When Unique Identifiers Fail: Using NLP Techniques to Associate Database Records to Individuals By Name

Ideally, individuals within the healthcare system can be uniquely identified by using their provincial healthcare numbers (PHNs.) However as data is collected, entered, and shared by healthcare professionals, errors such as typos, misaligned delimiters, or dropped values makes relying on PHNs not always possible. We present our approach to timely group together records associated to a single patient across several million records in PLOVER using natural language processing as well as techniques to validate the formed groups. TEAM: Kyle Clarkson, Phillip Dumitru, Chris Fjell

BCCDC POSTER PRESENTER

10/18/2021 2:40:00 PM