Toward biophysical markers of depression vulnerability

Area of Science:

• Neuroscience
• Computational Psychiatry
• Biophysics

Background:

• Psychiatric disorders, like depression, exhibit significant heterogeneity, complicating treatment due to diverse underlying neural causes manifesting as similar phenotypes.
• Current diagnostic and treatment approaches often struggle to capture the biological variability within depression.
• There is a need for objective biomarkers to subtype depression and guide personalized treatment strategies.

Purpose of the Study:

• To propose and validate the use of interpretable, biophysical parameters derived from neural models as biomarkers for psychiatric disorders.
• To investigate the potential of these parameters to capture the heterogeneity of depression at a biological level.
• To compare the efficacy of these biophysical parameters against traditional electroencephalogram (EEG) features for depression classification.

Main Methods:

• Analysis of electroencephalogram (EEG) data from a cohort of patients with depression and healthy controls.
• Construction of biophysical neural models using Dynamic Causal Modelling (DCM) to describe cortical network dynamics during a depression-related task.
• Extraction and analysis of interpretable, biophysical parameters from the constructed neural models.

Main Results:

• Biophysical model parameters were identified as significant biomarkers, enabling the subtyping of depression.
• These parameters created a low-dimensional, interpretable feature space that effectively described inter-individual differences in depressive symptoms.
• The biophysical parameters demonstrated superior classification performance compared to conventional EEG features, capturing the internal heterogeneity of depression.
• The study serves as a proof of concept for the superiority of combining biophysical models with machine learning over classical statistical methods and raw EEG data.

Conclusions:

• Interpretable, biophysical parameters derived from neural models represent a promising avenue for biomarker discovery in psychiatric disorders.
• This approach offers a more nuanced understanding of depression's heterogeneity, potentially leading to improved diagnostic and therapeutic strategies.
• The integration of biophysical modeling and machine learning shows significant potential to advance the field of computational psychiatry and personalized medicine.