(Yicai) March 18 -- Chinese tech giant Tencent Holdings has started helping Shenzhen Third People's Hospital create targeted health check-up plans through artificial intelligence.

The smart concierge service gathers health information from patients such as age, medical history, lifestyle habits, and work environment, and recommends targeted check-up plans for them, Yicai learned on the spot.

The system goes beyond the usual questions placed in hospital reception. Office workers who frequently stay up late are suggested screenings for cardiovascular and cerebrovascular risks. Moreover, people with a family history of a certain disease are proposed early screening for related diseases.

Feng Cheng, head of the health management department at the public hospital, said to Yicai that the system can interpret health reports, advising patients with health issues to visit relevant hospital departments for further consultation.

The program is built on Tencent's Hunyuan large language model and AI startup DeepSeek's LLM, combining the hospital's accumulated clinical experience in diagnosing and treating various diseases over the years.

Hospital President Lu Hongzhou said that AI can be widely applied in the medical field. Initially, it can take on relatively routine and repetitive tasks, helping to prevent human errors and saving time. In the future, AI will be deeply integrated into the entire medical process, including disease diagnosis and treatment, as well as new drug development, Lu added.

"Since the beginning of the year, we have clearly felt an increasing demand for AI applications in hospitals," said Wu Zhigang, general manager of the Shenzhen-based internet firm's health platform.

Wu attributed the change to the rapid development of LLMs. The trend of open-source LLMs has significantly reduced the deployment costs for hospitals, which are expected to continue to explore more AI application scenarios in the future, he added.

On the road to advanced medical AI, a key task is reducing hallucinations. Models need to integrate a vast amount of specialized medical knowledge for training and fine-tuning, ensuring quality control, Wu explained.

However, Wu emphasized that AI will not replace doctors but is designed to assist them in improving diagnostic efficiency. Ultimately, it is still the responsibility of doctors to ensure the professionalism and accuracy of each diagnosis.

Editors: Tang Shihua, Emmi Laine