Research Project
Clinical Text De-identification and Information Preservation
Removing identifying information from clinical records while keeping the medical information useful.
Research context
Research internship at Ruzivo Research Lab, within the AutomAI Research Program.
Problem
Electronic health records contain personally identifiable information alongside clinically useful information. De-identification needs to account for both.
Approach
Developing de-identification pipelines and training named entity recognition models to detect and redact personally identifiable information while preserving clinical utility.