Adaptive Framework for Habit Interventions Boosts Oral Health Coverage

A team led by Bhanu Teja Gullapalli (UCLA) has developed an online decision-making framework that continuously adapts when digital health interventions are delivered, addressing a key failure of static scheduling: as daily routines like tooth brushing gradually shift, fixed intervention times become misaligned with actual behavior, prompting either too early or too late to be effective. Their approach integrates timing adaptation into a sequential process that decides both when and whether to send a prompt, using a coverage-based metric that measures how close an intervention lands to a subsequent brushing event. Tested on data from a deployed oral health intervention trial, the adaptive strategy consistently outperformed fixed times derived from user-reported preferences, even as routines evolved.

The framework is currently live in an ongoing randomized controlled trial for digital oral health, with preliminary results mirroring the simulation and offline evaluations. While focused on brushing habits, the underlying algorithm is generalizable to any habitual behavior with flexible timing – such as eating, taking medication, or exercising. The work, published on arXiv and presented at HCI venues, offers a concrete method for making just-in-time adaptive interventions smarter. For digital health apps that rely on nudge-based behavior change, shifting from static to adaptive scheduling could significantly boost user engagement and habit formation by ensuring the right prompt at the right moment.