Evaluation of Resting-State EEG Patterns in Female Schizophrenia and Addiction Patients Using Convolutional Neural Networks

Machine learning aids in accurately diagnosing psychiatric disorders, impacting individuals' quality of life and society, by identifying and classifying them efficiently. Traditional diagnoses of psychiatric disorders relied on subjective assessments, which are time-consuming, costly, and human biased; machine learning offers faster, cheaper, and more accurate solutions. The aim of this study was to evaluate patterns of relationship in female schizophrenic and addictive disorder patients by the application of computational techniques to analyze resting-state electroencephalography (EEG) data signals obtained via machine learning with a view to identify distinctive patterns or features that can help medical practitioners to differentiate between the two patient groups. The study utilized data on beta waves collected from the forehead of the human subjects with the electrodes in positions fp1 and fp2. A review of published literature in the field of diagnosis of mental health disorders using machine learning was also conducted in order to identify current trends on the subject and identify possible research gaps for further research. The results show the relationships between age, education, IQ, and AB.D.beta a.FP1/FP2 assessed using correlation coefficients and p-values show a weak negative relationship between age and AB.D.beta a.FP1/FPI/FP2. The correlation coefficients are -0.1054, -0.1722, and -0.0975, respectively. The p-values for these relationships are 0.6407, 0.4435, and 0.4962, indicating that the correlations are not significant. Age, education and IQ did not appear to have a significant impact on the AB.D.beta a.FP1/FPI/FP2 patterns in the studied female patients. Our study investigates the efficacy of convolutional neural networks (CNNs) in diagnosing psychiatric disorders in females. We achieved an accuracy of 84.3% using CNNs for evaluating EEG patterns.