Key result
A machine learning approach using continuous wavelet transform on EEG signals achieved 99.26% diagnostic accuracy for combined brain hemispheres in distinguishing depressed from normal patients.
Why the study?
Current diagnostic techniques and treatments used by professionals for clinical depression are not always helpful.
Does a machine learning approach using continuous wavelet transform with EEG signals improve the detection of depression in patients?
Does a machine learning approach using continuous wavelet transform with EEG signals improve the detection of depression in patients?
A machine learning model using continuous wavelet transform on EEG signals achieved over 99% accuracy in distinguishing depressed patients from normal controls.
EEG-ML may aid depression diagnosis; hypothesis-generating and requires prospective validation before clinical adoption.
Nearly 264 million people around the globe currently suffer from clinical depression, according to the World Health Organization. Although there are diagnostic techniques and treatments presently used by professionals, they are not always helpful. Herein, we suggest the use of advanced technological methods to diagnose depressed patients correctly. A machine learning approach is presented, which uses the electroencephalogram for diagnostics. The model extracts multiple features by applying a continuous wavelet transform (CWT) for each recording. These recordings are employed to train and test the model, with data gathered from 15 depressed and 15 normal patients. After the features are extracted from these recordings, it is organized into matrix form. The features are dimensionally reduced using kernel‐principal component analysis and principal component analysis techniques, ranked using Student's t ‐test, and then labelled as normal or depressed with various classifiers. Accuracies of 99.33% and 99.13% were achieved for the right and left hemispheres of the brain, respectively, and 99.26% for the combined hemispheres of the brain. As compared to the discrete and empirical wavelet transform feature extraction methods, the CWT attained the best results. A depression severity index was also developed, using two features for discriminating the classes: normal versus depressed.
No takes yet. Share an insight, caveat, or question.
Raghavendra et al. (2021) studied clinical depression (n=30). Continuous wavelet transform (CWT) of electroencephalogram (EEG) signals vs. Discrete and empirical wavelet transform methods was evaluated on Diagnostic accuracy for combined brain hemispheres. A machine learning approach using continuous wavelet transform on EEG signals achieved 99.26% diagnostic accuracy for combined brain hemispheres in distinguishing depressed from normal patients.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: