MyMentorLLM: AI-Powered Psychotherapy Training Simulation
MyMentorLLM has been created by researchers as a multimodal environment for simulation that utilizes both voice and text, aimed at enhancing deliberate practice in psychotherapy. The platform produced 2,100 comprehensive training sessions in Cognitive Behavioural Therapy (CBT), each involving a DSM-5-TR-based patient suffering from major depressive disorder, generalized anxiety, or borderline personality disorder, alongside a trainee therapist and an expert supervisor. The structured nature of CBT and its focus on competency-based supervision were key factors in this analysis, which examined emotional dynamics, therapeutic skills, and diagnostic precision. The simulated patients displayed emotional profiles consistent with their disorders, which trainee therapists reflected, similar to actual counseling scenarios. Supervision quality varied among different LLMs, with many models inaccurately assessing their effectiveness. This initiative tackles the challenge of scaling psychotherapy training and expert supervision.
Key facts
- MyMentorLLM is a multimodal voice- and text-based simulation environment for deliberate practice in psychotherapy.
- The system generated 2,100 complete Cognitive Behavioural Therapy (CBT) training sessions.
- Each session links a DSM-5-TR-grounded patient, a therapist-in-training, and an expert supervisor.
- Patients had major depressive, generalised anxiety, or borderline personality disorder.
- CBT was adopted because its structured procedures facilitate standardised simulation and evaluation.
- Sessions were analysed for emotional dynamics, therapeutic competence, and diagnostic accuracy.
- Simulated patients expressed disorder-congruent emotional profiles mirrored by trainee therapists.
- Quality of supervision differed across LLMs, with most models overestimating their performance.
Entities
—