Discrete mathematical modelling for enhancing mental illness detection

In this paper a machine learning approach is used for diagnosis of mental illness such as Schizophrenia (SCZ) using Electroencephalogram (EEG) signals. EEG signals helps identification of normal and abnormal brain signals. In this approach, the methodology is divided in three major steps: data collection and pre-processing, feature extraction, and classification. The main focus of the work is feature extraction. For this multi-channel pattern correlation and feature merging (MCPCFM) is proposed that identifies and combines the linear and non-linear features using CNN model. Then data augmentation is also applied to handle data imbalance issue for training random forest classifier. This framework aims to efficiently represent EEG signals and reduce feature dimensionality issue to improve mental illness classification performance. The result analysis shows 90% of accuracy for detection. As compared to state-of-art models the proposed model outperforms best