Fraud, waste, and abuse are rampant in the healthcare industry. Tens of billions of dollars are lost to fraud in the United States each year, and the opioid crisis has made the problem worse.
In September, for example, the U.S. Department of Justice announced the arrests of 58 people in Texas, 16 of them doctors or other medical professionals, for their alleged involvement in Medicare fraud schemes and networks of “pill mill” clinics.
And, in 2017, the Boston Globe uncovered a massive network of brokers who specialized in enrolling multiple patients with phony addresses via Healthcare.gov, the Affordable Care Act portal. The result? Tens of thousands of dollars in kickbacks to brokers and physicians and no treatment for patients.
Fraudulent activities clearly are helping fuel the opioid crisis. But, rooting them out isn’t easy. It’s a huge big data exercise that requires analysis of internal patient, opioid treatment center and referring physician data. This data, in turn, must be connected with external data such as recently used addresses and phone numbers to find potential hidden connections among patients, doctors and treatment facilities.
We must also benchmark the end-to-end cost of care for every member across a network of doctors and opioid treatment facilities to identify higher-than-average cost of care, especially for new insurance policies issued under Obamacare or another government medical care plans.
Traditional analytics solutions built on relational databases aren’t up to the task. They require complex, slow, and expensive work with large tables containing prescriber, claims and patient data. And while graph database technology makes uncovering referral relationships much easier, first- and second-generation graph databases can’t scale as the data volume and number and complexity of relationships grow.
The latest graph databases and analytics, however, are designed to handle these challenges. Data scientists and business users, such as fraud investigators, can go 10 or more levels deep into the data in real-time across billions of claims and millions of members and prescribers.
The ability to look penetratingly into large data sets by traversing multiple connections in data is called deep link analysis. Consider a doctor who is referring to a large portion of patients to a specific substance abuse treatment center specializing in opioid addiction.
The latest technology can mine data from third-party sources such as Thomson Reuters or Dunn and Bradstreet to find all known administrators for the drug treatment center and their current and previous addresses.
Another way for a healthcare payer to spot fraud, waste or abuse is with hub and community detection. First, we find the most influential prescribers driving maximum referrals for a specific health condition. Then, we can establish a community of connected prescribers and the members availing the services from the providers.
Analysts can map out multiple communities of connected prescribers while comparing the average cost of care for a specific condition, such as high blood pressure. Which doctors have an above-average spend on the cost of care? The answers are in the data.
It’s never been more important for healthcare organizations to identify nefarious practices and take steps to eliminate them. Graph databases are a key ally in the fight.
Gaurav Deshpande is vice president of marketing and David Ronald is director of product marketing at TigerGraph.