Since the introduction of ChatGPT, the healthcare industry has been fascinated by the potential of AI models to generate new content. While the average person might be awed by how AI can create new images or re-imagine voices, healthcare is focused on how large language models can be used in their organizations. However, the effort to build, train, and evaluate this modeling is only a small fraction of what is needed to reap the vast benefits of generative AI technology.
Consider the iceberg analogy. The LLMs, algorithms, and structures that a healthcare payer or provider interacts with represent the visible part of the iceberg. But as we all know, what is below the surface and not visible is significantly larger and more important. For healthcare organizations, what’s below is data—vast amounts of data that LLMs will have to be trained on.
This is where the healthcare industry has a distinct advantage because payers and providers are sitting on an enormous amount of existing data. In fact, the average hospital produces 50 petabytes of data a year. That amount of data is more than twice the data currently housed in the U.S. Library of Congress. Nearly 80% of hospital data is unstructured and most of it has been underutilized until now.
To build effective and scalable generative AI solutions, healthcare organizations will have to think beyond the models that are visible at the surface. Payers and providers will need to create a data foundation that addresses elements such as bringing in the right data, how to classify it, and how to create a data lineage so data sources can be tracked to address potential AI hallucinations.
Key elements of this foundation are data strategy, data governance, and data engineering. A healthcare payer or provider must establish a data strategy to define its vision, goals, and roadmap for the organization to manage its data. This is the overarching guidance that drives digital transformation. Next is governance; the rules, policies, and processes to ensure data quality and integrity. Data engineering is the process of collecting and acquiring the right data in the right format from multiple sources.
All of this data management work must be completed with intention and planning so your generative AI solution will perform as expected. Additionally, healthcare organizations will need to look at existing workflows, problems that they are trying to solve, and applications that need to be launched. For any healthcare payer or provider, your strongest business asset will become your data and your model, especially when unstructured data becomes fully integrated.
The need for generative AI data management may seem daunting. The good news is that what once was an underwater mountain of unstructured data that was inaccessible can now be surfaced with generative AI. Organizations will be able to access information from sources such as claims, eligibility forms, enrollment data from providers, and policies, along with utilization management data, clinical data, medical records, and contracts.
Today, data scientists and business analysts are the end users of generative AI. One day soon, generative AI functionality will become mainstream and will be integral to healthcare delivery. Every part of the patient’s journey has the potential to benefit from this transformative technology, from finding the right provider to improving the hospital administration process to discovering insights into managing chronic conditions.
Generative AI has the potential to transform how healthcare is delivered, managed, and paid.
EXL, a leading data analytics and digital solutions company, can help your healthcare organization get there faster. We have developed an AI platform that combines foundational generative AI models combined with our expertise in data engineering, AI solutions, and proprietary data sets. Our platform features an AI workbench, plug-and-play generative AI accelerators, and security and compliance.
Our EXL Health team, with more than 8,000 healthcare professionals, including more than 2,000 health data analytics specialists and 2,100 clinical resources, works with many leading healthcare organizations and has a deep knowledge of healthcare customer data. We help payers and providers bring to the surface the data that matters to improve patient outcomes, optimize revenue, and maximize profitability across the care continuum.
To see how using generative AI technology can improve health outcomes, watch this short video. To learn more about generative AI in healthcare and why data management matters, watch our on-demand webinar. Learn more about how EXL can put generative AI to work for your business here.
About the Authors:
Jay Nambiar is the chief technology officer for the Healthcare business at EXL, a multinational data analytics and digital operations and solutions company. Arun Juyal is vice president and the global lead for digital transformation efforts for the Healthcare business at EXL.
