Automating Claims in the Pet Health Insurance Space
Trupanion is now automating 30% of its electronically submitted claims using machine learning models. As a result, invoices are paid directly to the veterinarian within seconds, rather than reimbursed to the policy holder after several weeks. A few of the challenges we’ve overcome in this project should be general enough to be useful in other applications.
I’ll be covering the following topics in my talk:
Methods for attaching medical meaning to free typed invoice line items
Replicating human decision making
Pain points of cross-departmental product teams
Communications addressing the implications of automating someone’s job
Dave created and currently leads the data science function at Trupanion. His first interaction with machine learning was during his graduate research 10 years ago, where he was tasked with extracting a control signal for moving robots using only human thoughts. He has been convinced that ml will change the world ever since. He is currently working through the challenge of integrating data science into an organization in a way that maximizes long term mutual benefit, while being the best leader possible for his team.