From March, 2022

Intelligent Contingency Management using Reinforcement Learning for Health Behavior Promotion in Everyday Life

Funded by National Research Foundation, Republic of Korea

This project aims to develop digital health services that provide personally tailored treatment and reinforcement to elicit behavior change in everyday life. For this, we formulate people's behavior change as the Markov Decision Process, in which states are contextual information recorded from multi-modal sensor data, actions are different treatments, and rewards are whether behavior change occurs. In this MDP, reinforcement learning agents can learn the optimal treatment for a given context to maximize the likelihood of behavioral change.