Personalized prediction of response to smartphone-delivered meditation training: A machine learning approach (Preprint)

Summary

An algorithm was developed to predict who benefits most from a meditation app. Using data from a randomized controlled trial of a 4-week meditation app versus a control condition in 662 school system employees, a machine learning model created a Personalized Advantage Index (PAI) that estimated each person's expected reduction in distress. The PAI scores significantly predicted which individuals improved more with the app than without. A simpler model using only repetitive negative thinking as a predictor performed similarly well. The algorithm could help individuals make informed decisions about whether a meditation app is right for them.

Study at a glance

Abstract

BACKGROUND Meditation apps have surged in popularity in recent years, with an increasing number of individuals turning to these apps to cope with stress, including during the COVID-19 pandemic. In fact, meditation apps now represent the most commonly used mental health apps for depression and anxiety. However, little is known regarding who is well-suited to these apps. OBJECTIVE The aim of this study was to develop and test a data-driven algorithm to predict which individuals are most likely to benefit from app-based meditation training. METHODS Using randomized controlled trial data comparing a 4-week meditation app (Healthy Minds Program; HMP) with an assessment-only control condition in school system employees (n = 662), we developed an algorithm predicting who is most likely to benefit from HMP. Baseline clinical and demographic characteristics were submitted to a machine learning model to develop a “Personalized Advantage Index” (PAI) reflecting an individual’s expected reduction in distress (primary outcome) from HMP vs. control. RESULTS A significant Group x PAI interaction emerged, indicating that PAI scores moderated group differences in outcome. A regression model including repetitive negative thinking as the sole baseline predictor performed comparably well. Finally, we demonstrate the translation of a predictive model to personalized recommendations of expected benefit. CONCLUSIONS Overall, results reveal the potential of a data-driven algorithm to inform which individuals are most likely to benefit from a meditation app. Such an algorithm could be used to objectively communicate expected benefits to individuals, allowing them to make well-informed decisions about whether a meditation app is right for them. CLINICALTRIAL clinicaltrials.gov (NCT04426318)