From Excel to AI: How Liberty Dental revolutionized care management

Liberty Dental Plan insures about 7 million people in the United States as a dental insurance company. We are owned by a private equity company and Anthem, which is now called Elevance Health. 

Our job is to take care of all 7 million members of our members. So, my primary job as senior VP of health plan operations as well as being the CIO is to take care of technology and cybersecurity. And over time I have been given more responsibility on the operations side: claims processing and utilization management, for instance, both of which are the key to any health insurance company (or any insurance company, really). So, those operations also fall under my belt as well. It’s a good balance between technology strategy and then applying that technology to operational areas as well. All of these are great opportunities to learn about and understand the whole business. 

For any health insurance company, preventive care management is critical to keeping costs low. The key to keeping costs low is that the number of claims must be low. So how much preventive care can you adopt to take care of your members to keep claims low and to keep costs low? 

Health insurance companies hire a bunch of nurses — call them care managers or case managers — who do the case management for all these people. 

At Liberty, this used to all be a manual process. based on a bunch of Excel files, even Microsoft Access and emails. So first and foremost, we automated all of it. Then, what care managers used to do is to create a care plan in detail, which used to take anywhere between an hour to 90 minutes to create a care plan for one member. 

Care plans are about setting goals. identifying barriers, doing interventions and identifying the patient actions that need to happen. So, we aggregated all this data, applied some machine learning algorithms on top of it and then fed it into large language models (LLMs) and now use generative AI (genAI), which gives us an output of these care plans. 

So, what had been taking 90 minutes per plan, per member, is available right there, with way higher accuracy than what a human could do — because they had to go through pages and pages of doctor’s notes, which is humanly impossible. There could be a small word about some socioeconomic barrier in those notes that someone could miss. But genAI doesn’t miss that. This project — for which we won a CIO 100 award this year — has, first and foremost, increased our accuracy. It’s also helped our care managers big time because they each get assigned thousands of members, which has significantly improved their output. 

The data factor 

I joined Liberty Dental about two and a half years ago, and the first big opportunity I saw was data, which was all over the place. We had a kind of small data warehouse on-prem. That was my first project. Like anyone else around that time, we moved it to the cloud, but in a way that it was only what was needed. We created our data model in a way that satisfied the requirements of what we had a vision of. 

So even before this whole ChatGPT/genAI became a big thing — like three months before that — we went live completely in the cloud on Azure. That was the foundation.  

But the biggest point is data governance. You can host data anywhere — on-prem or in the cloud — but if your data quality is not good, it serves no purpose. Data governance was the biggest piece that we took care of. That foundation was why we could go live with this within six months after genAI hit the market. And we’ve already seen a big ROI on this. 

The importance of education 

Another big part of this project was to educate everyone, including the board, not just on the data governance aspect, but overall, what we were planning to do and how it would benefit the organization.  

Even if I greenlighted the project and asked for, say, capex dollars, it was hard for them to envision how we could capture the right level of ROI. It was important to educate senior management on how things work, walking people through how ChatGPT works, going through how we would design our prompts and all that. Because it takes a bit of education to understand all of it. And it’s highly important that we as technologists share all the business knowledge. That is key. 

To anyone in IT, I tell them, you might know segment anything model (SAM) machine learning or Python or whatever the case might be. You’re still a commodity. It’s not until you learn the business and all the domain aspects that you become valuable. That’s always been our focus, and it’s helped us tremendously. 

Use cases matter 

Heading into 202, there’ll be more and more adoption of genAI in business because people are talking more and more about various use cases like mine, how they’re leveraging genAI within their business domain. So that gives people the idea that maybe they should investigate it, which is where knowledge of the whole business domain is important. Technologists can apply those tools and point out the use cases that are opportunities to use genAI. 

With genAI, it’s a matter of identifying the proper use case that gives you the ROI. In Liberty’s case, the ROI was like 400%…and it was realized very quickly because of all the work we’d done previously. 

Rajendra Kadam is the SVP, Health Plan Operations and CIO for Liberty Dental. He is responsible for Liberty’s information technology, security, claims operations and configuration. He joined Liberty in 2021 with more than 21 years of experience in technology and 15 years in healthcare.

This article was made possible by our partnership with the IASA Chief Architect Forum. The CAF’s purpose is to test, challenge and support the art and science of Business Technology Architecture and its evolution over time as well as grow the influence and leadership of chief architects both inside and outside the profession. The CAF is a leadership community of the IASA, the leading non-profit professional association for business technology architects.