A Hybrid Cnn-Lstm-Based Rnn Model for Classification of Schizophrenia

Schizophrenia is a mental health condition that affects the behavior and thought process of a person. With proper treatment, such as medication and therapy, many people with this medical condition are able to manage their symptoms and live meaningful lives. Thus, there is a need for early diagnosis of schizophrenia. However, due to the wide range of symptoms displayed by different schizophrenic patients, accurately identifying schizophrenia can proof challenging. Different conventional methods to try to classify schizophrenia are applied in many instances. Nevertheless, all aspects of the condition could not be sufficiently addressed by any of the conventional classification techniques. Researchers are currently testing the application of deep learning approaches for improved schizophrenia categorization. In this study, we present a novel hybrid RNN model that uses both CNN and LSTM technology for use in the classification of schizophrenia. The twelve layered model was developed using a total of 28 samples as input data. The dataset consisted of EEG signal recordings from 14 healthy controls and 14 schizophrenic patients. After training and testing, an accuracy of 79.35% was recorded by the model. This model has the potential to enable early detection of schizophrenia thereby leading to more optimized treatment and management methods. The study describes a new twelve layered CNN-LSTM hybrid framework for identifying schizophrenia employing EEG signals. Trained on data from 14 patients and 14 healthy controls, the framework obtained 79.35% accuracy, highlighting deep learning's promise for early success in schizophrenia detection.