Overview
JITAIs use changing information about a person and context to decide whether and how to offer support. This series moves from decision rules and signals to receptivity, burden, evidence, personalization, and trial design. It shows why a timely prompt can still miss the mechanism or be unusable, and why no delivery may be the better decision. Current findings are promising but mixed: controlled mental-health effects are small, while larger behavior estimates come mainly from pre-post designs.
Learning outcomes
- A JITAI is defined by a protocolized fast-timescale decision using changing information, not by a phone, sensor, AI system, or message alone.
- A useful rule connects a meaningful signal to an option that fits the mechanism and the person’s receptivity, burden, and context.
- Engagement, retention, adherence, proximal response, and long-term outcome are different measures and cannot substitute for one another.
- Current evidence supports promise, not universal superiority: mental-health effects are small in controlled comparisons, while larger behavior estimates rely mainly on pre-post data.
Keywords
Preview every chapter · Unlock full text, audio, and video from chapter 4
PlansFAQ
What makes an intervention a JITAI?
A JITAI uses changing information about a person or context in a fast, protocolized rule to decide whether and how to deliver an intervention option. Stable demographics alone, fixed schedules, or momentary support without state-based adaptation do not establish a JITAI.
Why can a timely JITAI prompt still fail?
A timely prompt can fail when its signal is a weak proxy, the option misses the target mechanism, or the person cannot use it under current conditions. Stress, driving, burden, classifier error, and habituation can reduce receptivity even when the rule fires correctly.
What does current evidence show about JITAIs?
Current evidence shows promising but bounded benefits, not universal superiority. A mental-health synthesis found small average controlled effects (Hedges’ g = 0.15), while a larger objective-behavior estimate (g = 0.77) came from pre-post data and is not directly comparable.
When is more JITAI adaptation justified?
More adaptation is justified when it resolves an important uncertainty or addresses a real limitation of a simpler rule and can be tested safely. Extra sensing, stochastic learning, human support, or social coordination also add burden, cost, and uncertainty; complexity alone is not evidence of benefit.