Identifying Symptoms in Women with Ovarian Cancer from electronic medical records

Start Date Apr 2021

Code W4-I

Status Ongoing

Project Lead
Senior Lead
Grace Turner, Assc Prof Meliha Yetisgen, DrMorhaf Al Achkar (all Washington), Dr Garth Funston (Cambridge)

Project Summary

Ovarian cancer often presents late, and has one of the worst outcomes of all gynecological cancers. There are no screening tests for ovarian cancer in widespread use, so women are diagnosed after they present to a primary care clinician with symptoms. Unfortunately the symptoms of ovarian cancer can be nonspecific, so cancer goes undiagnosed in many women sometimes for many months before diagnosis. We want to use a new tool, called Natural Language Processing (NLP) that looks into the electronic medical records of women with ovarian cancer. NLP lets us automatically pick up many more details of symptoms from their health records than has been possible before. We will compare the symptoms we find in women with ovarian cancer before diagnosis, to a control group of women without ovarian cancer. This research will help us find out if we can improve the early detection of ovarian cancer, and potentially create useful tools that primary care clinicians could use to improve the speed of diagnosis, and improve care for women with this cancer. 

Back to Research