On Nov 1, Shaanxi Daily reported that six provincial departments in Shaanxi, led by the provincial health commission, have jointly issued a work plan to promote and regulate "AI plus healthcare" applications, setting phased goals and eight key tasks alongside data-security and safeguard measures.
The plan lays out a two-stage timetable. By 2027, the province aims to build high-quality datasets and trusted data spaces, form a matrix of clinical specialty vertical large models and agents, bring AI-assisted decision support and smart outpatient services into wide use, and fold health data into the province's integrated government big-data system. By 2030, AI-assisted primary care is expected to reach near-full coverage, with tier-2 hospitals and above routinely running AI-assisted imaging diagnosis and clinical decision support, under a basically complete standards framework.
The eight tasks map the main touchpoints of medical AI. At the primary-care level, the plan reinforces the "Sanqin Smart Medical Assistant," pushes AI triage for common conditions and medication-safety prompts, and targets full coverage of the three shared centers for ECG, imaging and lab diagnostics, plus whole-journey smart health management for key and chronic-disease populations. On the clinical side, it calls for high-quality datasets, smarter review at tier-2 hospitals and above, and AI in visits, follow-ups and AR surgical training. Patient services include smart mutual recognition of test results.
Supporting measures follow the same pattern: a provincial health information platform, coordinated AI computing supply, AI skills folded into continuing education, and a security framework covering the full data lifecycle with explicit personal-privacy protection.
The title pairs "promote" with "regulate," and that is the notable design choice: the provincial regulator is drawing a roadmap and a boundary at the same time. Whether it works will show up first in primary-care usage rates and in how far data-sharing mechanisms actually get implemented.