Commentary: Holy grails, personalized medicine, and the public health burden of psychopathology – a reflection on Ahuvia et al. (2023)

Abstract

Clinical psychology and psychiatry have many ‘holy grails’ or research findings that are widely sought after but remain elusive. The use of machine learning (ML) models for treatment selection is one of these holy grails. Ahuvia et al. (Journal of Child Psychology and Psychiatry, 2023) recently analyzed a large trial (n = 996) of two distinct single‐session interventions (SSIs) for internalizing distress and found little evidence that an ML model could predict differential treatment response. I discuss potential avenues for advancing SSI research. One avenue is the dissemination and implementation of SSIs, including how they interact with other treatments in routine care. Quantifying and critically questioning the promises of holy grails like ML models is sorely needed. Using simulation modeling to evaluate the relative merits of using ML models for treatment selection or using SSIs versus other treatment strategies may be another path forward.

Identifying active ingredients in cognitive-behavioral therapies: What if we didn't?

“…While comorbidity is often dismissed as an artifact of our diagnostic systems, some comorbidity may be attributable to different stages of illness (e.g., much psychopathology may begin as non-specific distress, Zachar, 2023). Although this research should also be evaluated critically (Lorenzo-Luaces et al, 2021;Lorenzo-Luaces, 2023), mechanistically-informed personalized approaches that are tailored to different stages and targets of psychopathology may be one path towards improved outcomes.…”