Exploring Autism Spectrum Disorder Traits and Predictive Modelling using Optimized Feature Engineering in Machine Learning

Abstract

Autism Spectrum Disorder (ASD) is a neurodevelopmental disorder that affects a person's behavior, social interactions, communication, and interests. It typically manifests in early childhood, often before the age of two. ASD is characterized by a wide range of symptoms and severity, which is why it is referred to as a "spectrum" disorder.[1]According to the World Health Organisation, the prevalence of autism spectrum disorder (ASD) has been rising gradually, affecting about one kid out of every 160 worldwide. [2] ASD is a neurodevelopmental disorder that has a major effect on a person's emotional, social, cognitive, and physical well-being. In individuals with autism, neural function in specific regions of the brain is notably affected, particularly in areas such as the cerebral cortex, amygdala, basal ganglia, corpus callosum, and cerebellum. These brain regions play crucial roles in various cognitive and behavioral functions.[3] The obstacles faced by people with autism spectrum disorders (ASD) are numerous and include difficulties focusing, learning disabilities, mental health conditions like anxiety and despair, as well as mobility and sensory abnormalities. The range and severity of ASD symptoms include repetitive behaviours in social settings, obsessive interests, and communication problems.[4] Since autism spectrum disorder (ASD) is thought to be caused by genetic and environmental factors, early detection is essential. Early intervention, however, may help control the effects and possibly prevent further deterioration. The primary method for identification is through observation, involving parents, teachers, and special education teams recognizing potential symptoms. While identifying ASD symptoms in children can be relatively straightforward, the process is more challenging in adults, underscoring the importance of seeking healthcare for comprehensive testing. [5] The aim of this study was to conduct a comprehensive investigation to identify the complex factors contributing to autism spectrum disorder (ASD) in students.The paper unfolds with an exploration of related works, delving into prior studies on Autism Spectrum Disorder (ASD) traits in children and machine learning applications in ASD research. Following this, the paper introduces the comprehensive dataset utilized, curated specifically for investigating ASD development in children. The proposed system is structured into three main components: ASD Trait Analysis, Feature Engineering for machine learning, and Machine Learning Models. ASD Trait Analysis encompasses age distribution, trait prevalence calculation, and co-occurrence condition analysis. The subsequent section focuses on optimizing feature engineering through Recursive Feature Elimination. The final component introduces various machine learning models for predicting ASD, followed by a conclusion summarizing key findings and contributions