AI-Assisted Diagnostics Achieve 92% Accuracy in Identifying Autistic Adults
A recent study published in the journal Translational Psychiatry has demonstrated that AI-assisted diagnostics can identify autistic adults with 92% accuracy. The research utilized a combination of an autism questionnaire and online tests that assess emotion recognition, perception, and mental skills. The study involved 552 Dutch adults, including 286 with a formal autism diagnosis. Participants completed tasks measuring their ability to recognize emotions, control automatic responses, and switch between activities. The researchers extracted 54 measurements from these tasks and applied machine learning to classify participants based on their scores. The model achieved 81.8% accuracy using online tasks alone, but when combined with questionnaire scores, it reached 92% accuracy. This approach was particularly effective in identifying patterns across multiple tasks, even when individual test scores did not show significant differences. 
