An artificial intelligence model trained on 25 years of Welsh health records flagged bladder-cancer risk up to five years before diagnosis, outperforming NHS referral guidelines and potentially sparing thousands of patients unnecessary invasive cystoscopies, researchers at the University of Plymouth reported this week.
The PRECISE-AGZ system analyzed 48,261 variables — from smoking history to medication records — in nearly 70,000 anonymized patients from the SAIL database in Wales, distilling them to 38 predictive features. In validation, the model achieved 85 percent sensitivity and 91 percent specificity, identifying cases 12 months before formal diagnosis on average, with signal traces extending to five years, ScienceDaily reported.
“Our system goes far beyond traditional symptoms, and it’s great to have uncovered the results that we have,” said Xu Wang, the PhD student who led the data analysis. “The combination of multiple factors — from medication histories to seemingly unrelated conditions — can reveal patterns invisible to conventional screening methods,” Wang told ScienceDaily.
Dr. Helen Winter, Clinical Director for the Somerset, Wiltshire, Avon and Gloucestershire (SWAG) Cancer Alliance, emphasized the unmet need: “Bladder cancer has a major unmet need in early detection and the prospect of a screening tool is hugely exciting for patients and their families. It could identify those most at risk more accurately, reduce harm from invasive procedures, and catch cancer earlier, when survival outcomes and quality of life can be dramatically improved.”
AI Tool Flags Bladder Cancer Risk Years Before Symptoms
- An AI tool reading routine GP records could flag bladder cancer risk years before symptoms appear, allowing earlier, less invasive treatment.
- The system sorts patients into low, uncertain, and high-risk tiers, potentially reducing unnecessary cystoscopies — camera inspections of the bladder — for those at minimal risk.
- This is a research breakthrough, not yet a clinical tool; the model has only been validated on Welsh data and requires testing in diverse populations before NHS or global adoption.
From Hematuria to Holistic Risk Profiles
Current NHS referral pathways hinge largely on visible blood in the urine (hematuria), a symptom shared by kidney stones, benign prostatic hyperplasia, and urinary tract infections. This overlap drives both missed cancers and a high volume of negative cystoscopies. The Plymouth team, led by Professor Shang-Ming Zhou at the Centre for Health Technology, sought to replace this single-symptom trigger with a longitudinal, multi-variable risk score.
Using swarm-intelligence optimization — a computational method mimicking collective behavior — the model iteratively selected the 38 strongest predictors from the initial 48,261 candidates. Confirmed risk factors such as smoking and hematuria retained weight, but the AI surfaced counter-intuitive associations: patients with Parkinson’s disease or dementia showed lower bladder-cancer incidence, while long-term tamoxifen use correlated with higher risk, according to the study published in IEEE Transactions on Biomedical Engineering.
“This work represents a paradigm shift towards precision screening for bladder cancer. By harnessing the power of interpretable machine learning and comprehensive health records, we’re moving closer to detecting this disease at its earliest, most treatable stages.”
Professor Zhou said.
Risk Stratification: The Three-Tier Triage
| Risk Tier | Probability Range | Suggested Clinical Pathway |
|---|---|---|
| Low | Below 7% | Routine monitoring; defer invasive investigation |
| Uncertain (Grey Zone) | 7% – 55% | Enhanced surveillance; repeat non-invasive testing before cystoscopy |
| High | Above 55% | Urgent referral for cystoscopy and definitive diagnosis |
Validation Gaps and the Road to Clinic
Despite the headline performance, the study carries significant translational caveats. Every record originated from the Secure Anonymised Information Linkage (SAIL) databank in Wales, a relatively homogeneous, predominantly white population with a unified electronic health record infrastructure.
Professor Zhou stressed that external validation across ethnically diverse cohorts, different coding systems (e.g., ICD-10-CM in the US, SNOMED-CT in varied NHS trusts), and fragmented primary-care IT systems is essential before regulatory submission to the MHRA in the UK or the FDA in the United States.
Funding details were not disclosed in the primary publication; the research involved collaborators from North Bristol NHS Trust and University Hospitals Bristol NHS Foundation Trust.
Founded in 1862 as the Plymouth School of Navigation, the University of Plymouth traces its roots to maritime education before evolving through polytechnic status to become a university in 1992. Today it is a public research institution with 18,410 students, ranking 57th in the United Kingdom by enrolment, and hosts the Royal Statistical Society’s Centre for Statistical Education.
PRECISE-AGZ Risk Score Not Available to Patients
- No patient can currently request or receive a PRECISE-AGZ risk score.
- Visible blood in urine (hematuria) remains a red-flag symptom.
Bladder cancer remains the ninth most common malignancy globally, with 614,000 new cases and approximately 220,000 deaths recorded in 2022.
References
- Wang X, et al. Early Detection of Bladder Cancer Using Advanced Feature Engineering and Swarm Intelligence Optimization on EHRs. IEEE Transactions on Biomedical Engineering. Published online November 2024.
- NG12: Suspected cancer: recognition and referral. Updated 2023.
- Sung H, et al. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA: A Cancer Journal for Clinicians. 2021;71(3):209-249.
- SAIL Databank. Secure Anonymised Information Linkage.
- UK Medicines and Healthcare products Regulatory Agency.
Keep reading
- Trainer Justin Kraft Names 5 Standing Exercises to Restore Leg Power After 70
- Javier Marquina and Vicente Benavent Present Graphic Memoir 200 de PSA
- AI Identifies Bladder Cancer Risk Five Years Before Diagnosis, Study Finds (world-today-news.com)
- Pope Leo XIV Cleared for South America Tour After Lung Cancer Surgery (newsdirectory3.com)