«

»

Using Analytics to Prevent Fraud in Healthcare

Share

Using analytics to prevent fraud in healthcareWhile fraud may be a concern for many busy business professionals and organisational decision makers today, the evidence suggests it is becoming more  prevalent in the healthcare industry. The amount of data that goes through healthcare systems on a daily basis is phenomenal. Applying advanced analytics to all that data can assist in combatting fraud. We only cover two technologies here, but it showcases the possibilities of using analytics to combat fraud.

Did you know? Discovery Health¹ released a report in July 2012, which indicated that based on International data, losses due to healthcare fraud and abuse may account for between 3% and 15.4% of claims paid – with the average  approximately 7%.

There is a wealth of information captured in the collective healthcare ‘system’ on an on-going basis.  Consider that administrators have to monitor every patient; every visit; every antibiotic prescribed; the diagnosis the doctor gives the patient; every test undertaken; the amount of times a patient is submitted into hospital, as well as a mountain of information about each hospital visit. With so much data and information penetrating this sector, it is no wonder that the healthcare sector has become an easy target for fraud!

This got me thinking about technology and tools and how these can help to mitigate so of these risks and alleviate the pressures from an already stressful market. Looking at BI and advanced analytics and the benefits these technologies can bring to an organisation, how are these technologies deployed in the healthcare industry to try and stop or reduce fraud?

Advanced Analytics³ in its truest sense is defined as using extensive statistical and mathematical models to understand customer behaviour, predict retention, predict churn, and of course to predict fraud. At its heart, it provides the algorithms for complex analysis of either structured or unstructured data including sophisticated statistical models, machine learning, neural networks, text analytics, and other advanced analytics techniques.

By working closely with healthcare investigators, IBM⁴ have developed a fraud and abuse management solution, which employs advanced analytics for detecting and preventing incorrect claims before they are submitted to the patient’s insurance company. This management solution provides staff members with the tools and methods to analytically and scientifically find invalid/false claims that are often buried in the thousands of claims processed per year.

IBM states “The solution uses algorithms developed by IBM Research to analyse claims and pinpoint which claims are most likely to be erroneous. We use several types of analytics crucial to detecting fraudulent data, including outlier detection, comparing similar providers or claims to identify deviant or non-standard submissions; predictive models, defining patterns of abusive claims submissions and segmentation models, defining previously-unknown behaviour patterns. Our solution provides a graphical representation of aggregated claims data and detailed analysis identifying suspicious claims, including the specific reasons for why they are suspicious. This level of detail can be used to make smarter decisions about the amount of financial risks, due to fraud and error.””

SAP’s Fraud Management application, implemented on the SAP HANA in-memory database, enables healthcare organisations to detect, investigate, analyse, and prevent fraudulent activities in a high volume, high throughput environment. With the database and analytics performed in memory, fraud monitoring and detection is done in real-time as it is directly integrated and automated as part of the business processes. With the underlying data so readily available, and with very fast analysis turn-around times, investigators can analyse potentially fraudulent activities in near-time, predict fraudulent behaviours, and run real-time simulations to test hypotheses around fraud.

With any packaged application such as the two listed here, I always caution that one should carefully weigh up the degree of accuracy of models provided as part of a packaged solution, compared to purposefully developed models that are managed, evaluated and re-calibrated on an on-going basis. Packaged solutions often contain so-called industry best practice models, but the influential variables and their respective weights of importance, for example, may differ so much from where the model was developed to the culture and business environment where it is applied that the model outcomes are not always 100% correct. Of course, the outcomes of the models can be tested and evaluated on the real data to determine their accuracy and to adjust their variables, weights and algorithms. On a positive side, however, a packaged application can give you a very quick start to get going.

Some of the technological developments that will make a big difference in fraud management in healthcare are in-memory databases like SAP HANA and/or in-memory analytics  solutions, like SAP Predictive Analytics running against HANA or SAS High Performance Analytics running in a Hadoop cluster’s memory space. These highly speeded up analytical solutions make it possible to apply advanced analytical model to claims in near-time, before they are submitted for payment. Equally useful is streaming analytics technologies, also called Complex Event Processing (CEP) platforms, such as SAP Sybase’s Event Stream Processor, which applies analytical models to detect fraud “on the fly” as the data is processed through the system.

These solutions illustrate that technologies like business intelligence and advanced analytics should be used more actively in the healthcare industry to detect fraudulent claims. Although fraud will never completely disappear in this industry, by using in-memory or streaming advanced analytics, administrators will be given the option to evaluate claims in near-time and reduce fraud considerably.

References

¹ Current fraud trends within the healthcare industry

http://www.fanews.co.za/article.asp?FraudCrime~5,General~1094,Current_fraud_trends_within_the_healthcare_industry~12293

² Clampdown on healthcare fraud

http://www.fin24.com/Money/Clampdown-on-healthcare-fraud-20120726

³ What is advanced analytics?

http://fbhalper.wordpress.com/2010/12/20/what-is-advanced-analytics/

⁴ Smarter Healthcare

http://www.ibm.com/smarterplanet/za/en/healthcare_solutions/nextsteps/solution/Y649550I52610O70.html

 

Leave a Reply