A scoping review this week maps the design space of just-in-time adaptive interventions (JITAIs) for adult weight management and highlights a reality check: only a minority of studies use machine learning for decision-making, with many relying on rule-based logic instead [1]. That distribution is itself a trend signal—ML is creeping into personalization, but implementation constraints (data streams, transparency, safety, evaluation burden) still keep rule-based systems dominant [1].
Why it matters / what changes in tactics
-
Clinicians/program operators: Ask vendors/research teams a simple question: “What triggers the intervention—rules or ML—and how do we audit it?” If ML-driven, you’ll want guardrails (fallback rules, safety constraints, and clear escalation pathways).
-
Researchers: Comparative designs that pit rule-based vs ML-based adaptation on the same platform (with consistent outcome sets and engagement metrics) are likely to be especially informative and publishable now [1].
References
- Koh JYJ, Tan SX, Tan XM, et al. Just-In-Time Adaptive Interventions for Weight Management Among Adults With Excess Body Weight: Scoping Review. J Med Internet Res. 2025;27:e76625. doi: https://doi.org/10.2196/76625.
PubMed: https://pubmed.ncbi.nlm.nih.gov/41447266/