Between July 2022 and December 2024, a public-private partnership in Hubei province, China, delivered what a Lancet Primary Care comment calls the world's largest AI-assisted cervical cancer screening programme, reaching 9.34 million women. Of the 12.67 million eligible women aged 35-64 years, 9.34 million underwent AI-assisted screening within the 2.5-year span following the launch of the universal coverage campaign, a 73.7% coverage rate. Of those screened, 6.27 million (67.1%) lived in rural areas. The AI-assisted diagnostic system showed a 94.7% agreement rate with manual reading by cytopathologists.
The comment notes that population-wide screening is constrained by cytopathologist shortages, fragmented laboratory capacity and rural access barriers, and that evidence for AI has largely come from pilots rather than health-system roll-outs. It says the programme shows how workflow redesign, not just algorithms, can unlock rapid scale-up.
The Hubei model is a collaboration between the Chinese Government and a single contracted AI-technology company, Landing Med, which was selected through competitive bidding. Primary care centres handle community mobilisation, patient registration and collection of cervical exfoliated cells, then upload patient information and send the samples by logistics to Landing Med. Landing Med acts as a centralised, high-throughput diagnostic hub: it receives all samples, prepares the slides and performs automated AI analysis to identify negative samples. Its trained health workers and cytopathologists then manually review all positive or suspected positive samples, as well as a randomly selected 10% of negative samples for quality control.
Slide reading is centralised via cloud computing. Landing Med uploads reports of people with positive results to a shared cloud platform, where local specialist hospitals receive them, manage follow-up and provide colposcopy, biopsy and treatment. The AI-enabled diagnostic and reporting platform reported results to the Hubei Provincial Health Commission and was continuously updated in collaboration with institutions such as Wuhan University.
The programme was supported by 595 million yuan (US$83 million) from central and provincial governments. Of the funds, 55% went to Landing Med's sample screening and diagnosis services and 45% to primary and specialised care centres for personnel training, organisational management and programme implementation. Economic modelling found that AI-assisted liquid-based cytological screening every 5 years was more cost-effective than manually read liquid-based cytology. The authors also note that primary care workers' workloads might increase, creating a workforce paradox.
Details