Detecting Autism spectrum disorder with sailfish optimisation

Abstract

Autism Spectrum Disorder (ASD), a neurodevelopmental disorder, has been a bottleneck to several clinical researchers due to data modularization, subjective analysis, and shifts in the accurate prediction of the disorder amongst the sample population. Subjective clinical research suffers from a lengthy procedure, which is a time-consuming process. In this paper, Sailfish Optimization (SFO), a recently developed nature-inspired meta-heuristics optimization algorithm, is being utilized to detect ASD. The hunting methodology of sailfish inspires SFO. Classical SFO has examined the search space in only one direction that affects its converging ability. The Random Opposition Based Learning (ROBL) strategy enhances the exploration capacity of SFO and successfully converges the predictive model to global optima. The proposed ROBL-based SFO (ROBL-SFO) selects relevant features from autism spectrum disorder (child and adult) datasets. According to the results obtained, the proposed model outperforms the convergence capability and reduces local-optimal stagnation compared to conventional SFOs.