Related Experiment Video
Updated: May 28, 2025
Network Analysis of Foramen Ovale Electrode Recordings in Drug-resistant Temporal Lobe Epilepsy Patients
Published on: December 18, 2016
A novel ECG-based approach for classifying psychiatric disorders: Leveraging wavelet scattering networks
Hardik Telangore1, Nishant Sharma1, Manish Sharma1
1Department of Electrical and Computer Science Engineering, Institute of Infrastructure, Technology, Research and Management (IITRAM), Ahmedabad, India.
This study introduces an automated method using electrocardiogram (ECG) signals and wavelet scattering networks (WSN) to accurately detect neuropsychiatric disorders like bipolar disorder, depression, and schizophrenia. The novel approach achieves high accuracy, offering a potential tool for clinical diagnosis.
Area of Science:
- Biomedical Engineering
- Computational Neuroscience
- Machine Learning in Healthcare
Background:
- Neuropsychiatric disorders present significant diagnostic challenges due to subjective assessment methods and complex symptoms.
- Existing automated diagnostic systems often rely on electroencephalogram (EEG) signals, which are complex to analyze.
- Electrocardiogram (ECG) signals offer a viable alternative for diagnosing psychiatric conditions due to the heart-brain connection.
Purpose of the Study:
- To investigate the efficacy of using electrocardiogram (ECG) signals for the automated identification of neuropsychiatric disorders.
- To evaluate the performance of a wavelet scattering network (WSN) for analyzing ECG signals in psychiatric patients.
- To develop a reliable and accurate automated diagnostic tool for bipolar disorder (BD), depression (DP), and schizophrenia (SZ).
Main Methods:
- Utilized the Psychiatric ECG Beat Dataset comprising 233 subjects (198 with disorders, 35 controls).
- Applied wavelet scattering-based feature extraction on 3570 heartbeats to analyze ECG signals.
- Employed machine learning techniques, specifically the Fine K-Nearest Neighbor (FKNN) algorithm, with ten-fold cross-validation.
Main Results:
- Achieved an average classification accuracy of 99.8% and a Kappa value of 0.996 using the FKNN algorithm.
- Demonstrated high accuracy for automated identification: 99.78% for BD, 99.94% for DP, 99.98% for SZ, and 100% for controls.
- Reported F1 scores and precision values close to 1, indicating robust model performance.
Conclusions:
- The proposed wavelet scattering network (WSN) combined with FKNN provides a highly accurate automated method for detecting neuropsychiatric disorders using ECG signals.
- This approach overcomes limitations of traditional diagnostic methods and complex EEG analysis.
- The findings suggest a promising tool for objective and efficient clinical detection of psychiatric conditions.
