A new blood test uses artificial intelligence to detect cancer before the onset of symptoms

Representative Image

Scientists have developed a blood test that combines a "liquid biopsy" with an artificial intelligence that can detect cancer in its infancy. The results, published in the journal Nature, should help treat the disease well before the onset of symptoms, said Daniel De Carvalho, a senior scientist at the University Health Network in Canada.

"A major cancer problem is detecting it early. The discovery of a cancer-specific blood mutation, one in a billion, especially in earlier stages, where the amount of tumor DNA in the blood is minimal is a difficult problem to solve. Carvalho. By defining epigenetic modifications instead of mutations, the team was able to identify thousands of modifications specific to each type of cancer.

Then, using a Big Data-based approach, she applied machine learning to create clbadifiers able to identify the presence of cancers. DNA in the blood samples and determine what kind of cancer. Scientists tracked the origin and type of cancer by comparing 300 tumor samples from seven disease sites (lungs, pancreas, colorectal, bad, leukemia, bladder and kidney) and healthy donors at the site. 39, badysis of DNA without cells circulating in the blood. plasma.

In each sample, the "floating" plasma DNA corresponded to the tumor DNA. Since then, the team has expanded its research and has successfully badyzed and compared more than 700 blood samples and tumors from a greater number of cancer types.

Next Steps to Further Validate this Approach: Analyze Data from Extensive Population Health Research Studies Already in Progress In several countries, blood samples were collected months, or even years, prior to diagnosis. cancer, researchers said. Next, the approach will need to be validated as part of prospective studies on cancer screening.