I attended the INFORMS International Conference nearly a month ago and have been incredibly busy ever since. However, I attended a few talks that were, in my opinion, totally punk, so I need to blog about them, albeit belatedly.
Arnold Greenland from IBM discussed a real-time approach for determining dissability benefits using text mining. The Social Security Administration receives many requests for disability benefits, some of which are eventually funded or renewed and others which are not. Someone who could expect to receive benefits would have to wait about six weeks for a response (if my memory serves correct). Reviewing the cases is time consuming, especially when many of the applications and renewals are no-brainers. The algorithm determines which benefits cases are no-brainers and fast tracks these for review. After implementation, someone who could expect to receive benefits could expect a response in ten days (again, my memory is fuzzy here, but the point is that it made a huge impact!). The most amazing thing about this presentation is that it saved the Social Security Administration over $1B since its implementation 5-6 years ago. However, it looks like there is room for improvement since there is a long backlog on disability claims.