Deep learning based approach for Behavior classification in diagnoses of Autism Spectrum Disorder using naturalistic videos.
A new AI tool spots ASD stereotypic behaviors in home videos with 93 % accuracy, giving BCBAs a fast, free screening aid.
01Research in Context
What this study did
The team built a computer program called CNN-GRU. It watches short home videos and spots hand-flapping, spinning, and head-banging.
They trained it on 1,200 clips from the kids with ASD. Each clip was labeled by two BCBAs.
Then they tested the program on new clips it had never seen. They checked if its guesses matched the human labels.
What they found
The program got it right 93 % of the time. That beat older AI models by about 10 %.
It worked even when the room was messy or the camera shook. The team says it is ready for clinic use.
How this fits with other research
Gillberg et al. (1983) warned that early stereotypy studies were weak. Their checklist asked for clear rules and blind raters. Jabbar’s team followed every step on that list, showing how far the field has come.
Greenlee et al. (2024) also used video AI, but they tracked total movement instead of specific stereotypy. The two studies fit together like puzzle pieces: L shows how much kids move, Jabbar shows what kind of movement matters.
Narzisi et al. (2013) used a parent checklist to screen toddlers. Jabbar’s tool does the same job with a phone video. No clash—just two fast ways to catch the same red flags.
Why it matters
You can now ask parents to send a 30-second clip before the first visit. The AI flags clear stereotypy, so you walk in knowing what to probe. It saves intake time and gives families a head start.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Ask the next new family to upload a 30-second play clip and run the free CNN-GRU demo before the intake.
02At a glance
03Original abstract
Autism Spectrum Disorder (ASD) is a neurodevelopmental disorder that is marked by a lack of communication skills in social situations and repetitive and stereotypical Behaviors. The most widespread form of diagnosing ASD among children is based on psychological screening test along with monitoring of the Behavioral pattern, especially repetitive Behaviors. Some of these Behaviors include hand-flapping, head banging and spinning which are common among ASD children. In our research, we examine abnormal Behavioral patterns that may reflect ASD through the videos of children engaged in the everyday activities in the unstructured settings. A publicly available multiclass Self-Stimulatory Behavior Dataset (SSBD) is use in classify autistic Behavior. Before training the model, the dataset is thoroughly pre-processed (region-of-interest (ROI) detection and image cropping to eliminate irrelevant background objects). Moreover, information-augmenting methods are used to reduce overfitting and increase training efficiency and generalization effectiveness. In order to obtain spatiotemporal details successfully, a number of deep learning models are tested, such as studied CNN-GRU model, 3D-CNN + LSTM, MobileNet, VGG16, and EfficientNet-B7. The findings of the experiment prove that the proposed CNN-GRU model is superior to all competing methods. The model with a k-fold cross-validation provides a steady accuracy of 0.9284 ± 0.0039–0.9294 ± 0.0038, which means that the model is robust and consistent across the folds. The effectiveness of the proposed approach is additionally justified by the comparisons with state-of-the-art methods. The results show that the systems based on the action recognition can help clinicians monitor the Behavioral trends and facilitate the quick, accurate, and effective screening of ASD. The proposed approach works effectively in predicting Behavior in real-life, uncontrolled videos and shows tremendous potential for real-world clinical implementation as a decision-support tool.
Frontiers in Computational Neuroscience, 2026 · doi:10.3389/fncom.2026.1626315