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

Abstract

Identifying likely placebo responders can help design more efficient clinical trials by stratifying participants, reducing sample size requirements, and enhancing the detection of true drug effects. In response to this need, we developed a deep convolutional neural network (DCNN) model using resting-state EEG data from the EMBARC study, achieving a balanced accuracy of 69% in predicting placebo responses in patients with major depressive disorder (MDD). We then applied this model to two additional datasets, LEMON and CAN-BIND-which did not include placebo groups-to investigate potential relationships between the model's predictions and various clinical features in independent samples. Notably, the model's predictions correlated with factors previously linked to placebo response in MDD, including age, extraversion, and cognitive processing speed. These findings highlight several factors associated with placebo susceptibility, offering insights that could guide more efficient clinical trial designs. Future research should explore the broader applicability of such predictive models across different medical conditions, and replicate the current EEG-based model of placebo response in independent samples.