Deep Learning Approaches for Classifying Children With and Without Autism Spectrum Disorder Using Inertial Measurement Unit Hand Tracking Data: Comparative Study.

Researchers used special sensors to track hand movements in 41 children with and without autism during a simple cleaning task. They found that computer programs could tell the difference between autistic and non-autistic children's movements with over 90% accuracy. This suggests that movement patterns could potentially help with autism diagnosis in the future, as children with autism showed measurable differences in how they moved their hands and coordinated their movements.

This comparative study examined deep learning approaches to classify children with and without autism spectrum disorder using inertial measurement unit (IMU) hand tracking data during goal-directed arm movements. Researchers collected data from 41 school-aged children during a reach-to-clean task. Various deep learning models were evaluated, with a convolutional autoencoder combined with long short-term memory layers achieving the highest performance (90.21% accuracy, 90.02% F1-score in k-fold validation; 91.87% accuracy, 93.66% F1-score in patient-separated validation). The study validates that significant differences exist in physical movements between typically developing children and those with ASD, particularly in hand-eye coordination skills, which can be detected through computational analysis.

Results suggest potential for objective, technology-based autism screening tools using movement analysis. However, clinical validation and larger studies needed before implementation. Could complement existing behavioral assessments but not replace comprehensive diagnostic evaluation. Requires specialized equipment and technical expertise.

Small sample size of 41 children limits generalizability. Study type not specified in metadata. No comparison to existing diagnostic methods. Unclear if participants were matched for age, sex, or other demographic factors. Real-world clinical validation not demonstrated.

Autism spectrum disorder (ASD) is a prevalent neurodevelopmental condition that can be quite difficult to diagnose due to a lack of objective diagnostic methods in the currently used behavioral assessments. Recent work has shown that children with ASD have a higher incidence of motor control differences. A compilation of studies indicates that between 50% and 88% of the children with ASD have issues with movement control based on standardized motor assessments or parent-reported questionnaires. In this study, we assess a variety of deep learning approaches for the classification of ASD, utilizing data collected via inertial measurement unit (IMU) hand tracking during goal-directed arm movements.

IMU hand tracking data were recorded from 41 school-aged children both with and without an ASD diagnosis to track their arm movements during a reach-to-clean up task. The IMU data were then preprocessed using a moving average and z score normalization to prepare the data for deep learning models. We evaluated the effectiveness of different deep learning models using the preprocessed data and a k-fold validation approach, as well as a patient-separated approach. The best result was achieved with a convolutional autoencoder combined with long short-term memory layers, reaching an accuracy of 90.21% and an F1-score of 90.02%.

Once the convolutional autoencoder+long short-term memory was determined to be the most effective model for this datatype, it was retrained and evaluated with a patient-separated dataset to assess the generalization capability of the model, achieving an accuracy of 91.87% and an F1-score of 93.66%. Our deep learning approach demonstrates that our models hold potential for facilitating ASD diagnosis in clinical settings. This work validates that there are significant differences between the physical movements of typically developing children and children with ASD, and these differences can be identified by analyzing hand-eye coordination skills. Additionally, we have validated that small-scale models can still achieve a high accuracy and good generalization when classifying medical data, opening the door for future research into diagnostic models that may not require massive amounts of data.