A precision health approach to medication management in neurodevelopmental conditions: a model development and validation study using four international cohorts.

Abstract

Psychotropic medications are commonly used for children with neurodevelopmental conditions, but their effectiveness varies, making treatment selection challenging. This study developed artificial intelligence (AI) models to predict successful use of stimulants, anti-depressants, and anti-psychotics. Cross-sectional data from research cohorts (N=4,758) were used to predict medication use from Child Behaviour Checklist scores, validating the feasibility and generalizability of this approach. Longitudinal prediction of medication success was then evaluated using electronic medical records from the Psychopharmacology Program (PPP; N=312) at Holland Bloorview Kids Rehabilitation Hospital. Ensemble models achieved strong performance, indicated by high area under the receiving operating characteristic curve (median [IQR], stimulants: 0.84 [0.81,0.88], anti-depressants: 0.82 [0.77,0.87], anti-psychotics: 0.87 [0.83,0.91]). Findings demonstrate that AI can accurately learn expert prescribing patterns and predict treatment outcomes, supporting the potential of data-driven tools to guide personalized medication management for neurodevelopmental conditions and reduce the trial-and-error burden in clinical practice.