Bladder cancer detected up to five years before diagnosis—thanks to AI

Bladder cancer is a significant health concern, claiming 220,000 lives globally each year. Currently, detection relies heavily on invasive procedures like cystoscopy, which are not routine. Now, researchers from the University of Plymouth have developed an artificial intelligence (AI) tool capable of identifying bladder cancer patients up to five years before their official diagnosis.

Published in IEEE Transactions on Biomedical Engineering, the study analyzed electronic health records of nearly 70,000 patients from 1995 to 2020. Their AI system, PRECISE-AGZ, sifted through 48,261 health indicators, including smoking habits, exercise routines, and medication use, to identify 38 key features signaling a higher risk of bladder cancer.

The AI correctly detected 85% of cancer cases, correctly identified 91% of cancer-free individuals, and detected potential cases 12 months before diagnosis, with some signals appearing up to five years in advance. The tool also revealed hidden patterns, such as lower risk in patients with Parkinson's disease or dementia and higher risk in long-term tamoxifen users.

Furthermore, the significance of blood in urine varied depending on other health indicators, such as benign prostate enlargement in men, which lowered cancer risk. The AI system categorizes patients into low, uncertain, and high-risk groups, with the latter potentially avoiding unnecessary invasive procedures like cystoscopy. While promising, further validation across different healthcare systems is needed before wider implementation.