An ensembled two-phase deep learning approach for a psychiatric disorder detection

Computational Psychiatry represents a burgeoning realm within scientific inquiry, delving into the intricate interplay of neurobiology within the brain. The escalating prevalence of mental illness underscores the urgency to confront this challenge. Among the prevalent disorders, Schizophrenia and Bipolar Disorder loom large, affecting a significant portion of the population at some point in their lives. However, pinpointing psychiatric disorders poses a formidable challenge. Genetic predispositions significantly influence the development of mental illnesses, with intriguing overlaps observed among certain disorders. This convergence complicates accurate diagnosis. Here, a deep learning approach is considered for significant gene biomarker identification and classification of Schizophrenia and Bipolar disorder data which is collected from GEO (Gene Expression Omnibus) database. In this experimental model, an extraction approach, a kernel applied Fisher score (KFScore) method is presented to select the prominent genomes, and sine-cosine ensembled Monarch Butterfly algorithm (SC-MBO) optimized CNN (Convolutional Neural Network) strategy is implemented. Here, the SC-MBO ensembled approach is used to get the optimal value of the hyperparameters in CNN. The effectiveness of the presented model is estimated by accuracy% of classification, number of extracted prominent genomic feature, sensitivity, specificity, and ROC (Receiver Operating Characteristic) curve. The suggested strategy is based only on the current experimental conditions and the findings are preliminary. Consequently, the proposed method provides an initial and encouraging framework for bipolar disorder and schizophrenia detection and may be extended to other psychiatric disorder detection tasks with additional studies and validation.