Pancreatic cancer is one of the deadliest types of cancer in the world, and the number of people affected is expected to continue to rise.
The
findings, published in the journal Nature Medicine, suggest that AI-based
population screening could be useful in detecting people at risk of developing
the disease, which could speed up the diagnosis of the disease, which is often
in advanced stages with poor treatment and diagnosis efficacy.
Currently,
there are no population-based tools for pancreatic cancer screening on a large
scale. People with family history and certain gene mutations that predispose
them to pancreatic cancer are screened in a targeted way, but researchers say
such targeted screening may miss other cases that fall outside those
categories.
A new study suggests that artificial intelligence (AI) may contribute to early diagnosis of pancreatic cancer, according to the Washington Post.
The new
study, published last month in Nature Medicine, shows that AI screening could
enable earlier diagnosis of pancreatic cancer and lead to more effective
treatment.
In tests on
a large number of patients, the AI tool successfully identified people at risk
of pancreatic cancer, reviewed their medical records, and found evidence of
increased risk three years before they were diagnosed.
According
to the Mayo Clinic, pancreatic cancer is rarely detected in its early, more
curable stages, as symptoms often do not appear until it has spread to other
organs.
The need
for early detection is clear, as the disease is highly aggressive and usually
diagnosed late, leading to poor prognosis.
The
researchers used data from medical records of patients in both the United
States and Denmark from 1977 to 2020.
They
studied a group of 6.2 million Danish patients (of whom 23,985 were diagnosed
with pancreatic cancer) and a group of 3 million veterans receiving treatment.
Among the veterans, 3,864 were eventually diagnosed with the disease.
The
researchers analyzed the data using machine learning models and taught them to
predict cancer risk based on symptoms and various diagnosis codes in patients'
medical records.
“AI tools
that can focus on people at risk of pancreatic cancer can help improve clinical
decision-making,” said Chris Sander, a biologist at Harvard Medical School.
About 44%
of people diagnosed with pancreatic cancer at an early stage of the disease are
alive five years after diagnosis, but only 12% of those diagnosed are diagnosed
at an early stage. If the tumor has grown beyond the site of origin, survival
rates drop to between 2-9%.
If you have any questions, please leave a comment