Adnan Haque:
Hello, and welcome to our Digital Solutions podcast, The Future is Now, where we discuss the rapidly changing digital landscape in life insurance. I'm your host Adnan Haque, and I lead alitheia, which is our suite of instant decision tools we offer to our partners. So in our previous podcasts, we've discussed electronic health records, what they are, the rich digital health data they contain, their potential for fast, accurate risk decisions, and also the challenges the industry has faced with implementing EHRs. Today, we're going to talk about the new era of EHRs, specifically our Automated EHR Summarizer, which addresses many of the previous challenges that we've discussed. So joining me today is Rachel Eberle. Rachel, why don't you introduce yourself?

Rachel Eberle:
Thanks, Adnan. Rachel Eberle, and I have been an underwriter for almost 20 years now. And I have worked extensively on EHRs here at Munich Re, as well as the risk assessment part, alitheia.

Adnan Haque:
So, Rachel, you've been using the Automated EHR Summarizer on a daily basis, and we'll see how it's been going for you. Let's set the stage and really briefly bring everyone up to speed with EHRs and life insurance underwriting. So why should we care about EHRs?

Rachel Eberle:
Well, EHRs, the hit rates are increasing. So, for example, our company, Clareto, currently has hit rates at 50% to 60%, depending on the region and target market. About a year ago, hit rates were at 30% to 40%. So we do see significant gains in a relatively short period of time. And unlike other third-party data sources, where each one might provide one or two types of information, such as prescription info or clinical labs, EHRs really include all of that information in one source, which does provide the potential to replace other third-party data.

Adnan Haque:
So you mentioned these data elements, as you've looked through an EHR, what data do you think is most helpful to underwriters like yourself?

Rachel Eberle:
There's so much rich data in EHRs that underwriters need every day for risk assessment. EHRs provide us the ability to have an applicant's vitals, diagnoses, current medications, labs, as well as social history, such as smoking status, all in one place.

Adnan Haque:
So we have all this great data in EHR, but it can be sometimes unstructured, or semi-structured, and difficult to get to. So can you talk a little bit about how Munich Re addresses the problem?

Rachel Eberle:
Sure. Our teams at Munich Re have been working on normalizing EHR data into a structured format that not only can be used by a human underwriter to review the EHR, but can also be used in an AUW rule engine, as well as models. And we're applying natural language processing technology to the unstructured free text often found within an EHR to be able to consume that information in a more structured format.

Adnan Haque:
What are some of the common pitfalls that you've seen underwriters face, or these overarching automated programs face, when trying to leverage an EHR?

Rachel Eberle:
That's a great question. I think when our industry first started utilizing EHRs, we really thought of them as more of a replacement for an APS, or attending physician statement. And it's really not a one-to-one comparison. Both are a rich source of information, but they're not the same and they don't have the same use case. I believe that we should be thinking of an EHR more as another third-party data source, as opposed to a replacement for the traditional medical records.

Adnan Haque:
You mentioned the Automated EHR Summarizer, so not everyone is familiar with what it is and what it does. Why don't you give us a brief overview of what it really is?

Rachel Eberle:
Our Automated EHR Summarizer is Munich Re's EHR solution that gathers all of the underlying data within an EHR and presents the underwriting relevant information in two formats – a human-readable HTML format that an underwriter can go through in a very similar way that they would go through a set of medical records, as well as structured digital data. And that structured digital data is what we can utilize in rules engines and models.

Adnan Haque:
As both a developer and consumer in your day-to-day of the Automated EHR Summarizer, can you walk us through a little bit of how you and some of your peers are using the tool?

Rachel Eberle:
Like we just talked about for AUW programs, that is a great use case for the tool. The tool is able to get that data, and do it in a structured format, that can be used in a rules engine. So that would be one use case. The second would be for a more manual review or a light touch use case. So even in an AUW program, there are cases that are kicked out for an underwriter to review or gather more information, and that would be a perfect use case for an EHR. And that Automated EHR Summarizer would help the underwriter to flag what impairments were concerning, to find the most recent BMI or BP readings – all elements that are important when an underwriter is reviewing those cases.

I think the last way that an EHR Summarizer is very useful is in post-issue monitoring. That's a very large piece that needs to be done for AUW programs. However, it's very time-consuming, and is often pushed to the side because those are cases that are already issued and there's not an agent calling. Those tend to take priority. But it is important for AUW programs to do post-issue monitoring. We're seeing that a lot of companies are using EHRs for that. The Summarizer would really help the underwriter to be able to triage those cases much quicker and more easily with the Summarizer.

Adnan Haque:
With those three use cases you outlined, light touch underwriting, automated review, and then this post-issue monitoring, which of these three do you feel like the Summarizer has the most significant impact?

Rachel Eberle:
Right now, we're seeing probably the most significant impact on the post-issue monitoring and light touch side. Just because, with a lot of programs, there's still some more work to do for the automated rules engine side. But that's a very easy, quick win for the Automated EHR Summarizer, that post-issue monitoring and the light touch.

Adnan Haque:
I think you were getting into this, and I know it's relatively early stage, but talk us through a little bit more on the results you're seeing so far? And then how you see this progressing over the next six months, two years, five years, et cetera?

Rachel Eberle:
At Munich, we did a recent cost-benefit analysis and found that the mortality savings can vary widely, but can be anywhere from $100 to $800 per case.

Adnan Haque:
That's a big difference.

Rachel Eberle:
It really is. With hit rates continuing to increase, we're going to see higher adoption rates of EHRs as those hit rates continue to go up and up. We'll also, perhaps, start to see it replacing other third-party data sources, especially in AUW programs, and allow those programs to gather more information without pulling so many different third-party data sources.

Adnan Haque:
As a user, and you've gone through these results a few different times with different companies, what's your favorite section of the Summarizer as you go through it?

Rachel Eberle:
Mine would be the impairments. I think, as an underwriter, the top thing that you need to know when assessing a case is, what's going on with this applicant? What are their diagnoses? What else do I need to look for? That really informs what other information you need in the case. So being able to quickly go and assess and look at all the different diagnoses that have been pulled from the EHR in a quick spot is extremely helpful.

Adnan Haque:
So a lot of the pieces you mentioned with unstructured data, being able to parse out information, structure information, running NLP on it, potentially, a lot of that sounds like a service that may be best developed by a technology company. So why do you feel like Munich Re is the right partner to offer this type of service?

Rachel Eberle:
Ultimately, this tool is used for risk assessment, and Munich Re is an expert in risk assessment. This is our bread and butter and what we do every day. This tool has been developed with underwriters in mind, the end user in mind, developed for underwriters by underwriters. So I think, compared to other technology companies that maybe don't have that use case in mind or the risk assessment expertise, Munich is the right partner for that.

Adnan Haque:
I think that's an important point. We have strong technology expertise. We have a pretty sizable engineering team. We've always put risk assessment first and foremost because that's really the core DNA of our company on the life side. So how do you feel like, and this is a maybe more challenging question, how do you feel like this solution benchmarks against other tools out there in the market that solve a similar problem?

Rachel Eberle:
Yeah, I think there certainly are a lot of other tools out on the market. I do think the Automated EHR Summarizer is unique because of our focus on the data, and our focus on taking unstructured data and getting it into a structured format that can be used for a lot of different use cases. The biggest one being in rules engines. I don't see too many competitors out there putting a lot of effort towards that. The EHR Summarizer, it really helps, not only that human that's reading it, but will also help down the road as AUW programs continue to expand.

Adnan Haque:
We've talked about parsing of the information, showing it to an underwriter, taking digital data, feeding it into an automated solution. Let's talk about the risk assessment. Why don't you tease a little bit on the components you're working on and the team is working on, on assessing the risk from an EHR?

Rachel Eberle:
Our ultimate goal is to be able to make a point-of-sale decision using an EHR. Let's take an example of someone with a medical history of depression. We're able to pull out the type of medication that they're taking. That could be an antipsychotic medication, or something very benign, like an SSRI. We're able to pull out whether or not they've been hospitalized in the past, their current GAD or PHQ score, and use all that information in rules to determine a specific rating for the applicant.

Adnan Haque:
So say I'm a carrier, I say, "Hey, Rachel, you're doing some really cool stuff over there. How do I try it out?" What does that look like?

Rachel Eberle:
Well, good news, Adnan, we're offering a pilot for our clients to be able to get a sample of EHRs that they want to send us. We would provide the Automated EHR Summarizer to them, as well as our decisioning on that EHR. They can try that out for free.

Adnan Haque:
Perfect. And that decisioning, what kind of assessments are we talking? What comes back to you?

Rachel Eberle:
So, right now, you're going to get whether or not that EHR was flagged with concerning information in it or not flagged. Then you're going to get all of the impairments that were found, the current vitals, current BP readings, BMI, all of that is going to come back in the Automated EHR Summarizer.

Adnan Haque:
So, Rachel, I know you hang out with a lot of data scientists on the team. Why don't you tell us a little bit on what's coming on the data science front for EHRs?

Rachel Eberle:
Our data scientists have created two severity impairment models, one for hypertension and one for depression, that will be able to assess the severity of each of those impairments, using information from the EHR.

Adnan Haque:
I know, Rachel, the other model that's coming along the pipeline is one that really addresses the item you were talking about earlier. That's about how do you build trust that information is there in the EHR and you've extracted that information. So let's call it, we don't have a good name yet, but let's call it a confidence model for EHRs to really know, "Hey, I've actually gotten the information out, and I'm confident in the assessment."

Rachel Eberle:
Yep, so included in our Automated EHR Summarizer will be a confidence score, or something similar, that will let the user know how well that information was able to be pulled into the EHR Summarizer. As we know, the data structure in EHRs can vary greatly by provider, and even doctor. Some have better data than others. So that will let the user know what level of confidence they can have that that information was correctly pulled.

Adnan Haque:
Tell us a little bit about this confidence score.

Rachel Eberle:
The confidence score allows the user to trust that the information presented to them in the EHR Summarizer is correct.

Adnan Haque:
For me, trust is such an important word. Because I feel like everything we do at Munich Re is about providing trust and confidence for our partners, whether it's taking on risk on our balance sheet or all the work we do on the risk assessment side. So what will it take to really spur further adoption of these types of tools as EHRs continue to take off?

Rachel Eberle:
Further adoption, I think, as we see hit rates get higher and higher, that will certainly spur adoption. It's also important, at companies, that EHRs are really championed from the top down. There is some change management involved in that. But if an EHR can be utilized first, before a traditional set of medical records, I think that's a good way for companies to be able to see some of the value that an EHR can provide. Then, if an APS is still needed, that's fine, but there is so much rich information in an EHR that can be very valuable to the underwriter. We know that the time and cost savings of an EHR versus traditional medical records is certainly high.

Adnan Haque:
In almost every conversation I've been in with carriers, with different partners, they've seen value in having a summary, being able to have digital data, being able to triage a number of cases. One thing that comes up every once in a while with using the summary is, "Okay, what do you do if it gets it wrong? What do you do if some information is missing in the summary?" How do you manage that as part of building trust in this product? What do you say to people who have that concern?

Rachel Eberle:
I think it's important to note that no third-party data source has perfectly complete or perfectly correct information. We know, as underwriters, that there are often errors in traditional medical records that are received. And we have to deal with that. And I view all the other third-party data sources, including an EHR, no differently. I do think, as the user continues to use it and becomes more comfortable with it, some of that will wear off. I can remember third-party data sources that came out years ago. It's challenging for underwriters to sometimes adopt new things. But I think the more someone utilizes something, and the more comfortable they become with it, then that will happen naturally. EHRs are no different.

Adnan Haque:
We talked a lot about where we are today with this tool. We talked a little bit about where it's going over the next six months. How do you expect this tool, and adjacent tools to the Automated EHR Summarizer, to evolve going into the future?

Rachel Eberle:
They'll evolve as we're able to get better and better at processing some of that unstructured data, that will allow us to evolve. Ultimately, we need to be able to assess the full EHR in its entirety. As we improve and get better and better with that, we're very good at certain sections, a lot of the free text can be very challenging, but those are all things that Munich Re is working on. As that improves, the sky's limit in terms of what the use cases are for EHRs. I do think it could, and probably will, replace or minimize some of the use of all of the more individual third-party data we have out there. There's data that has prescription info, lab info -- EHRs really have that info in its entirety. To be able to use that in underwriting, in an automated fashion, will really allow AUW programs to expand far beyond where they are right now.

Adnan Haque:
In the industry as a whole, instant decision rates are roughly 15% to 20% on average as of the last survey we did. STP, where we're talking about fluidless decisions on close to fully under in mortality, that's roughly in the 45% range. If you were to pick a number, two years from now, where do you think we'll be, using EHRs?

Rachel Eberle:
I think I'll play it conservative, and say we'll be over 50%. But I think it's important to not just... It's a double-edged coin or two sides of the same coin. Because we certainly want STP rates to increase. But I think a tool like an EHR also provides us a lot of mortality savings, and will help with mortality slippage in AUW programs. We know that mortality slippage by AUW program can really vary. And an EHR is a great tool to help to minimize some of that.

Adnan Haque:
That makes sense. So I've been asked to... And this is a question that's slightly different than all the previous ones. I've been asked to ask this as a closing question. Is a hot dog a sandwich?

Rachel Eberle:
As a Chicago gal, I'm going to say no.

Adnan Haque:
No? There's no rationale at all.

Rachel Eberle:
It's different from a sandwich. It's its own category.

Adnan Haque:
All right, I'll take that. All right, so as life insurance lives on data, we, as a team, have been very, very data focused and data first approach to underwriting at Munich Re. We're really excited about the new era of EHRs as a data source. We think it's really impactful to dramatically shift what risk assessment looks like for the bulk of the industry. So, Rachel, thank you for taking the time to share how you're using the Automated EHR Summarizer to improve results.

Rachel Eberle:
Thank you for having me. It's exciting to be able to talk about our Automated EHR Summarizer, and to have clients and underwriters find success in utilizing it.

Adnan Haque:
Thank you so much. Next time we're here, we'll tackle another topic on everyone's mind, generative AI. So focusing on what AI can do for your business, and how you can develop a robust strategy, and go on your AI journey. So thanks for listening to our Digital Solutions podcast. We hope you join us next time.