Artificial intelligence will help to estimate even 3 years earlier who will get pancreatic cancer. Breakthrough news

Pancreatic cancer is a malignant tumor with an aggressive course. Currently, the cure rate is still very low (less than 9 percent). In our country, more than 3200 people suffer from it every year. This type of cancer accounts for 3 percent of all human malignancies and its incidence is increasing.

The first symptoms of pancreatic cancer may include abdominal pain, loss of appetite, weight loss, nausea, and jaundice. Initial symptoms are non-specificwhich, in turn, makes diagnosis difficult. The high mortality of people with this cancer is due to the fact that it usually found at an advanced stagein the presence of metastases. However, scientists have developed a method that will allow us to estimate who can get this cancer even 3 years earlier.

Pancreatic cancer can be detected much earlier

A study published in Nature Medicine shows that early detection of pancreatic cancer is possible thanks to appropriate review of the patient’s medical records. Researchers at Harvard Medical School and the University of Copenhagen have developed a method that can be determined using artificial intelligence tools. people at increased risk of pancreatic cancer. New research suggests that patients can be diagnosed with pancreatic cancer up to three years earlier than those with current diagnoses, and this, in turn, will help improve survival rates.

Several versions of the AI ​​model were developed to conduct the study, and the researchers “trained” them using medical data from 6.2 million Danish patients, 23,985 of whom developed pancreatic cancer. The algorithm recognized patterns that indicate cancer risk by examining patient disease codes and when they occurred. A higher risk of developing pancreatic cancer over the next 3 years has been associated with a diagnosis of:

• type 2 diabetes,

• anemia,

• problems associated with the digestive system,

• the appearance of gallstones.

With the onset of pancreatitis, the risk increased to 2 years. Although none of the identified diseases was the cause of cancer, scheme and sequencein which they occurred may indicate the risk of contracting the disease, and the task of artificial intelligence was to find these warning signs.

Re-survey confirms impressive results

Each version of the artificial intelligence algorithm very accurately predicted the risk of developing pancreatic cancer in patients at the time of testing. Moreover, the results were more accurate than currently available incidence estimates. The researchers repeated the study using the AI ​​algorithm that worked best. They did this on a different set of patients—they used a Veterans Health Administration data set of over 3 million people (including 3,864 people with pancreatic cancer). Also in this case artificial intelligence worked with very high accuracy.

According to experts, population screening based on artificial intelligence algorithms can help diagnose pancreatic cancer faster. The advantage of this method is that it can be applied to the entire population, and not just those people who have a genetic predisposition to this disease. This, in turn, can improve clinical decision making. However, it should be remembered that the research is currently in its early stages and requires many more tests, but the results even at this stage are very promising.