One of the devastating aspects of cancer is the disease often blindsides people – not only the patients, but their friends and loved ones as well.
Three of the most devastating words to hear together is, “You have cancer.” In many cases, patients are told there aren’t many treatment options available, either.
What if there was a way to predict if someone is going to get cancer later in life? This type of knowledge could prepare people for treatment, inspire them to get regular screenings, and lead to preventative lifestyle changes.
This future is possible through artificial intelligence, and doctors at Harvard Medical School are trying to create an AI capable of predicting cancer risk for people. If this technology materializes, it could go a long way to reducing cancer deaths and cancer rates, including for lung cancer and mesothelioma.
Explaining ‘Sybil’: Harvard University’s AI to Predict Cancer
Harvard researchers focused on lung cancer, which is the leading cause of cancer death in the United States and one of the most common types of cancer for people to get. There are a few causes, including smoking tobacco cigarettes, genetics, and exposure to asbestos.
Medical experts recommend people from 50-80 years old receive low-dose chest computed tomography (LDCT) screens, particularly if they have a history of smoking or a history of lung cancer in their family.
Lung cancer screening with LDCT has reduced lung cancer death rates by 24%, according to Harvard University, but it has not reduced the risk for non-smokers.
Harvard Medical School in collaboration with Massachusetts Institute of Technology created Sybil, an artificial intelligence, as a way of better predicting and diagnosing lung cancer in this population.
“Lung cancer rates continue to rise among people who have never smoked or who haven’t smoked in years, suggesting that there are many risk factors contributing to lung cancer risk, some of which are currently unknown,” said corresponding author Lecia Sequist, the Harvard Medical School Landry Family Professor of Medicine in the Field of Medical Oncology at Mass General.
Sybil can predict the risk of lung cancer for people up to six years before the cancer develops.
Testing Sybil as a Lung Cancer AI
The team tested Sybil in three studies, using scans from more than 6,000 people, 8,821 scans from Massachusetts General Hospital, and 12,280 scans from Chang Gung Memorial Hospital in Taiwan (China).
The last set included a range of smoking histories, including never-smokers. The measurement was distinguishing between disease and normal samples. In the three tests, Sybil scored a 0.92 out of a 1.0, 0.86 and 0.94 for lung cancer risk within one year. Sybil predicted lung cancer risk within six years with an accuracy of 0.75, 0.81 and 0.8.
“Sybil can look at an image and predict the risk of a patient developing lung cancer within six years,” said co-author and Jameel Clinic faculty lead Regina Barzilay, a member of the Koch Institute for Integrative Cancer Research. “I am excited about translational efforts led by the MGH team that are aiming to change outcomes for patients who would otherwise develop advanced disease.”
What Does This Mean for Mesothelioma?
Mesothelioma is one of the most difficult cancers to diagnose, and it’s similarly difficult to predict. However, there are risk factors, with the main one being exposure to asbestos. Fortunately, most people exposed to asbestos don’t develop mesothelioma. However, this means it’s challenging to predict who will get this rare and deadly cancer.
Sybil or an AI technology like it may be able to look at scans and predict mesothelioma cancer emerging years before it actually begins spreading. This type of knowledge can help patients begin treatment early and defeat mesothelioma before it spreads to the lungs or other organs.
If you’d like more news about mesothelioma and lung cancer – or cancer in general – sign up for the Mesothelioma Monthly email newsletter. This offering from Mesothelioma Guide delivers news about mesothelioma and asbestos cancers directly to your email each month.
Sources & Author