AI-Enabled Family Doctor Services for Community Blood Lipid Management in Shanghai

Editor's note:

The International Business Leaders' Advisory Council for the Mayor (IBLAC) brings together top executives from leading multinationals to exchange insights on Shanghai's development strategy. Each company brings forward a paper that draws from their global practices and local insights. Shanghai Daily has selected and summarized some of the astute suggestions to this city.

Cardiovascular disease remains a leading cause of death and disability in China, and elevated low-density lipoprotein cholesterol (LDL-C) is a well-established, modifiable risk factor for atherosclerotic cardiovascular disease (ASCVD).

Yet dyslipidemia is frequently asymptomatic and inadequately managed: A national meta-analysis reported an adult prevalence of 42.1 percent, while awareness, treatment and control rates were only 18.2 percent, 11.6 percent and 5.4 percent, respectively.

Shanghai's Community Health Management Work Guideline for Integrated Chronic Disease Prevention and Control, effective from January 1, 2026, represents an important policy advance by incorporating blood lipid management into community health services and moving the management gateway to residents aged 18 and above. The next challenge is to translate policy inclusion into continuous, effective and scalable management.

Novartis proposes a "Family Doctor + AI" model in which family doctors remain accountable for risk assessment, clinical decisions, treatment recommendations, referral and high-risk case management, while artificial intelligence supports repetitive and standardized tasks. These may include risk identification, follow-up reminders, health education, lifestyle support, retesting prompts, medication-adherence support and abnormal-result alerts. The objective is not to replace physicians, but to expand the reach, continuity and consistency of primary care.

This proposal is supported by an early health-economic evaluation conducted by the National Health Commission Key Laboratory of Health Technology Assessment at Fudan University and the Fudan University Center for Health Technology Assessment.

Using published parameters and explicit assumptions, researchers constructed a Markov cohort state-transition model with 13 health states, annual cycles and simulation to age 85. The model examined approximately 3.9 million Shanghai residents aged 20-59 with hyperlipidemia but without established ASCVD, comparing routine care, family doctor management, and family doctor management supported by AI.

The population estimate was derived from a Shanghai epidemiological study and the 2020 Shanghai Population Census. Outcomes included costs, quality-adjusted life years (QALYs), first myocardial infarctions, first ischemic strokes and ASCVD-free survival.

Under the baseline assumptions, family doctor management was less costly and more effective than routine care. Over the modeled time horizon, adding AI produced a further gain of 0.071 QALYs per person at an incremental cost of 2,715 yuan (US$495), yielding an incremental cost-utility ratio of 38,479 yuan per QALY.

This was below the study's conservative willingness-to-pay threshold of 99,665 yuan per QALY. At the population level, the model estimated that Family Doctor + AI could prevent approximately 12,967 additional first myocardial infarctions and 34,901 additional first ischemic strokes compared with family doctor management alone. Probabilistic sensitivity analysis indicated a 92.15 percent probability that the combined model would be cost-effective at the one-times-GDP-per-capita threshold.

These findings are projections rather than evidence of proven real-world effectiveness. Results depend on whether residents are reached, recommended interventions are initiated and sustained, LDL-C reductions are maintained, AI operating costs remain controlled, and family doctors can integrate the tools into routine workflows. Real-world pilots are therefore essential.

Novartis recommends selecting representative districts and community health centers for demonstration projects; establishing a closed-loop pathway from screening and assessment to intervention, retesting, adjustment and referral; integrating AI outputs into existing family doctor workstations; and creating unified standards for data, technical interfaces and quality evaluation.

Pilots should measure screening coverage, follow-up completion, treatment initiation, LDL-C target attainment, cardiovascular events, patient engagement, workload and total costs. Data security, algorithm transparency, professional accountability and resident acceptance must be addressed from the outset.

With its strong primary-care network, digital infrastructure and research capacity, Shanghai is well positioned to translate frontier technology into an evidence-based, evaluable and replicable model of chronic disease management. Blood lipid management could provide a practical starting point before broader application to the integrated management of hypertension, hyperglycemia and hyperlipidemia.