Human-Centred AI for Monitoring Therapeutic Fidelity in Coaching for Health-Promoting Behaviour

Uses ~2,000 hours of multimodal coaching recordings from the BetterBrains dementia-prevention trial to develop NLP/ML models -- grounded in humanistic psychology and self-determination theory -- that computationally assess coaching fidelity constructs (empathy, reflective listening, autonomy support), benchmarked against standard MITI fidelity coding, to support quality assurance in large-scale coaching trials.

Uses AI to automatically check whether health coaches are sticking to good coaching practice, using recordings from a large dementia-prevention study.