Artificial Intelligence For Personalized Mental Health Interventions: Opportunities, Effectiveness, and Challenges

Abstract

Background: Although artificial intelligence (AI) has been integrated into digital mental health interventions, the effect of personalised interventions has not been sufficiently described and the role of the individual in the process of personalisation in intervention effectiveness and engagement remains under-researched. This study examined the relationship between AI-assisted personalization and psychological outcomes and engagement as compared to standard digital mental health support.

Methods: Participants received either an AI-assisted personalized digital mental health intervention or standardized digital support. Depression, anxiety, perceived stress and psychological well-being were measured by the Patient Health Questionnaire-9, Generalized Anxiety Disorder-7, Perceived Stress Scale, and the World Health Organization-5 Well-Being Index, respectively. The engagement in and the intensity of interventions and personalization were examined, and multivariable analyses were conducted to determine factors associated with treatment response.

Results: AI personalised intervention was more effective than standardised intervention in improving depressive symptoms, anxiety, perceived stress, and psychological well-being. More personalization was correlated with more session completion, activity engagement, self-monitoring, and overall participation in the intervention. After controlling for important characteristics of participants and intervention, the greater the intensity of personalization the greater the improvement in psychological outcomes.

Conclusion: AI-driven personalization could help to enrich digital mental health interventions by increasing engagement with the interventions and enabling personalised psychological outcomes. Additional long-term and clinically vetted research is needed to confirm its long-term efficacy, safety, and generalizability.