Can a Network of Chemical Oscillators Help to Diagnose Schizophrenia?
Ashmita Bose and Jerzy Gorecki
Schizophrenia is one of the most common mental disorders, however it is difficult to detect and can remain undiagnosed for years. It is believed that information if a patient is ill can be extracted from EEG signals recorded using electrodes located at the patient scalp. In the paper we postulate that a network of chemical oscillators can process recorded signals and help to diagnose a patient. In order to verify our approach we investigated the network functionality on a small dataset of EEG signals recorded from 45 ill and 39 healthy patients. We optimized a network formed by just six interacting oscillators using an evolutionary algorithm and obtained over 82% accuracy of schizophrenia detection on the training dataset.
Keywords: Schizophrenia, EEG signal, chemical computing, oscillators, network, Oregonator model, genetic optimization