Plymouth PRECISE-AGZ Model Flags Bladder Cancer Signals in EHRs Up to Five Years Early
The PRECISE-AGZ AI system mined routine UK health records to detect bladder cancer earlier than current symptom-based rules, uncovering novel associations while outperforming existing referral criteria. Retrospective design and lack of external validation constrain immediate clinical use. Prospective trials are required to confirm net benefit and rule out bias.
Next steps include multi-site prospective cohorts within NHS England and linkage to genomic datasets to test whether the tamoxifen and neurological signals reflect shared biology or surveillance bias. Regulatory review by NICE and MHRA will hinge on demonstration of net reduction in invasive procedures without excess missed cancers.
VITALIS: Prospective NHS validation of PRECISE-AGZ will complete within 30 months and report at least 70% reduction in cystoscopies among gray-zone patients with no increase in missed cancers at 12-month follow-up.
Sources (3)
- [1]Primary Source(https://ieeexplore.ieee.org/document/10687234)
- [2]Supporting Source(https://www.bmj.com/content/385/bmj-2023-078943)
- [3]Supporting Source(https://www.thelancet.com/journals/lanonc/article/PIIS1470-2045(24)00112-3)