A Multimodal Data-Driven Assessment System for Autism Spectrum Disorder in Children: Development and Pilot Validation...

Autism Spectrum Disorder (ASD) is a neurodevelopmental condition affecting children's social communication, behavioral patterns, and emotional expressions. Conventional ASD diagnoses rely heavily on subjective observational tools such as the Autism Diagnostic Observation Schedule (ADOS) and the Childhood Autism Rating Scale (CARS), whose accuracy and efficiency can be limited. In this work, we develop a portable multi-modal data acquisition platform for ASD assessment and conduct a pilot validation of its effectiveness in early screening. The system integrates EEG, ECG, speech, facial expressions, and rating-scale data via wearable devices and smart terminals, employing an algorithmic framework for data fusion and analysis to generate individualized diagnostic reports. In a pilot study with seven participants, the platform identified significant feature differences between the ASD and control groups, including shorter speech pause duration (41.8% reduction, p < 0.001), higher EEG δ-band power (226% increase, p = 0.0015), and lower approximate entropy (38.5% reduction, p < 0.0001). These pilot findings support the effectiveness of multi-modal data fusion. Compared to existing diagnostic tools, this system has notable advantages in low cost, high accuracy, and user-friendliness, indicating strong potential for widespread application in family-based ASD screening and early intervention. With further data expansion and technical optimization, this multi-modal system holds broad clinical prospects.Clinical Relevance- The multi-modal ASD assessment system offers clinicians a low-cost and efficient tool for early screening and diagnosis of ASD. By integrating objective physiological and behavioral measurements, it helps practitioners develop individualized intervention strategies and improve diagnostic accuracy and treatment outcomes, thus showing great promise for clinical applications.