BI Keeps Health Care Healthy
Most discussions about health care tend to focus on who pays for it and how much it costs. But there are many other areas of rapid change in the industry, and Business Intelligence is playing an important role in helping physicians, hospitals, payers, and compliance officials get the most out of the resources spent.
BI and data analytics are a boon to practitioners of evidence-based medicine, both individual doctors treating individual patients and hospitals looking for better outcomes for all their patients. Hospitals are increasingly using BI systems for both clinical and financial purposes. Medicare fraud investigators use BI to find patterns that indicate improper or fraudulent claims. BI is even being used to assess and improve the general public’s access to health care in rural areas, and their outcomes.
Let’s take a look at how data analytics help doctors make faster, more accurate diagnoses, and how BI systems are a cost-effective way to take a bite out of the growing problem of Medicare and insurance fraud. Upcoming columns will discuss more health care issues and technology solutions.
The ability of BI to track and analyze post-treatment outcomes represents a significant advance in medical delivery. With Electronic Health Records (EHRs), it’s become much easier to study how different treatment protocols work -- in the field, not just in controlled trials. The transition to ICD-10 diagnostic codes will allow clinicians to make finer distinctions between conditions – and even today, EHRs and BI are giving doctors deeper insight into how to diagnose and treat specific combinations of symptoms.
The Boston chapter of The Data Warehousing Institute recently hosted a fascinating talk by Ted Slater, an authority on knowledge engineering and its use in molecular biology. Slater described techniques for aggregating medical information using semantics-based methods to establish relationships between information entities (such as medical records of patients with shared symptoms or diseases, taken from different databases). If you’re interested in how these relationships are established, a 2012 paper co-authored by Slater describes how life sciences researchers have tackled this problem – the aim is to get information out of silos and give doctors and researchers an integrated knowledge base to help them diagnose and treat patients. This type of “interoperable, semantically-rich data” can also lead to new uses for existing medicines.
As the cost of health care expands, so does the opportunity for miscreants to defraud the system. Fraud investigator Harry Markopolos, best known as the sleuth who exposed the giant Ponzi scheme run by hedge-fund manager Bernard Madoff, identifies Medicare fraud as one of the fastest-growing white-collar crimes. Addressing a meeting of the Massachusetts Society of CPAs, he observed that, “Drug dealers are going out of that business and into Medicare fraud, because there’s more money in it and the risk is lower.” Estimates of Medicare and Medicaid fraud typically put the proportion of improper claims at around 10 percent, or about $75 billion per year. Some experts, however, say the fraud total could be two or three times that amount. In any case, it’s a serious problem.
March’s 4Sight BI Perspective column discussed how BI helps auditors find anomalous transactions in any organization – things like padded expense reports, excessive corporate credit card limits, unauthorized electronic funds transfers, and other failures of internal controls. These are found via “supervised statistical methods,” so-called because they require samples from both known fraudulent and non-fraudulent records to arrive at a model for each.
If you don’t have a set of known fraudulent records, unsupervised methods can be used. The most obvious is simply to examine a set of records and flag the outliers for further investigation. Comparing a provider’s pattern of claims with those of similar providers in the same geographical area will often identify someone making fraudulent claims – if you see a home care provider ordering three times as many wheelchairs per hundred patients as anyone else, they may be up to something.
A Nevada study of Medicaid claims found that looking at claim patterns for such mundane items as disposable diapers could identify fraudsters – they were the ones who consistently claimed to have bought the maximum number that Medicaid would cover every month. The study found BI-based fraud detection a cost-effective way to identify fraudulent providers. And although drugstore-grade medical supplies wouldn’t seem like a big source of fraud, they are – because of the ease of setting up a phony pharmacy or medical office, submitting a few hundred thousand dollars in false claims, then disappearing before auditors figure out what’s going on. Faster detection is needed. That’s why BI – along with frequent data updates – has become an essential tool in the fight against fraud.
Health care generates so much data – financial, clinical, and administrative – that the potential Business Intelligence applications are almost unlimited. We’ll take a look at some more aspects of health care next month.
By John Kafalas
This monthly column covers Business Intelligence and data analysis issues. If you have questions, comments, or topic suggestions, please contact the author.
Copyright ©2014 4Sight Technologies, Inc. All rights reserved.