Diagnosing cancer is a depressingly common activity in hospitals—but spotting subtle variations in tumor types can require sifting through hundreds of previous cases. At MIT, though, researchers are developing a system that uses Artificial Intelligence to speed up that process.
A team from MIT’s Computer Science and Artificial Intelligence Laboratory has developed a software system that studies data from existing medical reports in order to suggest cancer diagnoses to doctors. In a paper published in the Journal of the American Medical Informatics Association, the team behind it describes how it can be used identify types of the cancer lymphoma.
There are in fact 50 subtypes of lymphoma, which can be hard to specifically diagnose; up to 15 percent are initially misdiagnosed, slowing treatment. The software developed at MIT taps into large banks of pathology reports, extracting medical data which can be tied together into relationships that the World Health Organization uses to define the sub-types of the cancer. The system also ties words commonly used in the medical records to each of these data points, providing an extra layer of information.
The system can then take a new set of test results and compare it to existing data on a number of levels. The result provides clinicians with both medical data and written descriptions of similar cases, helping suggest the types of lymphoma most likely to be present. There’s still some way to go for the AI system—it needs far more data and some comprehensive testing—but in the future, it could help make the lives of doctors far easier. [Journal of the American Medical Informatics Association via MIT]
Image by Sumlin under Creative Commons license