Pancreatic cancer survival rates are particularly low. A study shows that AI can be used to speed up the diagnosis of invasive disease, which is often diagnosed after a long delay.
You will also be interested
[EN VIDÉO] Cancer: From Theory to Practice Futura-Sciences met Norbert Vey, a physician at the Paoli-Calmettes Institute in Marseille, to better understand…
After breast and lung, pancreas is the target of artificial intelligence tool for early detection of cancer. The new study – conducted by researchers Harvard Medical School and the University of Copenhagen – found that AI could identify people at high risk of pancreatic cancer three years before the actual diagnosis.
Why use AI to diagnose pancreatic cancer?
Aggressive, pancreatic cancer is one of the deadliest cancers in the world. Researchers point to a condition that is often diagnosed at an advanced stage, when treatment is less effective and results are “disappointing.” Although about 44% of people diagnosed at an early stage of pancreatic cancer survive five years after diagnosis, only 12% of cases are diagnosed early. Because of this delay, tumors can grow outside the organ and patients’ survival rates drop from 2 to 9%.
However, patients with early-stage disease can be cured with surgery, chemotherapy, and radiation therapy. Therefore, early detection can go a long way in improving patient survival. A new study published in Natural Medicine AI-based population screening suggests early detection of people at high risk of developing pancreatic cancer.
An approach based on disease pathways
The researchers designed different versions of the AI models and trained them on clinical data from 6.2 million medical records from the Danish National Health System over a 41-year period. Of these patients, nearly 24,000 developed pancreatic cancer over time. ” During training, the algorithm identified patterns that indicated future risk of pancreatic cancer based on disease trajectories.The authors of the study explain. For example, a diagnosis of gallstones, anemia, type 2 diabetes and other gastrointestinal problems predicted the risk of pancreatic cancer within a three-year assessment. Less surprisingly, pancreatitis strongly predicted future pancreatic cancer within a shorter period of two years. »
According to the scientists, the developed tool was as low as current gene sequencing tests in predicting the onset of the disease. Currently, only people with a family history and some genetic predisposition to pancreatic cancer undergo targeted screening, which you should be aware may put other cases at risk. ” An AI tool to identify people at high risk for pancreatic cancer could ensure doctors are screening the right population while avoiding unnecessary tests and extra procedures for others. “, the researchers noted.