AI Autism & ADHD Diagnosis: Faster, More Accurate?
AI-Powered Motion Analysis Shows Promise for Early neurodivergence Detection
New Research Leverages Deep Learning and Biometrics to Improve Autism and ADHD Diagnosis
A groundbreaking study published in Scientific Reports details a novel approach to identifying neurodivergent conditions like autism and ADHD using deep learning analysis of subtle micromovements. Researchers have demonstrated the potential of this technology as an early screening tool, offering a faster, more accessible pathway to diagnosis, particularly in underserved communities. The study,led by Dr. KP doctor and colleagues, combines readily available sensor technology with elegant algorithms to quantify movement patterns associated with neurodevelopmental differences.
Identifying Neurodivergence Through Kinematic Data
The research team developed a deep learning model trained on kinematic data – measurements of motion – collected from participants using micro-electromechanical (MEM) sensors. These sensors, increasingly common in smartphones and smartwatches, track acceleration and other movement characteristics. The model analyzed these data to differentiate between neurotypical individuals and those with neurodivergence.
Notably, classification was most accurate for distinguishing patients who are neurotypical from those with neurodivergence. The tool was less reliable for identifying children who have both autism and ADHD, echoing clinical challenges with comorbid diagnoses. This highlights the complexity of overlapping conditions and the need for further refinement of the technology.
“After training on a larger and more extensive dataset, the Deep Learning approach could play an critically important role as an early screening tool for participants suspected of having a neurodivergent disorder, not onyl in the clinic but also in schools and other non-medical settings,” the study team wrote. “With rapid improvements in sensor technology, MEM [micro-electromechanical] sensors are becoming more affordable, reliable, and ubiquitous (such as in smartphones and smartwatches) making the study of kinematic data for applications such as this increasingly relevant.”
Quantifying Severity With Biometrics
Beyond simply identifying neurodivergent conditions, the study also explored novel biomarkers to assess symptom severity. Researchers focused on the Fano Factor and Shannon entropy,statistical measures that quantify randomness in movement. They discovered a correlation between these metrics and the intensity of symptoms.
Children with more severe autism or ADHD exhibited higher entropy and distinct fluctuation patterns in their acceleration data. For instance, participants with low-functioning autism demonstrated considerably greater variability in their hand motions compared to those with milder forms of the condition. This suggests that the technology can not only detect neurodivergence but also provide insights into its severity.
Potential Applications and Future Directions
While not intended to replace clinical diagnosis by qualified healthcare professionals, the authors envision this technology as a valuable triage or screening tool. It could be deployed in primary care offices, schools, or telehealth settings, particularly in areas with limited access to specialized care or long wait times. A 15-minute data collection session is estimated to be sufficient for initial screening, making it a practical solution for early intervention.
“Some patients will need a significant number of services and specialized treatments,” explained José JV, a researcher involved in the study, in a news release. “If, however, the severity of a patient’s disorder is in the middle of the spectrum, their treatments can be more minutely adjusted, will be less demanding and often can be carried out at home, making their care more affordable and easier to carry out.”
This research builds upon previous work by Wu et al. (2018) identifying a biomarker characterizing neurodevelopment with applications in autism, further solidifying the potential of kinematic data in understanding neurodevelopmental conditions. The development of accessible and affordable screening tools like this represents a significant step towards earlier diagnosis and more personalized support for individuals with neurodivergent conditions. Future research will focus on expanding the dataset and refining the algorithms to improve accuracy, particularly in differentiating between comorbid conditions.
References
- Doctor KP, McKeever C, Wu D, et al. deep learning diagnosis plus kinematic severity assessments of neurodivergent disorders. Sci Rep. Published online July 8, 2025. doi:10.1038/s41598-025-04294-9
- Artificial intelligence used to improve speed and accuracy of autism and ADHD diagnoses. News release. EurekAlert. July 8, 2025. Accessed July 8, 2025. https://www.eurekalert.org/news-releases/1090448
- Wu D, josé JV, Nurnberger JI, Torres EB. A biomarker characterizing neurodevelopment with applications in autism. Sci Rep. 2018;