PredPsych: A toolbox for predictive machine learning-based approach in experimental psychology research

Area of Science:

• Computational Psychology
• Machine Learning in Behavioral Science
• Quantitative Psychology

Background:

• Growing interest in machine learning (ML) for analyzing quantitative behavioral data in experimental psychology.
• Existing ML methods offer higher sensitivity than traditional univariate techniques but lack accessible implementations.
• Need for user-friendly tools to integrate advanced predictive analytics into psychological research.

Purpose of the Study:

• To develop and present PredPsych, an open-source R toolbox designed to democratize access to ML-based predictive methods for psychologists.
• To provide a practical framework for applying ML algorithms to behavioral data analysis.
• To illustrate the utility of PredPsych in addressing complex research questions beyond the scope of univariate analysis.

Main Methods:

• Development of an open-source R toolbox named PredPsych.
• Implementation of various machine-learning predictive algorithms within the toolbox.
• Demonstration of PredPsych's framework using a motion capture dataset from a recent study.

Main Results:

• PredPsych offers a user-friendly interface for applying sophisticated ML predictive algorithms.
• The toolbox enables the analysis of complex behavioral data, such as motion capture datasets.
• Examples are provided showcasing research questions addressable by PredPsych that are challenging for univariate methods.

Conclusions:

• PredPsych enhances the accessibility of advanced ML predictive methods for psychologists and clinical neuroscientists.
• The toolbox supports the computational assessment of potential bio-behavioral markers for diagnostic and prognostic purposes.
• PredPsych is expected to benefit researchers with limited programming experience, advancing quantitative behavioral analysis.