According to the World Health Organization’s (WHO) report in 2023, Autistic Spectrum Disorder (ASD) affects 1 in 100 children worldwide. ASD is a neurodevelopmental condition characterized by challenges in communication and social interaction. Currently used ASD screening tools face some challenges associated with biases and subjectivity in their assessments. Artificial Intelligence (AI) through machine learning methods is developing predictive models for ASD, given its capability to manage extensive datasets and identify hidden patterns within the data.
Karolinska Institute researchers, based in Sweden, introduced their latest AI model, a screening system named AutMedAI, which helps specialists identify toddlers who may be autistic with approximately 80% accuracy in children under the age of 2. The model utilized the largest datasets in ASD research, the SPARK, encompassing medical and behavioral data from individuals diagnosed with ASD. The tool is utilized to screen for ASD in infancy and early childhood based on minimal background, developmental, and medical information.
While it does not diagnose autism, it utilizes available information to identify individuals with elevated likelihood for autism at an earlier stage, enabling earlier identification and intervention. The significance of early diagnosis for targeted intervention and improved outcomes extends beyond ASD. AI models such as AutMedAI will aid the health system in identifying a myriad of diseases worldwide if deployed globally.
