By Alex Moore
The Center for Medicare and Medicaid Services (CMS) is borrowing the new approach for reducing waste, fraud, and abuse. It’s called the “Crushing Fraud Chili Cook-Off,” but it doesn’t involve any debates about whether beans belong in the pot. Instead, it’s a challenge for building machine learning models that can detect potential fraud in Medicare’s claims data.
How can you call it a chili cook-off without the chili?
The term comes from the world of data science, which is a profession for people who specialize in building models that predict different outcomes. Just like there’s no one right way to make chili, there’s usually not one right way to build a model. So, just like a chef might try a lot of chili recipes to see which one works best with a certain set of ingredients, data scientists often build a lot of models and have them compete when trying to solve a problem. Hence, the chili cook-off name.
How will the competition work?
The chili cook-off will unfold in two phases. The first phase, already complete, involved soliciting proposals from teams interested in participating in the competition. Submissions closed on September 19th and were required to propose a new way of applying machine learning to detect Medicare fraud, an explanation of how such an approach could improve CMS’s decision making, and a description of how the approach could scale across the agency.
After the end of the first phase, CMS announced 10 teams would advance to the next round of the competition. At this point, on October 30th, CMS provided a limited dataset to each team (in the interest of protecting user data) so they could begin experimenting. Each team had until December 1st to test their approach and submit a report showing the results of their efforts. CMS had not announced a winner at the time of this writing but is on track to do so on December 15th.
The team that wins the competition looks like it will receive a substantial advantage in federal contracting opportunities. While CMS will have the right to publish the work of all 10 finalists, CMS has said that “participants are welcome to leverage these innovations for any future contract bids.” Winners will also receive public recognition.
Why does the chili cook-off matter?
It represents a new way of cultivating innovation to fight Medicare Fraud. Top data science competitions often receive hundreds of participants, ranging from students and hobbyists to top experts in the field.
In 2024, the U.S. Patent and Trademark Office held a similar competition to improve its system for patent agents to search for related patent documents when reviewing applications. This is key for helping agents spend less time trying to find the relevant documents for a decision and more time reviewing the details that matter.