AB
AB
AiBoss

All Categories

project

Peking Union Medical College Hospital and Chinese Academy of Sciences jointly launched a large-scale AI model for rare diseases.

Peking Union Medical College Hospital's Taichu AI model, jointly developed by the Peking Union Medical College Hospital and the Institute of Automation, Chinese Academy of Sciences, is the first large-scale AI model in China dedicated to rare diseases and has officially entered clinical application. The model is based on years of accumulated knowledge of rare diseases in my country and the Chinese population...

What is the beginning of the Union?

Peking Union Medical College Hospital (PUMC) and the Institute of Automation, Chinese Academy of Sciences, have jointly developed China's first large-scale AI model for rare diseases, which has officially entered clinical application. Based on years of accumulated knowledge in my country's rare disease knowledge base and genetic testing data of the Chinese population, the model is the world's first large-scale rare disease model specifically tailored to the characteristics of the Chinese population. Employing a very small sample cold-start technique, it requires only a small amount of data and medical knowledge to integrate, enabling end-to-end decision support.

The main functions of Concord and Primordial

  • Initial consultation and advicePatients can obtain preliminary treatment suggestions in a short time through multiple rounds of interactive consultations with the model.
  • Assisting doctors in decision-makingThe model constructs a progressive reasoning chain of "symptoms - examination - differential diagnosis," which is highly consistent with doctors' clinical thinking and can help doctors quickly grasp the diagnostic and treatment approach.
  • Medical record writing and gene interpretationIn the future, it will support doctor-side service functions such as medical record writing, gene interpretation and genetic counseling.
  • Knowledge self-iterationBy recording interactions with patients, the diagnosis and treatment process can be evaluated, enabling proactive updates and evolution of decision-driven data, thus forming a closed loop of "clinical use - data feedback - model iteration".
  • Suppressing AI Illusions: Construct a multi-dimensional, traceable knowledge base to effectively suppress potential "illusions" in models and enhance the credibility of clinical decision-making.

The technical principles of Concord and Primordial

  • Extremely small sample cold start technologyTo address the issues of scattered rare disease cases and scarce data, the research team adopted a very small sample cold start approach, which requires only a small amount of data to be integrated with medical knowledge, enabling full-process auxiliary decision-making functions.
  • "Data + Knowledge" hybrid drivingThe model combines years of accumulated rare disease knowledge base data with gene testing data of the Chinese population. Through a hybrid "data + knowledge" approach, it enhances the model's decision-making logic and credibility.
  • Deep reasoning abilityIntroducing the deep reasoning capabilities of DeepSeek-R1, it constructs a progressive reasoning chain of "symptoms - examination - differential diagnosis," which is highly consistent with doctors' clinical thinking and can demonstrate the key nodes and branch logic from symptoms to diagnosis.
  • Proactive perception and interaction and closed-loop iterationThe model actively perceives changes in the patient's condition and updates the decision-making process through multiple rounds of interaction, forming a closed loop of "clinical use - data feedback - model iteration" to achieve autonomous knowledge iteration.
  • Multi-dimensional traceable knowledge baseTo suppress AI "illusions," the model has built a multi-dimensional, traceable knowledge base, integrated authoritative data, dynamically updated knowledge, and enhanced the credibility of clinical decisions.

Project address of Concord Taichu

Application scenarios of Concord Taichu

  • Initial patient consultation and rapid consultationPatients can obtain preliminary treatment suggestions in a short time through multiple rounds of interactive consultations with Peking Union Medical College Hospital and Taichu Hospital, which helps to quickly identify possible directions for rare diseases.
  • Promotion of the Rare Disease Diagnosis and Treatment Collaboration NetworkThe model has been piloted at the Rare Disease Joint Clinic of Peking Union Medical College Hospital for a year with good results. It will subsequently be integrated into the online medical services of Peking Union Medical College Hospital and gradually extended to hospitals within the national rare disease collaborative network, contributing to the construction of a tiered medical service system.
  • Knowledge self-iteration and closed-loop optimizationThe model evaluates the diagnosis and treatment process through interactions with patients, enabling proactive updates and evolution of decision-driven data, forming a closed loop of "clinical use - data feedback - model iteration" to continuously optimize diagnosis and treatment capabilities.
Back