Autism Spectrum Disorder Detection System For Childrens Using Multi-model Analysis

Abstract

Autism Spectrum Disorder (ASD) Is A Neuro Developmental Condition Characterized By Challenges In Social Communication, Emotional Expression, And Behavioral Flexibility. Early Screening Plays A Crucial Role In Enabling Timely Intervention And Improving Developmental Outcomes In Children. However, Traditional Diagnostic Procedures Depend Heavily On Expert Observation And Standardized Clinical Assessments, Which Can Be Time-consuming And Inaccessible In Many Regions.. This Paper Presents A Multi-modal Autism Spectrum Disorder Detection System For Children That Integrates Behavioral Screening, Facial Emotion Recognition, And Speech Pattern Analysis Within A Unified Artificial Intelligence Framework. The System Employs A Random Forest Classifier For Questionnaire-based Behavioral Screening, A MobileNetV2 Deep Learning Model For Facial Emotion Detection, And A Machine Learning Speech Analysis Model For Identifying Atypical Vocal Characteristics. Each Modality Is Processed Through Dedicated Preprocessing And Feature Extraction Pipelines Before Being Integrated Through A Decision-level Fusion Mechanism To Generate The Final ASD Risk Prediction.. A Web-based Application Built Using The Flask Framework Enables Users To Submit Questionnaire Responses, Upload Facial Images, And Record Speech Samples. Experimental Evaluation Demonstrates That The Multi-modal Approach Improves Predictive Accuracy Compared To Single-modality Methods. The Proposed System Provides A Scalable, Accessible, And AI-assisted Screening Tool That Supports Caregivers And Clinicians In Early ASD Risk Identification.

Keywords