Roland Westerlund

The doctoral project “From signal to action: Development and evaluation of a machine learning-based corrective model for monitoring and adjusting psychotherapeutic treatment” will involve all the clinical units at Modum Bad. Its core purpose is to help clinicians determine whether a patient’s therapeutic process is progressing favourably or deviating from an expected improvement trajectory. We know that traditional judgements of treatment progress rely heavily on clinical expertise, which is variable and prone to bias. By using machine learning, the project aims to provide a standardised, evidence-based way to monitor individual change during therapy, enabling earlier and more appropriate treatment adjustments.

The project forms part of a larger programme applying machine learning to both prescriptive (pre-treatment) and corrective (within-treatment) decision processes. This study focuses specifically on corrective modelling. It will proceed in three stages: model development using historical data, model validation, and a feasibility study in our clinical practice at Modum Bad. Expected benefits include better outcomes for patients, greater clinician confidence in personalised adaptations, and strengthened evidence for precision, feedback-informed mental health care.