Sybil—a new AI tool developed by KI member Regina Barzilay and Massachusetts General Hospital clinical collaborators Lecia Sequist and Florian Fintelmann—assesses a patient’s risk of lung cancer over six years by analyzing a single low-dose CT scan.
Unlike current methods, Sybil can make accurate predictions without using demographic or medical information or a radiologist’s annotation. The tool, described in the Journal of Clinical Oncology and profiled in The Washington Post, could be used to identify individuals who need additional testing and closer management and could be particularly useful given the rising incidence of lung cancer among non-smokers. MGH is launching a trial of Sybil, and researchers plan additional testing to ensure that it will maintain its accuracy across diverse populations.
Barzilay and Sequist will share project updates as part of SOLUTIONS with/in/sight: Algorithm & Views on May 4. The work is funded in part by the Bridge Project.