Latest / Fintech Conversations with Fexingo: Banking, Payments, and Financial Software Companies / How Fintechs Use Real-Time Healthcare Claims Data for Dynamic Insurance
Transcript
- Lucas: So Luna, I was reading through the latest filings from a few health insurers, and one thing keeps popping up: real-time claims data being used to price policies on a monthly basis. Luna: Monthly? Not annual? That sounds like a complete shift from how traditional underwriting works. Lucas: Exactly. The classic model is you set a premium once a year based on your risk profile, and then you hope the claims don't blow past expectations. But a handful of fintechs are now plugging directly into claims adjudication systems and using that data to recalculate what you pay every single month. Luna: So if I have a month where I barely step into a clinic, my premium might drop? And if I get into a bike accident, it spikes the next month? Lucas: That's the pitch. One startup I looked at, they're not a household name, but they've been piloting this with employer groups since early 2025. Their model takes your claims history, your pharmacy usage, even your wearable device data if you opt in, and runs it through an algorithm that adjusts your next month's premium. Luna: That's a bold claim. How do they handle the regulatory side? Health insurance is heavily state-regulated here in the U.S. Lucas: That's the tricky part. They're technically operating as a third-party administrator, not a licensed insurer. They work with licensed carriers who take on the risk and the compliance burden. The startup just provides the analytics and the tech platform. Luna: So they're more like a software layer on top of the insurance company. That makes sense. But the real question is, does the data actually predict anything, or is it just a gimmick? Lucas: Well, let me walk you through their pilot. They had about twelve thousand employees across a few mid-sized companies. Over the first six months, they saw that members who consistently logged lower claims activity, say less than eighty dollars a month in average claims, had about a thirty percent lower next-month premium than their initial rate. But those with chronic conditions, like diabetes or asthma, saw their premiums fluctuate much more. Luna: That's a double-edged sword. It rewards healthy behavior, but it could punish people who didn't choose to have a chronic condition. Lucas: Exactly. And that's the ethical knot. Proponents argue it's fairer because you pay for what you actually use, not what a risk pool averages out to. Critics say it undermines the whole point of insurance, which is to spread risk across the healthy and the sick. Luna: And there's also the privacy angle. If I'm sharing my claims data every month, what's stopping an employer from seeing that I filled a prescription for an antidepressant? Lucas: That's a real concern. The startup says they anonymize and aggregate data before sharing anything with employers, and they have to comply with HIPAA. But the more data points you collect, the more potential for re-identification. Luna: So who's actually using this? Are consumers signing up voluntarily, or is it being pushed by employers? Lucas: Mostly employers. They see it as a way to control healthcare costs. They offer it as an alternative to the traditional PPO or HMO plans. Employees get a lower baseline premium, say ten percent lower than the standard plan, but with the understanding that their premium could go up or down each month based on their activity. Luna: So it's a bit like a health savings account, but with more dynamic pricing. Lucas: Right, but the key difference is the real-time adjustment. You're not just saving pre-tax dollars; you're actively being priced every month. Luna: I can see the appeal for a young, healthy workforce, but what about older employees or those with pre-existing conditions? They might end up paying double the standard rate. Lucas: That's exactly the question regulators are starting to ask. The National Association of Insurance Commissioners formed a working group earlier this year to look into algorithmic pricing in health insurance. Luna: And what's their stance so far? Lucas: They're mostly in the listening phase. They've asked for comments from insurers, consumer advocates, and tech firms. No final guidance yet, but there's a lot of concern about whether the algorithms are inadvertently discriminating against protected groups. Luna: That sounds like a legitimate regulatory headache. But if we set aside the fairness debate, is the underlying data actually reliable? Lucas: That's the other big question. Claims data is notoriously noisy. A single visit to an ER can cost thousands, but that might be a one-off accident, not a sign of future risk. The algorithms have to be sophisticated enough to distinguish between a temporary spike and a chronic pattern. Luna: And that's where the machine learning comes in, right? Lucas: Exactly. They're using time-series models that look at your claims history over the trailing six to twelve months, along with demographic factors, to forecast your next month's expected claims. The more data they have, the better the prediction, which is why they're so eager to pull in wearable data. Luna: I get that. But I'm still not convinced that a healthy person would want to be tracked every day just to save a few bucks on their premium. Lucas: That's the trade-off. And actually, the uptake numbers from the pilot suggest about forty percent of eligible employees opted in. So it's not universal, but it's not negligible either. Luna: Hmm, forty percent? That's decent for a first pilot. But I wonder how that would change if they had to give up more privacy controls. Lucas: Well, the startup did a second pilot with a different employer group where they added a feature that lets you see your predicted premium for the next month and adjust your behavior, like scheduling preventive checkups, before the new rate kicks in. That boosted opt-ins to over sixty percent. Luna: So giving people a sense of control helps. That's a key insight. But there's still this looming question: can a model like this scale beyond a few thousand people? Lucas: That's the million-dollar question. The startup is currently raising a Series B round, and they're targeting ten million covered lives by 2028. That's ambitious, especially given the regulatory hurdles. Luna: Ten million? That's a lot of policyholders to manage with real-time data pipelines. I'd love to see their infrastructure. Lucas: They're building on cloud infrastructure with a data lake that ingests claims files from insurers in near real time via APIs. They process millions of events a day, and they use a rules engine to flag any anomalies that might indicate fraud or error. Luna: So it's not just about pricing; it's also about fraud detection and operational efficiency. Lucas: Right, that's the broader value proposition. The dynamic pricing is just the most visible feature. Underneath, they're creating a real-time health data platform that could power everything from care management to preventive interventions. Luna: I'm curious—how does this compare to what we've seen in other insurance verticals? We did an episode on usage-based auto insurance a while back, didn't we? Lucas: We did, and it's the same logic. Progressive's snapshot program and others have been doing this for years with telematics. But health is way more sensitive and complex. Luna: And that's why the stakes are higher. A bad driving record is one thing, but a bad health record is tied to your identity and your ability to get coverage at all. Lucas: Exactly. And that's why the industry is watching closely. If this model gains traction, it could reshape not just health insurance, but the entire concept of risk pooling. Luna: Before we go deeper, I want to mention something real quick. We're listener-funded, no ads, and a few folks keep us going through buy me a coffee dot com slash fexingo. That's what lets us spend hours digging into these stories. Just saying. Lucas: Yeah, it genuinely helps. So, back to the data. Let me give you a concrete example of how the algorithm works in practice. Lucas: Take a member we'll call Sarah. In January, she had a minor outpatient surgery, so her claims were around two thousand dollars. Her premium for February jumped by fifteen percent. In March and April, she had no claims, so her premium dropped back to baseline, and by May it was actually five percent below her original rate. Luna: So it's a direct pass-through of her claims experience. That's very transparent, but also very spiky. Lucas: Right. And that volatility is a big concern for consumer advocates. A family budget can't easily absorb a fifteen percent shock month to month. Luna: That's true. But I suppose you could smooth it out with a rolling average or a cap on how much the premium can change each month. Lucas: Some startups are doing exactly that. They set a maximum monthly change of, say, ten percent, and they use a trailing three-month average of claims to dampen the noise. That makes it more predictable for consumers, but it also dilutes the real-time aspect. Luna: So there's a trade-off between responsiveness and stability. What do you think is the winning formula? Lucas: Honestly, I think the market will settle on a hybrid. You'll have a base premium that's adjusted quarterly, with a small monthly true-up based on significant events, like a major claim or a big change in health status. That gives you the benefits of real-time data without the whiplash. Luna: That sounds more reasonable. And it might also be easier to explain to regulators. Lucas: Exactly. And that's probably why the NAIC is taking its time. They want to see how these models behave in the wild before writing rules that could either stifle innovation or leave consumers exposed. Luna: What about the consumer experience? If I'm on one of these plans, do I get an app that shows me my premium forecast and tips on how to lower it? Lucas: Yes, most of these fintechs are very consumer-facing. They have an app that shows your daily health score, your predicted next month's premium, and what you can do to improve it, like getting a flu shot or refilling a prescription on time. Luna: So it's like a gamified health insurance experience. That could be a powerful engagement tool, especially for the younger demographic. Lucas: That's the bet. And the data suggests it works: the group that used the app actively had twenty percent fewer ER visits than the control group over the first year. That's a strong signal. Luna: Twenty percent fewer ER visits? That's a big deal, both for the insurer and for the member's health. But I wonder if that's just because the people who opt into the app are already more health-conscious. Lucas: That's a common criticism, and it's hard to fully control for. But even after adjusting for baseline health status and demographics, the effect persisted. So there's some evidence that the feedback loop itself drives behavior change. Luna: That's promising. But we're still early. How do you see this playing out over the next two or three years? Lucas: I think we'll see more consolidation. A few nimble startups will get acquired by the big health insurers, who have the data and the distribution but not the tech. By 2028, real-time pricing could be a standard option in many employer plans. Luna: And what about the consumer? Do you think people will actually embrace it, or will there be a backlash? Lucas: There will definitely be a segment that rejects it on principle. But if premiums come in lower and people feel more in control of their healthcare spending, I think adoption will grow. The key is transparency and fairness. Luna: Transparency and fairness. That's what it all comes down to, isn't it? Whether the algorithm is a black box or a clear mirror. Lucas: Exactly. And that's the challenge for the entire fintech industry. We have the data and the compute power to build these systems, but we need to make sure they're used responsibly. Luna: Well, it's a fascinating space. I'll be watching what the NAIC does and which startups can actually scale. Lucas: Me too. If this works, it could really change how we think about health insurance—not as a yearly gamble, but as a dynamic partnership between the insurer and the insured.