Autism, a neurodevelopmental disorder, can be diagnosed as early as age 3. Often called “children from the stars,” autistic kids require early detection and intervention. Sustained rehabilitation training and special education are key to improving their social adaptability. Currently, China has over 13 million autistic patients. However, due to uneven distribution of medical resources and public misconceptions, the average age of diagnosis for autistic children in China is around 4, well beyond the critical pre-3-year intervention window.
Addressing this issue, the autism AI agent “StellarCare AI,” developed under the leadership of the Hangzhou Institute for Advanced Study, Chinese Academy of Sciences (CAS), was recently launched. With molecular medicine and artificial intelligence as its core engines, and integrated multimodal data analysis capabilities, StellarCare AI aims to provide innovative solutions for autistic patients and their families, including “precision screening, personalized intervention, and intelligent support.”
StellarCare AI originated from a public welfare research project. In November 2024, with support from the Fujian “Yu Ni” Autistic Children Service Center, a team of scientists from across biotechnology, molecular medicine, artificial intelligence, and big data analysis fields was formed. Led by academician Tan Weihong, director of the Hangzhou Institute for Advanced Study, CAS, and researcher Han Da from the same institute, the team included experts from Zhejiang University, the University of Hong Kong, Shanghai Jiao Tong University, and the Zhejiang Lab. They initiated the public welfare project “Application of Multimodal AI Technology in Early Autism Screening.”
In March this year, the research team launched a pilot program at the Licheng District Maternal and Child Health Hospital in Quanzhou, Fujian Province, collecting key data such as eye movements, facial features, and behaviors from over 1,300 children. The team integrated questionnaire scales, eye-tracking data, microexpression analysis, behavior videos, brain imaging, molecular diagnostics, and genomic data to build a multidimensional screening model. Based on a large language model, StellarCare AI has studied authoritative diagnostic guidelines and over 25,000 cutting-edge global research results. It can track the latest scientific research and clinical progress in real time, enabling rapid and accurate interaction with doctors, researchers, and parents.
Chen Guangyong, a researcher at the Hangzhou Institute for Advanced Study, CAS, stated that pilot data shows StellarCare AI has a screening accuracy rate exceeding 85% and has advanced the diagnosis window to 18 months. This breakthrough provides strong support for the early detection of autism, allowing more children to receive timely intervention during the critical period.
Han Da introduced three main application scenarios for StellarCare AI.
On the family side, StellarCare AI acts as a “family doctor,” offering full-chain services including “behavior analysis—personalized intervention—24-hour AI Q&A.” Parents can consult it about autism-related issues at any time to obtain professional advice and guidance.
In the medical field, StellarCare AI serves as a “medical assistant,” capable of interpreting multimodal data collected clinically. It helps doctors and scientists conduct in-depth research on autism and related cutting-edge topics, improving screening efficiency and accuracy, and breaking down barriers between clinical and research data.
At the societal level, StellarCare AI functions as an “expert think tank,” linking the government, hospitals, research institutions, and public welfare organizations. It gathers wisdom and resources from all parties to jointly build a full-life-cycle support network covering “screening—diagnosis—intervention—support.” It provides more comprehensive and cutting-edge professional knowledge and data support, promoting policy optimization and upgrading public awareness.
Currently, the research team is working to build a Chinese-specific multimodal database for autism to further enhance screening accuracy and intervention effectiveness.
“The warmth of technology lies in responding to social needs. I believe that in the future, more ‘children from the stars’ will better integrate into society and shine with their own light through the power of technology,” said Tan Weihong.
Related topics:
Young Adults with ADHD Symptoms Prefer Stimulating Music
Is It Time to Switch from Zoloft to Prozac? Here’s What You Need to Know
New Brainwave Biomarkers Offer Precision Target for OCD Treatment
