AI That Pays: Lessons from Revenue Cycle — Nathan Wan, Ensemble Health
Channel: aiDotEngineer
Published at: 2025-07-24
YouTube video id: TquUsN1QsWs
Source: https://www.youtube.com/watch?v=TquUsN1QsWs
[Music] So great to be here today. U excited to talk to you about a little bit of the healthare system that often gets overlooked. It's part of the healthare system. A system that actually continues to grow in multiple dimensions. Over the past couple of decades, its size, cost, and complexity outpace many other benchmarks. That's because right now 40% of hospitals operate at a negative margin. Let me put it another way. Almost half the hospitals in the country are losing money. It's uh and it's not because of the clinical costs. It's because of the broken and manual processes around the revenue cycle. There's delays, denials, a lot of rework, a lot of lost revenue. My name is Nathan. I'm the head of AI at Ensemble Health Partners and we work with hundreds of hospitals and health systems in the US to manage the revenue cycle. We're about 14,000 people and we've been a leader on the quality side within the industry. uh as an end-to-end solution that means we support the entire uh process every stage and gives us a really unique lens into uh all the problems and inefficiencies that occur throughout the entire process and also an opportunity to stop them um before they happen. Revenue cycle management or RCM refers to the financial process uh of the patient's journey uh within the healthcare system and it's traditionally thought of as a series of steps uh that go that goes from one to the next and just a little bit about how I I got here. Uh I started my career in tech working at Google building operational software for operational teams and then working on speech recognition and language modeling. Back in my day which is now like over 10 years ago we were comparing language models that took you know traditional models with Google scale data and compared them to deep learning models trying to make them compete with each other and see which one worked better. But one of the really interesting projects that we worked on was called what is now called ambient. Um this is one of the things that um where people are trying to use the technology to improve the administrative burden for um for for for doctors. Um that's because oftent times doctors will spend hours after they see a patient writing, documenting and creating notes for themselves. Um while we weren't successful back back then uh today there's multiple projects and multiple uh multiple groups launching this uh and making it commercially viable and so it's been really successful and really exciting to see that change. Then I spent a little time uh in the world of startups in the world of biotech. uh I changed both the scale and the domain that I worked in and it was very exciting. I was really excited to work on a very strong mission a really exciting mission where we had a a big opportunity to make a big impact. I started first in diagnostics. Uh this is where I built models uh that and and built teams to detect cancer from blood. Um the goal was to give early insight into whether or not a patient had cancer or not. And we use machine learning to look at the blood and look for biomarkers and look across multiple data sets and patients to identify you know where might be the signal for cancer. After spending some time there I and seen the company grow from 30 people to over 300 people. I ended up at a even smaller therapeutic startup. uh we worked on novel data sets looking for unique interactions in complex microbiome communities to try to identify compounds that could be unique and uh really valuable in uh drug discovery and that's the thing when most people think about AI in healthcare these are these are the things that most people think of right there's diagnostics there's imaging there's other ways to improve uh documentation and uh decision-m for clinicians and drug discovery and these really impactful and really important problems. Uh, you know, I I I really enjoyed working on them and but there are some of the toughest and hardest to solve. Um, but that's also because some of the benefits are going to be massive as we continue to see groups and organization crack parts of it and make make headway. Um, we're seeing a lot of change already. But there is another part of the system, the healthare system that's not nearly as flashy, not doesn't nearly have the same attention on research or in media. Um that uh that also has a huge measurable real impact and is also ripe for AI disruption. And that's the healthare the financial side of healthcare. Uh right now we estimate that 20% of the GDP uh is uh is is attributed to the healthare system. A large proportion of that is the administration of healthcare. Uh this is billing, insurance related things, but it starts from like the very beginning through to the very end of the patient's journey uh with any healthcare provider starting with eligibility checks, registration, there's documentation, medical coding, denial management, so on and so forth. Um often times these aren't things that you'll see when you visit a PA doctor's office. uh not unless you're you encounter like a really challenging situation or you lack coverage or um you face some complication with with your care. Um but this process is very complex because it is very manual. It is very rules driven and very inconsistent. Um and in over the past three decades uh this healthcare administration in general has the number of people working in healthcare administration has increased 30fold. Um but in the same time period the number of clinicians clinic clinicians excuse me has barely doubled. So it just goes to show how much faster how much more complex how much how how much how much more quickly this area grows compared to the clinical side of healthcare. And just another note about terminology I'll keep using these words patients providers and payers. Patients I think we all can understand those are the people who receive care and providers are the ones who deliver it. people like the hospitals, the specialty offices, the specialists, nurses and doctors that actually conduct and provide care medical care to the patients. And then payers are are those who provide the funding. So largely insurance companies, which would be private payers, but also other government institutions like Medicare and Medicaid. And just to make um and just to help make that more concrete, right, the the the cost and complexity of healthcare is really related uh is really correlated, we estimate a large amount of the the cost associated with healthcare is actually related to friction. And in his case, friction means the inefficiencies around communicating back and forth between payers and providers and patients. And often a lot of that actually results in um uh in the same situ same outcome, right? Either the claim gets paid or it doesn't get paid. Uh one of the things we'll talk about is denials because that's one of the biggest components of friction. And that's because it's both time and money for the provider to manage appeal and work through that process. And then with again very slim margins uh for these hospital systems um any any any impact or any change in the appeal uh denial rate can have a huge impact and ultimately AI has a big opportunity to shift resources from this bureaucracy this this friction towards hopefully something else right something more productive we might agree would be like the clinical care or anything else that we've uh we've described so just to make it concrete for you guys. This is an example of what a claim might look like and and uh also how much conversation occurs between the patient well largely the providers and the payers you know before during and after a patient visit or patient encounter. In this case this claim was denied four separate times and appealed and uh appealed four separate times as well. uh the the p the provider had to send uh documentation multiple times through multiple interfaces and probably through multiple different uh manual processes and for the provider they didn't receive payment for you know 200 days until after the uh procedure occurred and and with AI happening you know across the field you know providers aren't the only one looking to AI to make an improve improve their process. Pairs are also thinking about it as well. They've also increased their denial rates. They they are leveraging AI as well for increasing the volume that u they're able to adjudicate and to identify uh issues for denials um making this entire process more more strenuous and and creating a much larger backlog for all these providers. And the thing is that most of these denials aren't necessarily don't necessarily require a better appeal system. It's not like they need a a smarter appeal system. They just need a way to avoid errors that cause these in the first place. That is most of the time they're not necessarily medical agreements. They're just technical errors in registration or missing data that if we were to put them together in the right way the first time around, we could avoid a lot of this friction. So how at ensemble are we hoping to be able to solve this problem? Uh we think because we are a end-to-end uh RCM organization full service provider we have an opportunity to see the longitudinal data connect the dots between from the very beginning of the process to the very end of the process and and and really make uh make a change before the error occurs. One of the first situations I'll talk about is is prior off. Um this is uh this is an issue that affects the entire industry and prior arts occur because the prov um payers have required providers to um to ask for permission for certain procedures. Um but it's really challenging because it's often not clear when a prior author is required. You sometimes have to go to the payer portal and say you know is this procedure um is this does this procedure require prior authorization? And even when you do, sometimes they still might deny it because it was incorrect or uh uh it it just the policy had changed. And I think this is where we really think we have an opportunity to change uh to to basically correct the error before it happens because we can see that data from you know see all the historical data from the beginning part where prior offs are requested um and acquired um to the end of the process where we see the final denials. uh where AI can help uh not only can we try to predict the denial, we can also try to identify and correct the the dial. So an example is like if we if we see certain procedures and we know that often another denial reason is that uh the procedure was missing from the original document. And so we can try to flag that early and say if you're looking at these procedures, you actually might actually want this other one. And finally, even the actual process of acquiring prior authorization is a big opportunity. There's a it's a manual process. It requires documentation from different parts of the system to be put together by someone and and sent off to the payers um to to to make that uh request um for for power authorization. But sometimes we we were not as successful. Uh sometimes it's still it still may be the case that um uh denials may still occur. Um, we can't always avoid denials, but we're we're really but we're excited because Genai really has an opportunity to help us accelerate uh and improve that process as well. Uh, the the case study I'll talk about right now is called clinical denials and this occurs when the payer and the provider disagree about what was medically necessary um to to care for the patient. And when when this happens, uh the provider has uh in order for them to appeal, they have to go through a process where they have to build the um build the entire appeal packet. Um they'll need to look through the patient record. They'll look through guidelines or standard guidelines of care to identify what um uh what what care should have been provided to the patient. They look through payer policies to see what which would or would not be covered. And this is all a very timeconuming process. Some of these EMRs or electronic medical records are hundreds of pages long. Uh they have um they have all types of data in them. Text, images, labs, notes, tables. The clinical guidelines themselves have you know hundreds of clinical guidelines. And for different situations, we'll you'll need different different um different guidelines. And all this is done under tight deadlines to make sure that you respond in time to the payer after the denial. And all this means that there's a there's a very real and limiting factor of, you know, how many expert clinicians can we get um into the process to help us build and generate these ns appeals, excuse me. Uh you you as you might expect, you know, Genai can actually uh uh generate an appeal letter for you. an off-the-shelf one can if you prompt it will give you some appeal letter. Um but unfortunately that alone wasn't sufficient. We found that when we worked directly with our clinical experts that uh off-the-shelf models alone wasn't sufficient. And so we really worked hand inand to develop uh a model and a pipeline that allows uh the not only the the letter to be to meet the standard the quality standard um that we have as an organization uh but also to allow the clinical expert to make the final decision on whether or not uh the letter meets the meets the standard of quality before it gets submitted to the to the payer. And this is important because there there's also complexity around the clinical appeal process. There's different service lines, different uh different clients and all that all that gets put together in our Genai system to make sure that we can deliver these appeals uh more quickly and more consistently. We've seen already that we're uh increasing the speed of the process, a 40% reduction in time uh sometimes even more. Um we've also seen higher quality. We've been able to measure quality in terms of the overturn rate. How often are we seeing uh uh appeals being over denials being overturned and we have also seen the as a result the volume grow. But one thing I really want to point out here is that as as part of this operational and service team, we're able to really measure the the ROI directly. It's not just like a hand wavy this is value. We're tracking it very specifically and measuring it um uh very concretely. And that's one of the really exciting things about uh bringing bringing uh bringing AI to this RCM process. So I know AI won't I won't propo I won't purport to say that AI will be able to solve all of the problems overnight. This is an industry that's been reliant on a lot of processes um for a long time. You know, as I said, there's lots of rules. They're inconsistent. It's unstructured. Um you'll see data scattered across a wide range of systems. And this is one of the things that makes it really challenging to to bring together as a cohesive or um and and unified process. Um Ensemble has spent invested a long a lot of uh uh a lot already in building up a single consistent uh infrastructure to be able to do this. And one of the reasons we have been successful has been because of the the platform that we call EIQ where we bring together multiple formats, multiple data formats um within within a single platform. Um, but obviously there's uh still great opportunity to to be able to do that. You'll see that EMRs have many different format types and it will challenge any uh multimodal LLM to to parse correctly. We're excited because we already see AI deliver value as you saw with the clinopial case. Um, but also uh in the prior authorization case, we're built continuing to build agents for all aspects of the revenue cycle process. But we know automation alone isn't going to be enough. There's complexity and revenue cycle. Uh that you know clicking buttons and pushing things uh faster, pushing pieces through faster might not be the only way to do it, right? Like there's really an aspect of reasoning and connectivity that that we think about when we think about, you know, how to take errors that occur at the end of the process like the appeal process or the denial process and try to fix them upstream and early early on. And what we're really hoping for and what we're really excited about is being able to not just build better tooling, but also a smarter, more coordinated system that allows us to reduce and uh reduce waste in the overall revenue cycle process. So that's why I'm really excited to be on at Ensemble. I think we have a unique position to lead this transformation. We have the right data set. We've been building uh the the right team as well to bring all the experts from multiple disciplines to achieve this goal and we have the full scope of the RCM process to not only collect the data but also intervene and act on behalf of our clients. I thank you. I thank you all so much for your attention. Uh I hope this gives you a new way to think about AI and healthcare. And if if you have a chance, please find me and uh connect with me afterwards. Thanks so much. [Music]