Probabilistic Record Linkage of Hospital Patients - Chris Oakman

How can you tell if a patient is the same person across all the different electronic systems used in a hospital?
Can you be confident with messy data when lives are on the line?
Medical startup Luminare faced this challenge in a hospital setting and used Clojure to save the day and make the nurses happy again.
This talk will explore the challenge of record linking: dealing with dirty data sets, the pros and cons of different solution approaches, and using the Felligi-Sunter method to create a probabilistic algorithm to match records.