Experian Health VP: Using AI to Reduce Health Care Costs and Staff Burnout

Johnathan Menard, vice president of product, analytics at Experian Health, talks to Omdia analyst Andrew Brosnan about using AI to reduce health care costs by automating administrative tasks, which will also improve staff burnout rates. Deploying AI to bring out deeper insights from aggregated data will also lead to better decision-making among providers. But the cost of AI deployment and lack of workforce with relevant skills are hampering adoption.

Listen to the podcast below or read the edited transcript of the interview.

AI Business: Can you tell us about Experian Health and your role there?

Johnathan Menard: I'm head of analytics with Experian Health for the past few years now. … Experian Health serves as an administrative backbone for thousands of health care providers across the country for more than 25 years. Our mission is straightforward: We simplify health care. We provide products that help providers operate more quickly, smoothly, efficiently across the full continuum of the health care revenue cycle journey. We serve more than 60% of U.S. hospitals in this capacity, more than 7,700 medical practices, labs, pharmacies and the like.

We're also a part of Experian, the global information services company, and that gives us a breadth of resources from across our entire enterprise to bring many innovative solutions into the health care industry.

AI Business: How do you see AI fitting into what you’re doing with a lot of these health care organizations?

Menard: The clinical side has really moved faster to adoption, but we believe that there's a huge potential on the operations side of health care. There are many repetitive, tedious tasks, large amounts of data that's already collected, mostly structured and standardized, and that can be organized and analyzed with AI to help improve efficiency and accuracy.

You're talking about impacts the from a financial perspective: The wider adoption of AI in this way could lead to 5% to 10% (reduction) in total U.S. health care spending. That's roughly $200 billion to $360 billion annually within the next five years, and that's according to a recent paper by the National Bureau of Economic Research.

So if a third of all health care costs are associated with administrative tasks, that's why we committed to (addressing it with the development of) an AI platform. That will be the basis of many products, with our first launch in the market of AI Advantage recently focused just on improving the claims process because this was listed as one of the top contributors to wasted health care spending in the U.S. I think it was a Humana study that puts it at $265 billion.

It's even supported by our own surveys − last year, a third of health care executives said denials (of claims) were increasing at a rate of 10% to 15%. The average denial rate is already above 11%. That's one in 10 patients potentially having to deal with the uncertainty of who (will pay the health care bill) when they should be focusing on wellness. That's where we see Experian Health being able to lean in and drive some value and change in the health care industry with AI.

AI Business: You mentioned that there’s greater adoption on the clinical side than the operational side. What do you think are some of the challenges in operations that makes this so?

Menard: I think it's safe to categorize the start of 2023 in general as bringing about a wave of AI advancement. …  We hear daily about the launches of new AI solutions across industries and health care’s no different. Health care in general has seen AI adoption primarily led by the clinical side (but) we're starting to see signs of shifts in that approach. More priority and funding is being directed to reducing the administrative and operational burdens. Specifically for the health care revenue cycle space, it's just starting to gain momentum.

But there's definitely some skepticism due to early offerings not really meeting expectations. So the challenge is that we are in this very important early phase of educating and aligning our subset of the industry on establishing healthy expectations of AI − identifying the critical and urgent jobs to be done but also where is AI as a tool is best suited or not to deliver value.

As for what are some of the challenges, there are some universal truths like general resistance to something new. It requires learning new skillsets, new ways of doing business. It even takes personal vulnerability by those leading the change, choosing to take risks in what is a heavily risk-averse (industry) culture. … Just like other industries, we have the other issues such as data integrity, regulations, ethics, and costs that impede adoption.

For health care, that last one weighs as one of the heaviest. A recent survey shows that the biggest barrier to AI adoption was actually cost. In health care specifically, it's not just a matter of implementing the technology or solution but also maintaining it on a yearly basis with talent. Organizations are going to have to recruit an AI-competent workforce.

A similar study by KPMG found the same: The high cost and lack of workforce talent were the biggest barriers to adoption of AI in the industry. This is particularly a more challenging barrier in health care, because anywhere from 53% to 68% of the nation's hospitals are operating in the red, according to the AHA (American Hospital Association). So they just can't compete with the salaries that other industries can (offer), with their razor thin operating margins − if they're even in the positive.

AI Business: Can vendors help address the cost issue since they are embedding AI into their existing products? Will that help address the skills and cost challenges?

Menard: Yes, it will. Providers should look to explore different partners and vendors to make the right choices for them specifically. Vendors have the economy of scale, the data and hard lessons learned across broader markets (and can use these advantages to help clients as opposed to) having individual providers tackle some of these pain points themselves.

I see it becoming necessary for vendors specifically to embed their AI directly into the points of interactions they already have established with providers to drive a different style of work. It's no longer about exception-based work. It's about how you drive insights earlier into the process to avoid the need for the work in the first place. How are you getting more predictive? … How are you using data at scale to drive insights to build decision engines and let AI do what it does best with human-in-the-loop augmentation?

AI Business: By bringing together different sources of data, can providers open up not just insights but also potential new revenue streams once they have integrated the data − and can this help offset the costs of doing these AI initiatives?

Menard: I absolutely do. It's actually a partnership between providers and their vendors. Data is helping every industry become more efficient and productive. Health care should be no exception. There is a mounting pressure for health care organizations to deliver better outcomes while minimizing operational costs − do more with less, and do it better than ever before.

At Experian Health, we have health data spanning eligibility and benefits, address, identity, claims remittance payments; take that and you couple it with the insights we have on 300-plus million consumers, 126 million households, we're able to offer providers one of the most holistic views of today's health care consumer. That's where it gets really exciting when you think about partnering with providers in augmenting those kinds of capabilities (such as in electronic health records, medical insights) that they have in their systems to be able to drive a different style of care.

In the value-based care model, it's about costs, right? (That means it also has to be) about prevention − that's an exciting path for health care to mature towards, and it's going to take data assets outside of what providers currently have, and merging all those disparate data assets into one place and towards one common purpose that will also unlock brand new value opportunities that haven't existed before.

AI Business: Pulling together all these different data sources can actually lead to keeping people from getting sick in the first place with proactive intervention. How far away do you do you think that capability is?

Menard: It's probably in the two- to four-year range. These assets all exist; there are other industries that are doing this and Experian Global is really starting to bend the maturity of the data adoption curve, to be able to understand fit-for-purpose data, and the ability to merge these things. So I think we're very close. And that becomes exciting on the revenue cycle side, being able to reallocate some of that cost that goes from there and funnel it back into letting providers do what they're supposed to do − provide care versus trying to collect for the care that's provided.

AI Business: On the operational side, do you see some prime areas for greater adoption? What are some use cases the industry should be moving on first?

Menard: Some key areas include claims denials – you’ve got to get paid for the services that you're providing. When you focus on that problem, it ‘unlocks the safety’ and the potential to be able to explore different areas.

In health care, it started on the revenue cycle side; a lot of our adoption was in the robotic process automation (RPA) front. What comes next acts as a springboard to inject more smart process automation capabilities – making automation more robust, less onerous and less costly to maintain.

AI Business: How do you see generative AI making an impact on the health care industry?

Menard: Like most industries, something like ChatGPT can be beneficial for health care. It can be used with patient interfacing and engagement, like answering patient queries, providing relevant information, assisting in scheduling appointments, and managing appointment reminders. At Experian Health, we call it the ‘digital front door’ to help providers with tasks that can be conducted online in advance.

Adoption is important because it’s what patients expect. They expect those digital options, especially when they're seeing hundreds of apps hitting the market, this starts to build the expectation of what service looks like from outside markets into our own.

When we did our own study, The State of Patient Access, the results showed that both providers and patients want access to digital functions, such as scheduling and registration. But unfortunately, they also agree that digital services in the health care space are not improving when asked if the experience of accessing health care is better or worse.

In the prior year, only 17% of patients surveyed said it was better. That's a small subset of the population that feels that digital services are getting better for them. And (their expectation of user-friendly interfaces is) only going to grow as they get more such services from other industries.

AI Business: How do you see generative AI being rolled out in the health care industry? Will it be internally facing first and then patient-facing? Or will these things happen in parallel?

Menard: There will be a crawl, walk, run approach, starting internally, being able to screen test some of these capabilities safely inside to improve efficiency before it's released more externally.

I think some of the most important digital services for patients in order to have a positive experience including being able to schedule appointments online or via mobile devices, having an online mobile option for payments, and more digital options for managing just health care in general. I would anticipate that those become the first venues where we move away from the internal and expose capabilities that are consumer-facing.

AI Business: What do you think about some of the privacy issues and concerns that have been raised since ChatGPT became public? How can health care organizations and vendors work towards addressing some of those concerns?

Menard: Those become the primary reason for the crawl, walk, run experience. We will be very safe and secure stewards of that data, making sure that in any capacity these capabilities are developed and released with the utmost security at the forefront.

In health care, that slows down our adoption and our speed-to-market compared to other industries. But as we sit and watch others learn from their own mistakes, we get the benefit of a second entrance, being able to learn from the experts that have gone in before.

Being responsible first is going to be a trend across most of the health care industry, both on the provider side as well as the vendor side, making sure that all the data security concerns are shored up and that any activity we engage in is compliant with current industry security standards. And that'll take some time to sort out.

AI Business: Do you see generative AI helping to bring together those disparate data sources and allowing users to generate insights from across the data landscape through an easy-to-use chat interface?

Menard: I think it'll work. As more and more industries use those interfaces, the trust in the capabilities (will grow) for key stakeholders in health care. And as they learn to trust the power of big data, the power of bringing these data assets together, the permissible use of permission rights will be embraced. The ability to make a large impact in other industries is what's going to unlock it in the health care industry.

As we watch other first actors do great things with the capabilities, being able to use low code, no code and generative solutions, these will make it faster, easier and cheaper to develop solutions that will help with speed-to-market and competitiveness, and our ability to also generate more solutions at scale.

We’re very excited about the potential of AI and our innovation at Experian Health. That’s why we built the AI Advantage platform – to launch other products in the future, to solve other issues throughout the health care journey. … We talked a little bit earlier about automation, adoption and health care. To me, the best way to automate a process is to eliminate the need for it in the first place.

AI Business: Hospitals are under a lot of revenue pressure. Also, in the U.K., the NHS talks a lot about employee burnout because they’re overloaded with work. Do you think health care organizations will find a way to make these investments that will ultimately bring about greater efficiency and address some of the staffing concerns?

Menard: I don’t really see there being a big choice as we progress in the current environment. Providers need to make sure staff see the benefits of what this technology can bring. They also have to make sure that they give them the proper training on how to embrace these capabilities. They do not replace your job; they augment you to do more, or they allow you to focus on doing the right thing, not the right thing that needs their specific level of expertise. It all comes back to the health care industry being able to measure those improvements.

On the operations side, I'm very confident that staff can be shown the ROI (return on investment) in the form of reduced administrative burden since they're the ones bearing that burden. They will be focused on higher priority stuff, that they're making a difference. Health care is a care industry; it's what most people that work there are there for − to be able to help and provide that care in whatever way their skills allow.

It's going to have several benefits that they can measure: Spending more time with patients, seeing more of their activities result in positive outcomes, and on the back end, recouping more dollars. ... I think that that will drive staffing satisfaction and we'll see less turnover of this highly skilled workforce.

AI Business: Do we need more proof of concept (PoC) projects around AI in the industry or do you think that senior management in health care organizations recognize the potential return on these investments? Are will still in the PoC phase?

Menard: I think we're in the PoC phase. These executives don't have the ability to take gambles, to take really big risks if they're already in the red, or if they're operating on a 3% margin. They have to be very certain about where they invest their budget, where are they trying to assign their limited funds to being able to drive the most value.

That was at the core of what we released with AI Advantage. We can show a client how the model is going to act on their data, before they make a purchase decision. This is the value we can very directly give you. If this is the total opportunity, this is the expected ROI, here's how early adopters are seeing it and when we apply that to your actual data, this is what it translates to you − it takes the risk out of it.

You can't do that without the proof of concept. There are too many competing priorities, especially in the revenue cycle, and (executives) need to be laser-focused and very confident in their decision-making. And that's what makes that possible.