Early Detection of Schizophrenia Prodromal Symptoms using Dense Neural Networks

Schizophrenia is a serious psychiatric disorder, which causes hallucinations, delusions, dangerous disorganized behaviours and social disengagements in patients. Specifically, 1% of world population i.e. 20 million people and in India nearly 1 million patients are suffering from schizophrenia disorder, due to which their lifetime suicidal rate increased by 10%. The existing literature on Schizophrenia diagnosis are primarily focusing on different types of scan investigations such as MRI and PET, which involves complicated procedures. However, predicting schizophrenia prodromal symptoms in patients is very much essential to fasten the treatment process and thereby save the patient's lifetime. Based on these aspects, this research study presents a new Dense Neural Network (DNN)-based framework, for the early detection of Schizophrenia prodromal symptoms by means of employing both the clinical features and demographical factors of patients. Experimental evaluations clearly demonstrate the enhanced efficiency of the proposed framework in terms of accuracy, precision, recall and F1-score metrics, when compared with baseline techniques such as MLP, LGBM and GB methods and thereby suggests it for clinical application developments.