Quantifying Distances Between Non-Elliptical Clusters to Enhance the Identification of Meaningful Emotional...

Area of Science:

• Psychophysiology
• Computational Statistics
• Clinical Psychology

Background:

• Coordinated emotional responses are crucial for adaptive functioning.
• Clustering can identify subgroups with similar emotion profiles for novel treatments.
• Traditional clustering methods struggle with non-normal psychophysiological data.

Purpose of the Study:

• To evaluate distance measures for assessing cluster separation in model-based clustering.
• To explore the utility of non-elliptical finite mixture models for psychophysiological data.
• To provide practical guidance on applying clustering techniques to emotional response data.

Main Methods:

• Summarized and computationally suggested multivariate distance measures for cluster separation.
• Conducted a simulation study evaluating distance measures across various cluster distribution scenarios (location, scale, skewness, rotation).
• Applied proposed methods to psychophysiological and subjective emotional response data from the Transdiagnostic Anxiety Study.

Main Results:

• Non-elliptical models offer advantages over traditional methods for non-normal data.
• Specific distance measures effectively quantify cluster separation in complex scenarios.
• Demonstrated practical application in identifying emotional response subgroups.

Conclusions:

• Advanced clustering techniques, including non-elliptical models and robust distance measures, enhance the analysis of psychophysiological data.
• These methods can reveal meaningful subgroups in emotional response patterns.
• Findings offer guidance for clinical applications in understanding and treating emotional dysregulation.