Machine learning has taken a significant step forward in identifying potential risk factors for autism in infants. A recent study published in JAMA Network Open by researchers from the Karolinska Institutet has introduced a new machine learning model designed to predict autism in young children. This innovative algorithm has the potential to enhance early detection of autism, a crucial factor in providing timely and effective support to affected individuals.
Utilizing a vast US database called SPARK, which includes information on approximately 30,000 individuals, both with and without autism spectrum disorders, the research team developed four distinct machine-learning models by examining 28 different parameters. These parameters can be collected without the need for extensive assessments or medical tests before a child reaches 24 months of age. The most effective model, named ‘AutMedAI,’ showed an accuracy of nearly 80 percent for children under two years old, making it a valuable tool for healthcare professionals.
According to Kristiina Tammimies, an associate professor at KIND, the study successfully identified around 80% of children with autism from a group of approximately 12,000 individuals using the AutMedAI model. Factors such as the age of the child’s first smile, the age of the first short sentence spoken, and the presence of eating difficulties were found to be strong predictors of autism when combined with other parameters.
Shyam Rajagopalan, the study’s first author and researcher at the Karolinska Institutet, emphasized the importance of early diagnosis for children with autism to ensure effective interventions that support their optimal development. The AI model was found to perform well in identifying children with challenges in social communication, cognitive abilities, and broader developmental delays.
While the research team acknowledges the potential of the AutMedAI model, they emphasize that additional enhancements and clinical validation are essential before implementing it in healthcare settings. The ultimate goal is for the model to complement clinical assessments of autism rather than replace them.
Moving forward, efforts are underway to incorporate genetic data into the algorithm to improve the accuracy of autism forecasts. The study’s findings mark a significant advancement in the early identification of autism risk factors and the potential for improving the quality of life for individuals and their families.
References:
– Rajagopalan, S. S., et al. (2024) Machine Learning Prediction of Autism Spectrum Disorder From a Minimal Set of Medical and Background Information. JAMA Network Open. doi.org/10.1001/jamanetworkopen.2024.29229.
Source:
Karolinska Institutet
