TAIPEI (Taiwan News) — National Taiwan University Hospital has partnered with Academia Sinica to develop an AI model, PanMETAI, which can screen for pancreatic cancer through a blood test with 90% accuracy, CNA reported.
The hospital said the method analyzes metabolites in the blood, creating an effective cancer screening tool. It requires a single drop of blood, which can provide up to 260,000 metabolic signals. The AI model then analyzes these signals to identify features closely related to the cancer and assess the risk.
The hospital has also tested the method on participants at a hospital in Lithuania, achieving high accuracy. The study was published in the international journal Nature Communications.
Chang Yu-ting (張毓廷), an internal medicine doctor at the hospital, said that early-stage pancreatic cancer is difficult to detect due to the lack of obvious symptoms. About 85% of cases are found at an advanced stage, significantly reducing their chances of survival.
The hospital emphasized that surgery to remove pancreatic tumors is challenging, as the organ is surrounded by dense blood vessels and nerves. Only around 30% of people with the condition can undergo surgery. In some cases, fibrotic tissue forms around the tumor, blocking drugs and making chemotherapy less effective.
About 2,000 new cases are diagnosed in Taiwan annually, with an average age of 70, Chang said. The condition is also appearing more often in younger people.
Traditional screening methods focus on detecting specific markers in the blood, Chang added. However, the cancer is complex, making it easy to misdiagnose when relying on a few indicators.
Chang said the cancer can cause changes in metabolism and body tissue, such as high blood sugar, low blood fat, and loss of fat under the skin that leads to weaker muscles. He urged people with these unexplained changes to get screened.
The hospital has also applied the AI model and screening tool to stomach, colorectal, and liver cancers, Chang added. Early results are promising, though further validation is needed to confirm its effectiveness for other cancers.
