Entering ChinaEnter symptoms such as “found that the child has been significantly behind in development, language and movement since the age of 2, and cannot complete the communication”, and after a few seconds, the artificial intelligence (AI) model will give a judgment that “it is necessary to be vigilant against rare genetic diseases (such as Rett syndrome, Angelman syndrome, etc.) or complex neurodevelopmental disorders”, and give medical advice such as department visits and supplementary examinations.
This is the scene where the reporter tried the country’s first artificial intelligence model in the field of rare diseases, “Concord Taichu”. The reporter learned from Peking Union Medical College Hospital on the 19th that the large model, jointly developed by the hospital and the Institute of Automation of the Chinese Academy of Sciences, has recently opened the test to patients for the initial consultation and appointment function.
Although there are few patients with a single disease, there are many kinds of rare diseases. It is easy to misdiagnose, miss, and diagnose, which is a difficulty faced by many patients with rare diseases, and artificial intelligence models are expected to become the “key to solving the problem”.
Zhang Shuyang, President of Peking Union Medical College Hospital, introduced that the research and development of artificial intelligence-assisted diagnosis and treatment tools for rare diseases is a key research direction that the rare disease expert team has adhered to for many years, and the research and development of the “Xiehe Taichu” rare disease model is based on the accumulation of China’s rare disease knowledge base for many years and the genetic testing data of the Chinese population, which is the first rare disease model in the world that meets the characteristics of the Chinese population, which can help doctors identify and diagnose rare diseases more accurately and quickly, and further shorten the diagnosis time.
Xu Bo, director of the Institute of Automation of the Chinese Academy of Sciences, said that AI models often need to be trained with massive data, but rare disease cases are scattered and data is scarce, making it difficult for traditional AI models to be effectively trained. In this regard, the R&D team innovates the technical path and adopts the method of cold start of very small samples, which only requires a small amount of data and medical knowledge to achieve the auxiliary decision-making function covering the whole process.
“The ‘Concord Taichu’ model is an important breakthrough in the ‘Chinese plan’ for the diagnosis and treatment of rare diseases.” Zhang Shuyang said that Peking Union Medical College Hospital, as the national leading hospital of the National Rare Disease Diagnosis and Treatment Collaboration Network, will continue to promote the deep integration of AI technology and clinical needs, better help improve grassroots capabilities and build a hierarchical diagnosis and treatment guarantee system, and let the “glimmer” of rare disease diagnosis and treatment illuminate more families.
At present, the initial consultation and appointment function of this large model has been open to patients for testing, and patients can obtain preliminary diagnosis and treatment suggestions through multiple rounds of interactive consultation. In the next step, it will support doctor-side service functions such as medical record writing, gene interpretation, and genetic counseling, and will conduct the second phase of online testing.
It is reported that the follow-up “Xiehe Taichu” rare disease model will be connected to the online diagnosis and treatment service of the rare disease joint outpatient clinic of Peking Union Medical College Hospital, and will be gradually promoted to the hospitals of the National Rare Disease Diagnosis and Treatment Collaboration Network in the future.
