Scientific Publications

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

Neuroinformatics, Volume 23, article number 32 (2025)

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. 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).

DOI: 10.1007/s12021-025-09725-6

Predicting age from resting-state scalp EEG signals with deep convolutional neural networks on TD-brain dataset

Khayretdinova M, Shovkun A, Degtyarev V, Kiryasov A, Pshonkovskaya P and Zakharov I (2022)

The current study aimed to create a new deep learning solution for brain age prediction using raw resting-state scalp EEG. The architecture and training method of the proposed deep convolutional neural networks (DCNN) improve state-of-the-art metrics in the age prediction task using raw resting-state EEG data by 13%. Given that brain age prediction might be a potential biomarker of numerous brain diseases, inexpensive and precise EEG-based estimation of brain age will be in demand for clinical practice.

DOI: 10.3389/fnagi.2022.1019869

Prediction of brain sex from EEG: using large-scale heterogeneous dataset for developing a highly accurate and interpretable ML model

Khayretdinova M, Zakharov I, Pshonkovskaya P, Adamovich T, Kiryasov A, Zhdanov A, Shovkun A.

This study presents a comprehensive examination of sex-related differences in resting-state electroencephalogram (EEG) data, leveraging two different types of machine learning models to predict an individual's sex. The best-performing model achieved an accuracy of 85% and a ROC AUC of 89%, surpassing all prior benchmarks set using EEG data and rivalling the top-tier results derived from fMRI studies.

DOI: 10.1016/j.neuroimage.2023.120495

Optimization of the Deep Neural Networks for Seizure Detection

A. Shovkun, A. Kiryasov, I. Zakharov and M. Khayretdinova

The goal of the present study was to optimize model selection and data preparation procedures for seizure detection in patients with epilepsy on wearable EEG data for the "ICASSP Signal Processing Grand Challenge". We tested more than 100 deep convolutional neural networks (DCNN) architectures and hyperparameter combinations to achieve the most accurate, robust, and generalizable performance in seizure detection tasks.

DOI: 10.1109/ICASSP49357.2023.10094645

Data Leakage Problem in Large Multi-site EEG Datasets

Zakharov I., Kiryasov A., Shovkun A., Pshonkovskaya P., Khayretdinova M.

In the current study, we show that data leakage is an important problem in the existing large-scale EEG datasets. Using the DCNN model we demonstrate such non-physiological information of EEG as the recording location can be predicted with 99% accuracy. Our results show that crucial for advances in neuroscience large-scale EEG projects studies urgently require tools to harmonize data and eliminate the data leakage problem.

"Brain sex" prediction from EEG data using tree-based algorithms

Ilya Zakharov, Alexey Shovkun, Andrey Kiryasov, Polina Pshonkovskaya, Timofey Adamovich, Andrey Zhdanov, Mariam Khayretdinova

Current research on sex-related electrical signatures of the brain shows that some of these features are more common in females and others are more common in males. Overall, sex-related brain variance is better described as a continuous rather than a binary variable. The "brain sex" phenotype may act as a biomarker to mark certain mental health disorders.