Personalised & optimised therapy (POT) algorithm using five cognitive and behavioural skills for subthreshold depression

Abstract

Personalising psychotherapies for depression may enhance their efficacy. We conducted a randomised controlled trial of smartphone cognitive-behavioural therapy (CBT) among 4,469 adults in Japan (RESiLIENT trial, UMIN-CTR UMIN000047124). Participants received one of nine CBT skills or combinations, or a health information control (HI), over six weeks. All interventions were found efficacious. We developed prescriptive models using machine learning to forecast changes on the Patient Health Questionnaire-9 (PHQ-9) at week 26 and created a personalised and optimised therapy (POT) algorithm that recommended the most suitable CBT for each participant. In a simulated randomised comparison, the effect of POTs over HI was a difference by −1.41 (95%CI: −1.91 to −0.90) points on the PHQ-9 corresponding with a standardised mean difference of −0.37 (−0.49 to −0.23), which was 35% greater than that of the group-average best intervention. A new randomized trial to confirm the external validity and applicability of the algorithm is warranted.