In this episode, Marc Jeffreys, General Manager, Health, Gradient AI, discusses how AI can streamline group health underwriting while keeping human judgment at the center of decision-making. He explores risk assessment, model transparency, and what health plans should look for in AI partners.
Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.
Transcript
From the show. Tap a time to play from there.
- Hi, everyone. This is Lucas Voss with Becker's - Healthcare. Thanks so much for tuning in to - the Becker's Healthcare podcast series. It's fantastic - to have you. Today, we're talking about - speed - versus - judgment, - rethinking AI's role - in - underwriting. And joining me to discuss just that. - It's so great to have you is Mark - Jeffries. He's the general manager health at Gradient - AI. Mark, it's so great to have you. - Thanks for being here. Yeah. Appreciate it, Lucas. - Appreciate the opportunity.
- Before we head into our conversation here, I - just wanna give, our listeners a sense of - what you do for work. If you could - just introduce yourself a little bit, talk a - little bit about your work in health care. - Yeah. Yeah. Happy to. So again, Mark Jeffries. - I'm the general manager for the health practice - here at Gradient AI. I've been with Gradient, - almost two years now. I'm accountable for strategy, - execution, and growth of the health business, - and I have a keen focus on developing - our portfolio of solutions to meet customer needs. - Prior to joining Gradient, I was at Cigna
- for the last ten years where I helped - build and then ultimately led their supplemental health - business, which was part of the US health - care organization. - I've also, in my career, spent time at - Aetna and the Medicare Advantage business, MetLife, and - the Hartford in group benefits and financial services. - So I have a - a broad and varied background. - Again, it's so great to have you. I - wanna talk about the title of our episode - today or part of it, speed versus judgment, - because this is something that's extremely prevalent in - health care right now, especially when it comes - to
- AI and where the humans are involved still - and the robots take over. Again, especially in - underwriting, as AI takes more of that group - health underwriting process over, which is certainly this - case, - from your perspective, which decisions - are still - important where you need an underwriter to have - that judgment in? - Yeah. That's... Frankly, it's an interesting question. Right? - Personally, I'd argue that all decisions still call - for an underwriter's judgment. Yeah. That probably sounds
- kind of odd coming from a guy whose - company develops AI powered tools for underwriting, - but the underwriter still plays a fundamental role - in determining the context - in which an AI generated result or output - should be relied on for a decision. - You know, the underwriter uses their judgment to - decide about the the level of risk they're - willing to assume when using that AI solution - to help manage their overall book of business. - You know, it's also important to recognize that - the use of AI tools - isn't an all or nothing proposition. - Right? And I think a lot of the
- conversation that's happening in the industry today is - kind of looking at it through that binary - lens. You either use it or you don't. - I don't think that's really the reality or - where the opportunity lies. - Using various features in our tools, underwriters can - enable multiple tiers of assessment and adjudication - that... To drive efficiencies in their business processes. - You know, for example, using an integrated rules - engine, an underwriter can enable - sophisticated workflow processes based on the presence of - certain conditions or drugs. If the conditions or
- drugs are present, the quote can be routed - to different levels of review within the process. - Right? So that type of approach helps the - underwriter allocate their valuable time and expertise to - quotes that need a deeper level of review. - I wanna tap just on your experience a - little bit here specifically. Again, you you spent - so much time in this space too. - Is there an example that you can think - of where this was... The process that you've - just described - happened where an AI generated assessment looked really - straightforward, right, but an underwriter took... Then took - a closer look and said, okay. Hey. I
- I I need to take a look here. - We might need a review. And and then - from your perspective again, what prompts those reviews? - What prompts the underwriter - to say, hey, I I need to take - another look? - Yeah. Yeah. That's a that's a great question - as well. - First first, I'd say it's important to keep - in mind that, you know, AI tools, - are providing a point in time assessment - based on, you know, an employee census, - that's provided at the time of that initial - assessment. Right? - So I've seen situations - where, - you know, significant changes have occurred within a
- census population between the time of the initial - assessment and the time of the final quote - being released. - So in one case, it was simply an - error. The employees from a single site in - a multi site organization were left off the - original census. - This stuff happens. Right? And in another case, - though, there was a material change in the - health of an employee that emerged - after the initial assessment. Right? So if you - think about, you know, how care is delivered - and how claims - appear within, you know, an ecosystem.
- They happen, you know, - can happen between, you know, in that thirty - to sixty day window from an initial assessment - to a final quote. But in both cases, - the underwriter - prompted the review - because they were seeking the most current information - prior to releasing the final quote. - Right? That that quest for the most recent - information - is a key role that the underwriter plays - in the process. - Right? So they are, - you know, sitting in the driver's seat. They're - the ones, who are making a determination
- to look for more information before they make - a decision to put a quote in market. - So you mentioned a a lot of it - is dependent on, - actually, the situation, obviously, where where they're at - at the point of time that will determine, - okay, I need to look at it again. - I I need to verify something. But when - we're going into that review, when we're going - into that that personal underwriter review, - what does it mean to have a meaningful - review of a model's recommendation, and what does - that look like in practice? And, again, what - are that in... What's that information that you're - looking for that an underwriter needs
- to make that independent decision? - Yeah. Yeah. That's a... That's also a really - good good question here. So - in the case of our products, you know, - we provide multiple levels of information to help - the underwriter assess the risk of a group. - So at the highest level, we're providing - a risk score that's made up of a - demographic and morbid... Or morbidity score. The score - is expressed as a number where one is - viewed as an expected risk of groups like - the one that's being assessed. - A score of 1.2 then says that particular
- group is 20% - more risky than expected. - That scores a... Is a quick way to - initially determine the depth of review need... Needed - for a group. The next level of information - within our tools highlights specific conditions, drugs, - and whether they're present, within that particular group. - You know, these can be, - ultimately customized by the underwriter to to use - flags so that the things they care about, - or concerned about can always be elevated and - highlighted, - so that it speeds up their initial review - process.
- And then from there, you know, depending on - the level of information, - that they need to inform their decision, they - can drill down into - further details to understand prevalence of conditions and - drugs and the predicted costs, - expectations for that group on a forward looking - basis. - All that information is is great. Right? And - is super powerful, - to enable the underwriter. - However, what's not included - are the subjective inputs - that are also important components in making a - decision. For example,
- tenure of client, - multiple product ownership, broker relationship, - size of the group, etcetera. Right? - Also are factors that that may - influence a decision. - So it's the underwriter's role to balance the - art, the subjective, and the science, the objective - to effectively manage their book of business. - Now this, might come as a surprise to - some, but we can disagree with AI. - It is it is it is allowed. Again, - when that disagreement happens, right, you you just
- talk through the process. When an underwriters disagrees - with the model's recommendations, says, hey. There's some - factors I wanna consider here. I wanna change - that. How should that decision then be resolved - and documented? - Yeah. So the use of the word recommendation - here is interesting. You know, it almost feels - like the underwriter and the model are, you - know, on this... On a same level playing - field. - They're they're really not. Right? - You know, the model is providing information - which has been developed and curated to provide - inputs to the underwriter. - The underwriter makes the ultimate recommendation.
- Right? It does vary by organization, but but - most organizations - have a workbench solution that their, their underwriters - are working in. And and typically, right, if - there is, - you know, disagreement - or additional information that they're factoring in that's - influencing their decision, they're documenting, - those types of things within that workbench so - that there's a kind of a longitudinal history - of the relationship with that particular prospect or - customer. - And I would expect, you know, they would - make note of, of a model, you know,
- a a quote unquote, a model disagreement - like this. It's actually a a really important - part of the process because tracking - that type of disagreement - can help highlight if there's, if there's drift - in the model output or the underwriting parameters - that they need to address. Right? And so - it's an important feedback loop. - And when we engage with our customers and - and deliver our solutions, we're working with them, - right, to to gather that feedback over time - to ensure that, you know, we're calibrating - our output from our solutions that... To specifically
- meet their needs. - Mark, it's so great to have you. I - I wanna hit on... Before we close our - conversation, I wanna hit on on one more - important thing here because you have a very - unique perspective on this. There is a lot - of vendor fatigue around AI right now, specifically - in health care. There are so many solutions. - People are bombarded - with - options when it comes to artificial - intelligence - for group health plans specifically. - When they are evaluating these AI underwriting tools - that can clearly help them as we've outlined,
- what should they be looking for - in a vendor, in a partner, and what - are some of the red flags - that you're looking for specifically that a platform - isn't really optimized for speed over judgment? - Sure. Sure. I think, - it it really starts with, do the vendors - understand the market? Right? And so if you - truly understand the underwriting of the health care - market, you know it's a balance between art - and science. - Right? It's not a clear cut binary - yes, no decision in most cases.
- A vendor that's solely focused on straight through - processing, - you know, to drive kind of OPEX savings - doesn't fully appreciate the subjective realities - of the underwriting within this world. So I'm - not saying there isn't a need or a - place for automation in health care underwriting. There - absolutely is. But the recognition that the underwriter - sits at the center of the ecosystem is - really important. - AI powered models and automation are ultimately - empowering them to focus their time and experience - on the quotes that need their expertise,
- which is the art in the process. - So my advice, right, partnering with a vendor - who understands this dynamic will ensure - that you find that kind of that right - balance to maximize the value of the relationship - between the underwriter and then the technology that - you you use to empower them. - Mark, I feel like we struck that balance - here today between speed and judgment in our - episode. - It's it's so fantastic to have you. Thanks - so much for taking some time for us. - Oh, absolutely. I really appreciate it. Thank you. - And we also want to thank our podcast
- sponsor, Gradient AI. You can tune in to - more podcasts from Becker's Health Care by visiting - our podcast page at beggershospitalreview.com.
In this episode, Jon Grandstaff, Senior Vice President of Payment Integrity at Zelis, discusses how payers can reduce payment friction, use AI to support more accurate claims reviews and improve payer-provider...
In this episode, Amy Conlon, MPH, Population Health Sr. Manager at Exact Sciences, and Jeff Huebner, MD, FAAFP, Executive Medical Officer, Division of Medicaid Services, Wisconsin Department of Health Services, discuss...
In this episode, Denis Mrkva, Senior Vice President of Analytics Solutions at Data Axle, discusses how health plans can combine consumer, employer, geographic and healthcare data to identify growth opportunities, reduce...
In this episode, Bob Tavernier, Solutions Executive at Quest Analytics®, discusses how health plans can leverage AI, provider data and strategic partnerships to improve network accuracy, strengthen regulatory compliance...
In this episode, Martina Lee Strickland, Head of Marketing, PR and Brand at Clever Care Health Plan, joins the podcast to discuss the importance of building trust when serving underserved communities. She also shares...