Researchers at Karolinska Institutet have developed a machine learning model that can predict autism in young children with nearly 80% accuracy, using limited information. This breakthrough could lead to earlier interventions and improved outcomes for children with autism spectrum disorders.
The study, published in JAMA Network Open, introduces a new AI model named ‘AutMedAI’ that analyzes 28 different parameters to identify patterns associated with autism. These parameters include readily available information about children under 24 months of age, without requiring extensive assessments or medical tests.
Leveraging Machine Learning for Early Diagnosis
The research team, led by Associate Professor Kristiina Tammimies, used data from approximately 30,000 individuals with and without autism spectrum disorders from a large US database called SPARK. By applying machine learning techniques to this vast dataset, they developed four distinct models to identify autism-related patterns.
“With an accuracy of almost 80 percent for children under the age of two, we hope that this will be a valuable tool for healthcare,” says Tammimies, emphasizing the potential impact of the AutMedAI model.
The study revealed that certain combinations of parameters were particularly strong predictors of autism. These include the age at which a child first smiles, speaks their first short sentence, and the presence of eating difficulties.
Implications for Early Intervention and Support
Early diagnosis of autism is crucial for implementing effective interventions that can help children develop optimally. The AutMedAI model showed promising results in identifying children with more extensive difficulties in social communication and cognitive ability, as well as those with general developmental delays.
Shyam Rajagopalan, the study’s first author and an affiliated researcher at Karolinska Institutet, explains the significance of these findings: “This can drastically change the conditions for early diagnosis and interventions, and ultimately improve the quality of life for many individuals and their families.”
The researchers are now planning to further improve and validate the model in clinical settings. They are also working on incorporating genetic information into the model, which could lead to even more specific and accurate predictions.
However, Tammimies cautions that the model is not intended to replace clinical assessments: “To ensure that the model is reliable enough to be implemented in clinical contexts, rigorous work and careful validation are required. I want to emphasize that our goal is for the model to become a valuable tool for health care, and it is not intended to replace a clinical assessment of autism.”
Why it matters: Autism spectrum disorders affect approximately 1 in 36 children in the United States, according to the Centers for Disease Control and Prevention. Early detection and intervention can significantly improve outcomes for these children, enhancing their social skills, communication abilities, and overall quality of life. The AutMedAI model’s ability to predict autism with high accuracy using readily available information could lead to earlier diagnoses, allowing for timely interventions and support for children and their families.
This research represents a significant step forward in the use of artificial intelligence in healthcare, particularly in the field of developmental disorders. By harnessing the power of machine learning and large datasets, researchers are opening new avenues for early detection and personalized treatment approaches in autism spectrum disorders.
As the field of AI in healthcare continues to evolve, tools like AutMedAI could become invaluable assets for healthcare professionals, potentially reducing diagnostic delays and improving access to early intervention services for children with autism spectrum disorders.
