Protecting people’s health and well-being with AI
Faster data pipelines
Faster insurance risk assessment
The rapid advancement of AI is transforming the healthcare industry and improving patient outcomes. But in order to deliver high-quality care, the ability to efficiently process payment and reimbursements is critical to sustaining the financial viability of healthcare providers, ensuring timely access to care for patients, and supporting the overall functioning of the healthcare ecosystem. With AI-centric solutions, healthcare organizations can better understand trends and proactively identify risks and areas of improvement. Milliman's MedInsight group is trusted across every segment of the healthcare industry — providing actuarial insights and consulting. With data volume and complexity rising, the company's outdated on-prem database systems proved too complex to maintain and rigid to scale. The need to transform this process was imperative, and Milliman saw this early on and turned to Databricks Data Intelligence Platform to serve as the foundation for their innovation. With their ongoing investment in AI technologies such as large language models (LLMs) and a unifying approach to data, analytics and AI, Milliman MedInsight understood that the Databricks Data Intelligence Platform had everything the company needed to reinvent how they deliver value to their clients. With faster processing power that doesn't slow with large workloads, Milliman MedInsight can now deliver better insights to their healthcare customers in a shorter amount of time, which they can use to serve their patients better while maintaining financial stability.
Legacy solutions can't keep up with modern healthcare data needs
Milliman is a data company at their core, and their MedInsight product group is focused on helping U.S. healthcare customers — payers, providers, nonprofits, municipalities, etc. — assess risks, evaluate insurance portfolios, optimize pricing strategies and make smarter decisions. The healthcare ecosystem is governed by how reimbursements are made and how efficiently payments flow. To bring clarity to organizations, Milliman MedInsight needed to deliver clean and complete data to their clients. The process of aggregating and cleansing data for clients became increasingly time-consuming and burdensome as healthcare data sets grew in both volume and complexity. And in more recent years, timeliness became an increasingly important factor. A small team tried building out a 50-node cluster on Apache Spark™ but found it overwhelmingly laborious to maintain, equating every hour of valuable customer-impacting work with four hours needed to maintain the cluster.
"The healthcare system needs all types of clean and accurate data in near real-time to function effectively," said Iyibo Jack, the Principal, Senior VP of Product Development at Milliman MedInsight. "As that volume of data and the number of attributions within the data increased, our systems and team lacked the flexibility and resources to handle the scale and timeliness needed to provide value to our healthcare customers."
It was clear that Milliman MedInsight needed to rethink how technology could best support their business requirements and future-proof their operations as they evolve into more advanced technologies such as LLMs to ensure they continue to serve as the trusted partner for the world's largest healthcare companies.
Transforming healthcare with AI today and into the future
One of the big reasons why Milliman turned to Databricks Data Intelligence Platform was because they looked to the future and planned with AI in mind. This ensured Milliman had the support and AI-centric capabilities they needed to bring valuable insights to the 300+ healthcare companies they supported.
"The world is changing fast, and so are the demands on data teams," said Jack. "If I ran an internal data warehousing team, Snowflake would have been a good fit for those specific requirements. But we're planning for the future that included AI, so we wanted to partner with the company that has the pedigree and technical track record to take us where we need to be."
Milliman MedInsight uses Delta Lake as their optimized storage layer, providing different data tiers for various use cases from raw data to highlight cleansed and complete data for reporting and machine learning. As data increases, Jack's team relies on Auto Loader to incrementally and efficiently process new data files without any additional setup. And their data science team is focused on managing ML processes directly in MLflow. With the decoupling of compute and storage, Jack's team is able to tap into the full power of machine learning without overburdening operational budgets.
Beyond bringing data and analytics together, the lakehouse has also brought Milliman MedInsight's various data teams together, improving collaboration to develop new solutions such as the Data Science Portal (DSP). The DSP is a data science center of excellence, providing the company's customers with self-service access to a massive library of data and models, all while removing the complexities of provisioning clusters, preparing data and ensuring security. "Our Data Science Portal helps our clients cut through the noise and answer more questions with enhanced data and analytical models," explained Jack. "Facilitated through a workspace in the lakehouse, we've created a culture of efficiency and productivity that has empowered our citizen developers like never before."
Looking ahead to AI innovations in healthcare
Through Databricks Data Intelligence Platform, Milliman MedInsight has found a unifying platform that can help them transform healthcare for the better. With a unified architecture that combines data, analytics and AI under a single roof, they are poised to lead in AI innovations that will help them deliver better patient outcomes faster and more efficiently. And it all starts with the data.
With the lakehouse, Milliman MedInsight has seen anywhere from 15x to 35x improved workload performance. What does that mean? As data is processed faster, the insights and predictions that are served to the end users also accelerate. As a result, Milliman MedInsight has seen a 5x to 7x increase in the delivery of insights into health insurance risk to their customers, allowing them to make smarter decisions faster.
Milliman MedInsight is solving healthcare data needs not only now, but with complete trust in Databricks to support a strategic road map to an AI-driven future. Part of their plan? To build an AI assistant, powered by large language models (LLMs), for MedInsight that focuses on all the data unique to the company — domain expertise, documentation and training — and then training the LLMs with terabytes of healthcare claims data to help their clients make smarter decisions that impact the millions of healthcare patients that Milliman MedInsight's customers serve.
What sets Databricks Data Intelligence Platform apart is that it's built with an open reference architecture that works at scale. That's what Milliman MedInsight is putting their trust and future into. "If I were to sum up Databricks in a single word, I would say 'transformative,'" concluded Jack. "They've evolved tremendously, and we are taking full advantage of it to address the challenges our customers will face not only today but years from now."
