Using Extreme Value Statistics to Reconceptualize Psychopathology as Extreme Deviations From a Normative Reference...

Area of Science:

• Neuroimaging
• Statistics
• Clinical Neuroscience

Background:

• Neuroimaging studies often overlook extreme values in data distributions, which are crucial for understanding psychiatric disorders.
• Existing statistical methods in neuroimaging primarily model the central data, failing to capture informative tail distributions.
• Normative models are increasingly used in neuroscience to quantify individual deviations from a reference group, but methods for modeling extreme deviations are lacking.

Purpose of the Study:

• To develop a statistical framework for accurately modeling extreme deviations in neuroimaging data using normative models and multivariate extreme value statistics.
• To provide a method for quantifying individual neurophenotypic deviations from a reference cohort.
• To enhance the understanding of risk, stratification, and anomaly detection in psychiatric and neurological disorders.

Main Methods:

• The study proposes a framework integrating normative modeling with multivariate extreme value statistics.
• It adapts principles from extreme value statistics, commonly used in meteorology, for neurobiological data analysis.
• The approach includes a non-technical introduction to extreme value statistics and mapping multivariate tail dependence.

Main Results:

• The framework successfully models extreme deviations in neuroimaging data, capturing tail distributions more effectively than traditional methods.
• Demonstration on the UK Biobank dataset showed that extreme values can accurately estimate risk and detect atypicality.
• The method allows for mapping the tails of normative distributions for biological markers and assessing multivariate tail dependence.

Conclusions:

• This novel framework offers a robust tool for statistical modeling of extreme deviations in neurobiological data.
• It has the potential to lead to more accurate and effective diagnostic tools for neurological and psychiatric disorders.
• By focusing on informative tail distributions, the approach enhances the utility of normative models in clinical neuroscience.