Brain Sciences · 2024 · 12 citations · 47 references
EngineeringBiomedical EngineeringPsychiatryBiomedical Signal ProcessingMachine LearningTime Series AnalysisDimensionality ReductionElectroencephalogram AnalysisElectroencephalogram Signal ProcessingElectroencephalographyNeural Signal ProcessingSupport Vector MachinesDiscrete Wavelet TransformEarly DiagnosisEeg Feature ExtractionEeg Signal ProcessingFuzzy EntropyK-nearest NeighborsLinear Discriminant AnalysisNaive Bayes
Background: Mental health issues are increasingly prominent worldwide, posing significant threats to patients and deeply affecting their families and social relationships. Traditional diagnostic methods are subjective and delayed, indicating the need for an objective and effective early diagnosis method. Methods: To this end, this paper proposes a lightweight detection method for multi-mental disorders with fewer data sources, aiming to improve diagnostic procedures and enable early patient detection. First, the proposed method takes Electroencephalography (EEG) signals as sources, acquires brain rhythms through Discrete Wavelet Decomposition (DWT), and extracts their approximate entropy, fuzzy entropy, permutation entropy, and sample entropy to establish the entropy-based matrix. Then, six kinds of conventional machine learning classifiers, including Support Vector Machine (SVM), k-Nearest Neighbors (kNN), Naive Bayes (NB), Generalized Additive Model (GAM), Linear Discriminant Analysis (LDA), and Decision Tree (DT), are adopted for the entropy-based matrix to achieve the detection task. Their performances are assessed by accuracy, sensitivity, specificity, and F1-score. Concerning these experiments, three public datasets of schizophrenia, epilepsy, and depression are utilized for method validation. Results: The analysis of the results from these datasets identifies the representative single-channel signals (schizophrenia: O1, epilepsy: F3, depression: O2), satisfying classification accuracies (88.10%, 75.47%, and 89.92%, respectively) with minimal input. Conclusions: Such performances are impressive when considering fewer data sources as a concern, which also improves the interpretability of the entropy features in EEG, providing a reliable detection approach for multi-mental disorders and advancing insights into their underlying mechanisms and pathological states.
47
Feature-level fusion approaches based on multimodal EEG data for depression recognition
Hanshu Cai, Zhidiao Qu, Zhe Li et al. · Information Fusion · 2020 · 334 citations
Nursing, Neuroscience, Psychiatry +16
A wavelet-based technique to predict treatment outcome for Major Depressive Disorder
Wajid Mumtaz, Likun Xia, Mohd Azhar Mohd Yasin et al. · PLoS ONE · 2017 · 264 citations · Full text
Computing, Pharmacology, Psychiatry +17