Use of Machine Learning to Identify Children with Autism and Their Motor Abnormalities

Area of Science:

• Neuroscience
• Developmental Psychology
• Biomedical Engineering

Background:

• Autism spectrum disorder (ASD) is a complex neurodevelopmental condition characterized by social-communication deficits and restricted, repetitive behaviors.
• Early and accurate diagnosis of ASD is crucial for timely intervention and improved outcomes, but current diagnostic methods can be challenging, especially in very young children.
• Clinical heterogeneity within ASD presents a significant challenge for research and treatment.

Purpose of the Study:

• To investigate the potential of using upper-limb kinematic analysis to classify young children with ASD.
• To develop and validate a machine learning model for discriminating between children with ASD and typically developing children based on motor task performance.
• To explore the existence of a distinct motor signature associated with ASD in early childhood.

Main Methods:

• A proof-of-concept study involving 15 preschool children diagnosed with ASD and 15 typically developing controls (aged 2-4 years).
• Development of a supervised machine learning algorithm.
• Kinematic analysis of a simple reach-to-drop upper-limb movement task.

Main Results:

• The machine learning model achieved a high classification accuracy of 96.7%.
• Seven specific features related to the goal-oriented phase of the movement were key discriminators.
• The findings suggest a potential motor signature differentiating children with ASD.

Conclusions:

• Simple upper-limb movements, analyzed via kinematic analysis and machine learning, can accurately classify low-functioning children with ASD aged 2-4.
• This approach may aid in identifying a specific subgroup of patients with ASD, potentially reducing clinical heterogeneity.
• These preliminary findings highlight a promising avenue for objective, early identification of autism spectrum disorder.