People with autism are typically diagnosed by clinical observation of their behaviour but since the clinical decision process can be influenced by subjectivity, AI learning models (LLMs) have entered the field to help characterise the symptoms and channel them down to a diagnosis. According to ANI, researchers found that the AI model suggested that signs of repetitive behaviour, special interests and perception-based behaviour are also indicators of autism. This result significantly decreases the focus on social factors as a primary autism indicator, which the DSM-5 emphasises.
So, in a world where the power of AI is increasing significantly, could it relieve the burden of medical diagnosis out of doctors' hands? Danilo Bzdok of the Mila Quebec Artificial Intelligence Institute and McGill University in Montreal clarified, “Our goal was not to suggest that we could replace clinicians with AI tools for diagnosis. Rather, we sought to quantitatively define exactly what aspects of the observed behaviour or patient history a clinician uses to reach a final diagnostic determination. In doing so, we hope to empower clinicians to work with diagnostic instruments that are more in line with their empirical realities."
ALSO READ | Scientists develop tests to indicate autism in children
Furthermore, scientists have fine-tuned and customised an LLM, pre-trained on a whopping 489 million unique sentences. This model will be able to recognise the diagnostic outcome of over 4,000 medical reports written by medical professionals and clinicians. The reports would mainly consist of behaviour observed by the professionals and the patient history. After running the report through the learning model, the machine analyses specific sentences and matches them with the most relevant ones to predict a diagnosis. The subsequent numerical results were then compared to the DSM-5 under specific diagnostic criteria. Researchers were surprised by the accurate the LLM’s accurate findings.
ALSO READ | Five documentaries for parents of autistic children
The primary indicators of autism remain the lack of communication skills and social interaction but scientists hope that the LLM will help medical professionals working with various psychiatric, mental health, and neurodevelopmental disorders in their clinical judgements.
