A novel AI test created by UCL and the Institute of Cancer Research helps determine which men with high-risk, localized prostate cancer will respond positively to the hormone therapy drug abiraterone.
Researchers say the test could maximize the possibility of cure and reduce NHS costs by ensuring treatment is only given when it is likely to work, avoiding unnecessary side effects and wasted resources.
How the AI test works
Researchers at University College London and the Institute of Cancer Research developed an AI test that analyzes routine prostate tumor biopsy images using advanced algorithms to detect features invisible to the human eye. The tool classifies patients based on their likely benefit from abiraterone treatment.
The AI test was tested on biopsy samples from over 1,000 men in the STAMPEDE trial. It identified about 25% of patients whose five-year risk of death dropped from 17% to 9% when treated with abiraterone alongside standard hormone therapy. For the other 75%, abiraterone showed no significant survival benefit, indicating standard therapy alone is sufficient.
Professor Gert Attard, co-lead of the trial from UCL Cancer Institute, said the study demonstrates how AI algorithms can reduce overtreatment while increasing the chance of cure for patients with advanced prostate cancer.
Accurate selection of treatment saves lives — and money
Abiraterone is a hormone therapy widely used for prostate cancer treatment, but it carries risks including high blood pressure and an increased chance of diabetes.
Accurately targeting treatment is vital to avoid exposing patients to these side effects unnecessarily. With the AI tool precisely identifying those who benefit from abiraterone, it spares others from ineffective treatment and potential harm.
By limiting drug use to patients who truly need it, the tool also reduces expenses by cutting drug costs and lowering healthcare visits caused by side effects. This targeted approach delivers clear savings for health systems.
AI’s breakthroughs across healthcare
Artificial intelligence is transforming healthcare by accelerating drug development and improving diagnostic accuracy. For example, AI has cut cancer drug delivery times by half, significantly speeding up access to new treatments. It has also demonstrated an impressive 81% accuracy in predicting cancer patients’ survival, helping clinicians make more informed treatment decisions.
In neurology, AI models have achieved a 78% success rate in forecasting Alzheimer’s disease, enabling earlier diagnosis and timely intervention.
These advances underscore AI’s growing role in enhancing both the speed and precision of medical care, ultimately leading to better health and longer lives for patients across many areas of medicine.