Identifying Autism with Head Movement Features by Implementing Machine Learning Algorithms
Zhong Zhao1, Zhipeng Zhu1, Xiaobin Zhang2
1Institute of Human Factors and Ergonomics, College of Mechatronics and Control Engineering, Shenzhen University, 3688 Nanhai, Avenue, Shenzhen City, Guangdong Province, China.
Journal of Autism and Developmental Disorders
|July 12, 2021
Summary
Head movement analysis shows potential for identifying autism spectrum disorder (ASD). This study found specific head rotation patterns could accurately distinguish children with ASD from those with typical development (TD).
Area of Science:
- Neuroscience
- Biomedical Engineering
- Developmental Psychology
Background:
- Autism spectrum disorder (ASD) diagnosis relies on behavioral observation, lacking objective biomarkers.
- Head movements are complex motor outputs potentially reflecting neurological differences.
Purpose of the Study:
- To explore the feasibility of using head movement features for identifying autism spectrum disorder (ASD).
- To determine if head movement dynamics can serve as objective biomarkers for ASD detection.
Main Methods:
- Children with ASD and typical development (TD) answered yes-no questions while encouraged to move their heads.
- Head rotation range (RR) and amount of rotation per minute (ARPM) were calculated for pitch, yaw, and roll.
- Machine learning classifiers were trained using these head movement features.
Main Results:
- A decision tree classifier achieved a maximum accuracy of 92.11%.
- The most effective features were RR_Pitch and ARPM_Yaw.
- Head movement dynamics demonstrated significant differences between ASD and TD groups.
Conclusions:
- Head movement dynamics present objective biomarkers for identifying individuals with autism spectrum disorder.
- This approach offers a potential non-invasive method for ASD detection.
- Further research can refine these biomarkers for clinical application.
