JMIR has published a 6-arm RCT protocol for Shared Strength, an open-access mental health screener and self-guided CBT course for displaced Ukrainians, now live in multiple languages.
The new study protocol published in JMIR Research Protocols outlines an ambitious plan to use behavioral economics and artificial intelligence (AI) to improve engagement in self-guided digital mental health interventions. The trial responds to one of the biggest challenges in digital health: while online programs can reduce barriers such as cost, stigma, and accessibility, many users disengage before completing them, limiting their effectiveness.
The research team is focusing on a digital resiliency course called Shared Strength (Спільна Сила), adapted for Ukrainian refugees affected by displacement from the ongoing war. Built on the EvolutionHealth.care platform, the course integrates evidence-based strategies such as cognitive behavioral therapy, motivational interviewing, and gamification. To test strategies for boosting adherence, the investigators will run a six-arm randomized controlled trial (RCT). Each trial arm will expose participants to different combinations of behavioral nudges—including directive tips, social proof, present-bias framing, to-do checklists, and gamified progress tracking.
The study has several goals:
- Primary objective: determine which combinations of prompts and checklists drive higher engagement (measured through click rates, completion, and session duration).
- Secondary objectives: identify demographic and behavioral predictors of engagement, and generate a foundational dataset of behavioral phenotypes that can fuel AI-driven personalization in future digital health programs.
Recruitment will take place through partnerships with nonprofit and refugee-serving organizations. Participants will enroll anonymously, ensuring privacy and scalability. The design also emphasizes cultural adaptation, with Ukrainian clinicians and developers contributing to translations, tone adjustments, and technical usability to ensure that prompts and interventions feel supportive and relevant to displaced users.
The trial is expected to begin alpha and beta testing in mid-2025, with a soft launch later this year. Researchers anticipate that arms featuring structured checklists and gamification will yield the highest engagement, based on prior studies. The data collected will not measure clinical outcomes directly, but will instead build a predictive analytics foundation to tailor future interventions. Over time, this approach may support real-time personalization, improving adherence and extending the reach of digital mental health care to underserved populations worldwide.
The authors highlight both strengths and limitations. Conducting the study in a real-world, open-access environment increases ecological validity but also raises challenges in verifying participant identity. While engagement is not a clinical endpoint, it is closely tied to long-term effectiveness, making it an important precursor for future outcome studies.
Ultimately, this trial positions itself as a blueprint for scalable, AI-informed digital mental health interventions. By applying methods commonly used by commercial platforms like LinkedIn or Duolingo—continuous A/B testing and nudging strategies—the study aims to close the gap between digital health and other industries in optimizing user engagement. If successful, the framework could be adapted to other languages, contexts, and displaced populations, addressing a global shortage of mental health resources.
Full article: Developing a Behavioral Phenotyping Layer for Artificial Intelligence–Driven Predictive Analytics in a Digital Resiliency Course: Protocol for a Randomized Controlled Trial, JMIR Research Protocols, 2025.
Screener: https://sharedstrength.care/wb-dat
