Risk Assessment for Parents Who Suspect Their Child Has Autism Spectrum Disorder: Machine Learning Approach

Area of Science:

• Pediatric Neurology
• Developmental Psychology
• Computational Linguistics

Background:

• Parents often use online communities to investigate potential autism spectrum disorder (ASD) markers.
• Automated tools can support parental decision-making in identifying developmental concerns.

Purpose of the Study:

• To evaluate the feasibility of using automated text analysis and minimal questioning to assess ASD risk from parental concerns shared online.
• To develop a predictive algorithm for ASD risk based on online parental reports.

Main Methods:

• 115 parents of children aged 16-30 months with social-communication concerns provided online reports.
• Parents completed the Modified Checklist for Autism in Toddlers-Revised, with Follow-up (M-CHAT-R/F) and Ages and Stages Questionnaire (ASQ).
• An algorithm predicted ASD risk using combined text analysis and a single, algorithmically selected screening question.

Main Results:

• Screening identified 58-88% of children at risk for ASD.
• Children with a family history of ASD were 3 times more likely to screen positive.
• Text combined with a single M-CHAT-R question significantly improved ASD risk prediction accuracy compared to text alone.

Conclusions:

• Web-based platforms can be utilized for prescreening autism spectrum disorder risk.
• Integrating automated text analysis with standardized screening questions enhances early identification of ASD concerns.