In order to diagnose patients, doctors will typically need to talk to the patients to find out a bit more about the symptoms, observe them, run tests, and etc., all of which will be then be written on a patient’s chart and medical record. Pneumonia, for example, can happen to patients who stay in the ICU of a hospital, but it is usually through a thorough chart review that doctors are able to diagnose them.
This means that there is a chance that sometimes doctors might miss a patient or two due to oversight. Perhaps looking to combat that possibility and also to make things much more efficient, the University of Washington has teamed up with Microsoft researcher, Lucy Vanderwende, to work on a project called deCIPHER.
This project will use natural language processing on computers to try and see if machines are capable of diagnosing critical illnesses from medical records. In order to test it out, they used the electronic medical records of 100 patients who were treated in the ICU at Seattle’s Harborview Medical Center and applied the NLP tools to it. They then took the data and ran it through a machine learning framework to see if the software could be trained to identify cases of pneumonia.
According to Meliha Yetisgen, the Assistant Professor of Biomedical Information at the University of Washington, “The results were so promising—the software achieved a correct diagnosis with correct time-of-onset for positive cases in 84 percent of the patients—that our clinical collaborators are considering the addition of our pneumonia-detection models to the dashboard they use to monitor ICU patients.” Pretty cool, huh? Makes us wonder if more automated diagnoses could be done in the future.
Next Story: Sense 6.0 Update For HTC One M7 On AT&T Released