Designing a Comprehensive System for Screening Children with Autism Disorder

visits: 51

Among every 160 children, one is diagnosed with autism spectrum disorder (ASD). While prevention remains elusive, early intervention during childhood can significantly alleviate its impact. ASD manifests across a spectrum, exhibiting varying degrees of severity, thereby demanding meticulous and resource-intensive diagnosis. Despite recent technological strides introducing a plethora of tools for ASD screening in children, their singular focus may undermine reliability.

Our research seeks to tackle these challenges through a comprehensive multi-modal system employing technology-based screening tools. By amalgamating diverse modalities, our system endeavors to bolster the accuracy and confidence of ASD screening compared to individual tool screening. Our subsystems encompass a parrot-like robot, an intelligent toy car, and a light wheel, bolstered by an expert system for ASD screening that integrates parental observations of the child's behavior.

This holistic approach facilitates the collection of behavioral data through child interactions, subsequently processed via image, signal processing, and machine learning techniques. Ultimately, the system aims to provide parents with an accurate assessment of their child's behaviors, flagging potential autism red flags to prompt timely specialist visits.
In our preliminary study, published in the ICRoM conference of 2023, we focused on analyzing three distinct ASD symptom groups using three modalities: the sensorized toy car for assessing repetitive behaviors, the Lightwheel for examining sensory input sensitivity, and the Roboparrot for simulating social interaction initiation. Through the integration of outcomes from each modality, we achieved a significant enhancement in ASD screening accuracy and reliability, surpassing individual modality accuracy by over 10% on average.

In conclusion, our research underscores the pivotal role of multi-modal frameworks in optimizing ASD screening outcomes, with our initial findings showcasing a significant enhancement in accuracy and reliability. Integration of additional intelligent tools holds promise for further enhancing the reliability and accuracy of ASD screening.

Members: Bijan Mehralizadeh, Dr Hadi Moradi, Dr Hamidreza Pouretemad, Pegah Soleiman, Shahab Nikkhoo, Farid Anjidani, Negin Azizi, Javad Razi, Bahar Baradaran

Collaborating Centers: Tehran Autism (CTAD)

Related Papers: