cd /news/artificial-intelligence/ai-uncovers-hidden-bladder-cancer-wa… · home › topics › artificial-intelligence › article
[ARTICLE · art-174343] src=sciencedaily.com ↗ pub= topic=artificial-intelligence verified=true sentiment=↑ positive

AI uncovers hidden bladder cancer warning signs years before diagnosis

Researchers at the University of Plymouth developed an AI model called PRECISE-AGZ that identified 85% of bladder cancer cases and correctly classified 91% of patients without the disease, using electronic health records from nearly 70,000 patients spanning 1995 to 2020. Led by Professor Shang-Ming Zhou at the university's Centre for Health Technology, the system screened 48,261 possible health indicators down to 38 predictive features and detected warning signals as early as 12 months before diagnosis, with some traceable up to five years, outperforming existing NHS referral guidelines. Bladder cancer kills approximately 220,000 people worldwide each year, and the researchers say the tool could reduce reliance on invasive cystoscopy by flagging at-risk patients earlier.

key takeaway

PRECISE-AGZ screened 48,261 health indicators from records of nearly 70,000 patients down to 38 features, hitting 85% sensitivity and 91% specificity for bladder cancer and beating NHS referral guidelines up to 12 months before diagnosis.

read7 min views2 publishedNov 22, 2026
AI uncovers hidden bladder cancer warning signs years before diagnosis
Image: Sciencedaily (auto-discovered)
  • Date:

  • November 21, 2026

  • Source:

  • University of Plymouth

  • Summary:

  • Scientists have developed an AI tool that can uncover hidden warning signs of bladder cancer up to five years before diagnosis, identifying 85% of cases in a major study involving nearly 70,000 patients. By revealing subtle patterns that conventional screening misses, the technology could help doctors detect cancer earlier and spare patients unnecessary invasive procedures.

  • Share: Bladder cancer kills approximately 220,000 people worldwide each year, yet recognizing the disease early can be difficult. Its symptoms can resemble those of far less serious conditions, including kidney stones, allowing warning signs to go unnoticed.

Researchers at the University of Plymouth have now developed an artificial intelligence system that could help doctors identify people at risk of bladder cancer years before they receive a diagnosis. In some patients, the technology detected subtle warning signals dating back as far as five years.

The new approach uses AI to examine electronic health records, searching for combinations of symptoms, medical conditions, risk factors, and healthcare activity that might otherwise be overlooked. By recognizing these patterns, researchers hope to make bladder cancer screening more accurate and help doctors find the disease when treatment is most likely to succeed.

AI Finds Hidden Bladder Cancer Clues in Medical Records

Under current medical referral guidelines, one of the main warning signs of bladder cancer is visible blood in the urine (hematuria). However, blood in the urine can also result from noncancerous problems, including kidney stones and prostate conditions. This overlap makes it difficult to determine which patients need further investigation and contributes to missed cases.

Confirming bladder cancer currently requires a procedure called cystoscopy (where a long, thin tube with a small camera inside is moved up the urethra and into the bladder). Although effective, the examination is invasive, highlighting the need for better ways to identify patients who are most likely to have the disease.

To explore a possible solution, a team led by Professor Shang-Ming Zhou at the University of Plymouth's Centre for Health Technology studied electronic medical records from nearly 70,000 patients. The records covered a 25-year period, from 1995 through 2020.

The researchers created an AI model called PRECISE-AGZ and used it to investigate 48,261 possible health indicators (from smoking and exercise habits to medication use). From this enormous collection of information, the system narrowed the list to 38 features that were especially useful for predicting bladder cancer risk.

AI Identifies 85% of Bladder Cancer Cases

The results suggest that AI could substantially improve the way patients are assessed for bladder cancer.

The model correctly identified 85% of patients who had bladder cancer and accurately classified 91% of those who did not have the disease. It also performed better at detecting bladder cancer than existing NHS referral guidelines.

Importantly, the system demonstrated effective detection as early as 12 months before a patient's official diagnosis. Some of the warning signals it recognized could be traced back up to five years, suggesting that valuable clues may exist in medical records long before cancer is discovered.

The AI confirmed the importance of established risk factors, including smoking and blood in the urine. However, it also uncovered unexpected associations that conventional screening approaches would be unlikely to detect.

For example, patients diagnosed with Parkinson's disease or dementia appeared to have a lower risk of bladder cancer. Meanwhile, long-term use of tamoxifen (a breast cancer medication) was associated with a higher risk. These findings do not establish that Parkinson's disease or dementia protects against bladder cancer, or that tamoxifen causes it. Instead, they raise questions about possible shared biological mechanisms that researchers believe deserve closer examination.

Another surprising finding involved blood in the urine. Although this symptom is often considered a major warning sign, its significance appears to depend partly on a patient's other medical conditions. Among men with benign prostate enlargement, for instance, blood in the urine was associated with a lower bladder cancer risk.

Understanding these relationships more fully could eventually help doctors distinguish patients who need urgent investigation from those who may not require invasive testing.

A New Approach to Early Cancer Screening

Beyond identifying warning signs, the AI system could also divide patients into three groups according to their estimated likelihood of having bladder cancer: low-risk (below 7% probability), uncertain (grey zone: 7-55%), and high-risk (above 55%).

This approach could help healthcare providers decide which patients need immediate attention and which may benefit from additional monitoring.

For example, patients in the uncertain category might be monitored before undergoing procedures such as cystoscopy. If further research confirms the system's reliability, this could reduce unnecessary examinations while allowing doctors to concentrate resources on people at greatest risk. PhD student Xu Wang, who led the data analysis, said: "Bladder cancer ranks as the ninth most common cancer worldwide, with 614,000 new cases diagnosed in 2022."

"Despite its prevalence, no routine screening program exists for the general population, and - following symptoms - detection relies heavily on invasive cystoscopy procedures. Our system goes far beyond traditional symptoms, and it's great to have uncovered the results that we have. The research shows that the combination of multiple factors - from medication histories to seemingly unrelated conditions - can reveal patterns invisible to conventional screening methods."

Could AI Help Doctors Catch Bladder Cancer Earlier?

The potential benefits extend beyond improving diagnostic accuracy. Detecting bladder cancer earlier could mean more treatment options, better survival outcomes, and fewer patients undergoing invasive procedures unnecessarily.

Dr. Helen Winter, Clinical Director for the Somerset, Wiltshire, Avon and Gloucestershire (SWAG) Cancer Alliance, emphasized the importance of developing better screening methods.

"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."

Professor Shang-Ming Zhou believes the findings point toward a more individualized approach to cancer screening, in which AI helps doctors interpret a patient's broader medical history rather than relying primarily on individual symptoms.

"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."

More Research Needed Before Clinical Use

Despite the promising results, the researchers caution that the technology is not yet ready for widespread clinical use.

All the patient information used in the study came from the SAIL database in Wales. Before the model can be introduced into other healthcare settings, researchers will need to determine whether it performs equally well with different patient populations and medical record systems.

Professor Zhou explained why broader testing is essential.

"It's important to emphasize that further validation across different healthcare systems is essential before anything is rolled out more widely - for example, all of the data analyzed came from the SAIL database in Wales, so we'd want to investigate other datasets too.

"But it's a very exciting early study. Additional studies would also be needed to confirm the causal relationships suggested by the model's predictions."

The findings offer an encouraging glimpse of how artificial intelligence might improve cancer screening by uncovering patterns hidden within years of routine medical information. With further validation, such tools could help doctors recognize bladder cancer sooner, prioritize patients who need testing, and potentially improve outcomes.

The study, "Early Detection of Bladder Cancer Using Advanced Feature Engineering and Swarm Intelligence Optimization on EHRs," was published in IEEE Transactions on Biomedical Engineering.

The research also involved collaborators from North Bristol NHS Trust and University Hospitals Bristol NHS Foundation Trust.

Story Source:

Materials provided by University of Plymouth. Original written by Mrs. Amy King. Note: Content may be edited for style and length.

Journal Reference:

  1. Xu Wang, Andrea Preston, Jonathan Aning, Michael Loizou, Shang-Ming Zhou. Early Detection of Bladder Cancer Using Advanced Feature Engineering and Swarm Intelligence Optimization on EHRs .IEEE Transactions on Biomedical Engineering , 2026; 73 (11): 3680 DOI:10.1109/TBME.2026.3658230

Cite This Page:

ScienceDaily. Retrieved November 21, 2026 from www.sciencedaily.com

── more in #artificial-intelligence 4 stories · sorted by recency
── more on @university of plymouth 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
→ Live at https://your-agent.zahid.host ✓
Get free account → Pricing
from €0/mo · no card required
LIVE [news/ai-uncovers-hidden-b…] indexed:0 read:7min 2026-11-22 · —