The normative modeling framework for computational psychiatry.

Area of Science:

• Computational psychiatry
• Clinical neuroscience
• Neuroimaging analysis

Background:

• Traditional patient vs. healthy control studies in mental health research struggle with biological heterogeneity.
• Existing analytical approaches lack tools to effectively model individual variations in mental disorders.
• Normative modeling presents an innovative framework to address these limitations.

Purpose of the Study:

• To define a standardized protocol for conducting normative modeling analyses.
• To guide users through the entire process using the Predictive Clinical Neuroscience toolkit (PCNtoolkit).
• To demonstrate downstream applications of normative modeling in clinical neuroscience.

Main Methods:

• Utilizing the Predictive Clinical Neuroscience toolkit (PCNtoolkit) for normative modeling.
• Standardized protocol covering input data selection and modeling choices.
• Charting population centiles for mapping biology and behavior at the individual level.

Main Results:

• The protocol facilitates statistical inferences at the individual level by mapping biological and behavioral variations.
• Demonstrates downstream analyses including high-risk individual stratification and subtyping.
• Enables behavioral predictive modeling based on normative models.

Conclusions:

• Normative modeling provides a robust solution to analyze biological heterogeneity in mental disorders.
• This framework moves beyond simplistic case-control comparisons, offering more nuanced insights.
• The standardized protocol allows for efficient (1-3 hours) and comprehensive normative modeling analysis.