Leveraging machine learning to uncover the hidden links between trusting behavior and biological markers

Area of Science:

• Neuroscience
• Computational Psychiatry
• Biomarker Discovery

Background:

• Understanding trust is crucial, especially for individuals with mental health conditions like major depressive disorder (MDD) who face challenges in interpersonal trust.
• Quantifying internal decision-making processes, particularly for trust, is difficult due to their unobservable nature.

Purpose of the Study:

• To explore biomarkers associated with trust-based decision-making using quantitative analysis.
• To develop and apply a machine learning model to infer latent decision-making parameters from behavioral data.

Main Methods:

• Developed a machine learning approach using a Bayesian hierarchical model to quantitatively infer trust-related decision-making parameters.
• Applied the model to behavioral data from patients with major depressive disorder (MDD) and healthy controls (HCs) during a trust game.
• Correlated estimated individualised model parameters with specific blood metabolites.

Main Results:

• The machine learning model successfully predicted participants' behaviors in the trust game.
• No significant group-level differences in estimated decision-making parameters were found between MDD and HC groups.
• Specific blood metabolites, including 5-aminolevulinic acid, acetylcarnitine, and 2-aminobutyric acid, were significantly associated with individual differences in trusting behavior.

Conclusions:

• Identified novel links between trust-based decision-making processes and specific blood biomarkers.
• Demonstrated a framework for integrating behavioral modeling with biomarker discovery to understand social functioning.
• Findings may inform targeted interventions to improve social interaction and well-being in clinical populations.