The Challenge

Insurers often have a large variety of sources and levels of complexity when it comes to data. With nearly 50 existing data sources and BI tools, this modernization initiative was ambitious. These sources included patient, financial, policy, payment, claim, and administrative data, among others. Accuracy was crucial, as claim rework led to higher costs and detrimental outcomes for patients and providers. In response, the company employed a team of analysts and strategists, all working to identify trends and patterns within the data signifying causes of claim rework. Eliminating major causes of rework was a direct line to better patient outcomes and lower costs.

“The environment was data rich, and analysts were already hard at work looking through it,” explains BlueGranite Healthcare and Life Sciences Solution Architect CJ Knapp. “The problems were the volume and variety of data. With so much data from so many sources, effectively using analytics time and resources became critical.”

Delivering the results of all this analysis to the proper stakeholders – including C-level decision-makers, team leaders, and various busines unit report users – is critical to long-term success. To tackle this issue, the organization sought to create a data platform and analytics solution that was nimble on the edge, with powerful analytics, insightful visuals, and a single source of truth at its core.

BlueGranite utilized its Catalyst for Modern Data Platform to quickly deploy a new Azure-based analytics and data platform to automate the formerly tedious, inefficient report building process; it refreshes data weekly and monthly, pulling information for multiple clients from five source systems into a consolidated data model to create in-depth product overviews. A set of client-agnostic, metadata-driven code allows consultants here to analyze at least 160 categories of data, such as weekly distribution and individual sales for countless products. While the solution currently incorporates several major manufacturers, it can scale out to accommodate any number of clients.

Outcomes

“It was important for our team to meet with both the analysts and the business users to review all of the ways they used data, but even more crucial for us to understand the insights and business outcomes they were hoping to find,” explained Emily Depasse BlueGranite Senior Consultant for Modern Business Intelligence. With the number of analysts working on the problem already, this transformed BlueGranite’s role. Here, our solution needed to allow analysis without fractured data sources, different reports in different systems, and the uncovering of deeper insights not visible to the human eye.

The first step was listening to what our client needed most: insights into how to reduce claims rework rate. “Conversations with business unit leaders and analysts from the different teams allowed us to find key factors that, when addressed, would speed up time to insight and allow more space for advanced analytics,” CJ says, recounting the conversations. “We found fractured data sources, multiple reports with many manual steps, and not a lot of space for predictive analytics. Creating a modern business intelligence platform that creates one single source of truth for all these reports, while easing management concerns, was critical to creating a healthy data estate for our client.”

Setting up for Success
After considering the complexities, BlueGranite’s Catalyst Framework for Modern BI utilized the best architectural practices necessary to address the growing data source and tool issues quickly, while smoothing the transition from QlikView and Excel reports to Power BI. As the complete set of data products is slated to be available in the Azure Data Lake at a later time, a phased approach was appropriate: first, BlueGranite would develop reports in Power BI using the existing on-premises data source, and at the same time create a plan to adjust and expand the data source to Azure Synapse.

This allowed for reporting tool and source centralization, as well as the start of predictive analytics using AutoML in Power BI. “Sometimes we get so used to the routine of how something has always been that it helps to have an outside opinion, and our BlueGranite team provided a fresh perspective on how to migrate their analytics to Power BI and make them even more powerful,” recounts Emily. With this in mind, BlueGranite guided the team’s Modern BI migration effort in three main areas:

1 – Data Centralization: Working with the analysts, business units, IT, and enterprise data management, we were able to uncover that most of the reporting came from the same data. This involved identifying data sources, creating a semantic layer to bring them all into Power BI in an easy to navigate data model for analysts, and creating a data refresh schedule to keep data recent and performant.

2 – Report Creation and Empowerment: The team worked diligently to convert the array of existing reports in QlikView and Excel into automated Power BI reports that had drill-through capabilities, powerful visuals, and a mix of traditional business intelligence reporting and powerful, modern AI visuals.

3 – Predictive Analytics with AutoML: Centralizing the data underlying the claims rework analysis enabled the team to start looking into powerful machine learning techniques to identify underlying causes of rework. “We held a series of meetings to demo our progress and receive feedback, allowing us to leverage our technical understanding of the tools with the domain knowledge from the client. This was critical to understand how to best represent the complicated health insurance claim data inherent in this business and deliver results shaped by the key stakeholders,” recounts BlueGranite Data Scientist Tom Winenandy.

These meetings led to a three-part approach. “First, we built a machine learning model that estimates the percent probability any given claim will be reworked. Second, we built a machine learning model that predicts what will be the likely cause behind a claim being reworked,” states Tom.

Tom continued, “Finally, we also demonstrated the interpretability of the above two AI models to understand which variables contribute to if an insurance claim is predicted to be reworked and why. By knowing the variables associated with claim rework, the organization can better identify areas of improvement, and intervene when possible to reduce expensive processing costs.”




Working from Centralized and Organized Data
“Before the migration, our end users would follow a monthly pattern of extracting data from one or more sources and performing manual analysis in Excel. Our Power BI solution makes the ad hoc analysis they’re used to even easier, and it makes better use of the data and their time with rich interactive visuals and AI capabilities,” Emily recounts of the value added by our solution.

The key to reaching goals for analytics is establishing a data source and a set of data practices that are centralized and consistent across all team members. Using dataflows in Power BI brings all the data together in common, verified data models, thereby streamlining all future reporting and analytics.

Emily explained it best, saying, “The data analysts and consumers we worked with had an invaluable knowledge of data available, which we used to design Power BI datasets, reports, and machine learning models. It was important for us to work closely with the stakeholders to make the most of their data.”

BlueGranite sought a variety of perspectives on the legacy solutions and how improvements could be made in this first approach. Working with the technical team offered an important understanding of the dataset and how to build a data model designed for performance. The analysts who work in the detailed data day to day solidified the important use cases, while leadership offered a vision on the future of analytics, as well as the types of questions they hope this dataset can answer.

With a variety of use cases and requirements in mind, BlueGranite replaced existing QlikView functionality in Power BI and created new visuals for more advanced analysis. Some key features include:

• Bookmarks and buttons in Power BI for easy navigation and optimal user experience.
• Dynamic measures which group attribute values outside of a “top n” number of values, giving users power to control the level of detail displayed.
• AI features in Power BI, such as Key Influencers and Decomposition Tree, which can replace manual root cause analysis in Excel.
• Interactive visuals, drill-through, and export functionality.
• Data refresh triggered by a Power Automate flow to work seamlessly with irregular data source refresh cycle.
  
  

The Results

Rapidly, BlueGranite was able to help this insurance provider reach the intended goals by identifying causes of rework, centralizing data analysis and reporting, implementing some initial predictive analytics, and lowering the number of hours spent on organizing reporting data. The proof of success was in the vaunted time-savings in reporting and analysis creation, as well as having automated reporting with machine learning-driven analytics identifying and categorizing rework causes. Report viewership and understanding rose with each subsequent effort.

The specialized, precise deployment of this well-architected solution ensures rapid ROI realization. As skilled analysts continue to extend and promote the analytics created in this Power BI environment, more people across the organization can glean actionable insight to provide better claims processing and operational efficiency. With the new claims reprocessing reporting suite in Power BI:

• Advanced, modern business intelligence visualizations enable deeper understanding of claims rework.
• Users spend less time performing manual analysis and more time using insights to establish the important cost-saving process changes that prevent claims reprocessing.
• Analysts can feel more confident in the accuracy of data without having to visit various systems to find what they need.

The Power BI migration is a big improvement over legacy visuals, but it’s just the beginning. While the business team adjusts to the new reports and visuals in Power BI, the groundwork is being laid for a more detailed and enhanced data source. A modern, scalable solution built using Databricks and Azure Synapse will be better equipped to handle a large volume of data. With this expanded dataset, data scientists can build and train models to identify claims as candidates for rework before it even occurs – allowing for earlier intervention and better outcomes for patients, providers, and the healthcare insurer alike.

To learn how BlueGranite can speed up Power BI deployment and adoption for your organization, connect with us today.