Analytics in Healthcare: How to Survive the ICD-10 Transition

Posted by Sarah Anderson on Thu, Oct 01, 2015

Is now the right time to add machine-learning analytics into your charge-capture management system?

ICD-10ICD-10 is here and chances are, as a provider, you’re as ready for it as you can be, knowing there could be some hiccups and impact on revenues. Most predict that the impact will be confined to inpatient revenues, driven by significant adjustment issues in grouper methodologies. But what if the impact extends well beyond that, to outpatient revenues?

Most hospitals today rely on rules-based charge-capture systems that require humans to manually input thousands of rules to ensure that all outpatient charges are captured correctly and completely. Most systems come with a global set of rules and require manual effort to maintain and customize to the hospital. However, ICD-10 has nearly five times the number of diagnosis codes as ICD-9. Since many rules rely in part on diagnoses and procedures, much of the rule logic may break down in the early stages of this transition, and it will take time for new rules to be coded, tested, and implemented. In the meantime, your charge capture may suffer.

A machine-learning, pattern-based analytics approach, however, addresses the entire issue differently. Instead of manually developing new rules, the system starts with the hospital billing data and looks for charging patterns. It’s a living, breathing system that continues to learn as more billing data passes through it, and it learns rapidly — in a matter of weeks or months. Furthermore, the machine-learning algorithms have evolved to a point where they can draw conclusions irrespective of diagnosis codes just by looking at the patterns among charges, a massive advantage with the change to ICD-10.

In addition, a machine-learning solution prioritizes each flagged outlier charge based on the payer plan reimbursement amount combined with the likelihood that it’s actually missing or incorrect, so auditors start every day investigating the most valuable claims. This is a key advantage during this transition period, when the impact of the increased workload will be felt the most. What’s more, instead of just one-way communication that provides only lists of recommendations, it’s a self-adapting system that learns from the feedback provided by the auditors via an easy-to-use Web interface. 

A machine-learning system also provides visibility into why errors are occurring. Management can quickly pinpoint problem areas and address them immediately with hospital staff, further reducing the ICD-10 learning curve for humans. If you’ve been waiting for the right time to implement an advanced analytics solution into your revenue cycle management platform, this is it. You can shorten your transition time, find more missing charges than ever before, increase auditor productivity, and get your charge capture back on track.

To learn more about how to get started, download our article, published by HFM, on Hospital Revenue Leakage, our revenue cycle management solution for providers just like you.



Sarah Anderson is the Director of Marketing at Opera Solutions.

Download Now



Topics: Healthcare, Machine Learning