In a session at the Becker's Hospital Review 2nd Annual CIO/HIT + Revenue Cycle Conference, July 27 and 28, in Chicago, Paul Bradley, PhD, chief data scientist at ZirMed, discussed data analytics and how developing a data-driven system with predictive analytics can improve revenue cycle management.
The current revenue cycle climate is rife with decreasing reimbursement and disparate systems used by healthcare providers. Many systems don't offer real-time updates leading to a lack of actionable workflow.
According to Dr. Bradley, data-driven systems have a number of benefits in this environment:
• Proactive issue identification that allows for tracking trends
• Clinical and financial data integration that includes recommendations on how to deal with difficult claims cases
• Reporting that allows you to get the information to the right people
• Use of predictive analytics to improve revenue cycle processes
"With predictive analytics, accounts become visible," said Dr. Bradley. "We can look back at a case or issue and see how it was handled. Then we can use that information to predict issues that may come up in the future. You can focus your staff on accounts and issues where their expertise is needed most."
Using predictive models can allow for improved revenue cycle processes. Organizations can use the model to figure out if a charge is missing; a patient will pay their part of the bill; a patient will be readmitted; or a diagnosis-related group code was assigned in error.
With regard to patient payment predictions, algorithms can be used to examine historic data and make correlations. Data-driven systems can be used to create self-pay predictive models that include a number of criteria, such as self pay type, patient class (whether outpatient or inpatient), payment history and debt history. The model can then be used to predict not only if a patient is likely to pay, but also how much they are likely to pay.
"This helps providers and billing offices figure out which accounts to focus on and ascertain what options they can provide," said Dr. Bradley.
However, it is important that the data used for predictive analytics is clean and validated. This means making the data consistent and removing any duplicates. Inconsistent data leads to bad decision-making.
Building and integrating a data-driven system that uses predictive analytics in a healthcare organization can be challenging. Some common barriers include:
• Concurrent initiatives competing for resources
• Lack of appropriate focus on the installation of the data mining and predictive analytics
• High volume but poor quality of data
• Limitations of host systems creating silos of data and lack of reporting capabilities
"But, ultimately, having this [predictive analytics] data layer improves revenue cycle functions because it ensures that everyone is speaking the same language," says Dr. Bradley. "Also, automation takes out non-value add tasks."