Healthcare organizations that take the following five steps can ensure successful implementation of a predictive analytics tool.
1. Secure commitment across the business office. Get on the same page. Finance, revenue cycle functions, and IT should collaborate to identify criteria for success, resources, implementation strategies, and if and how the existing process will be affected. Consider formalizing an approval process to ensure that checks and balances are in place. This is the first critical step to ensuring that expectations are on target for the delivery of accurate and consistent analytics.
2. Acquire an all-access pass to data. Predictive models are only as accurate as the data being modeled. Ensuring the data integrity and validity is a vital prerequisite to predictive modeling. Executives need intimate knowledge of the data required to model (e.g., the timeframe, fields, and sources). IT is often an understaffed department, so having detailed data requirements in place ahead of time goes along with expediting the data extracts needed for predictive analytics.
3. Review models. Using sampling to review the analysis from the models is the most important step in the process. It is important not to rely on technical data experts to analyze data, but instead to work directly with the line-of-business managers and staff who are tasked with specific end goals to randomly sample accounts to review and validate that the model has accurately predicted the right outcome. Depending on the purpose of your analysis, answer the following questions: Does the account truly have a missing charge? Did the model predict correctly the accounts that are contractually underpaid? Did the model flag the accounts that were sitting in bad debt although the patient had insurance that covered the procedure?
4. Establish metrics. Predictive analytics is highly measurable, so it is easy to establish metrics, set performance targets, and monitor performance. The benefits of predictive analytics can be twofold: identifying revenue opportunity and increasing staff efficiency to recover the revenue. First, determine baseline metrics for cash collections or staff productivity, for example, before applying a predictive analytics tool. Once the tool is applied, measure against the baseline to determine the impact. For example, how many accounts were auditors able to review, resolve, and collect against in one week before and after implementing a predictive analytics tool?
5. Automate model scoring. The data must always be current. The power of predictive analytics is that once an approved model has been built, it can continuously score new data 24/7. The key to making this process efficient is to establish a consistent data extract schedule with IT, automate the process of scoring the new data, and validate.
Publication Date: Monday, February 01, 2010