This study examined whether machine learning could predict individual responses to and engagement with a smartphone-based digital intervention among people with schizophrenia and related psychotic disorders. Researchers used data from an earlier open trial involving 38 participants with psychosis, primarily schizophrenia spectrum disorders, who used the App4Independence (A4i) psychosocial smartphone intervention for one month. Machine learning models incorporated participants’ pre-intervention symptom data, A4i usage data, and information from semi-structured interviews to predict changes in psychiatric symptoms, patterns of app engagement, and participants’ qualitative impressions of the intervention. The models demonstrated a moderate ability to predict participants’ interaction and experience with A4i, as well as changes in psychosis-related symptoms. The findings suggest that engagement with digital mental health interventions varies considerably between individuals and that baseline symptom information may help identify people who are more likely to engage with and respond positively to digital interventions. The study highlights the potential for machine learning to support more personalized approaches to digital mental health treatment for people with psychosis.
Publications
August 19, 2026