Predictive modeling and the elusive 1–2 percent
Amidst the multiple major changes occurring in healthcare today, there is a persistent theme across conversations in healthcare boardrooms and executive offices:
“We know we’re leaving money on the table.”
To say there are dozens of instances when hospitals are uncompensated or underpaid for the care they provide is laughable. There are thousands.
These opportunities are scattered across millions of claims and tens of millions of potentially relevant patient-, procedure-, and provider-level data points. The only way to uncover them is through data-mining and predictive analytics – and it is increasingly vital that they be uncovered.
For hospitals, these hidden opportunities translate to millions of dollars a year – in the realm of charge-capture alone, hospitals on average leave 1–2 percent of net revenue on the table. Yet even the traditional processes for preventing and capturing more of this net revenue can drive up costs – and are almost impossible to optimize without machine-learning and predictive modeling.
Understanding the basis for predictive modeling – patterns, connections, and inevitabilities within your own data
Insights gleaned though data-mining and machine-learning give hospitals the ability to draw upon exponentially greater information points that in turn feed data-driven predictions of charging anomalies. The outcome of this process is a rapid, accurate identification of charge capture issues specifically and instantly tailored to your health system and its unique elements. These identifications can then be intelligently and automatically routed to the right person at the right time – allowing issues to be prioritized and corrected in a manner aligned with a hospital or health system’s unique strategies.
If the technology incorporates the charge description master (CDM) as well as the commercial contracts and applicable government rates, all charging anomalies can be assigned a net revenue impact specific to your health system. Further, machine-learning algorithms underlying these identifications can continue to adapt to changes in a hospital or health system’s charging data, clinical practices, payer contracts, and HCIS configuration as these elements evolve over time. This is not possible with manual intervention and rules-based logic alone – again, due to the volume, variety, and non-uniform forces that drive changes across healthcare data-sets.
In addition to detecting missing charges, predictive modeling can identify over-charging risk, DRG anomalies, and other coding variances.
The bottom line is ensuring fundamentally sound bottom-line optimization
Let me be clear: this is about hospitals and health systems being paid for the care they provide – being paid the full agreed-upon amounts per their contracts with payers. This is about the need to ensure the core financial components of healthcare are fundamentally sound and working as intended.
Without enhanced revenue cycle integrity – and without the ability to accurately model the outcome of a financial process and understand how it will impact the organization – hospitals and health systems are being asked not only to do more with less, but also to do more with a perpetually unknown amount and perpetually limited ability to control what that amount is.
We can’t improve a system if the mechanisms by which it operates aren’t consistent and reliable. We should all support ironclad revenue integrity for hospitals – and that starts with empowering them to model the true impact of uncompensated care, unnecessary denials, and unrealized revenue.