Predicting Placebo Responses Using EEG and Deep Convolutional Neural Networks: Correlations with Clinical Data Across...

Area of Science:

• Neuroscience
• Clinical Psychology
• Biomedical Engineering

Background:

• Clinical trials for major depressive disorder (MDD) face challenges in participant stratification.
• Identifying placebo responders can optimize trial efficiency and reduce sample sizes.
• Predictive models for placebo response are needed to enhance trial design.

Purpose of the Study:

• To develop and validate a deep convolutional neural network (DCNN) model for predicting placebo response in MDD patients.
• To investigate the correlation between DCNN predictions and clinical features in independent datasets.
• To identify factors associated with placebo susceptibility for improved clinical trial strategies.

Main Methods:

• A DCNN model was developed using resting-state electroencephalography (EEG) data from the EMBARC study.
• The model achieved 69% balanced accuracy in predicting placebo response.
• The model was applied to independent datasets (LEMON, CAN-BIND) to assess correlations with clinical features.

Main Results:

• The DCNN model demonstrated predictive capability for placebo response in MDD.
• Model predictions correlated with age, extraversion, and cognitive processing speed in independent samples.
• These factors are known to be associated with placebo response in MDD.

Conclusions:

• The study presents a novel EEG-based DCNN model for predicting placebo response in MDD.
• Findings suggest that age, extraversion, and cognitive processing speed are associated with placebo susceptibility.
• The developed model and identified factors offer potential for more efficient clinical trial designs in MDD research.