AI Learns to Detect Signs of Pancreatic Cancer Long Before Diagnosis
The REDMOD model identified early changes an average of 475 days before diagnosis and surpassed experienced radiologists in sensitivity.
Scientists have developed an artificial intelligence system called REDMOD that can detect early signs of pancreatic cancer on computed tomography (CT) scans. Moreover, the changes may be so subtle that even experienced doctors do not notice them, reports ScienceDaily.
Researchers examined scans from 219 patients who were later diagnosed with cancer and compared them with scans from 1,243 individuals without the disease.
It turned out that the AI can recognize suspicious changes an average of 475 days before a diagnosis is made. The system detected 73% of cases, whereas radiologists identified 39%.
This is particularly important because pancreatic cancer is often detected at late stages, when treatment becomes more difficult.
However, the technology is not yet ready for widespread use. Scientists will need to verify its effectiveness in new clinical trials. In the future, such systems could help doctors detect this dangerous disease earlier and improve patients' chances of successful treatment.
