Detecting Receptivity for mHealth Interventions in the Natural Environment
summary
A chatbot app called Ally+ intercepts silent push messages and uses in-the-moment context to decide delivery, comparing a static random-forest model, an adaptive logistic-regression model that retrains per user, and a random-timing control. A 3-week deployment with 83 US participants shows the static model yields up to 40% improvement in timely responses versus control; the adaptive model starts worse but improves over time, and the app retries every 5 minutes for up to 30 minutes before forcing delivery.
AI-written summary · about
related talks
09mEnhance Healthcare Chatbot Transparency
16mEthical AI for Emotional Micro-Moments
22mAll Work and No Play? Conversations with a Question-and-Answer Chatbot in the Wild
17mSession: AI Powered Community Relations In App Management
32mHow we use AI to deliver Telehealth at Transcarent
15mZenZone: AI-Powered Peace of Mind
30mElsa: AI Conversational Agent Aimed At Improving Mental Health For Women
02mInterview with Mireia Scholz