Health and Addiction 65–77 (Academic Press, 2023). Mishra, V. et al. Detecting Receptivity for mHealth Interventions in the Natural Environment. Proc. ACM Interact. Mob. Wearable Ubiquitous Technol. 5 , 74:1-74:24 (2021). Tables Table 1. Recruitment channel analysis. Channel Type Impressions Clicks Click through rate (%) Installs Install rate (%) Spend (S$) Cost per install (S$) Google Ad campaign 233,667 2,220 0.95 219 9.87 1,131.31 5.17 Facebook and Instagram Ad campaign 107,528 1,042 0.97 0 N/A 637.87 N/A LinkedIn, Twitter and Facebook Research Centre’s Page Post 904 29 3.21 1 3.45 0.00 0.00 Telegram Channel Posts on #SGResearchLobang and #paidstudiesNTU 4,600 704 15.30 44 6.25 0.00 0.00 E-Mail Research Centre’s Mailing list 220 145 66.00 4 2.76 0.00 0.00 Flyer Distribution University campus (student interns) 2,000 171 8.55 32 18.71 1,244.07 113.10 Flyer Distribution Residential Area (Little Ants) 10,100 5 0.05 0 0 1,668.00 N/A Friend Referral Link sharing unknown 3 N/A 2 66.67 0.00 0.00 Table 2. Participant demographic profile. Total n (%) Age (years) (N=99) 21-35 36-50 51+ Prefer not to say 69 (69.7%) 21 (21.2%) 6 (6%) 3 (3%) Sex (N=99) Female Male Prefer not to say 63 (63.6%) 33 (33.3%) 3 (3%) Ethnicity (N=99) Chinese Indian Malay Other Prefer not to say 74 (74.7%) 6 (6.1%) 2 (2.0%) 12 (12.1%) 5 (5.1%) Education Level (N=99) University Degree Diploma A-level O-level/N-level Prefer not to say 68 (68.7%) 15 (15.1%) 6 (6.1%) 4 (4.0%) 6 (6.1%) Monthly Household Income (S$) (N=63) =11,000 Prefer not to say unknown 6 (10.0%) 17 (27.0%) 14 (23.0%) 7 (11.0%) 16 (25%) 3 (5.0%) Employment Status ( N=63) Employed Student Homemaker Unemployed Retired Prefer not to say 38 (61.0%) 12 (19.0%) 3 (5.0%) 4 (7.0%) 1 (2.0%) 3 (5.0%) Marital Status (N=63) Single Married Divorced or separated Prefer not to say 38 (60.0%) 18 (29.0%) 2 (3.0%) 5 (8.0%) Self-rated health (N=19) Excellent Very Good Good Fair Poor 0 (0.0%) 6 (31.6%) 12 (63.2%) 1 (5.3%) 0 (0.0%) Table 3. Participant vulnerability. Total n (%) Mental Health Vulnerability Score (N=99) No Risk Low Risk Moderate Risk High Risk Very High Risk 20 (20.2%) 22 (22.2%) 40 (40.4%) 9 (9.1%) 8 (8.1%) Diet Vulnerability Score (N=99) No Risk Low Risk Moderate Risk High Risk Very High Risk 56 (56.4%) 43 (43.4%) 0 (0.0%) 0 (0.0%) 0 (0.0%) Physical Activity Vulnerability Score (N=99) No Risk Low Risk Moderate Risk High Risk 62 (62.6%) 8 (8.1%) 5 (5.1%) 24 (24.2%) Additional Declarations Competing interest reported. This study was funded by the National Research Foundation, Prime Minister’s Office, Singapore under its Campus for Research Excellence and Technological Enterprise (CREATE) programme. The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; or decision to submit the manuscript for publication. JLM, TK, OC, and FW are affiliated with the Centre for Digital Health Interventions, a joint initiative of the Institute for Implementation Science in Health Care, University of Zurich; the Department of Management, Technology, and Economics at ETH Zurich; and the Institute of Technology Management and School of Medicine at the University of St Gallen. CDHI is funded in part by CSS, a Swiss health insurer; Mavie Next, an Austrian health insurer; and MTIP, a Swiss digital health investor. However, none of these companies were involved in this study. All other authors declare no potential conflict of interest. Supplementary Files AppendixLvLUPFeasibilityStudyCopy.docx SupplementaryFile1Copy.docx Cite Share Download PDF Status: Published Journal Publication published 22 Dec, 2025 Read the published version in Scientific Reports → Version 1 posted Editorial decision: Revision requested 10 Nov, 2025 Reviews received at journal 07 Jul, 2025 Reviewers agreed at journal 20 Jun, 2025 Reviewers invited by journal 20 May, 2025 Editor assigned by journal 20 May, 2025 Editor invited by journal 28 Apr, 2025 Submission checks completed at journal 28 Apr, 2025 First submitted to journal 19 Apr, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6484027","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":460552124,"identity":"e7b15941-b09a-423e-9ea9-58aa6ff7d9dc","order_by":0,"name":"Jacqueline Louise Mair","email":"data:image/png;base64,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","orcid":"","institution":"Future Health Technologies, Singapore-ETH Centre, Campus for Research Excellence And Technological Enterprise (CREATE)","correspondingAuthor":true,"prefix":"","firstName":"Jacqueline","middleName":"Louise","lastName":"Mair","suffix":""},{"id":460552126,"identity":"acf92a1e-97f3-4843-8575-0a8550484fd9","order_by":1,"name":"Ahmad I. Jabir","email":"","orcid":"","institution":"Future Health Technologies, Singapore-ETH Centre, Campus for Research Excellence And Technological Enterprise (CREATE)","correspondingAuthor":false,"prefix":"","firstName":"Ahmad","middleName":"I.","lastName":"Jabir","suffix":""},{"id":460552130,"identity":"9247fdbd-fd21-409a-b7bb-248a2d8e9990","order_by":2,"name":"Alicia Salamanca-Sanabria","email":"","orcid":"","institution":"Future Health Technologies, Singapore-ETH Centre, Campus for Research Excellence And Technological Enterprise (CREATE)","correspondingAuthor":false,"prefix":"","firstName":"Alicia","middleName":"","lastName":"Salamanca-Sanabria","suffix":""},{"id":460552131,"identity":"697bace1-8c55-4476-86bb-d91e700aa8ab","order_by":3,"name":"Oscar Castro","email":"","orcid":"","institution":"Future Health Technologies, Singapore-ETH Centre, Campus for Research Excellence And Technological Enterprise (CREATE)","correspondingAuthor":false,"prefix":"","firstName":"Oscar","middleName":"","lastName":"Castro","suffix":""},{"id":460552133,"identity":"5fac1726-cff2-45df-9a77-f05ccca5d901","order_by":4,"name":"Shenglin Zheng","email":"","orcid":"","institution":"Future Health Technologies, Singapore-ETH Centre, Campus for Research Excellence And Technological Enterprise (CREATE)","correspondingAuthor":false,"prefix":"","firstName":"Shenglin","middleName":"","lastName":"Zheng","suffix":""},{"id":460552136,"identity":"599924c0-1b13-42bf-a4ba-b41954f5d7fa","order_by":5,"name":"Roman Keller","email":"","orcid":"","institution":"Future Health Technologies, Singapore-ETH Centre, Campus for Research Excellence And Technological Enterprise (CREATE)","correspondingAuthor":false,"prefix":"","firstName":"Roman","middleName":"","lastName":"Keller","suffix":""},{"id":460552139,"identity":"7096873f-7690-4059-961c-4e9b0cdea68f","order_by":6,"name":"Bea Franziska Frese","email":"","orcid":"","institution":"Future Health Technologies, Singapore-ETH Centre, Campus for Research Excellence And Technological Enterprise (CREATE)","correspondingAuthor":false,"prefix":"","firstName":"Bea","middleName":"Franziska","lastName":"Frese","suffix":""},{"id":460552143,"identity":"f8676901-7595-474d-83ae-c6a5aaed0997","order_by":7,"name":"Chiang Siang Lim","email":"","orcid":"","institution":"Future Health Technologies, Singapore-ETH Centre, Campus for Research Excellence And Technological Enterprise (CREATE)","correspondingAuthor":false,"prefix":"","firstName":"Chiang","middleName":"Siang","lastName":"Lim","suffix":""},{"id":460552147,"identity":"86e51b60-bb9f-47aa-959b-333a105332b8","order_by":8,"name":"Aishah Alattas","email":"","orcid":"","institution":"Future Health Technologies, Singapore-ETH Centre, Campus for Research Excellence And Technological Enterprise (CREATE)","correspondingAuthor":false,"prefix":"","firstName":"Aishah","middleName":"","lastName":"Alattas","suffix":""},{"id":460552149,"identity":"ac5a06ea-b114-4bbb-8f85-16fe7377252b","order_by":9,"name":"Samarth Negi","email":"","orcid":"","institution":"Future Health Technologies, Singapore-ETH Centre, Campus for Research Excellence And Technological Enterprise (CREATE)","correspondingAuthor":false,"prefix":"","firstName":"Samarth","middleName":"","lastName":"Negi","suffix":""},{"id":460552153,"identity":"1288f410-c086-4cb3-b635-118ae7046c27","order_by":10,"name":"Akshaye Shenoi","email":"","orcid":"","institution":"Future Health Technologies, Singapore-ETH Centre, Campus for Research Excellence And Technological Enterprise (CREATE)","correspondingAuthor":false,"prefix":"","firstName":"Akshaye","middleName":"","lastName":"Shenoi","suffix":""},{"id":460552156,"identity":"c8abe7d7-a87b-47f6-b211-2b5687f1d4b7","order_by":11,"name":"Rob Martinus van Dam","email":"","orcid":"","institution":"Milken Institute School of Public Health, George Washington University","correspondingAuthor":false,"prefix":"","firstName":"Rob","middleName":"Martinus van","lastName":"Dam","suffix":""},{"id":460552158,"identity":"3e841711-1fad-4306-9a62-c141bbebd009","order_by":12,"name":"E Shyong Tai","email":"","orcid":"","institution":"Yong Loo Lin School of Medicine, National University of Singapore","correspondingAuthor":false,"prefix":"","firstName":"E","middleName":"Shyong","lastName":"Tai","suffix":""},{"id":460552160,"identity":"70154286-1570-443e-b788-e8d820cfeef1","order_by":13,"name":"Elgar Fleisch","email":"","orcid":"","institution":"ETH Zurich","correspondingAuthor":false,"prefix":"","firstName":"Elgar","middleName":"","lastName":"Fleisch","suffix":""},{"id":460552162,"identity":"b75fa091-6e9e-4035-bdf9-ae2bf7919bb1","order_by":14,"name":"Florian von Wangenheim","email":"","orcid":"","institution":"ETH Zurich","correspondingAuthor":false,"prefix":"","firstName":"Florian","middleName":"","lastName":"von Wangenheim","suffix":""},{"id":460552164,"identity":"32d78bfc-fe24-4bb9-87c5-78349cf38820","order_by":15,"name":"Lorainne Tudor Car","email":"","orcid":"","institution":"Lee Kong Chian School of Medicine, Nanyang Technological University Singapore","correspondingAuthor":false,"prefix":"","firstName":"Lorainne","middleName":"Tudor","lastName":"Car","suffix":""},{"id":460552167,"identity":"215833b2-ac7e-4315-81d8-d7b24b253ee1","order_by":16,"name":"Falk Müller-Riemenschneider","email":"","orcid":"","institution":"Saw Swee Hock School of Public Health, National University of Singapore","correspondingAuthor":false,"prefix":"","firstName":"Falk","middleName":"","lastName":"Müller-Riemenschneider","suffix":""},{"id":460552170,"identity":"28c9e149-bca6-42ca-8aa5-de4555fec310","order_by":17,"name":"Tobias Kowatsch","email":"","orcid":"","institution":"University of Zurich","correspondingAuthor":false,"prefix":"","firstName":"Tobias","middleName":"","lastName":"Kowatsch","suffix":""}],"badges":[],"createdAt":"2025-04-19 10:08:09","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6484027/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6484027/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1038/s41598-025-30960-z","type":"published","date":"2025-12-22T15:57:51+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":83373474,"identity":"dbf9161d-1c9c-4592-a63c-681af8e3f647","added_by":"auto","created_at":"2025-05-24 04:30:32","extension":"jpeg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":609402,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eScreenshots of key LvL UP components, including conversational agent personas, cockpit, chat dialogue, life hack, journal, and notifications.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-6484027/v1/86dbbb95c298a26be2addc67.jpeg"},{"id":83373704,"identity":"793d535d-338a-493e-a574-07bacbe6c453","added_by":"auto","created_at":"2025-05-24 04:38:32","extension":"jpeg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":143680,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eLvL UP app component usage since installation.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-6484027/v1/2d103ac899aa4c9c11643670.jpeg"},{"id":83373706,"identity":"b031265d-29f4-4991-8bb9-e4bf24a9cf97","added_by":"auto","created_at":"2025-05-24 04:38:32","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":27418,"visible":true,"origin":"","legend":"\u003cp\u003eNumber of tasks completed per time of day.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eNote. Morning = 6am – 12pm, Afternoon = 12pm – 6pm, Evening = 6pm – 12am, Night = 12am – 6am\u003c/em\u003e\u003c/p\u003e","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-6484027/v1/8cdbd2545a7fa69ddcc6f4d7.png"},{"id":83373481,"identity":"240b9782-0276-40f7-a172-5ba324faea53","added_by":"auto","created_at":"2025-05-24 04:30:32","extension":"jpeg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":129223,"visible":true,"origin":"","legend":"\u003cp\u003eNumber of tasks completed per day of the week.\u003c/p\u003e","description":"","filename":"floatimage4.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-6484027/v1/7d313e2d1eea62f6196d885a.jpeg"},{"id":99172469,"identity":"41d9e656-ff77-4373-86f6-cf2440b4cdd8","added_by":"auto","created_at":"2025-12-29 16:09:58","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1913013,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6484027/v1/0d0d71bf-81c1-411b-95f0-ac4fbd06d47c.pdf"},{"id":83373479,"identity":"068e9d87-2762-4720-afd5-3c8a6dd8983b","added_by":"auto","created_at":"2025-05-24 04:30:32","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":1966302,"visible":true,"origin":"","legend":"","description":"","filename":"AppendixLvLUPFeasibilityStudyCopy.docx","url":"https://assets-eu.researchsquare.com/files/rs-6484027/v1/179ed8fa717e14838dcf8bfe.docx"},{"id":83373477,"identity":"97c71281-1442-4e9f-8042-7784de0a8f01","added_by":"auto","created_at":"2025-05-24 04:30:32","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":36136,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryFile1Copy.docx","url":"https://assets-eu.researchsquare.com/files/rs-6484027/v1/ae65b03ab21be558089d544a.docx"}],"financialInterests":"Competing interest reported. This study was funded by the National Research Foundation, Prime Minister’s Office, Singapore under its Campus for Research Excellence and Technological Enterprise (CREATE) programme. The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; or decision to submit the manuscript for publication.\n\nJLM, TK, OC, and FW are affiliated with the Centre for Digital Health Interventions, a joint initiative of the Institute for Implementation Science in Health Care, University of Zurich; the Department of Management, Technology, and Economics at ETH Zurich; and the Institute of Technology Management and School of Medicine at the University of St Gallen. CDHI is funded in part by CSS, a Swiss health insurer; Mavie Next, an Austrian health insurer; and MTIP, a Swiss digital health investor. However, none of these companies were involved in this study. All other authors declare no potential conflict of interest.","formattedTitle":"LvL UP 1.0, a holistic mHealth lifestyle coaching intervention for the prevention of non-communicable diseases and common mental disorders: a mixed methods feasibility study","fulltext":[{"header":"Background","content":"\u003cp\u003eThe growing burden of non-communicable diseases (NCDs) and common mental disorders (CMDs), along with their associated risk factors, contribute significantly to escalating healthcare and socio-economic costs worldwide [\u003cspan additionalcitationids=\"CR2\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. Addressing the prevention, management, and treatment of NCDs and CMDs requires a paradigm shift toward holistic interventions that prioritise health-promoting behaviours, mental well-being, and overall wellness [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. However, the initiation and long-term maintenance of health-promoting behaviour change remains challenging and is often adopted by only a fraction of those in need [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eLifestyle coaching, which focuses on enhancing behaviours related to physical activity, nutrition, stress management, and sleep, has demonstrated its potential to improve health outcomes and reduce chronic disease risks [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. Yet, traditional delivery models, typically involving in-person or group-based sessions with a single practitioner, are time- and resource-intensive, limiting accessibility and scalability [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eTo address these challenges, innovative delivery methods are required, particularly those that can overcome barriers related to cost, accessibility, and sustained engagement. Advances in mobile health (mHealth) technologies and artificial intelligence present new opportunities for scalable and user-friendly interventions. Conversational agents (CAs), or chatbots, have emerged as promising tools for delivering tailored health behaviour support [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]. With their ability to engage in interactive, real-time dialogue, CAs can mimic human coaching interactions, delivering tailored advice, reminders, and motivation while being available 24/7. Despite their potential, the use of CAs in holistic lifestyle coaching, that is, addressing multiple interconnected domains related to physical and mental health, is still relatively underexplored [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. Moreover, few CA interventions are culturally tailored to Asian population [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eIn response to this need, we have undertaken a rigorous theory-informed and user-centred intervention development process to create \u0026ldquo;LvL UP\u0026rdquo; [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e], an app-based holistic lifestyle coaching intervention designed specifically for the Singapore context. The intervention employs a \u0026lsquo;talk and tools\u0026rsquo; approach [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e] to deliver health literacy coaching sessions via an automated CA focused on three key areas\u0026mdash;physical activity, diet, and mental well-being\u0026mdash;which are complemented by behavioural tools, including a journal, life hacks, and a slow-paced breathing training game.\u003c/p\u003e \u003cp\u003eThe LvL UP intervention is set in Singapore, a highly urbanised and densely populated city-state in Southeast Asia, home to a culturally diverse population comprising mainly Chinese, Malay, and Indian ethnic groups. As a nation facing rising rates of NCDs [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e] and CMDs [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e], Singapore presents a compelling case for preventive health interventions, particularly among young adults. The country\u0026rsquo;s advanced digital infrastructure, high smartphone penetration, and strong government support for digital adoption make it an ideal setting for deploying mHealth interventions.\u003c/p\u003e \u003cp\u003eThis study aimed to assess the technical feasibility, user satisfaction, and recruitment capability of the LvL UP intervention in a real-world setting, with the goal of informing future trials and implementation strategies.\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003e\u003cstrong\u003eIntervention Development\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe LvL UP intervention has been developed as part of a 5-year programme of research under the Future Health Technologies Programme in Singapore. An iterative and user-centred development process has been carried out in line with the multiphase optimization strategy [13] and the UK Medical Research Council guidelines for developing and evaluating complex interventions [14].\u0026nbsp;Based on findings from initial market analyses\u003csup\u003e\u0026nbsp;\u003c/sup\u003e[15,16], evidence syntheses [5,17,18] and qualitative studies with potential stakeholders [19,20], a behavioural diagnosis was conducted using the Capability, Opportunity, Motivation, Behaviour (COM-B) model of behaviour change [21] to develop a conceptual model underpinning the intervention [9]. We then engaged in a user-centred prototyping process to develop LvL UP version 1.0; a smartphone app containing a motivational interviewing-inspired CA to act as a \u0026lsquo;digital lifestyle coach\u0026rsquo;, \u0026lsquo;life hacks\u0026rsquo; to encourage healthy habit formation, and \u0026lsquo;tools\u0026rsquo; supported by a transdiagnostic cognitive behavioural therapy approach including a slow-paced breathing training game and a journaling tool to promote emotional literacy. Four different CA personas were developed to represent typical Singaporean lifestyles, and users could choose one to communicate with throughout the intervention (Figure 1). Additionally, gamification and storytelling features were built into the intervention design to encourage sustained engagement with the app. This included rewards for task completion and unlocking new features as the intervention progresses. Each of these components is described in detail elsewhere [9]. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eIn brief, the CA was built using an open-source platform called MobileCoach (www.mobile-coach.eu) [22] that operates rule-based with scripted sessions and pre-defined answer options developed by domain experts of the research team. This ensures the fidelity and safety of the intervention content in the health context and has been used in similar interventions previously\u003csup\u003e\u0026nbsp;\u003c/sup\u003e[23-26]. Coaching sessions deliver evidence-based psychoeducational content with a motivational interviewing-inspired communication strategy focused on three lifestyle behaviour pillars:\u003c/p\u003e\n\u003cul type=\"disc\"\u003e\n \u003cli\u003eMove More: Increasing physical activity (e.g., daily step count), exercise (home-based training), reducing sitting time/screen-based activity.\u003c/li\u003e\n \u003cli\u003eEat Well: Encouraging a nutrient-rich diet, reduced snacking, food purchasing and meal preparation suggestions, links between diet and well-being.\u0026nbsp;\u003c/li\u003e\n \u003cli\u003eStress Less: Activities based on Cognitive Behavioural Therapy (CBT) evidence, including emotional regulation, behavioural activation, cognitive flexibility, and problem-solving therapy.\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003eLife hacks are actionable health and well-being-related tips, delivered either in the morning, at lunchtime or in the evening, that can be easily implemented into the daily routine. They serve to (i) ensure a holistic approach to improving health-related behaviours regardless of the coaching pillar chosen by the participant; (ii) target improvements in self-efficacy; and (iii) facilitate routine during everyday life to encourage positive habit formation [27]. Breeze is an intervention component that uses the smartphone\u0026rsquo;s microphone to continuously detect breathing phases in semi-real time (i.e., inhale, exhale, and pause between inhale and exhale), which are then used to trigger a gamified biofeedback-guided breathing training [28,29]. Gamified biofeedback is shown as a sailing boat moving down a river which speeds up when the user performs slow-paced breathing. This targets experiential outcomes in addition to instrumental outcomes of psychological well-being and heart rate variability [30]. The free-text Journal tool allows users to reflect on their thoughts, feelings, behaviours, and learnings throughout the intervention and report moments of gratitude to help them reduce stress and solve problems [31]. All journal entries were stored on the device and, therefore, remained strictly confidential to the user, with only the time stamp and mood rating assigned to each journal entry being recorded on the server. Users could view and edit their journal entries at any time.\u003c/p\u003e\n\u003cp\u003eIn addition to these therapeutic components, gamification, storytelling, and just-in-time intervention notifications were included as engagement strategies.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eLvL UP 1.0 is intended to be used daily for 21 days, with users completing one coaching session per day, after which continued use is optional. Progress was monitored through the LvL UP Shield on the app\u0026rsquo;s home screen. Each time users completed a task in the app they received a puzzle piece to fill the LvL UP Shield. Tasks were split across four categories: Coaching Sessions, Life Hacks, Tools (Breeze or Journal), and Basics (intermediary dialogues and in-app surveys to gather evaluation data). When the shield was filled with the required puzzle pieces, users would unlock an episode from the LvL UP Storytelling approach and progress to the next level with a new shield to fill. \u0026nbsp;In total, there were three levels (See Supplementary File).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eStudy Design\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe tested the feasibility and acceptability of LvL UP 1.0 in a one-group, in-the-wild study using a mixed-methods approach. The study was conducted entirely remotely in Singapore between March and August 2023. As this was a feasibility study, a formal sample size calculation was not appropriate [32]. To ensure saturation in identifying usability issues, a minimum of 20 participants is typically required based on heuristics in usability engineering [33], but to account for the retention challenges common in remote digital health studies [34], we aimed to recruit 200 participants. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthical approval\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe Institutional Review Board of the National University of Singapore (NUS-IRB-2021-623) and the Ethics Commission of ETH Zurich (EK-2021-N-109) approved the study. The study was conducted in accordance with the Declaration of Helsinki and all participants provided written informed consent.\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eParticipant Recruitment\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eIn an attempt to address diversity issues in digital health research [35,36], the recruitment strategy aimed to reach a demographically diverse sample of adults. Both offline and online approaches were employed to maximise outreach. Offline efforts included door-to-door and on-street flyer distribution in residential areas, complemented by six student interns distributing flyers in high-footfall locations around the university campus over a 10-day period. Online recruitment leveraged various channels, including posts on Telegram groups dedicated to research participation, electronic direct mail (EDM) marketing and social media posts via the research centre\u0026rsquo;s mailing list and social platforms, and ad campaigns through a dedicated LvL UP social media page. Social media ad campaigns on Meta and Google Ads began in March 2023 for iOS and expanded to both iOS and Android from May to July 2023. Campaigns targeted English-speaking adults aged 21 and older living in Singapore. Initial ad variations were tested using Meta\u0026rsquo;s A/B testing feature, comparing a video post and a static image with positive framing (\u0026ldquo;Do you want to improve your well-being?\u0026rdquo;) against negative framing (\u0026ldquo;Are you struggling with your mental and physical well-being?\u0026rdquo;) (see Supplementary File). Over seven days, the positively framed ads outperformed the negatively framed ones, with a lower cost per click (S$0.34 vs. S$0.41), prompting their use for the remainder of the campaign. Additionally, a project website provided detailed information about the LvL UP app, the research program and study team, and a link to request an offline version of the LvL UP life hacks and journal template. To further extend reach, a referral link within the app enabled participants to invite friends and colleagues to download and engage with the app.\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eProcedure\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eApp Onboarding\u0026nbsp;\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eOn opening the app, participants were prompted to register with an email address and complete an e-consent to participate in the study. They then proceeded to choose a digital coach (CA); one of four coaching personas designed to represent the target group of young and middle-aged adults (see Supplementary File). Participants completed a \u0026lsquo;Welcome Day\u0026rsquo; dialogue which provided an overview of the goals and intended use of the intervention, a vulnerability assessment, and a shared decision-making process on which pillar (Move More, Eat Well, or Stress Less) to begin with. When the vulnerability assessment identified someone as having severe depression or anxiety (PHQ-4 score \u0026gt;9) they were advised to seek mental health resources in Singapore (e.g., SOS Samaritans) and face-to-face counselling from a health professional. They were not prohibited from using the app or continuing with the study. Still, they were advised that the LvL UP intervention is designed to prevent disease and not treat a pre-existing condition.\u0026nbsp;After completing the \u0026lsquo;Welcome Day\u0026rsquo; dialogue, participants were free to use the app as they wished, although they were prompted to complete a coaching session and a life hack daily through the app\u0026rsquo;s home screen and in-app notifications.\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eOutcome measures\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSocio-demographic characteristics, including age group, gender, ethnicity, educational level, marital status, and household income were obtained via in-app questionnaires. Health status was also measured by the single-item Self-Rated Health Index [37] and the World Health Organization-Five Well-Being Index (WHO-5) [38]. The health vulnerability assessment included the Patient Health Questionnaire-4 (PHQ-4) for symptoms of depression and anxiety [39], the International Physical Activity Questionnaire (short form) [40] and a dietary guideline adherence questionnaire developed for this study based on Singapore\u0026rsquo;s My Healthy Plate (see Supplementary File).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe efficacy of each recruitment channel was assessed using dynamic link tracking and QR codes in all promotional materials.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe feasibility of the LvL UP app was assessed by (i) user engagement in terms of app usage (i.e., response to notifications, timestamps and duration of app component use, and app usage patterns), (ii) session alliance with the CA, self-reported on a six-point Likert scale anchored from 0 (\u0026lsquo;not at all\u0026rsquo;) to 5 (\u0026lsquo;completely) [41], and (iii) technology acceptance, including perceived enjoyment, ease of use, usefulness, and control, self-reported on a six-point Likert scale anchored from 0 (\u0026lsquo;strongly disagree\u0026rsquo;) to 5 (\u0026lsquo;strongly agree\u0026rsquo;) [42] (see Supplementary File).\u003c/p\u003e\n\u003cp\u003eUser satisfaction with LvL UP was assessed using a short feedback survey containing two open-response questions that asked participants what they liked about the app and what needed to be improved. Additionally, the cultural relevance of the app content was assessed across three dimensions of equivalence: functional (whether it contains familiar behaviours, relevant cultural context, and culturally tailored goals and examples), conceptual (whether the underlying ideas and expressions are meaningful to the target. group),\u0026nbsp;and linguistic (whether the language, including regional variations, idioms, and slang, is appropriate) [43]. These were rated on a six-point Likert scale ranging from \u0026lsquo;not at all\u0026rsquo; (0) to \u0026lsquo;completely\u0026rsquo; (5).\u0026nbsp;For those completing the short feedback survey, an option to participate in an interview\u0026nbsp;was offered. Depending on participant preference, interviews were conducted virtually using Zoom or WhatsApp chat by two members of the research team (AIJ and BF). The topic guide for the interview focused on motivation to use the app, experience using the app\u0026rsquo;s features, and perceived impact of the app. It was piloted with members of the study team and consisted of semi-structured, open-ended questions, with more specific probes used to solicit additional information (funnelling technique).\u003c/p\u003e\n\u003cp\u003eAfter completing at least one coaching session, all self-report surveys were available in the app and tied to the LvL UP Shield progress to encourage completion. Participants were reimbursed with a S$5 shopping e-voucher upon completing the short feedback survey and a S$20 shopping e-voucher for interviews.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData Analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe used Firebase Analytics to analyse dynamic links and summarise installs from each recruitment channel. Data from the Google and Meta Ads Manager were used to summarise ad performance including impressions, click through rate, cost per click, link clicks, and ad spend.\u003c/p\u003e\n\u003cp\u003eVulnerability assessment cut-off scores were established as follows: PHQ-4 score = 3 (indicating symptoms of anxiety and depression); IPAQ-short form total physical activity in mins per week = \u0026lt; 120 (below physical activity guidelines); dietary guideline adherence score = 18 (not consuming recommended portions of key food groups).\u003c/p\u003e\n\u003cp\u003eApp usage data were analysed using R (version 4.4.2) and Jupyter Notebook. Quantitative survey data were summarised using descriptive statistics and open-ended responses were coded and analysed together with the interview data by one researcher (JLM). Data are reported as mean \u0026plusmn; standard deviation throughout unless otherwise specified.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eInterviews were automatically transcribed verbatim by the Zoom platform or downloaded directly from WhatsApp. Two members of the research team (AIJ and XL) reviewed the transcripts for errors and de-identified the files. Transcripts were then uploaded to Atlas.ti (V.9) for analysis. Template analysis, a form of thematic analysis, was used to analyse the data [44]. The initial template (codebook) was defined \u003cem\u003ea priori\u003c/em\u003e and covered four key domains based on our research aims (i.e. to improve the LvL UP app): (i) participants\u0026rsquo; perceptions of the app, including coaching sessions, life hacks, the journal, Breeze, storytelling (animations), gamification (Shield), in-app notifications, and overall look and feel; (ii) suggestions for improvement; (iii) barriers and facilitators to app use; and (iv) perceived impact on health and behaviour. To ensure rigor in the coding process, two researchers (AIJ and SZ) independently coded an equal subset of transcripts using a deductive approach. Each researcher maintained a reflective notebook to document emerging codes and analytical insights. Following independent coding, they reviewed each other\u0026rsquo;s transcripts, compared codes, and discussed any discrepancies or additional codes that emerged beyond the initial template. Disagreements in coding or code definitions were resolved through discussion, with a third reviewer (JLM) acting as an arbiter when consensus could not be reached. The codebook was iteratively refined until all coders reached an agreement. The template was then applied to the open-ended survey responses by JLM. The final template covered three overarching topics: (i) app components, including coaching sessions, life hacks, Breeze, and the journal; (ii) engagement strategies; and (iii) overall user experience.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003e\u003cstrong\u003eReach\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe average click through rate across all social media ad campaigns was approximately 1%, which is considered reasonable for social media news feed ads [45]. While EDM achieved the highest click through rate, friend referrals proved most effective in converting clicks into app installs (Table 1). A technical issue with link tracking on Meta prevented the accurate tracking of app installs through that platform.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eRecruitment\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eIn total, 307 individuals downloaded the app, 249 opened the app and registered an email address, 215 verified their email address, 204 started the welcome dialogue, 105 completed the vulnerability assessment, and 99 (herein defined as \u0026lsquo;active users\u0026rsquo;) completed the \u0026lsquo;Welcome Day\u0026rsquo; dialogue to unlock the coaching sessions. The recruitment cost per install was S$15.92 and per active user was approximately S$47.29. The main reasons people download the app were (1) to see what the app was about (n=52; 41.6%), (2) to improve overall well-being (n=36, 28.8%), (3) to lose weight (n=13; 10.4%), (4) to learn about healthy living (n=7; 5.6%), (5) to get more active (n=5; 4.0%), (6) to improve diet (n=3; 2.4%) and (7) to manage emotions (n=3; 2,4%).\u003c/p\u003e\n\u003cp\u003eThe active users were mostly female, aged 21-35 years, of Chinese ethnicity, single, educated to university degree level, and in employment (Table 2), with mild to moderate health vulnerabilities in terms of physical activity, diet, and mental well-being (Table 3). Of the 19 active users completing the health status survey, average mental well-being, (measured by the WHO-5) was rated 16.76 out of 25 (67.05%) and the majority self-rated their health as good (Table 2).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTechnical Feasibility\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTechnology acceptance was rated good overall, with most people agreeing the app was enjoyable, easy to use, and provided useful information (n=31; mean \u0026plusmn; SD; 3.73 \u0026plusmn; 1.07). The cultural relevance of the app content was rated moderate (n=10; functional equivalence 3.6 \u0026plusmn; 0.97; conceptual equivalence 3.6 \u0026plusmn; 0.97; and linguistic equivalence 3.5 \u0026plusmn; 1.18) and users\u0026apos; session alliance ratings with the CA were good (n=19; 3.9 \u0026plusmn; 1.22).\u003c/p\u003e\n\u003cp\u003eDaily engagement with the LvL UP app features (coaching sessions, life hacks, journal, and Breeze) since installation is summarised in Figure 2. Usage was highest in the first eight days\u003cem\u003e.\u0026nbsp;\u003c/em\u003eBy week two, 77.8% (77/99) of the initial users were lost to attrition. Of those who spent more than one day on the app, they completed an app activity approximately every three days (3.21\u0026nbsp;\u0026plusmn;\u0026nbsp;7.23 days). Users were most active in the afternoons (Figure 3) and on Tuesdays (Figure 4).\u003cem\u003e\u0026nbsp;\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eOn average, active users completed three coaching sessions (mean \u0026plusmn; SD; 3.35 \u0026plusmn; 4.53), five life hacks (4.92 \u0026plusmn; 6.52), two journal entries (2.34 \u0026plusmn; 2.30) and four slow-paced breathing training sessions (4.29 \u0026plusmn; 4.09). Active users spent between 1 and 28 days (4.07 \u0026plusmn; 5.02 days) completing at least one task on the app and 35.4% (35/99) were active for less than one day. Among the active users, 9.1% (9/99) completed Level 1, 4.0% (4/99) progressed to Level 2, and 3.0% (3/99) completed the intervention. Only one user managed to complete the intervention within 21 days. Five users (5%) were considered super users, engaging in the intervention regularly for 21 days or longer.\u003c/p\u003e\n\u003cp\u003eA total of 405 coaching session notifications were successfully delivered. Of those, 77.8% (315/405) were responded to, and 28.9% (117/405) were within 10 minutes (1.53 \u0026plusmn; 2.41 min). The median response time between notification delivery and observed coaching session engagement was 57 minutes. However, the notification response rate could be higher as not all activities on the app were tracked (i.e. the time user opened the app) due to system limitations.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcceptability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFifty participants completed the short feedback survey, and 14 interviews were conducted (n=11 on Zoom and n=3 on WhatsApp chat). The interviews took, on average, 33 minutes (15.8 \u0026ndash; 50.0 min). Feedback was centred around the positive and negative aspects of (i) the app components, including the coaching sessions, life hacks, Breeze, and journal, (ii) engagement strategies, and (iii) overall user experience.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eLvL UP intervention and app components\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eMany participants had positive experiences with the CA and the coaching sessions, describing the content as helpful, comprehensive, and interactive. Some mentioned that the intervention provided support and served as a reminder to make healthier choices in everyday life:\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e\u0026ldquo;It had a lot of tips on how to eat well, and not just like cooking wise. I was quite impressed with the social situation one, because yeah, I did really think about how the social situation would affect our eating habits a lot. And then in particular\u0026hellip;communicating how to say no in a way that doesn\u0026apos;t affect the other party.\u0026rdquo; [Interviewer: like, how to reject the food?] \u0026ldquo;Yeah\u0026hellip;that stood out for me\u003c/em\u003e.\u0026rdquo; (ID09, female, 21-35 years old)\u003c/p\u003e\n\u003cp\u003e\u0026ldquo;\u003cem\u003eI think LvL UP serves as an accountability partner in terms of your well-being journey because you are required to check in every day. And if you don\u0026apos;t then [it will] give you a notification. So, I think there is a good reminder that you have to be mindful of these new habits to incorporate\u0026nbsp;and learn new things every day.\u0026rdquo;\u003c/em\u003e (ID08, female, 21-35 years old)\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e\u0026rdquo;It helps that I don\u0026apos;t feel alone when I don\u0026apos;t have a friend to work out with. And I don\u0026apos;t get judged by others. I felt I looked forward to communicating with Kai [CA] again and he is very personal despite it being a chatbot only.\u0026rdquo;\u0026nbsp;\u003c/em\u003e(ID12, female, 36-50 years old)\u003c/p\u003e\n\u003cp\u003eSome felt the content was personalised which made it more motivating, while others felt the content could be more relevant to their needs.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e\u0026rdquo;I like the chatbot, it makes it personalised and creates more motivation with the regular check ins and all. The exercises and concepts discussed via chatbot are comprehensive and second only to live discussions with coaches, I feel.\u0026ldquo;\u0026nbsp;\u003c/em\u003e(ID13, female, 21-35 years old)\u003c/p\u003e\n\u003cp\u003eOn the other hand, many participants mentioned that the coaching sessions were too long and responses from the CA were too slow, and that daily coaching may be too demanding. Additionally, elements of the motivational interviewing style of coaching did not seem to work well in a chat-based format, for example, asking for permission before progressing with advice.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e\u0026ldquo;\u0026hellip;.they [CA] would pose a question and then I\u0026apos;ll have to click \u0026lsquo;yes, I want to hear more\u0026rsquo;, before they put the next topic in. Well, personally, I feel that if you just put the topic itself and carry on talking, without putting in the random interjecting statements like, \u0026lsquo;yes, I want to hear more\u0026rsquo;, it\u0026apos;s a bit easier for me to process the information because it comes to me directly. I do not need to keep on clicking the button\u003c/em\u003e.\u0026rdquo; (ID07, male, 21-35 years old)\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e\u0026ldquo;I think it\u0026apos;s really cool that there\u0026apos;s the badges that you get to earn. The icon that I was talking about when you complete an activity. So, you get to see your progress. That is something that resonates with me because I\u0026apos;m somebody that really likes to see gamification integrated into whatever I\u0026apos;m doing, because it just makes the overall experience much more enjoyable, much more fun.\u0026rdquo; (ID02, male, 21-35 years old)\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eEngagement strategies\u0026nbsp;\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThere was some appreciation for the storytelling approach which participants found interesting and potentially effective in keeping people engaged, but it was also perceived as time consuming.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e\u0026ldquo;\u0026hellip;the storyline thing\u0026hellip;the characters that you can choose, and they have the daily check-in with you. So, it feels like a game but not really a game where you can\u0026nbsp;follow the character\u0026rsquo;s development\u003c/em\u003e\u0026hellip;\u0026rdquo; (ID09, female, 21-35 years old)\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e\u0026ldquo;LvL UP tries to give you a story. I mean, it\u0026apos;s not bad lah, the story, because maybe you identify with a certain character like the lady, Dana, who drinks bubble tea and whatever, right? But when I\u0026apos;m pressed for time, then it becomes very irritating.\u0026rdquo;\u0026nbsp;\u003c/em\u003e(ID06, female, 36-50 years old)\u003c/p\u003e\n\u003cp\u003eMany participants talked about the need for incentives and rewards, tied to the gamification features of the app, to encourage its continued use as well as motivate health behaviour change.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e\u0026ldquo;I think incentive-based health care provision is actually quite effective. Like anecdotally, I\u0026apos;ve seen people going to lengths walking their 10,000 steps a day to clock their Healthy365 [step counting app] steps, you know, so that they can get rewards. Yeah, so I do think that the use of incentives might motivate people to pursue a healthier lifestyle\u003c/em\u003e.\u0026rdquo; (ID07, male, 21-35 years old)\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e\u0026ldquo;Perhaps once you manage to accumulate all these icons, then you get some sort of monetary incentives or something like that. I think that would really incentivize people to want to do the tasks.\u0026rdquo;\u003c/em\u003e (ID02, male, 21-35 years old)\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e\u0026ldquo;Maybe you can connect with other partners like merchants to give discount vouchers, or like you can collaborate with I don\u0026apos;t know, like speaker or gym source to provide more services or something. (\u003c/em\u003eID08, female, 21-35 years old)\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eOverall user experience\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThe LvL UP app was generally described as easy to use although some participants mentioned that there were too many components to the app and the levelling up criteria was not always clear. There was a preference for straightforward functionality, the option to tailor settings in the app (such as timing of coaching sessions), and an app that required minimal time investment.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e\u0026ldquo;The ability to schedule our coaching sessions was a good move to help integrate it into my lifestyle.\u0026rdquo;\u0026nbsp;\u003c/em\u003e(ID14, male, 21-35 years old)\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e\u0026ldquo;\u0026hellip;because Lumihealth is kind of like touch and go once you see the quest, then you pretty much know what to do. As for LvL UP, for the most part I would have to stay on the app for maybe let\u0026apos;s say, 10 to 15 min before I get the task done.\u0026rdquo;\u0026nbsp;\u003c/em\u003e(ID02, male, 21-35 years old)\u003c/p\u003e\n\u003cp\u003eRegarding interactions with the CA, the predefined answer options were generally appreciated from an ease of use perspective, however participants would prefer more autonomy over their responses, through either more diverse answer options or the ability to type their own response.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e\u0026ldquo;Offer more options to disagree with the coach. I\u0026apos;m going through the Move More pillar, and there have been various instances where I had to select an option equivalent to \u0026lsquo;Yes\u0026rsquo; of varying degrees to agree with the chatbot because there is no negative option, even though I do not agree with the statement put forth. The conversation then ends up feeling shallow, like not true to self or lying.\u0026rdquo; (WA03, female, 36-50 years old)\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e\u0026ldquo;But sometimes I think I am just very lazy to type, lah. That\u0026apos;s good to have an option. But I think it can be another option to let you type yourself\u0026rdquo;. [Interviewer: So, having a variety of options to respond to the coach is something that you would prefer. Am I right to say that?] \u0026ldquo;Yeah, correct.\u0026rdquo; (ID05, female, 21-35 years old)\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eFeedback on the user interface was mixed; some liked the colourful look and feel while others said a cleaner interface would be preferable. The animated design of the digital coaches and the story were sometimes described as \u0026ldquo;childish\u0026rdquo;, with one participant suggesting that videos of real people would be preferred.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThere was an expectation that the app would better integrate with existing apps and devices for a more personalised experience.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e\u0026ldquo;Also, consider syncing with wearable devices, allow data sharing or integrate with health-related apps to have a comprehensive experience. (ID15, female, 36-50 years old)\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eFinally, suggested improvements were related to usability (improving the registration process and the speed of the app, fixing bugs and app crashes).\u0026nbsp;\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eThis study provides insights into the feasibility and acceptability of the LvL UP 1.0 intervention delivered to adults in Singapore. The findings align with broader trends in mHealth research, highlighting both the potential and challenges of digital behavioural interventions, particularly those delivered via CAs. LvL UP 1.0 was technically feasible; all intervention components were successfully delivered, technology acceptance and session alliance were rated positively, and notifications were delivered as planned. However, active users were not representative of the broader target population, and engagement beyond eight days was low, possibly due to shortcomings in the overall user experience.\u003c/p\u003e\n\u003cp\u003eRecruitment and engagement strategies play a crucial role in the success of remotely delivered mHealth interventions. In line with previous studies, we found that positively framed health messages were more effective in promoting engagement than fear-based messaging [45]. The overall click through rate of approximately 1% was consistent with industry standards for social media advertising [46], but conversion from advertisement clicks to active app users remained a challenge. Friend referrals were the most effective in converting interest into active participation, supporting existing evidence that peer recommendations enhance trust and motivation in digital health adoption [36].\u003c/p\u003e\n\u003cp\u003eThe demographic profile of active users in this study suggests that LvL UP appealed primarily to young, well-educated, employed women with mild to moderate health vulnerabilities. This is consistent with broader digital health trends and underscores the ongoing challenge of reaching individuals from lower socioeconomic backgrounds or ethnic minority groups [47,48], who may stand to benefit the most from such interventions. Expanding recruitment strategies to include community-based outreach and partnerships with trusted local organizations could enhance engagement among those at higher risk of developing NCDs or CMDs but who may not (yet) be actively seeking support [36]. Additionally, while LvL UP attracted initial interest, the attrition rate\u0026mdash;from app downloads to active participation\u0026mdash;suggests potential barriers during early engagement, including digital literacy, perceived relevance, and trust [36]. The application of user-centered design principles in LvL UP provides a strong foundation, and future iterations can build on this by incorporating additional accessibility and culturally adapted features to ensure inclusivity and sustained engagement.\u003c/p\u003e\n\u003cp\u003eChallenges with uptake and attrition may have been exacerbated by the study\u0026rsquo;s fully remote, \u0026lsquo;in-the-wild\u0026rsquo; design, where users received no researcher-led guidance beyond the app\u0026rsquo;s welcome dialogue. While this approach enhanced external validity by simulating a real-world app launch, sustained usage was low, despite the reported positive experiences with the LvL UP content. Although the usage patterns align with trends seen in other app-based [49]\u0026nbsp;and CA interventions [50], the drop in adherence presents challenges for evaluation the therapeutic effectiveness of remotely delivered interventions. User experience, gamification, progress tracking, and rewards are key features that can enhance user retention in mHealth interventions [17,36]. Therefore, future iterations of LvL UP could explore simplifying navigation, and offering clearer onboarding guidance, feedback on activities, and achievement-based rewards, to help improve usability and retention.\u003c/p\u003e\n\u003cp\u003eSolving the problem of attrition in mHealth is an ongoing effort. However, findings from this study point to three potential solutions: (i) access to human support for those who need it, (ii) more personalised support to improve therapeutic alliance, and (iii) delivering relevant intervention support at the most optimal moments.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFirstly, adherence to mHealth interventions improves when usage expectations are clearly communicated, and human support is available [17]. Participants in this study emphasized the importance of a straightforward user experience in creating positive first impressions and the need for clear guidance on the intended use of the intervention. Offering a human connection during onboarding and throughout the intervention may foster supportive accountability [51] and sustained engagement. Embedding tools like LvL UP within existing healthcare or community support systems, such as primary care, mental health services, or peer-led community programmes, could strengthen this accountability while offering additional opportunities for engagement. For example, a health professional or community worker introducing the app, guiding the initial setup, and providing periodic encouragement may bridge the gap between digital and human support. This blended model could enhance retention, particularly for users with greater needs or lower digital confidence. Nevertheless, relying on human involvement at scale may be impractical, and further research into optimising when and how to provide such support within mHealth interventions is needed [52].\u003c/p\u003e\n\u003cp\u003eSecondly, as shown elsewhere, there was a clear expectation and appreciation for personalised health behaviour support [17,19]. The CA in the LvL UP app is based on a rule-based system whereby\u0026nbsp;coaching sessions are predefined based on evidence-based content, and personalization is achieved through branching. While this approach ensures fidelity of the content and offers a more interactive delivery of health literacy support, branching capability is limited and it is impossible to adapt content to an individual\u0026rsquo;s changing needs. Furthermore, delivering motivational interviewing in a chat-based format with pre-defined answer options is challenging because it relies on reflecting one\u0026rsquo;s own words. In this study, attempts to \u0026ldquo;ask, offer, ask\u0026rdquo; when providing advice or guidance were also perceived as annoying. Moving forward, strategies employing large language models to personalise the interaction further using more natural language while adhering to the spirit and principles of MI may offer a solution [53,54].\u003c/p\u003e\n\u003cp\u003eThirdly, the\u0026nbsp;engagement data suggest that LvL UP was most frequently used in the afternoons and on Tuesdays, which could inform future \u0026lsquo;just-in-time\u0026rsquo; strategies by tailoring intervention delivery to peak usage periods when people are most receptive [55]. Recent research has shown the potential for machine learning approaches to predict these moments based on features such as device usage, phone battery level, location, and physical activity [56]. However, more research is needed in the context of complex health behaviour interventions such as LvL UP.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eRecommendations for future work\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThe findings from this study highlight several key considerations for future iterations of LvL UP:\u0026nbsp;\u003c/p\u003e\n\u003col\u003e\n \u003cli\u003eExpand and diversify recruitment strategies by complementing social media advertising with peer referral systems, community-based outreach, and partnerships with trusted local organisations. Clearer value propositions tailored to diverse user groups may also improve conversion from interest to active use.\u003c/li\u003e\n \u003cli\u003eEmbed LvL UP within existing care and community ecosystems to enhance supportive accountability and engagement. Future iterations could explore blended models where LvL UP is introduced and supported by healthcare providers, community health workers, or peer mentors.\u003c/li\u003e\n \u003cli\u003eLeverage machine learning and just-in-time adaptive interventions to deliver support during periods of peak user receptivity, informed by real-world usage data and contextual cues such as time of day, activity level, and device interaction.\u003c/li\u003e\n \u003cli\u003eEnhance the CA experience by improving dialogue flow and incorporating hybrid interaction models\u0026mdash;combining structured choices with optional free-text input\u0026mdash;to allow more natural, personalised, and engaging conversations.\u003c/li\u003e\n \u003cli\u003eRefine onboarding and retention strategies to create good first impressions by streamlining the initial user journey and introducing elements such as progress tracking, social or peer-based accountability, and achievement-based rewards to promote sustained engagement.\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Conclusion","content":"\u003cp\u003eWhile LvL UP demonstrated feasibility and acceptability, engagement challenges highlight the need for further optimization. The findings align with broader trends in mHealth research, reinforcing the importance of personalization, human coaching support and accountability, gamification, and rewards in sustaining long-term engagement. Future work should explore how to balance scalability with tailored support, leveraging emerging technologies such as large language models to enhance personalisation and engagement, and subsequently, intervention effectiveness.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAuthor Contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eJLM \u0026ndash; conceptualisation, methodology, formal analysis, investigation, writing-original draft, project administration\u003c/p\u003e\n\u003cp\u003eAS - conceptualisation, methodology, writing-reviewing \u0026amp; editing\u003c/p\u003e\n\u003cp\u003eOC \u0026ndash; methodology, investigation, writing-reviewing \u0026amp; editing\u003c/p\u003e\n\u003cp\u003eAJ \u0026ndash; methodology, formal analysis, investigation, writing-reviewing \u0026amp; editing\u003c/p\u003e\n\u003cp\u003eSZ - methodology, formal analysis, investigation, writing-reviewing \u0026amp; editing\u003c/p\u003e\n\u003cp\u003eRK \u0026ndash; methodology, formal analysis, investigation, writing-reviewing \u0026amp; editing\u003c/p\u003e\n\u003cp\u003eBF \u0026ndash; methodology, investigation, writing-reviewing \u0026amp; editing\u003c/p\u003e\n\u003cp\u003eCSL \u0026ndash; software\u003c/p\u003e\n\u003cp\u003eAA - methodology, investigation\u003c/p\u003e\n\u003cp\u003eSN \u0026ndash; data curation, writing-reviewing \u0026amp; editing\u003c/p\u003e\n\u003cp\u003eAS - data curation, writing-reviewing \u0026amp; editing\u003c/p\u003e\n\u003cp\u003eRVD - methodology, writing-reviewing \u0026amp; editing, funding acquisition\u003c/p\u003e\n\u003cp\u003eEST - supervision, funding acquisition\u003c/p\u003e\n\u003cp\u003eEF - funding acquisition\u003c/p\u003e\n\u003cp\u003eFW - supervision, funding acquisition\u003c/p\u003e\n\u003cp\u003eLTC - conceptualisation, writing-reviewing \u0026amp; editing, supervision, funding acquisition\u003c/p\u003e\n\u003cp\u003eFMR - conceptualisation, writing-reviewing \u0026amp; editing, supervision, funding acquisition\u003c/p\u003e\n\u003cp\u003eTK \u0026ndash; conceptualisation, methodology, resources, writing-reviewing \u0026amp; editing, supervision, funding acquisition\u003c/p\u003e\n\u003cp\u003eAll authors read and approved the final manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData Availability Statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe datasets collected and analysed in the current study are available in anonymised form from the corresponding author on reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAdditional Information\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors acknowledge the important contribution of Mr Prabhakaran Santhanam in the technical development of LvL UP 1.0 and Ms Xiaowen Lin for her assistance in processing interview transcripts.\u003c/p\u003e\n\u003cp\u003eThis study was funded by the National Research Foundation, Prime Minister\u0026rsquo;s Office, Singapore under its Campus for Research Excellence and Technological Enterprise (CREATE) programme. The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; or decision to submit the manuscript for publication.\u003c/p\u003e\n\u003cp\u003eJLM, TK, OC, and FW are affiliated with the Centre for Digital Health Interventions, a joint initiative of the Institute for Implementation Science in Health Care, University of Zurich; the Department of Management, Technology, and Economics at ETH Zurich; and the Institute of Technology Management and School of Medicine at the University of St Gallen. CDHI is funded in part by CSS, a Swiss health insurer; Mavie Next, an Austrian health insurer; and MTIP, a Swiss digital health investor. However, none of these companies were involved in this study. All other authors declare no potential conflict of interest.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eMurphy, A. \u003cem\u003eet al.\u003c/em\u003e The household economic burden of non-communicable diseases in 18 countries. \u003cem\u003eBMJ Glob. Health\u003c/em\u003e \u003cstrong\u003e5\u003c/strong\u003e, e002040 (2020).\u003c/li\u003e\n\u003cli\u003eMurray, C. J. L. \u003cem\u003eet al.\u003c/em\u003e Global burden of 87 risk factors in 204 countries and territories, 1990\u0026ndash;2019: a systematic analysis for the Global Burden of Disease Study 2019. \u003cem\u003eThe Lancet\u003c/em\u003e \u003cstrong\u003e396\u003c/strong\u003e, 1223\u0026ndash;1249 (2020).\u003c/li\u003e\n\u003cli\u003eGBD 2019 Mental Disorders Collaborators. Global, regional, and national burden of 12 mental disorders in 204 countries and territories, 1990\u0026ndash;2019: a systematic analysis for the Global Burden of Disease Study 2019. \u003cem\u003eLancet Psychiatry\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e, 137\u0026ndash;150 (2022).\u003c/li\u003e\n\u003cli\u003eKatz, D. L. \u003cem\u003eet al.\u003c/em\u003e Lifestyle as Medicine: The Case for a True Health Initiative. \u003cem\u003eAm. J. Health Promot.\u003c/em\u003e \u003cstrong\u003e32\u003c/strong\u003e, 1452\u0026ndash;1458 (2018).\u003c/li\u003e\n\u003cli\u003eZheng, S. \u003cem\u003eet al.\u003c/em\u003e Effectiveness of holistic mobile health interventions on diet, and physical, and mental health outcomes: a systematic review and meta-analysis. \u003cem\u003eeClinicalMedicine\u003c/em\u003e \u003cstrong\u003e66\u003c/strong\u003e, (2023).\u003c/li\u003e\n\u003cli\u003eOlsen, J. M. \u0026amp; Nesbitt, B. J. Health Coaching to Improve Healthy Lifestyle Behaviors: An Integrative Review. \u003cem\u003eAm. J. Health Promot.\u003c/em\u003e \u003cstrong\u003e25\u003c/strong\u003e, e1\u0026ndash;e12 (2010).\u003c/li\u003e\n\u003cli\u003eWerntz, A., Amado, S., Jasman, M., Ervin, A. \u0026amp; Rhodes, J. E. Providing Human Support for the Use of Digital Mental Health Interventions: Systematic Meta-review. \u003cem\u003eJ. Med. Internet Res.\u003c/em\u003e \u003cstrong\u003e25\u003c/strong\u003e, e42864 (2023).\u003c/li\u003e\n\u003cli\u003eSchachner, T., Keller, R. \u0026amp; Wangenheim, F. von. Artificial Intelligence-Based Conversational Agents for Chronic Conditions: Systematic Literature Review. \u003cem\u003eJ. Med. Internet Res.\u003c/em\u003e \u003cstrong\u003e22\u003c/strong\u003e, e20701 (2020).\u003c/li\u003e\n\u003cli\u003eCastro, O. \u003cem\u003eet al.\u003c/em\u003e Development of \u0026ldquo;LvL UP 1.0\u0026rdquo;: a smartphone-based, conversational agent-delivered holistic lifestyle intervention for the prevention of non-communicable diseases and common mental disorders. \u003cem\u003eFront. Digit. Health\u003c/em\u003e \u003cstrong\u003e5\u003c/strong\u003e, (2023).\u003c/li\u003e\n\u003cli\u003eBuen, R. \u003cem\u003eet al.\u003c/em\u003e Talk and Tools: the best of both worlds in mobile user interfaces for E-coaching | Personal and Ubiquitous Computing. \u003cem\u003ePers. Ubiquitous Comput.\u003c/em\u003e \u003cstrong\u003e21\u003c/strong\u003e, 661\u0026ndash;674 (2017).\u003c/li\u003e\n\u003cli\u003ePhan, T. P. \u003cem\u003eet al.\u003c/em\u003e Forecasting the burden of type 2 diabetes in Singapore using a demographic epidemiological model of Singapore. \u003cem\u003eBMJ Open Diabetes Res. Care\u003c/em\u003e \u003cstrong\u003e2\u003c/strong\u003e, (2014).\u003c/li\u003e\n\u003cli\u003eSubramaniam, M. \u003cem\u003eet al.\u003c/em\u003e Tracking the mental health of a nation: prevalence and correlates of mental disorders in the second Singapore mental health study. \u003cem\u003eEpidemiol. Psychiatr. Sci.\u003c/em\u003e \u003cstrong\u003e29\u003c/strong\u003e, (2020).\u003c/li\u003e\n\u003cli\u003eCollins, L. M. \u003cem\u003eOptimization of Behavioral, Biobehavioral, and Biomedical Interventions: The Multiphase Optimization Strategy (MOST)\u003c/em\u003e. (Springer International Publishing, 2018). doi:10.1007/978-3-319-72206-1.\u003c/li\u003e\n\u003cli\u003eSkivington, K. \u003cem\u003eet al.\u003c/em\u003e A new framework for developing and evaluating complex interventions: update of Medical Research Council guidance. \u003cem\u003eBMJ\u003c/em\u003e \u003cstrong\u003e374\u003c/strong\u003e, n2061 (2021).\u003c/li\u003e\n\u003cli\u003eKeller, R. \u003cem\u003eet al.\u003c/em\u003e Digital Behavior Change Interventions for the Prevention and Management of Type 2 Diabetes: Systematic Market Analysis. \u003cem\u003eJ. Med. Internet Res.\u003c/em\u003e \u003cstrong\u003e24\u003c/strong\u003e, e33348 (2022).\u003c/li\u003e\n\u003cli\u003eCastro, O. \u003cem\u003eet al.\u003c/em\u003e Top-funded companies offering digital health interventions for the prevention and treatment of depression: a systematic market analysis. \u003cem\u003eArch. Public Health\u003c/em\u003e \u003cstrong\u003e82\u003c/strong\u003e, 200 (2024).\u003c/li\u003e\n\u003cli\u003eJakob, R. \u003cem\u003eet al.\u003c/em\u003e Factors Influencing Adherence to mHealth Apps for Prevention or Management of Noncommunicable Diseases: Systematic Review. \u003cem\u003eJ. Med. INTERNET Res.\u003c/em\u003e \u003cstrong\u003e24\u003c/strong\u003e, (2022).\u003c/li\u003e\n\u003cli\u003eMair, J. L. \u003cem\u003eet al.\u003c/em\u003e Effective Behavior Change Techniques in Digital Health Interventions for the Prevention or Management of Noncommunicable Diseases: An Umbrella Review. \u003cem\u003eAnn. Behav. Med.\u003c/em\u003e \u003cstrong\u003e57\u003c/strong\u003e, 817\u0026ndash;835 (2023).\u003c/li\u003e\n\u003cli\u003eMair, J. L. \u003cem\u003eet al.\u003c/em\u003e Exploring the potential of mobile health interventions to address behavioural risk factors for the prevention of non-communicable diseases in Asian populations: a qualitative study. \u003cem\u003eBMC Public Health\u003c/em\u003e \u003cstrong\u003e23\u003c/strong\u003e, 753 (2023).\u003c/li\u003e\n\u003cli\u003eSalamanca-Sanabria, A. \u003cem\u003eet al.\u003c/em\u003e Exploring the Perceptions of mHealth Interventions for the Prevention of Common Mental Disorders in University Students in Singapore: Qualitative Study. \u003cem\u003eJ. Med. Internet Res.\u003c/em\u003e \u003cstrong\u003e25\u003c/strong\u003e, e44542 (2023).\u003c/li\u003e\n\u003cli\u003eMichie, S., Atkins, L. \u0026amp; West, R. \u003cem\u003eThe Behaviour Change Wheel: A Guide to Designing Interventions\u003c/em\u003e. (Silverback Publishing, 2014).\u003c/li\u003e\n\u003cli\u003eKowatsch, T. \u003cem\u003eet al.\u003c/em\u003e Design and Evaluation of a Mobile Chat App for the Open Source Behavioral Health Intervention Platform Mobilecoach. in \u003cem\u003eDesigning the Digital Transformation. Desrist 2017. Lecture Notes in Computer Scienc\u003c/em\u003e vol. 10243 485\u0026ndash;489 (Springer, Berlin, 2017).\u003c/li\u003e\n\u003cli\u003eOllier, J. \u003cem\u003eet al.\u003c/em\u003e Can digital health researchers make a difference during the pandemic? Results of the single-arm, chatbot-led Elena+: Care for COVID-19 interventional study. \u003cem\u003eFront. Public Health\u003c/em\u003e \u003cstrong\u003e11\u003c/strong\u003e, (2023).\u003c/li\u003e\n\u003cli\u003eUlrich, S., Lienhard, N., K\u0026uuml;nzli, H. \u0026amp; Kowatsch, T. A Chatbot-Delivered Stress Management Coaching for Students (MISHA App): Pilot Randomized Controlled Trial. \u003cem\u003eJMIR MHealth UHealth\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, e54945 (2024).\u003c/li\u003e\n\u003cli\u003eStasinaki, A. \u003cem\u003eet al.\u003c/em\u003e Effects of a novel mobile health intervention compared to a multi-component behaviour changing program on body mass index, physical capacities and stress parameters in adolescents with obesity: a randomized controlled trial. \u003cem\u003eBMC Pediatr.\u003c/em\u003e \u003cstrong\u003e21\u003c/strong\u003e, 308 (2021).\u003c/li\u003e\n\u003cli\u003eStieger, M. \u003cem\u003eet al.\u003c/em\u003e Changing personality traits with the help of a digital personality change intervention. \u003cem\u003eProc. Natl. Acad. Sci.\u003c/em\u003e \u003cstrong\u003e118\u003c/strong\u003e, e2017548118 (2021).\u003c/li\u003e\n\u003cli\u003eArlinghaus, K. R. \u0026amp; Johnston, C. A. The Importance of Creating Habits and Routine. \u003cem\u003eAm. J. Lifestyle Med.\u003c/em\u003e \u003cstrong\u003e13\u003c/strong\u003e, 142\u0026ndash;144 (2018).\u003c/li\u003e\n\u003cli\u003eShih, C.-H. (Iris) \u003cem\u003eet al.\u003c/em\u003e Breeze: Smartphone-based Acoustic Real-time Detection of Breathing Phases for a Gamified Biofeedback Breathing Training. \u003cem\u003eProc. ACM Interact. Mob. Wearable Ubiquitous Technol.\u003c/em\u003e \u003cstrong\u003e3\u003c/strong\u003e, 152:1-152:30 (2019).\u003c/li\u003e\n\u003cli\u003eLukic, Y. X., Teepe, G. W., Fleisch, E. \u0026amp; Kowatsch, T. Breathing as an Input Modality in a Gameful Breathing Training App (Breeze 2): Development and Evaluation Study. \u003cem\u003eJMIR Serious Games\u003c/em\u003e \u003cstrong\u003e10\u003c/strong\u003e, e39186 (2022).\u003c/li\u003e\n\u003cli\u003eLukic, Y. X. \u003cem\u003eet al.\u003c/em\u003e Physiological Responses and User Feedback on a Gameful Breathing Training App: Within-Subject Experiment. \u003cem\u003eJMIR Serious Games\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e, e22802 (2021).\u003c/li\u003e\n\u003cli\u003eParks, A. \u0026amp; Schueller, S. \u003cem\u003eThe Wiley Blackwell Handbook of Positive Psychological Interventions\u003c/em\u003e. (John Wiley \u0026amp; Sons, Ltd, 2014).\u003c/li\u003e\n\u003cli\u003eBillingham, S. A., Whitehead, A. L. \u0026amp; Julious, S. A. An audit of sample sizes for pilot and feasibility trials being undertaken in the United Kingdom registered in the United Kingdom Clinical Research Network database. \u003cem\u003eBMC Med. Res. Methodol.\u003c/em\u003e \u003cstrong\u003e13\u003c/strong\u003e, 104 (2013).\u003c/li\u003e\n\u003cli\u003eNielsen, J. \u003cem\u003eDesigning Web Usability\u003c/em\u003e. (New Riders Publishing, Indianapolis, USA, 2000).\u003c/li\u003e\n\u003cli\u003ePratap, A. \u003cem\u003eet al.\u003c/em\u003e Indicators of retention in remote digital health studies: a cross-study evaluation of 100,000 participants. \u003cem\u003eNpj Digit. Med.\u003c/em\u003e \u003cstrong\u003e3\u003c/strong\u003e, 1\u0026ndash;10 (2020).\u003c/li\u003e\n\u003cli\u003eClark, L. T. \u003cem\u003eet al.\u003c/em\u003e Increasing Diversity in Clinical Trials: Overcoming Critical Barriers. \u003cem\u003eCurr. Probl. Cardiol.\u003c/em\u003e \u003cstrong\u003e44\u003c/strong\u003e, 148\u0026ndash;172 (2019).\u003c/li\u003e\n\u003cli\u003eMair, J. L. \u003cem\u003eet al.\u003c/em\u003e Understanding and overcoming barriers to digital health adoption: a patient and public involvement study. \u003cem\u003eTransl. Behav. Med.\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e, ibaf010 (2025).\u003c/li\u003e\n\u003cli\u003eWare, J. E. \u0026amp; Sherbourne, C. D. The MOS 36-item short-form health survey (SF-36). I. Conceptual framework and item selection. \u003cem\u003eMed. Care\u003c/em\u003e \u003cstrong\u003e30\u003c/strong\u003e, 473\u0026ndash;483 (1992).\u003c/li\u003e\n\u003cli\u003eWorld Health Organization. \u003cem\u003eThe World Health Organization-Five Well-Being Index (WHO-5)\u003c/em\u003e. https://cdn.who.int/media/docs/default-source/mental-health/who-5_english-original4da539d6ed4b49389e3afe47cda2326a.pdf?sfvrsn=ed43f352_11\u0026amp;download=true (2024).\u003c/li\u003e\n\u003cli\u003eKroenke, K., Spitzer, R. L., Williams, J. B. W. \u0026amp; L\u0026ouml;we, B. An Ultra-Brief Screening Scale for Anxiety and Depression: The PHQ\u0026ndash;4. \u003cem\u003ePsychosomatics\u003c/em\u003e \u003cstrong\u003e50\u003c/strong\u003e, 613\u0026ndash;621 (2009).\u003c/li\u003e\n\u003cli\u003eTran, V. D. \u003cem\u003eet al.\u003c/em\u003e Validity of the International Physical Activity Questionnaire\u0026ndash;Short Form for Application in Asian Countries: A Study in Vietnam. \u003cem\u003eEval. Health Prof.\u003c/em\u003e \u003cstrong\u003e43\u003c/strong\u003e, 105\u0026ndash;109 (2020).\u003c/li\u003e\n\u003cli\u003eFalkenstr\u0026ouml;m, F., Hatcher, R. L., Skjulsvik, T., Larsson, M. H. \u0026amp; Holmqvist, R. Development and validation of a 6-item working alliance questionnaire for repeated administrations during psychotherapy. \u003cem\u003ePsychol. Assess.\u003c/em\u003e \u003cstrong\u003e27\u003c/strong\u003e, 169\u0026ndash;183 (2015).\u003c/li\u003e\n\u003cli\u003eKowatsch, T. \u003cem\u003eet al.\u003c/em\u003e Conversational Agents as Mediating Social Actors in Chronic Disease Management Involving Health Care Professionals, Patients, and Family Members: Multisite Single-Arm Feasibility Study. \u003cem\u003eJ. Med. Internet Res.\u003c/em\u003e \u003cstrong\u003e23\u003c/strong\u003e, e25060 (2021).\u003c/li\u003e\n\u003cli\u003eSalamanca-Sanabria, A., Richards, D. \u0026amp; Timulak, L. Adapting an internet-delivered intervention for depression for a Colombian college student population: An illustration of an integrative empirical approach. \u003cem\u003eInternet Interv.\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e, 76\u0026ndash;86 (2019).\u003c/li\u003e\n\u003cli\u003eBrooks, J., McCluskey ,Serena, Turley ,Emma \u0026amp; and King, N. The Utility of Template Analysis in Qualitative Psychology Research. \u003cem\u003eQual. Res. Psychol.\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, 202\u0026ndash;222 (2015).\u003c/li\u003e\n\u003cli\u003eGallagher, K. M. \u0026amp; Updegraff, J. A. Health message framing effects on attitudes, intentions, and behavior: a meta-analytic review. \u003cem\u003eAnn. Behav. Med. Publ. Soc. Behav. Med.\u003c/em\u003e \u003cstrong\u003e43\u003c/strong\u003e, 101\u0026ndash;116 (2012).\u003c/li\u003e\n\u003cli\u003eChaffey, D. 2024 average ad click through rates (CTRs) for paid search, display and social media. \u003cem\u003eSmart Insights\u003c/em\u003e https://www.smartinsights.com/internet-advertising/internet-advertising-analytics/display-advertising-clickthrough-rates/ (2024).\u003c/li\u003e\n\u003cli\u003eYao, R. \u003cem\u003eet al.\u003c/em\u003e Inequities in Health Care Services Caused by the Adoption of Digital Health Technologies: Scoping Review. \u003cem\u003eJ. Med. Internet Res.\u003c/em\u003e \u003cstrong\u003e24\u003c/strong\u003e, e34144 (2022).\u003c/li\u003e\n\u003cli\u003eWoolley, K. E. \u003cem\u003eet al.\u003c/em\u003e Mapping Inequities in Digital Health Technology Within the World Health Organization\u0026rsquo;s European Region Using PROGRESS PLUS: Scoping Review. \u003cem\u003eJ. Med. Internet Res.\u003c/em\u003e \u003cstrong\u003e25\u003c/strong\u003e, e44181 (2023).\u003c/li\u003e\n\u003cli\u003eBaumel, A., Muench, F., Edan, S. \u0026amp; Kane, J. M. Objective User Engagement With Mental Health Apps: Systematic Search and Panel-Based Usage Analysis. \u003cem\u003eJ. Med. INTERNET Res.\u003c/em\u003e \u003cstrong\u003e21\u003c/strong\u003e, (2019).\u003c/li\u003e\n\u003cli\u003eJabir, A. I. \u003cem\u003eet al.\u003c/em\u003e Attrition in Conversational Agent\u0026ndash;Delivered Mental Health Interventions: Systematic Review and Meta-Analysis. \u003cem\u003eJ. Med. Internet Res.\u003c/em\u003e \u003cstrong\u003e26\u003c/strong\u003e, (2024).\u003c/li\u003e\n\u003cli\u003eMohr, D., Cuijpers, P. \u0026amp; Lehman, K. Supportive Accountability: A Model for Providing Human Support to Enhance Adherence to eHealth Interventions. \u003cem\u003eJ. Med. Internet Res.\u003c/em\u003e \u003cstrong\u003e13\u003c/strong\u003e, e1602 (2011).\u003c/li\u003e\n\u003cli\u003eCastro, O. \u003cem\u003eet al.\u003c/em\u003e The LvL UP trial: Protocol for a sequential, multiple assignment, randomised controlled trial to assess the effectiveness of a blended mobile lifestyle intervention. \u003cem\u003eContemp. Clin. Trials\u003c/em\u003e \u003cstrong\u003e150\u003c/strong\u003e, 107833 (2025).\u003c/li\u003e\n\u003cli\u003eShenoi, A. \u003cem\u003eet al.\u003c/em\u003e MICoach: LLM-driven motivational interviewing for physical activity behaviour change. Under review, in \u003cem\u003eProceedings of the ACM International Conference on Intelligent Virtual Agents \u003c/em\u003e(ACM, BERLIN Germany, 2025)\u003c/li\u003e\n\u003cli\u003eSteenstra, I., Nouraei, F., Arjmand, M. \u0026amp; Bickmore, T. Virtual Agents for Alcohol Use Counseling: Exploring LLM-Powered Motivational Interviewing. in \u003cem\u003eProceedings of the ACM International Conference on Intelligent Virtual Agents\u003c/em\u003e 1\u0026ndash;10 (ACM, GLASGOW United Kingdom, 2024). doi:10.1145/3652988.3673932.\u003c/li\u003e\n\u003cli\u003eKeller, R., von Wangenheim, F., Mair, J. L. \u0026amp; Kowatsch, T. Receptivity to mobile health interventions. in \u003cem\u003eDigital Therapeutics for Mental Health and Addiction\u003c/em\u003e 65\u0026ndash;77 (Academic Press, 2023).\u003c/li\u003e\n\u003cli\u003eMishra, V. \u003cem\u003eet al.\u003c/em\u003e Detecting Receptivity for mHealth Interventions in the Natural Environment. \u003cem\u003eProc. ACM Interact. Mob. Wearable Ubiquitous Technol.\u003c/em\u003e \u003cstrong\u003e5\u003c/strong\u003e, 74:1-74:24 (2021).\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Tables","content":"\u003cp\u003eTable 1. Recruitment channel analysis.\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"100%\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eChannel\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 15px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eType\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eImpressions\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eClicks\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eClick through rate (%)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eInstalls\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eInstall rate\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003e(%)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSpend (S$)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eCost per install (S$)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11px;\"\u003e\n \u003cp\u003eGoogle\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 15px;\"\u003e\n \u003cp\u003eAd campaign\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e233,667\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e2,220\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e0.95\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e219\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e9.87\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e1,131.31\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e5.17\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11px;\"\u003e\n \u003cp\u003eFacebook and Instagram\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 15px;\"\u003e\n \u003cp\u003eAd campaign\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e107,528\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e1,042\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e0.97\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003eN/A\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e637.87\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003eN/A\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11px;\"\u003e\n \u003cp\u003eLinkedIn, Twitter and Facebook\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 15px;\"\u003e\n \u003cp\u003eResearch Centre\u0026rsquo;s Page Post\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e904\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e29\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e3.21\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e3.45\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e0.00\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e0.00\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11px;\"\u003e\n \u003cp\u003eTelegram Channel\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 15px;\"\u003e\n \u003cp\u003ePosts on #SGResearchLobang and #paidstudiesNTU\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e4,600\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e704\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e15.30\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e44\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e6.25\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e0.00\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e0.00\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11px;\"\u003e\n \u003cp\u003eE-Mail\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 15px;\"\u003e\n \u003cp\u003eResearch Centre\u0026rsquo;s Mailing list\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e220\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e145\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e66.00\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e2.76\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e0.00\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e0.00\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11px;\"\u003e\n \u003cp\u003eFlyer Distribution\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 15px;\"\u003e\n \u003cp\u003eUniversity campus (student interns)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e2,000\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e171\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e8.55\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e32\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e18.71\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e1,244.07\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e113.10\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11px;\"\u003e\n \u003cp\u003eFlyer Distribution\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 15px;\"\u003e\n \u003cp\u003eResidential Area (Little Ants)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e10,100\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e0.05\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e1,668.00\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003eN/A\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 11px;\"\u003e\n \u003cp\u003eFriend Referral\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 15px;\"\u003e\n \u003cp\u003eLink sharing\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003eunknown\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003eN/A\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e66.67\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e0.00\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 10px;\"\u003e\n \u003cp\u003e0.00\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u0026nbsp;\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTable 2. Participant demographic profile.\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"100%\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 302px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 299px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eTotal\u0026nbsp;\u003c/strong\u003en\u003cem\u003e\u0026nbsp;\u003c/em\u003e(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 302px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eAge\u0026nbsp;\u003c/strong\u003e(years)\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e(N=99)\u003c/p\u003e\n \u003cp\u003e21-35\u003c/p\u003e\n \u003cp\u003e36-50\u003c/p\u003e\n \u003cp\u003e51+\u003c/p\u003e\n \u003cp\u003ePrefer not to say\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 299px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e69 (69.7%)\u003c/p\u003e\n \u003cp\u003e21 (21.2%)\u003c/p\u003e\n \u003cp\u003e6 (6%)\u003c/p\u003e\n \u003cp\u003e3 (3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 302px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSex\u0026nbsp;\u003c/strong\u003e(N=99)\u003c/p\u003e\n \u003cp\u003eFemale\u003c/p\u003e\n \u003cp\u003eMale\u003c/p\u003e\n \u003cp\u003ePrefer not to say\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 299px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e63 (63.6%)\u003c/p\u003e\n \u003cp\u003e33 (33.3%)\u003c/p\u003e\n \u003cp\u003e3 (3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 302px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eEthnicity\u0026nbsp;\u003c/strong\u003e(N=99)\u003c/p\u003e\n \u003cp\u003eChinese\u003c/p\u003e\n \u003cp\u003eIndian\u003c/p\u003e\n \u003cp\u003eMalay\u003c/p\u003e\n \u003cp\u003eOther\u003c/p\u003e\n \u003cp\u003ePrefer not to say\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 299px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e74 (74.7%)\u003c/p\u003e\n \u003cp\u003e6 (6.1%)\u003c/p\u003e\n \u003cp\u003e2 (2.0%)\u003c/p\u003e\n \u003cp\u003e12 (12.1%)\u003c/p\u003e\n \u003cp\u003e5 (5.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 302px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eEducation Level\u0026nbsp;\u003c/strong\u003e(N=99)\u003c/p\u003e\n \u003cp\u003eUniversity Degree \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n \u003cp\u003eDiploma\u003c/p\u003e\n \u003cp\u003eA-level\u003c/p\u003e\n \u003cp\u003eO-level/N-level\u003c/p\u003e\n \u003cp\u003ePrefer not to say\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 299px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e68 (68.7%)\u003c/p\u003e\n \u003cp\u003e15 (15.1%)\u003c/p\u003e\n \u003cp\u003e6 (6.1%)\u003c/p\u003e\n \u003cp\u003e4 (4.0%)\u003c/p\u003e\n \u003cp\u003e6 (6.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 302px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eMonthly Household Income\u0026nbsp;\u003c/strong\u003e(S$) (N=63)\u003c/p\u003e\n \u003cp\u003e\u0026lt;2,999\u003c/p\u003e\n \u003cp\u003e3,000 \u0026ndash; 6,999\u003c/p\u003e\n \u003cp\u003e7,000 \u0026ndash; 10,999\u003c/p\u003e\n \u003cp\u003e\u0026gt;=11,000\u003c/p\u003e\n \u003cp\u003ePrefer not to say\u003c/p\u003e\n \u003cp\u003eunknown\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 299px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e6 (10.0%)\u003c/p\u003e\n \u003cp\u003e17 (27.0%)\u003c/p\u003e\n \u003cp\u003e14 (23.0%)\u003c/p\u003e\n \u003cp\u003e7 (11.0%)\u003c/p\u003e\n \u003cp\u003e16 (25%)\u003c/p\u003e\n \u003cp\u003e3 (5.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 302px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eEmployment Status (\u003c/strong\u003eN=63)\u003c/p\u003e\n \u003cp\u003eEmployed\u0026nbsp;\u003c/p\u003e\n \u003cp\u003eStudent\u003c/p\u003e\n \u003cp\u003eHomemaker\u003c/p\u003e\n \u003cp\u003eUnemployed\u003c/p\u003e\n \u003cp\u003eRetired\u003c/p\u003e\n \u003cp\u003ePrefer not to say\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 299px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e38 (61.0%)\u003c/p\u003e\n \u003cp\u003e12 (19.0%)\u003c/p\u003e\n \u003cp\u003e3 (5.0%)\u003c/p\u003e\n \u003cp\u003e4 (7.0%)\u003c/p\u003e\n \u003cp\u003e1 (2.0%)\u003c/p\u003e\n \u003cp\u003e3 (5.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 302px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eMarital Status\u0026nbsp;\u003c/strong\u003e(N=63)\u003c/p\u003e\n \u003cp\u003eSingle\u003c/p\u003e\n \u003cp\u003eMarried\u003c/p\u003e\n \u003cp\u003eDivorced or separated\u003c/p\u003e\n \u003cp\u003ePrefer not to say\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 299px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e38 (60.0%)\u003c/p\u003e\n \u003cp\u003e18 (29.0%)\u003c/p\u003e\n \u003cp\u003e2 (3.0%)\u003c/p\u003e\n \u003cp\u003e5 (8.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 302px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSelf-rated health\u0026nbsp;\u003c/strong\u003e(N=19)\u003c/p\u003e\n \u003cp\u003eExcellent\u003c/p\u003e\n \u003cp\u003eVery Good\u003c/p\u003e\n \u003cp\u003eGood\u003c/p\u003e\n \u003cp\u003eFair\u003c/p\u003e\n \u003cp\u003ePoor\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 299px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e0 (0.0%)\u003c/p\u003e\n \u003cp\u003e6 (31.6%)\u003c/p\u003e\n \u003cp\u003e12 (63.2%)\u003c/p\u003e\n \u003cp\u003e1 (5.3%)\u003c/p\u003e\n \u003cp\u003e0 (0.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTable 3. Participant vulnerability.\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"100%\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 302px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 299px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eTotal\u0026nbsp;\u003c/strong\u003en\u003cem\u003e\u0026nbsp;\u003c/em\u003e(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 302px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eMental Health Vulnerability Score\u003c/strong\u003e (N=99)\u003c/p\u003e\n \u003cp\u003eNo Risk\u003c/p\u003e\n \u003cp\u003eLow Risk\u003c/p\u003e\n \u003cp\u003eModerate Risk\u003c/p\u003e\n \u003cp\u003eHigh Risk\u003c/p\u003e\n \u003cp\u003eVery High Risk\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 299px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e20 (20.2%)\u003c/p\u003e\n \u003cp\u003e22 (22.2%)\u003c/p\u003e\n \u003cp\u003e40 (40.4%)\u003c/p\u003e\n \u003cp\u003e9 (9.1%)\u003c/p\u003e\n \u003cp\u003e8 (8.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 302px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eDiet Vulnerability Score\u003c/strong\u003e (N=99)\u003c/p\u003e\n \u003cp\u003eNo Risk\u003c/p\u003e\n \u003cp\u003eLow Risk\u003c/p\u003e\n \u003cp\u003eModerate Risk\u003c/p\u003e\n \u003cp\u003eHigh Risk\u003c/p\u003e\n \u003cp\u003eVery High Risk\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 299px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e56 (56.4%)\u003c/p\u003e\n \u003cp\u003e43 (43.4%)\u003c/p\u003e\n \u003cp\u003e0 (0.0%)\u003c/p\u003e\n \u003cp\u003e0 (0.0%)\u003c/p\u003e\n \u003cp\u003e0 (0.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 302px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ePhysical Activity Vulnerability Score\u0026nbsp;\u003c/strong\u003e(N=99)\u003c/p\u003e\n \u003cp\u003eNo Risk\u003c/p\u003e\n \u003cp\u003eLow Risk\u003c/p\u003e\n \u003cp\u003eModerate Risk\u003c/p\u003e\n \u003cp\u003eHigh Risk\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 299px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e62 (62.6%)\u003c/p\u003e\n \u003cp\u003e8 (8.1%)\u003c/p\u003e\n \u003cp\u003e5 (5.1%)\u003c/p\u003e\n \u003cp\u003e24 (24.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Digital health, chatbot, behaviour change, lifestyle medicine, conversational agent, preventive health","lastPublishedDoi":"10.21203/rs.3.rs-6484027/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6484027/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eLvL UP is a smartphone-based holistic lifestyle coaching intervention aimed at improving health behaviours, mental well-being, and preventing noncommunicable diseases and common mental disorders. It features a \u0026lsquo;talk and tools\u0026rsquo; approach, combining automated health literacy coaching via conversational agent with digital tools such as journaling, life hacks, and slow-paced breathing exercises. An \u0026lsquo;in-the-wild' mixed-methods study was conducted in Singapore to evaluate LvL UP\u0026rsquo;s feasibility and acceptability to inform a future definitive trial. The app was available on iOS and Android from March to August 2023 and was promoted through online and offline strategies. Data collection included in-app surveys, usage metrics, and interviews, summarised using descriptive statistics and template analysis. The app was downloaded 307 times. Data from 99 active users were analysed. Most users were female and aged 21\u0026ndash;35 years with mild to moderate vulnerabilities in physical activity, diet, and depressive symptoms. Engagement was highest during the first eight days, with 9% remaining engaged for up to 50 days. Users rated technology acceptance highly, finding the app enjoyable, easy to use, and informative. Suggested improvements included streamlined onboarding, fixing bugs, shortening dialogues, and adding rewards. The findings support LvL UP\u0026rsquo;s feasibility and have informed enhancements for future trials.\u003c/p\u003e","manuscriptTitle":"LvL UP 1.0, a holistic mHealth lifestyle coaching intervention for the prevention of non-communicable diseases and common mental disorders: a mixed methods feasibility study","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-05-24 04:30:27","doi":"10.21203/rs.3.rs-6484027/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-11-10T06:45:55+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-07-07T16:17:35+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"34508367642339095953940034485669445422","date":"2025-06-20T12:13:42+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-05-20T10:20:37+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-05-20T10:18:34+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2025-04-28T08:48:17+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-04-28T06:01:09+00:00","index":"","fulltext":""},{"type":"submitted","content":"Scientific Reports","date":"2025-04-19T09:58:29+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"4bf7fdc4-cb7e-4692-a08f-e7e481c28058","owner":[],"postedDate":"May 24th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[{"id":48922697,"name":"Biological sciences/Psychology/Human behaviour"},{"id":48922698,"name":"Health sciences/Health care/Disease prevention/Lifestyle modification"}],"tags":[],"updatedAt":"2025-12-29T16:05:12+00:00","versionOfRecord":{"articleIdentity":"rs-6484027","link":"https://doi.org/10.1038/s41598-025-30960-z","journal":{"identity":"scientific-reports","isVorOnly":false,"title":"Scientific Reports"},"publishedOn":"2025-12-22 15:57:51","publishedOnDateReadable":"December 22nd, 2025"},"versionCreatedAt":"2025-05-24 04:30:27","video":"","vorDoi":"10.1038/s41598-025-30960-z","vorDoiUrl":"https://doi.org/10.1038/s41598-025-30960-z","workflowStages":[]},"version":"v1","identity":"rs-6484027","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6484027","identity":"rs-6484027","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}