According to MedCityNews: “With fraud estimates as high as $272 billion annually across the healthcare industry, there’s always been good reason for payers to devote significant resources to better detecting and preventing fraud, waste, and abuse.” This means data, and data means the integration of data so we can abstract out indictors that are able to spot, and prove fraud.
“For payers, this starts with thinking about fraud analytics tools as one piece of a more comprehensive, data-focused prevention strategy. By focusing on technologies that enable easier data integration and sharing between stakeholders, payers can begin to move away from simply flagging abnormal claims to linking together individuals and funds in ways that uncover fraudulent networks lurking below the surface, as it were.”
The problem? If we keep data in silos, and don’t share it, there is no ability to spot patterns that appear to be, or are, fraud. This is not about combining data for the purposes of working with shared business processes; this is about abstracting certain data so it can be understood in the context of other data.
For instance, let’s say there are a series of claims to a payer from one individual or perhaps a healthcare provider. Those are not suspicious unto themselves, but when we add the data from other payers into the mix, and leverage the right analytics, we can quickly discover that those who make requests for payment have done so several times, for the same health care provided.
Not all cases are that easy. If we have access to data inter-payer, we need to consider subtler patterns that can be found in the abstracted data. For instance, perhaps the same name is not used, but they do use the same address, or the same e-mail. While this scenario may not be direct fraud per se, it could lead to closer examination to determine if fraud really exists.
The use of machine learning and other AI capabilities as applied to the abstracted data holds great promise as well. As we build data analysis processes that can both think and learn, we remove the humans from the mundane task of looking at thousands of pieces of suspicious activity that may not lead anywhere. These intelligent systems can spot patterns, and refine their understanding of what the patterns represent, based upon experience.
“Stakeholders have staggering volumes of siloed data spread across their enterprises that can be leveraged to augment these efforts. A data-focused approach enhances the utility of fraud algorithms because they are able to pull from more complete datasets, creating less false positives before investigation.”
If you think this is very “Big Brother”ish, not at all. This is about stopping those who remove resources from a system that everyone needs, and thus allows users to leverage those resources in better ways. Payers attempt to spot these issues as best they can, but their greatest chance of success comes with integrated data that allows them a more holistic view of the transactions.