AI model detects signs of bladder cancer up to five years before diagnosis
Researchers at the University of Plymouth in the UK have developed an AI model that can detect hidden indicators of bladder cancer up to five years before the disease is diagnosed.
Translated from Arabic. Read the original بالعربية
Researchers at the University of Plymouth in the United Kingdom have developed an advanced artificial intelligence model that can detect early indicators associated with bladder cancer years before the disease is formally diagnosed.
The researchers analyzed the electronic medical records of about 70,000 patients, with the aim of finding health patterns that are difficult to detect using traditional methods.
The study produced promising results that could help improve early detection of the disease in future and direct people at higher risk toward specialized medical testing.
Study reveals hidden health indicators of cancer
The study was led by Professor Shang-Ming Zhou, a digital health specialist at the University of Plymouth's Centre for Health Technology.
The research team used a model called PRECISE-AGZ to analyze health data collected between 1995 and 2020.
The researchers examined 48,261 potential health indicators before identifying 38 key factors that can be used to assess the risk of bladder cancer.
The data included medical history, daily habits, medications used and symptoms patients showed before their diagnosis.
Promising results for early detection of bladder cancer
The new model produced encouraging results, showing that artificial intelligence could be used to improve how the disease is detected.
The main findings of the study were as follows:
It identified 85% of bladder cancer cases.
It correctly classified 91% of people without the disease.
It detected some indicators five years before diagnosis.
It showed notable effectiveness in the 12 months before diagnosis.
The model also showed a better ability to identify cases than the medical referral criteria currently used in the United Kingdom.
Unexpected links to certain diseases and medications
Analysis of the data revealed possible links between certain health conditions and medications and the risk of bladder cancer.
Among the findings that drew the researchers' interest, the data studied showed lower probabilities of the disease among patients with Parkinson's disease and dementia.
The model also detected a link between long-term use of tamoxifen, a drug used to treat breast cancer, and a higher likelihood of bladder cancer.
However, the researchers stressed that these findings do not prove a direct causal relationship and do not justify changing the treatments prescribed to patients.
These links require further studies to understand their nature and possible causes.
Classifying patients by risk level
The new model allows people to be classified into three categories, based on the probabilities calculated by the algorithm:
Low risk: probability of the disease under the model is less than 7%.
Medium or uncertain risk: from 7% to 55%.
High risk: more than 55%.
This classification could help in future to prioritize medical tests and reduce unnecessary diagnostic procedures.
However, these percentages do not amount to a final diagnosis, and they still require clinical validation before being used widely.
Bladder cancer affects thousands of Canadians each year
Bladder cancer is one of the most common types of cancer in Canada, particularly among men.
The Canadian Cancer Society estimates that about 13,200 new cases and 2,500 related deaths could be recorded in 2026.
The five-year net survival rate in Canada is about 78%, though chances of survival vary depending on the stage at which the disease is found.
These figures underline the importance of investing in early diagnostic technologies and improving access to specialized medical testing.
Bladder cancer symptoms that should not be ignored
Blood in the urine is one of the main warning signs associated with bladder cancer.
Other symptoms can include frequent urination, pain or burning during urination, and a sudden, repeated urge to use the toilet.
However, these symptoms can also be linked to urinary tract infections and other health conditions that are not cancer.
Health authorities recommend seeing a doctor if blood appears in the urine, even without pain, to determine the cause and arrange appropriate tests.
Can AI replace medical tests?
Cystoscopy and tissue biopsies remain the main procedures for confirming the disease.
The new model is intended to help doctors make decisions, not to replace established diagnostic tests.
The technology must also be tested using data from different countries and health systems to verify that it can deliver reliable results for diverse populations.
Success in these tests could open the way to developing similar tools for the early detection of other types of cancer, improving treatment chances and easing the burden on health institutions.
More from Hala Canada
- Canada is headed toward missing its Paris Agreement targets. Why won't the government say so?
- Luxury home sales fall in Toronto and Vancouver, but $10-million-plus sales hit a 10-year high in Toronto
- Canada expects clarity on Chinese auto investment in 2027, favours joint ventures
- New drug shows promise against pancreatic cancer in clinical trials, raising hopes in Canada
- Two people killed, third seriously injured in Scarborough shooting
- Pentagon imposes new limits on military co-operation with Canada, U.S. officials say