Researchers at Yonsei University College of Medicine in South Korea have successfully trained artificial intelligence to identify autism in children, achieving a remarkable 100% accuracy rate. The innovative algorithm utilizes high-resolution photographs of the patients’ eyes, providing a non-invasive and efficient screening method.
Study Details and Methodology:
The study involved 958 participants, averaging between 7 to 8 years old, whose retinas were photographed. Half of the participants were diagnosed with Autism Spectrum Disorder (ASD), while the other half formed a control group, matched for age and gender. The severity of ASD symptoms was evaluated using calibrated Autism Diagnostic Observation Scale (ADOS-2) and Social Responsiveness Scale (SRS-2) severity scores. A comprehensive collection of 1,890 images was gathered, forming the dataset for the development of a sophisticated deep learning algorithm.
The Advancements and Implications:
The resultant AI-driven model serves as a significant milestone in the field, offering an objective and accurate screening tool for childhood autism. Moreover, the utilization of retinal photographs extends the potential of assessing symptom severity in adults, indicating broader applications beyond childhood diagnosis, adds NIX Solutions. For further insights into the study, detailed information can be accessed via this link.
The study represents a pivotal breakthrough in utilizing AI technology for autism detection, showcasing promising prospects for enhanced diagnosis and understanding of the condition.