正文 · AI 翻译
On the afternoon of October 19, 2026, the Medical and Health World Models session was held in Zhongshan Hall at the Nanjing International Expo Center.
On the afternoon of October 19, 2026, the Medical and Health World Models session, one of the key parallel events of the 2026 Physical AI and World Models Conference, was held in Zhongshan Hall at the Nanjing International Expo Center. Under the theme “Foreseeing Disease, Simulating Life: World Models Reshape Medical Artificial Intelligence,” the session brought together academicians and experts, clinical practitioners, university researchers, and representatives of the artificial intelligence, pharmaceutical, and data industries to discuss key issues including how disease states evolve, how responses to interventions can be simulated, medical knowledge modeling, real-world validation, and medical data security. Unlike traditional artificial intelligence, which focuses on recognizing individual features and making static predictions, medical world models also explore how to understand changes in a patient’s condition over time and the possible outcomes of different interventions.
This year’s conference, themed “Modeling the Physical World, Intelligently Enabling a Better Future,” was jointly organized by Jiangsu Data Group, Nanjing Big Data Group, and the Chinese Association for Artificial Intelligence. According to information officially released by the conference, it is China’s first high-profile, comprehensive industry conference focused on physical AI and world models. Structured around “1 opening ceremony + 1 accompanying exhibition + N thematic seminars,” it provides a forum for exchanges on world model theory, healthcare, industrial applications, data development, autonomous driving, and embodied intelligence. The Medical and Health World Models session was hosted by Nanjing Big Data Group Co., Ltd. and Jiangsu Shuxintong Data Services Co., Ltd.; organized by Nanjing Trusted Data Services Co., Ltd., the Nanjing Data Elements Innovation Center (Ningshufu), and the Nanjing Data Association; and supported by the Nanjing Office for Advancing the Biomedical Industry and the Nanjing Office for Advancing the Artificial Intelligence (Software) Industry. Teng Gaojun, an academician of the Chinese Academy of Sciences and director of the Division of Medicine and Life Sciences at Southeast University, chaired the session; Nanjing Vice Mayor Sun Baijun attended and delivered remarks.
Eight keynote speeches followed a progression from “clinical needs to modeling methods to research validation to application exploration.” Liu Yun of Jiangsu Provincial People’s Hospital presented “Artificial Intelligence Transforms the Paradigm of Chronic Disease Management,” focusing on long-term chronic disease management and intervention research; Jiang Qing of Nanjing Drum Tower Hospital discussed “A New Method for Generalization Across the Medical Continuum,” examining how medical models apply as populations and settings change; Song Jian of Peking Union Medical College Hospital, Chinese Academy of Medical Sciences, spoke on “Using World Models to Rehearse the Future of Rehabilitation: A Paradigm Upgrade for Continuous Diagnosis and Treatment Platforms,” focusing on continuous assessment of rehabilitation status and simulation of future outcomes; Jiao Yun of Zhongda Hospital, Southeast University, shared “Building an Ontology-Based Medical World Model,” exploring how medical concepts and relationships can be organized into computable representations of knowledge. Turning to scientific discovery and industry practice, Tang Peifeng of Suzhou Mega Technology Co., Ltd. presented “From AI Models to Wet-Lab Validation: Accelerating Innovation Driven by Clinical Needs,” focusing on a research loop linking AI-generated hypotheses with experimental feedback; Shao Yang of Nanjing Geneseeq Technology Inc. shared “Using AI to Mine Genomic Big Data for New Drug Targets,” discussing the use of genomic data to identify potential drug targets; Ji Jianmin, a collaborating expert with Tuodao Medical Technology Co., Ltd., introduced “The Evolution and Practice of Embodied Models for Endoscopic Surgery,” focusing on perception, decision-making, and action in surgical settings; and Liu Shengping of Unisound AI Technology Co., Ltd. concluded with “Medical World Models: Simulating Clinical Processes Based on Synthetic Diagnosis and Treatment Trajectories,” exploring methods for modeling continuous diagnosis and treatment. The eight presentations spanned medical understanding, validation, and action, while also underscoring that the reliability of medical world models must be continually tested in real-world settings.
A presentation of major achievements showcased innovations in six areas, covering high-quality datasets, an integrated medical and health world model system, disease-specific world models, a data foundation, and intelligent applications. The first initiative announced, the “Joint Initiative to Build High-Quality Datasets for Medical World Models,” involves the Information Center of the Nanjing Municipal Health Commission, Jiangsu Provincial People’s Hospital, Nanjing Drum Tower Hospital, Nanjing Hospital of Integrated Traditional Chinese and Western Medicine, Nanjing Maternity and Child Health Care Hospital, Nanjing Brain Hospital, Nanjing Central Hospital, Nanjing Jiangning Hospital, and other organizations. It focuses on collaboration among institutions, data standards, and data quality management to provide sustained data support for the training, evaluation, and clinical research of medical world models, subject to legal compliance and the protection of patient privacy.
Next, Jiangsu Provincial People’s Hospital and Nanjing Trusted Data Services Co., Ltd., a subsidiary of Nanjing Big Data Group, together with teams from H3C, UnionTech Software, and Yashan Database, unveiled an “Integrated Medical and Health World Model System.” It deeply integrates world model technology with domestically produced computing equipment, operating systems, and databases to build a secure, trustworthy, independently controlled intelligent computing platform deployed locally in hospitals. It aims to shift medical research from a paradigm of “humans propose hypotheses and AI assists with validation” to one in which “AI proposes scientific hypotheses and humans validate them scientifically,” helping usher medical research into a new era of AI-driven scientific discovery. Jiangsu Provincial People’s Hospital and Nanjing Trusted Data Services Co., Ltd. jointly released five chronic disease world models covering insulin resistance, diabetes medication, continuous hypertension management, blood lipids and atherosclerosis, and chronic kidney disease risk, with a focus on the dynamic study of changes in disease states and responses to interventions. A “Thalassemia Screening World Model,” jointly developed by Wuzhou Workers’ Hospital and the Guangxi Precision Medicine Industrial Technology Research Institute, was also unveiled. Based on the principle that a hospital’s raw data remains within its own domain, the project explores training and validating the model within the hospital and, subject to security and regulatory compliance and local clinical validation, gradually using its screening capabilities in medical cooperation with ASEAN. It advances an innovative practice described as “data stays in the hospital, intelligence reaches ASEAN.”
In addition, Zhongda Hospital, Southeast University, and Yidu Tech unveiled a “High-Quality Data Foundation for Medical World Models,” exploring support for calculating states and training and validating models through data quality management, unified medical terminology, and the organization of time-series information. Neusoft Group unveiled its “Neusoft Citywide Digital and Intelligent Health Solution Based on the Tianyi Medical Large Model,” focusing on applications of medical large models in regional health services, coordination of hospital operations, and other settings. Nanjing Hanwei Public Health Research Institute Co., Ltd. unveiled an “Intelligent Medical Research Product Based on World Models,” exploring intelligent assistance with medical research questions and support for the research process. Each of these achievements has a different focus; together, they show a direction for the coordinated development of data, models, applications, and research.
Discussions at the session indicated that the value of medical artificial intelligence is expanding beyond answering medical questions and identifying isolated risks to describing disease states, studying changes in disease progression, and comparing possible intervention paths. The key to world models is not only whether their predictions are accurate, but also whether the basis for those predictions is clear, whether simulations can be tested against real-world data, whether they are reliable across different healthcare settings, and whether doctors can understand and use them prudently. External validation, clinical evaluation, and ongoing oversight are still needed to gradually clarify where these technologies can be used; model simulations must not replace doctors’ decisions on diagnosis and treatment.
Jiangsu brings together strong medical institutions, universities and research institutes, and software and information industry resources, providing a sound foundation for collaboration among industry, academia, and research institutions on medical world models. Drawing on medical institutions, research teams, and data service companies, Nanjing is conducting joint research to connect the supply of high-quality data, medical knowledge modeling, model training, clinical validation, and wider adoption. With China–ASEAN healthcare cooperation in view, the thalassemia screening world model presented at the session also proposed a cooperative approach in which “raw data does not leave its domain, while model capabilities serve across regions,” raising new research questions about the secure use of medical data and the wider adoption of artificial intelligence advances.
From “foreseeing disease” to “simulating life,” this session’s eight keynote speeches and presentations of achievements in six areas showed medical artificial intelligence being explored at multiple levels, from theoretical research and trusted data to clinical settings. Next, the organizations involved will continue to pursue interdisciplinary collaboration, model validation, and application evaluation based on real clinical needs, so that world models can gradually move from research concepts toward medical innovations that can be tested, deployed, and continually improved, better serving doctors and patients.
This article was provided by Nanjing Big Data Group and republished by QbitAI with permission. The views expressed are those of the original author.