User Engagement Strategies in Digital Mental Health Interventions for Stress Management: A Narrative Review (Preprint)

BACKGROUND Stress is a ubiquitous public health concern associated with increased risk of cardiovascular disease, depression, anxiety, and reduced quality of life. Digital mental health interventions (DMHIs), delivered via smartphones, web platforms and wearables, offer scalable support for stress management but often struggle to sustain user engagement which is a prerequisite for their effectiveness. Emerging adaptive mechanisms, such as personalisation and gamification, aim to enhance contextual relevance, emotional resonance and user adherence but the evidence describing how these mechanisms function together within stress focused DMHIs, remains fragmented. OBJECTIVE This review aims to synthesise evidence on engagement strategies in stress-focused DMHIs, identifying effective adherence techniques, the influence of personalisation, biometric feedback, Just-In-Time Adaptive Interventions (JITAIs), gamification and Artificial Intelligence (including large language models, LLMs), and methodological approaches to studying engagement. METHODS This narrative review was informed by a systematic search of PubMed, IEEE Xplore, and Scopus (January 2018–April 2025) for peer‑reviewed studies on stress‑focused DHIs reporting at least one engagement feature. Eligible studies included randomised controlled trials, qualitative investigations, simulation or modelling work, and implementation research. Data were extracted using a structured analysis capturing study characteristics, engagement strategies, outcomes, and alignment with the research questions. Descriptive statistics summarised study types and feature frequencies, while a narrative synthesis integrated descriptive mapping with interpretive, theory‑informed analysis organised by research question. RESULTS 22 studies met the inclusion criteria. Randomised controlled trials (10) were most common, followed by qualitative investigations (5) and design or simulation studies (7). 14 studies incorporated personalisation and JITAI features, 2 included biometric feedback, and 3 used gamification elements. 11 studies integrated AI, 7 of which employed generative LLM chatbots. Interventions combining tailored content with contextually timed delivery achieved higher adherence than static programmes. Personalisation often co‑occurred with AI, while biometric and JITAI features appeared mainly in technologically advanced systems. Adaptive components enhanced perceived relevance and emotional engagement but raised concerns about transparency, privacy, and burden. Wearable‑triggered prompts and reinforcement learning demonstrated potential, yet remained sensitive to privacy risks. Gamification boosted short‑term motivation when embedded in meaningful narratives, but showed mixed results when poorly implemented. Qualitative analyses further identified barriers related to power dynamics, cultural relevance, and data transparency. CONCLUSIONS Adaptive personalisation and timely delivery were the most effective engagement strategies, while biometric feedback and gamification remain underused. AI and reinforcement learning enable contextual support, but require safeguards for transparency and fairness. Diverse methods across RCTs, qualitative and implementation research provided complementary insights. Stress management DHIs are evolving toward intelligent, user-centred systems that adapt to emotional and contextual states. Future work should emphasise interpretable AI, culturally grounded co-design and longitudinal, mixed-methods evaluation to sustain engagement and equitable stress reduction.