Skip to main content

Ping An launches AI avatars of top Chinese doctors

A new chatbot on its mobile health app features AI-based avatars that can communicate synchronously with patients through text, voice, or video chat.
A patient wearing mask at home in an online consultation.
Photo: Edward Jenner/Pexels

Shanghai-based Ping An Health, the health technology unit of Chinese insurance firm Ping An, has launched a generative AI-powered chatbot on its mobile health application featuring the avatars of real physicians. 

The new feature, Ping An Xin Yi, on the Ping An Health app, provides round-the-clock on-demand access to AI-assisted health consultations delivered through digital avatars. It also offers simplified interpretation of medical reports and laboratory results and personalised medication reminders. 

The chatbot currently covers general practice and will be later expanded to gynecology, internal medicine, pediatrics, surgery, andrology, traditional Chinese medicine, dermatology, and general medicine.

HOW IT WORKS

Ping An Health explained that the chatbot was built using a three-tiered data training structure. The base layer is Ping An Medical Master, the company's large language model based on its five medical databases covering 37,000 diseases and 420,000 disease-related terms; the second layer is a knowledge base from the doctor whom the avatar is modeled from, including their social media content and published materials. The third layer involves fine-tuning from the doctor themselves, including manual annotation and training with their video content. 

Besides genAI, Ping An Xin Yi also utilises natural language processing, machine learning, and medical document recognition. The avatars currently replicate the image and voice of top specialists in proctology, hepatobiliary surgery, and obstetrics and gynecology in China. Users can interact with them synchronously through text, voice, or video chat.

THE LARGER TREND

Digital avatars representing real-life doctors are also featured in the virtual hospital project of Tsinghua University Institute for AI Industry Research. Set for public pilot some time this first quarter, the autonomous and self-evolving virtual healthcare setting will be run by these genAI-driven doctors, which have shown high accuracy in examining, diagnosing, and treating patients in a recent study. 

Ping An Health has doubled down on AI integration in recent years. Last year, it introduced its medical LLM, Ping An Medical Master, and the Ping An Doctor's Home doctor's dashboard. Early in February, it completed the deployment and partial verification of DeepSeek, which it expects to help boost accuracy in health consultations and disease diagnoses. Ping An Health claims its current AI-powered diagnosis and treatment system demonstrates near full accuracy in triage and assisted diagnosis. 

Meanwhile, Ping An Health's telemedicine standards – as applied in its AI-powered online family doctor service Ping An Family Doctor – reportedly served as the model for nationwide standards for virtual care delivery being piloted in China.