Pancreatic cancer has one of the lowest five-year survival rates of any cancer, in part because late diagnosis is common. Could artificial intelligence change that?
Research published in Nature Medicine in May suggests that AI screening of large groups of patients could make earlier diagnosis possible, which in turn could lead to earlier and more effective treatment of the disease. In the analysis, an AI tool successfully pinpointed people at elevated risk for pancreatic cancer by looking back at their medical records, finding evidence of heightened risk up to three years before they were diagnosed.
Researchers used data from the medical records of patients in both the United States and Denmark dating from 1977 to 2020. They looked at a group of 6.2 million Danish patients, 23,985 of whom were eventually diagnosed with pancreatic cancer, and 3 million military veterans receiving care through Veterans Affairs, of whom 3,864 were ultimately diagnosed.
The researchers used a machine-learning model to analyze the data, teaching it to predict cancer risk based on symptoms and the various diagnosis codes contained in patients’ medical records.
Some of the symptoms associated with a higher risk prediction aren’t traditionally linked to pancreatic cancer. Gallstones, Type 2 diabetes, anemia and gastrointestinal symptoms such as vomiting and abdominal pain were all linked to a higher risk score as long as three years before diagnosis.
In a real-world scenario, the researchers write, about 320 of every 1,000 people the AI model identified as high risk would go on to develop pancreatic cancer. By targeting surveillance to high-risk patients, the tool could make screening more affordable, they write.
Right now, the U.S. Preventive Services Task Force doesn’t recommend screening symptom-free individuals for pancreatic cancer. Screening of high-risk patients is associated with a higher chance of long-term survival, however.
“An AI tool that can zero in on those at highest risk for pancreatic cancer who stand to benefit most from further tests could go a long way toward improving clinical decision-making,” said study co-author Chris Sander, a biologist who directs a Harvard Medical School lab devoted to using machine learning and other technologies to solve biological problems, in a news release.
If applied at scale, such a tool could prolong life spans and improve treatment outcomes, Sander said.
The Washington Post · Erin Blakemore
You don’t need AI to predict cancer. Any dummy who took the vax can unfortunately be “predicted” for it.