Summary: Researchers have identified that slight hand movements during grasping tasks could be effective in diagnosing autism spectrum disorder (ASD) with notable accuracy. By utilizing machine learning techniques, they analyzed finger movements in both autistic and non-autistic individuals, achieving around 85% accuracy in classification.
These motor discrepancies, often observable in early development, may complement existing diagnostic approaches that tend to focus on later behavioral signs. The results could pave the way for simpler, more scalable diagnostic tools, facilitating earlier interventions and better support for those on the spectrum.
Key Facts:
- 85% Accuracy: High precision in classifying autism through analysis of grasping motions.
- Early Motor Signals: Subtle differences in movement may enable earlier ASD diagnosis compared to current methods.
- Scalable Potential: Analysis of natural hand movements offers a more straightforward and accessible diagnostic approach.
Source: York University
Obtaining a prompt autism spectrum disorder diagnosis can be quite challenging, but recent findings from York University suggest that the way young adults, or possibly even children, grasp objects might simplify the diagnostic process.
This team, part of an international collaboration, employed machine learning to examine natural hand movements—specifically, finger actions during grasping—among autistic and non-autistic individuals.
“Our models were able to classify autism with about 85 percent accuracy, indicating that this technique could lead to simpler, scalable diagnostic tools,” explained Associate Professor Erez Freud from York’s Department of Psychology and the Centre for Vision Research.
“Currently, approximately one in 50 Canadian children are affected by autism, and achieving timely and accessible diagnosis presents significant challenges. Our research adds to the growing body of evidence suggesting that subtle motor patterns could be valuable diagnostic indicators—something not yet heavily utilized in clinical settings.”
In addition to social and communication difficulties, autism, a neurodevelopmental disorder, can include motor abnormalities often apparent in early childhood. The researchers believe that evaluating these motor movements early on could promote quicker diagnoses and interventions.
“The primary behavioral markers for diagnosis typically relate to late-onset signs, so capturing these early motor markers may effectively reduce the age of diagnosis,” stated Professor Batsheva Hadad from the University of Haifa, an autism research expert and a key collaborator in the study.
Participants, both autistic and non-autistic, were asked to grasp blocks of different sizes using their thumbs and index fingers, which had tracking markers. They lifted each block and returned it to the same spot before resetting their hand position.
Machine learning was employed to analyze the finger movements during these grasping tasks. Both participant groups had normal IQs and were matched for age and intelligence, with young adults chosen to eliminate developmental differences from the results.
The research indicated that subtle variations in motor control could be captured with over 84% accuracy. Additionally, distinct kinematic features in the grasping motions between the two groups were observed.
Freud noted that prior studies rarely utilized naturalistic precision grasping tasks, but machine learning provides a robust tool for analyzing motor patterns, introducing new methodologies for assessing autism spectrum disorder.
The findings, Freud emphasized, could lead to the creation of more accessible and trustworthy diagnostic tools, alongside timely interventions that could significantly enhance outcomes for autistic individuals.
About this Autism research news
Original Research: Open access.
“Effective autism classification through grasping kinematics” by Erez Freud et al. Autism Research
