Digital and Intelligent Health Forum | Hu Jianping explores path to building AI-native hospitals

The core transformation of healthcare informatization in the digital-intelligent era lies in reconstructing information architecture and reshaping service models, said Hu Jianping, vice-president of the China Health Information and Big Data Association, at the Digital and Intelligent Health Sub-Forum of the 2026 World Internet Conference Digital Silk Road Development Forum on July 22.

Hu highlighted four key shortcomings in the current digitization efforts of domestic hospitals: the attachment of data to functional modules, processes embedded rigidly in codes, AI largely functioning as external add-ons, and human-computer interactions relying heavily on command-based codes.

He outlined an innovative concept of AI-native hospitals that use AI to reconstruct their underlying systems. Interaction methods shift to semantic dialogues, enabling systems to automatically identify needs and generate diagnostic workflows. The framework integrates five core dimensions: native perception, native logic, native collaboration, native interaction, and native security. It also introduces AI training methods to minimize hallucinations by embedding safety and ethical rules into all business processes.

To implement the system, a "five layers and two verticals" architecture is adopted. The five layers include data, capability, orchestration, multi-agent systems, and interaction hierarchy, while the two verticals encompass safety audits and end-to-end monitoring mechanisms.