Fuzzy Matching Saves Lives
November 02, 2013
Did you know fuzzy matching logic can potentially save lives? Did you know deceased Boston Bomber Tamerlan Tsarnaev’s 2011 trip to Russia was not detected because of a misspelling of his name? (according to Sen. Lindsay Graham). I have to wonder if the FBI and CIA’s systems are talking to each other and if they have identity resolution and fuzzy matching logic to merge disparate data bases. On a mass scale, true, raw fuzzy matching and entity resolution requires a supercomputer if the data are too massive to process. But – don’t we, as the United States, have access to these supercomputers and the best scientists in the world?
For those of you who are new to fuzzy matching, it is simply the algorithms and techniques involved in comparing similar but distinct phrases, names, and addresses. Using the Levenshtein distance and the TriGram and PctTriGram functions that Automated Auditors built using SAS, fuzzy matching is not only possible, but easy. SAS has canned fuzzy matching functions, and recently added the power to develop user-defined functions, which we used to develop the TriGram functions.
We recently worked on a model that required employing fuzzy matching logic to merge the List of Excluded Individuals and Entities (LEIE) with Medicare claims data. The LEIE file does not always contain the provider’s NPI number, so we matched on the provider name and date of birth. Fuzzy matching was required to find the most accurate matches. We then were able to develop a predictive model that compares excluded Medicare providers with active providers, thereby increasing our power to detect potential Medicare fraud.
Comments are closed