Optimized for Fraud Detection with 5 Separate Name Matching Algorithms
Medicaid fraud is costing the US government over 5 billion dollars per year. How does it happen? Why can’t this be stopped?
First, it’s easy to defraud a system with insufficient capabilities. Before TeraMatch® was installed for the State of Texas , workers were looking through 5 lists from different federal and state agencies with over the counter matching technologies like EXCEL and SQL2007. Deterministic matching. Yes, or no—nothing in between. The computer does what you tell it to—match the data records. If they are slightly different, do not match them. Is this how you would compare data? Of course not.
Enter TeraMatch®, the closest we can get to having a computer think for you.
Look at this example:
A state worker is tasked with enrolling doctors and equipment providers for Medicaid reimbursement. Here’s some information from a fraudulent doctor trying to enroll:
file1.csv
Last Name: Lee
First Name: Jenkins
SSN: 449112963
Address: 2207 Far West Blvd
Address2: Bldg 2, Suite 240
City: Lincoln
State: Nebraska
Zip: 68501
Before enrolling the provider, the worker logs on to the secure website and enters the above information into the Interactive search form. TeraMatch® compares the information he inputs with 5 different files and finds no matches with a perfect 100/100 score. But, it does find one match from the On-going Investigations file with a score of 68/100. Take a look at the information from the On-going Investigations file and notice the similarities:
file2.csv
Name1: Leroy
Name2: Jerkins
Social Security Number: 449-11-2936
Street: 2270 FarWest Street
Other: Ste 240
City: Houston
State: TX
Zip: 73245
And here’s the TeraMatch® result for the state worker to review. File 1 is the Medicaid provider file and File 2 is the On-going Investigations file:
TeramatchID: 122
Score: 68.445
Last Name
file1.csv: Jenkins
file2.csv: Jerkins
First Name
file1.csv: Lee
file2.csv: Leroy
SSN
file1.csv: 449112963
file2.csv: 449-11-2936
Address1:
file1.csv: 2207 Far West Blvd
file2.csv: 2270 FarWest Street
and so forth...
You see that TeraMatch® is comparing each bit of information just as you would. If the first name is sort of the same, that really doesn’t mean that much to you. Last name one letter off? Getting closer, but it still doesn’t mean it’s the same person. Address? Very close. Social Security number? The last 2 digits are transposed. Taken all together, this is definitely fishy and warrants a closer look and maybe a phone call. You see how TeraMatch® is making the same kind of judgment about the data record as you would?
So TeraMatch® takes each bit of information on any kind of data record, inventory item, name, circuit id, order id, and scores it for similarity. Then it adds up the score and provides it to you along with all the other potential matches. For names, TeraMatch® uses 5 sub-algorithms which add to its score. And with the newest version, a user can match on a "one-to-many" basis instead of "big batch to big batch".
Fraudulent doctors and equipment providers can’t get out of Texas fast enough. Bad data causing high costs, missed revenue, lost inventory items, and bad equity or options fills are a thing of the past for S3’s clients . With S3’s software-as-a-Service deployment approach, you can start Teramatching your data in as little as 2 months. Why wait?
For more information please contact S3