Project Lead: Aparna Ramanathan, UW Medicine

Data Science Lead: June Yang, Spencer Wood

Currently in the United States, postmarket surveillance of medical device safety is conducted via the Manufacturer and User Facility and Device Experience (MAUDE) database.  MAUDE is targeted at device manufacturers and does not engage patient or provider users of new medical devices; this lack of stakeholder engagement has led to the injury of hundreds of thousands of gynecology patients.  With the Essure® system of permanent birth control as an example, attention was not drawn to MAUDE database reporting pathways until after patients themselves gained traction by reporting complications through social media platforms. 

Using Essure® as a model, we aim to acquire social media data related to patient experience with Essure®, build out a natural language processing pipeline to facilitate analysis of these data, and build and test models for identifying posts that express negative outcomes.  We would use this testing to prototype a model for detecting early warning signals for safety issues regarding Essure® and other implantable medical devices.