Using an AI-Based System for Dimensional Deliberate Practice in Psychodynamic Psychotherapy

Abstract

Despite decades of progress in psychotherapy research, the dynamic core of clinical dialogue has remained largely inaccessible to empirical refinement. In this article, we present a novel framework that integrates artificial intelligence with a dimensional model of the psyche to operationalize deliberate practice in psychodynamic psychotherapy. Grounded in the MATRIX — a validated system for coding therapeutic discourse— - complex psychoanalytic constructs are reconfigured into five psychological dimensions and five relevant transitional principles. Through this lens, therapist–patient interactions can be coded, evaluated, and restructured using large language models (LLMs), enabling high-resolution, theory-informed training loops. By identifying critical deviations from dimensional transition rules and generating alternative interventions that restore clinical coherence, the system transforms reflective insights into actionable techniques. This model not only bridges theory and practice, but redefines what it means to train for expertise in psychotherapy, offering a scalable pathway for deliberate, data-driven professional growth.