Neurological and psychiatric disorders show significant variability in their onset, progression, and response to treatment. Normative modeling is a statistical method that can help us parse the heterogeneous signals associated with these disorders, bringing us closer to understanding and possible individualized treatment of these brain disorders. Similar to growth charting in pediatrics for height and weight, these models estimate the population means and centiles of variation, allowing for the calculation of deviation scores from this norm for each individual. There has been a considerable increase in the application of normative models for brain disorders, partly due to the recognition that heterogeneity in our patient groups muddles standard case-control analysis. However, most of these applications encounter challenges due to improper use and assessment, often lacking a reliable framework for longitudinal tracking1. In this narrative review, we first provide an overview of the objectives and applications of current normative models within neuroimaging, from parsing heterogeneity and identifying clusters or subtypes, to mapping deviation scores to clinical and behavioral variables. We then outline a roadmap for future methodological advancements for individual longitudinal tracking, focusing on velocity models and thrive lines.
Fraza et al. (2025) studied this question.