A leading national health insurance company used HGS’s intelligent machine learning solution, leading to cognitive intake automation cutting costs by up to 50%.
For healthcare payers, today’s winning solutions are a delicate balance of controlling costs, while simultaneously enhancing customer experience and adhering to strict regulatory requirements. A leading national health insurance plan sought to shore up manual reading of unstructured/structured information on grievances—a key weakness in their claims resolution process.
The client had several disparate, manual claims-processing systems and relied on a human claims processor to pull data to deliver a service intake of faxes and scanned document images into manual queues. The inputs consisted of different sources and format types, and working through repetitive processes was a less-than-optimal use of employees’ time. As a result, the client was facing an 84% upheld rate of denial decisions in several markets. With the client’s average auto-forward rate of four per month, this led to a negative impact on their Star ratings. There were over 5000 claims/week and seasonality in data added further complexities to the issue.
HGS implemented a proactive approach to automate steps in the intake process to improve process efficiencies.
Before deploying the analytics and insights as-a-service solution, the process flow looked like this:
HGS delivered an intelligent machine learning solution to help improve the client’s Star rating, specifically by improving grievances and appeals efficiencies. The team addressed the entire workflow with a cognitive content processing solution, leveraging an image analytics engine to process the source. The solution featured a natural language processing (NLP) engine to analyze keywords and context in client grievances and appeals correspondence.
Documents were classified as expedited or non-expedited on the basis of rules or associate judgment. The solution leveraged an NLP engine to minimize subjectivity and improve process efficiency and addressed all types of text—unstructured, semi-structured, and fully structured. The NLP engine sent the documents to the machine learning text classifier, which tagged and queued documents for further processing. With self-learning, this cognitive solution recognized patterns of subjectivity and replicated human-associated decision-making over time.
The process re-engineering solution included the following:
- Intelligent content processing through an HGS-developed internal image analytics engine for source and format-agnostic extraction and processing
- NLP engine that analyzed keywords, context, and intent in the documents
- ML engine to classify the documents based on image analytics and NLP to push the documents into the required queue
After deploying the analytics and insights as-a-service solution, the innovation workflow was defined as below:
The project was launched within two months with an objective of reducing overturned denial decisions and auto-forward rates by 50-75%.
While the solution has enabled significant process enhancements, it also offers additional value from all the case data gathered. The system analyzes exceptions in cases from the past to determine the direction of the workflow through a feedback learning loop. The loop helps tackle similar scenarios in the future, which ensures that case exceptions are always taken care of without requiring human intervention.
The solution has improved the ratings in the following areas: appeals auto-forward; appeals upheld; timely decisions about appeals; and reviewing appeals decisions. Key outcomes of the HGS solution included:
- An average of 0 auto-forwards, thereby improving the Star ratings of the MA plan by an 8% weighted average.
- FTE reduction of 30-50%, with a 40% workflow efficiency improvement.
Call to Action
Are you looking for analytics and insights as-a-service solution? Learn more.