Transcript

Welcome back to Dishing Data. Today we're talking about one of the most complex applications of data. science in the government, healthcare data. Whether you're working with medical claims, investigating fraud, waste, and abuse, or trying to understand utilization and outcomes, healthcare generates enormous volumes of data and tremendous opportunities for analysis. Joining me today is Kyle Crane, who works at the intersection of healthcare analytics and data science. So let's jump into it. Healthcare is one of the largest and most data-intensive sectors in government. What makes healthcare data especially challenging and valuable from a data science perspective? Uh, I would say that healthcare data's unique, not just because it's high volume, but it's very complex- Okay ... in the way in that it's structured. Okay. So, you' know, you think about when you or I go to the doctor, it's not just, you know, who is Liv Hebner? Who is Kyle Crane? It's what is the service done? When was it completed? What are the diagnoses? How do those all fit together into a singular visit, right? Mm-hmm. And if you're trying to put all of those things into one position, that data becomes really, really messy very quickly, particularly when you're thinking about it from a data science perspective. Okay. So when we're trying to really work backwards, it's about organizing and operationalizing the data. Okay. So what are some practical ways data science is applied to healthcare data? So some of the things that we do include program integrity, some of the fraud, waste, and abuse, that you guys have talked about on other episodes. Okay. Um, cost utilization, you know, where are patients or providers, you know, doing the same types of services or where are costs increasing or decreasing over time. Uh, policy evaluations, things like mental health awareness or preventive, preventative care. Okay. Um, and then, you know, tools or, or dashboarding, so things like repeat analyses or types of procedures that are done on a recurring basis. How can we put that in a place, take all this messy data, and have it done automatically? Um, so you just mentioned dashboards and tools. What distinguishes like a useful tool versus this like one-off analysis? So that's a good question. There are some things that are done, you know, you, you call it a dashboard that, that can look really pretty on a screen. Sure. But we're... What we really wanna be able to do is. take those things that are repeatable- Mm-hmm ... and automate it. Okay. So think about, I'll, I'll use the mental health case again, right? So every time that you, you or any, you know, patient goes to see some mental health care provider, how can we track to see if you are being, you know, reimbursed correctly? Are, are you paying too much out-of-pocket for the same type of service that, you know, you would go to see a psychiatrist or to go to see a surgeon? Right. Okay. Um, how do we check that so it's done consistently every single time? Okay. Um, if an agency or client wants to strengthen its healthcare analytics capabilities, what can your team add for, add for them, add value for them? Sure. Uh, so obviously we just talked about some of the dashboarding, so I, I won't repeat that, but we're talking- That's a big one. It. is. It... And it's one that is super, super popular- Right ... of course. Uh, but we're talking about reproducible pipelines, ETL, things like that. Um, transparent methodologies, trying to set something up from start to finish- Mm-hmm ... from a, uh, more of a background or subject matter expertise- Okay ... kind of perspective. Um, more the reproducible codes and documentation that goes with that, and ultimately a lot of thought leadership is something that a lot of the clients we work with are really looking for in this kind of space. Okay, great. Um, well, healthcare data- presents some of government's toughest analytical challenges, but also some of its greatest opportunities. Um, if you're looking to modernize healthcare analytics, improve program integrity, or build scalable tools, we'll be glad to talk. Thanks for joining us, and we'll see you next time. All right.