EMR + AI cuts screening errors
An EMR study from rural China found integrating AI-driven recall and tracking measurably reduced both over‑ and under‑screening for cervical cancer—coverage rose and inappropriate repeat tests fell. That underscores the power of data integration across cytology, HPV results, and clinical workflows to tighten screening quality and follow‑up timing. (nature.com)
The study analyzed 33,362 women aged 35–64 in Wuxiang County and found only 19.9% of screening events were guideline‑adherent, 29.5% were over‑screened, and 50.6% were under‑ or unscreened. (nature.com) A county‑wide EMR platform deployed in 2022 coincided with a sharp decline from pre‑EMR over‑screening peaks of 62.3% measured during 2018–2021. (nature.com) A separate multi‑county deployment described the same digital approach as an OCR‑enabled “One‑ID” platform that used deterministic identity linkage and real‑time duplicate‑test alerts to manage population screening across 153,978 encounters in six rural counties. (cancerbiomed.org) Performance metrics after One‑ID deployment show over‑screening fell from 12.64% in 2023 to 0.17% in 2024 (absolute reduction 12.17%, 95% CI 11.94–12.40; P<0.001), the share of women with a first screen within the prior 3 years rose from 78.3% to 88.2%, and colposcopy completion improved from 64.1% to 84.9%. (cancerbiomed.org) Case‑level outcomes also improved: CIN2+ detection increased from 0.35% (2021–2023 pooled) to 0.67% in 2024, and CIN2+ management completion rose from 56.0% to 76.2% (95% CI: 13.3–27.2; P<0.001). (cancerbiomed.org) The npj Digital Medicine manuscript (Zhu et al., corresponding authors Dr. Hanyue Ding and Prof. Youlin Qiao) was accepted 20 February 2026 and posted online 20 March 2026, with declared funding from the Bill & Melinda Gates Foundation and Tencent Inclusive Health Lab. (crossref.org)
