Developing an efficient personalized therapist fit index to optimize psychotherapy case assignments

Psychotherapists vary in their effectiveness, with differences being more pronounced for certain patients. These therapist effectiveness data can plausibly inform treatment personalization. For example, predictive analytics can determine which patients-based on their presenting characteristics-are more likely to improve when treated by a historically effective clinician and which are likely to improve regardless of who treats them. Accordingly, this study first developed a personalized therapist fit (PTF) index. Second, we examined whether patients who had been naturalistically assigned to a PTF-recommended therapist had better treatment outcomes. Patients were 2,160 adults treated by 72 therapists. We used outcomes data from a first patient wave to classify therapists as highly effective (top third of the sample based on their average patient's improvement and percentage of their patients who reliably improved) or average/below average (middle or bottom third of the sample on at least one of these indices). With a second patient wave, machine learning predicted which patients would have better outcomes when working with a highly effective therapist versus comparable outcomes when working with any therapist. Finally, with a third wave of patients, we examined the stability of the PTF algorithm and the effect of being naturalistically matched to a PTF-recommended therapist. Therapists' effectiveness level and patients' PTF were moderately stable and reliably estimated across the training and hold-out samples. Moreover, patients matched versus unmatched to their PTF had better outcomes (d = 0.48). Results preliminarily establish an efficient, effective, and readily scalable form of personalized case assignment. (PsycInfo Database Record (c) 2026 APA, all rights reserved).