The "15th Five-Year Plan for National Health" proposes to use digital intelligence technology to empower medical health. The National Health and Wellness Conference also made it clear that medical service capabilities should be improved, medical experience should be optimized, and people's health needs should be met. On November 5, Wang Xiaohang, chief technology officer of Ping An Group, delivered a speech at the Wuzhen Summit of the 2026 World Internet Conference, sharing Ping An’s exploration of using AI medical care to open up full service links for complex diseases.
Wang Xiaohang said that for patients with complex diseases, obtaining medical information is only the first step. What is more important is to obtain high-level professional diagnosis and treatment suggestions, find suitable experts, and receive continuous rehabilitation and health management after treatment.
Ping An relies on its self-developed large-scale medical model to implement AI doctor-related services. So far, Ping An's AI doctors have provided nearly 300,000 professional medical services on average every day. In the first three quarters of 2026, the AI doctors have been used more than 62 million times, providing a complete service chain before, during and after diagnosis. Relying on its professional resources such as Peking University Medical Care and Concordia Innovation Port, it connects online and offline professional medical services through the "four-to" service network of online, hospital, enterprise and home.
AI participates in the diagnosis and treatment of complex diseases, experts control key decisions
According to Wang Xiaohang, in terms of technical capacity building, Ping An AI doctors have built three-level AI capabilities: general practice, specialist, and MDT (multidisciplinary consultation). The consistency of multidisciplinary diagnosis and treatment plans for complex diseases with expert conclusions is about 90%, and the medical red line rate is controlled at 0.3%. In April this year, Ping An Medical's large model 3.5 scored 57.27 points in the internationally authoritative HealthBench Hard evaluation, ranking first in the world. As of the end of September this year, Ping An's AI doctors can accurately diagnose more than 11,300 diseases, with an accuracy of 96% in assisting diagnosis and treatment.
In terms of diagnosis and treatment services, Ping An uses multi-agent collaboration technology to integrate key clinical information such as patient history, pathology, imaging, and molecular testing, sort out key diagnosis and treatment issues, and output multidisciplinary diagnosis and treatment reference plans. All key judgments are reviewed by clinical experts, and expert consultation is initiated when necessary, forming a service model of "AI problem discovery, multidisciplinary collaboration, and expert control."
AI accurate matching expert, Ping An coordinates professional medical resources
According to Ping An's third quarter data, Ping An Group currently serves 256 million individual customers, has 180 million health files and more than 2,000 expert-labeled data on complex diseases. The huge amount of high-quality data provides a good foundation for model training.
As the medical and health traffic portal of Ping An Group, Ping An AI doctors conduct multi-dimensional matching based on disease stage, clinical evidence, and expert expertise, and select matching experts based on disease types, molecular test results, and professional practical experience. The professional team then reviews the matching logic and coordinates and implements the entire medical process including consultation, registration, examination, and hospitalization. At present, 25% of the demand for offline medical auxiliary services such as Ping An Green Pass and accompanying medical services are first received and accurately distributed by Ping An AI doctors.
At the same time, in terms of offline medical resource construction, Ping An has more than 38,000 domestic cooperative hospitals, and its overseas medical network covers 35 countries and more than 1,300 medical institutions, including 40 national-level high-level experts. More than 70% of service needs are distributed by AI.
AI dynamic assessment and proactive intervention empower critical illness health and rehabilitation management
Ping An's AI doctors have also expanded their capabilities to include postoperative rehabilitation, specialized disease follow-up, and long-term disease course management. The model combines the patient's treatment history, reexamination results, and changes in physical indicators to carry out risk monitoring and assist doctors in completing continuous follow-up. Taking gastric cancer surgery as an example, AI can not only identify risks such as weight loss and insufficient nutritional intake, but also identify easily overlooked danger signs such as "weight regain but abnormal increase in abdominal circumference" and provide timely warnings for professional assessment.
Relying on the clinical and rehabilitation expert resources of Peking University Medical Care, Ping An has built a specialized disease management system around major diseases such as gastric cancer and breast cancer, integrating AI into the complete process of "professional assessment - rehabilitation plan - active intervention". Ping An's health management community currently has 2 million members. Empowered by AI, the patient compliance and health management compliance rate has reached 80%.
Wang Xiaohang said that Ping An's AI doctors do not replace clinicians, but amplify the accessibility of high-quality medical services. Ping An will continue to leverage its strategic advantages of "comprehensive finance + medical and elderly care" dual-wheel drive and technology empowerment to promote AI medical care from "able to answer" to "competent", continue to deepen the diagnosis and treatment of complex diseases, serious illness rehabilitation, and long-term health management services, and provide more accessible and high-quality medical services to the society.
