Our largest claims extract is 250,000 claims for one year for one company. It took our system nearly 9 hours to import it into our SQL data base using custom-built Ruby code. Talend does the same job in about 10 minutes. I'm actually astonished at how quickly Talend manages to crunch through this data.George Sibble, Director of Finance and Strategy
Building a better health care system
change:healthcare was founded in 2006 to bring order to the staggering amount of billing paperwork generated by today's healthcare system. The technology firm is dedicated to educating healthcare consumers with the belief that if people know the price, they can save money. Its unique internet-based solutions are available as an employee benefit through employers and Third-Party Administrators (TPAs) or directly to consumers interested in controlling their healthcare costs through true consumerism.
The company's premier product, the Healthcare Consumerism Index™ (HCI), helps employers and employees better understand the cost-effectiveness of their healthcare consumer decisions. The HCI report presents actionable information to employees to help them modify future purchase behaviors towards smarter, more value-driven healthcare decisions, thanks to a complete, real-time, HIPAA-compliant view of their past and current healthcare expenses.
Building with open source
A small, but quickly growing start-up, change:healthcare collects medical claims data from various TPAs and converts it into a single format in order to run comparables and reports. The company wanted a highly scalable ETL solution that could handle daily dumps from multiple companies in a variety of formats, each containing hundreds of millions of dollars in claims. The cost savings associated with open source was also a factor in the decision process. Research on the internet led them to Talend Open Studio for Data Integration.
Industry-related security concerns
There are unusually strict privacy considerations within the healthcare industry. change:healthcare typically works with self-funded companies, which are organizations that pay their own medical claims. As the law is written, these companies have a right to all their claims information, but can't view it themselves because they are not allowed to see their employees' claims. The intermediary is a TPA which operates like a credit-card company, handling all the processing and then telling the company how much they owe for that month's healthcare claims.
"These self-funded companies need to get their own information processed and presented to them," said George Sibble, Director of Finance and Strategy at change:healthcare. "We deal primarily with small, regional TPAs who then upsell us to their clients, because we allow them to save a tremendous amount on their healthcare costs. We're dedicated to transparency and help both employers and employees - as well as general consumers - leverage healthcare experiences in a way that both engages and aids them in choosing more cost-effective healthcare for themselves and their families. Just as an example, a monthly prescription might cost $500 from one venue and only $150 at another. We show people how they can save money."
"Since we're dealing with such sensitive information," added George Sibble, "we don't allow any direct access to the database. We output everything into our own text file format, use it in our internal software, and then import them into the system. Eventually, we'll integrate Talend directly into the SQL database and smooth out this process."
Multiple formats and codes
At present, change:healthcare is dealing with three formats for its medical claims - a positional text file; a crypt type code, presented as eight different positional text files; and UMR which is a CSV. In addition, there are five or six different types of Pharmacy claims. "Companies send us their data in PDF format, or Excel spreadsheets, or something entirely different with a description of where the data fields are," said George Sibble. "As the claims come in, Talend Open Studio for Data Integration converts them automatically into our target formats. The solution provides a very intuitive user interface and we use its graphical mapping and transformation capabilities extensively."
"As soon as we downloaded the product we realized it was the perfect solution for us," added George Sibble. "The speed at which we can operate is just invaluable."
Saving time and money
"Time and dollars are critical in the start-up world," said George Sibble. "Our largest claims extract is 250,000 claims for one year for one company. It used to take our system nearly 9 hours to import it into our database using custom-built Ruby code. But running the same frame through an amazingly complicated ETL means that I need to take every claim apart and rebuild it from scratch to get correct variables on it. Talend does the same job in about 10 minutes. I'm actually astonished at how quickly Talend manages to crunch through this data."
Because Talend is both fast and scalable, change:healthcare was able to save a great deal of money on programming. "Now, any time I change the schema of a file in any degree," added George Sibble, "I can do it myself in five minutes, instead of paying a programmer. Talend is a prepackaged solution so we save quite a bit on deploying and testing software. I estimate that with Talend we've saved close to $100,000 in development costs, which is huge for a company of our size."
"I had no experience with data integration software prior to Talend," said George Sibble, "and found Talend Open Studio for Data Integration to be remarkably easy to learn. I'm self-taught. Whenever I didn't clearly understand something, I went to TalendForge.org - Talend's community web site that contains the Talend Exchange, the Forums, the Bugtracker, etc. - and usually got a response within 10 minutes. It's an amazing resource. I know that whatever ideas I come up with, they'll help me find a way to implement them."
"Talend Data Integration is in our near future," concluded George Sibble. "We're excited about the enterprise features, such as automatic scheduling, triggers, and reports. We're also going to move to a centralized repository and are looking forward to setting up multiple users."