The Potential Impact of Intelligent Systems for Mobile Health Self-Management Support: Monte Carlo Simulations of Text Message Support for Medication Adherence

BACKGROUND Mobile health (mHealth) services cannot easily adapt to users' unique needs. PURPOSE We used simulations of text messaging (SMS) for improving medication adherence to demonstrate benefits of interventions using reinforcement learning (RL). METHODS We used Monte Carlo simulations to estimate the relative impact of an intervention using RL to adapt SMS adherence support messages in order to more effectively address each non-adherent patient's adherence barriers, e.g., forgetfulness versus side effect concerns. SMS messages were assumed to improve adherence only when they matched the barriers for that patient. Baseline adherence and the impact of matching messages were estimated from literature review. RL-SMS was compared in common scenarios to simple reminders, random messages, and standard tailoring. RESULTS RL could produce a 5-14% absolute improvement in adherence compared to current approaches. When adherence barriers are not accurately reported, RL can recognize which barriers are relevant for which patients. When barriers change, RL can adjust message targeting. RL can detect when messages are sent too frequently causing burnout. CONCLUSIONS RL systems could make mHealth services more effective.