Innovation Lab
PROTOTYPES
Identify Medicare Improper Payments – CORMAC’s Innovation Lab Prototype using Machine Learning on Skilled Nursing Facility (SNF) Claims Data
The Unsupervised Machine Learning Model provides a highly scalable machine driven way to identify SNF claims that have a high potential of being improper claims. The Model analyzes billions of claims and identifies risk score based on data patterns and correlations. Focusing on the high-risk cohort reduces administrative efforts going towards examining records which were paid correctly. This approach saves time and money, while providing a path to curtail improper payments before they are paid.
Simplifying Health Quality Measure Compliance—Using RPA to Streamline Data Extraction
CMS's Post-Acute Care (PAC) application maintains a list of Quality Measures (QM) that are tracked for reporting purposes in order to provide better healthcare and improve spending efficiency. Currently, the QMs are manually tracked by visiting various Federal Registry websites where the government publishes the measures that need to be tracked and the PAC team gathers those measures in an Excel spreadsheet to review and track compliance.
To bring in efficiency and timeliness, CORMAC built Robotic processes using Automation Anywhere to convert a human process to a robotic process. The process Robots automatically extract the data from various federal registries and save the data to an Excel workbook for further processing.
The successful creation of Robotic processes helped in gaining a 90% time saving over the human process and minimized the chances of error.
Microservices to Modernize a Medicare Legacy System’s Final Action Claim Function
CORMAC redesigned one of the Medicare Legacy systems functions named the Final Action Claim. We used MicroServices-based architecture and modern data framework including a Data Lake for processing files, along with DevOps tools for faster integration and deployment. These MicroServices can be used as a plug-and-play code-based architecture so that they can be put together in varying orders to achieve the desired functionalities. This approach minimized the complexity of the code resulting in decreased maintenance requirements.
