The findings, announced in late January by a team of researchers at Harvard Massachusetts General Hospital and the Massachusetts Institute of Technology, use algorithms to predict everything from breast and prostate cancer to the likelihood of tumor regrowth. It is part of a growing medical trend that Research is on the rise, but scientists say more testing is needed before these products can be fully released into clinical practice.
This tool is called Sybil, named after the prophet in ancient Greek literature. It is a deep learning model in which computers analyze huge datasets to identify and classify patterns. Sybil was trained on her six years of lung scans of patients in the United States and Taiwan, researchers said.
The study results showed that Sybil achieved scores that are scientifically considered “good” and “strong” in predicting lung cancer over six years. Researchers note that the one-year prediction rate was stronger.
Florian Fintelmann, an interventional radiologist at Massachusetts Comprehensive Cancer Center and study co-author, said lung cancer “is the number one killer of cancer because it’s relatively common and relatively difficult to treat. ‘ said. “If lung cancer can be detected early, the long-term prognosis is greatly improved.”
Cancer is the second leading cause of death in the world, and advances in artificial intelligence software and computing power have made it a ripe area for researchers to apply this technology in hopes of helping doctors make diagnoses. became.
Researchers are using artificial intelligence to track prostate cancer progression, breast cancer, and even tumor regrowth after treatment.
Many of our technologies involve analytics Collect large amounts of medical scans, data sets, or images and feed them to complex artificial intelligence software. From there, the computer is trained to find images of tumors and other abnormalities that researchers claim are more accurate and faster than the human eye.
Although there has been a surge in new treatments to combat lung cancer in recent years, researchers say many patients still die because of the barrier.
Due to limited federal funding, older and poorer people may not be screened. According to MIT researchers, many patients diagnosed with lung cancer have never smoked or are ex-smokers who quit smoking more than 15 years ago and are not eligible for screening in the United States.
For those who can be screened, the most common method is to use a low-dose computed tomography scan called an LDCT. The researcher created Sybil to speed up the screening process. This allowed the software to analyze her LDCT images without the help of a radiologist and predict cancer risk up to six years in advance.
But building Sybil was a challenge, say the study authors. Peter Mikhael, a researcher and affiliate of his Jameel Clinic at MIT and its Computer Science and Artificial Intelligence Laboratory, described it as “trying to find a needle in a haystack.”
largely Because early stage lung cancer is in a small part of the lung and can be difficult to see with the naked eye, the image data for training Sybil did not include any overt signs of cancer. to the spot. To ensure that the software can assess cancer risk, the research team “labeled hundreds of CT scans with visible cancerous tumors” and input them into Sybil, then The software was unleashed on CT scans that were limited to signs of cancer, researchers said.
The team used datasets from the National Lung Screening Trial, Massachusetts General Hospital, and Taiwan’s Chang Gung Memorial Hospital. Some data, according to the survey, were overwhelmingly biased toward whites.
Medical experts warn that more research is needed before cancer software can be used clinically, according to government scientists and research studies.
Researchers from Harvard University and the Netherlands say the skill of translating information generated by AI algorithms is still in its “nascent stages.” Moreover, the benefits that AI can bring to healthcare are currently very limited. Even with these detection tools, physicians still need to make diagnoses, plan treatment, and manage overall care.
Other medical experts point out that more testing is needed to see how well the software works with different patients using different scanners and tools. Also, more work needs to be done to show that this software is actually benefiting people. For example, it can extend life, prevent cancer, and save time and money. How the algorithm works, they said, should be transparent, not a “black box.”
The MIT researchers say they will continue their research.
said Lecia Sequist, director of the Center for Early Cancer Detection Innovation at Massachusetts General Hospital.