Leverage Health: Thanks for talking with us today, Rebecca. Could you give our readers a snapshot of your background and how it led you to co-found Astrata?
Rebecca Jacobson: I’m happy to, Richard. First, though, let me say how honored I am to be working with Leverage Health.
An MD with clinical training in pathology, I began my career as an academic physician. My passion for computer science led me to get further training in medical informatics and an MS in information science. I then joined the faculty at the University of Pittsburgh. I taught at Pitt for 15 years, eventually as a full professor of Biomedical Informatics and CIO of the Personalized Medicine Institute. During this time, working across many academic health systems that used natural language processing (NLP) to glean information from clinical notes, my team and I created one of the earliest clinical NLP engines.
As a national leader in the field of clinical informatics, I saw firsthand the importance of taking our innovations out of the technology lab and into healthcare. To help accomplish this, I joined the University of Pittsburg Medical Center as Vice President of Analytics, where my team and I established both NLP and machine learning programs. We solved a range of fascinating healthcare problems, but the area that spoke to me most was quality. It’s so fundamental to value-based care. Today, as president of Astrata, I and my team are on a mission to help payers and value-driven providers achieve new standards for quality healthcare programs.
Leverage Health: Could you explain Astrata’s model as well as the market trends that have created tailwinds for what you offer?
Rebecca Jacobson: Over the last five years, the market for artificial intelligence (AI), machine learning (ML), and healthcare analytics has exploded. Despite the recent hype, it’s clear that analyzing data isn’t enough. Companies, like Astrata, that leverage insights gained through AI, ML, and analytics into end-to-end solutions will have an edge. The tailwinds are 1) rapidly changing technology that provides quality measurement, 2) huge advances across the field of NLP that have led to greater accuracy, and 3) the keen interest of regulators, such as the National Committee for Quality Assurance (NCQA), as well as federal agencies like the FDA, in using NLP to solve real-world healthcare problems.
Leverage Health: What’s behind the growing focus on Health Effectiveness Data Information Set (HEDIS) scores, and Star from the Centers for Medicare and Medical Services (CMS)?
Rebecca Jacobson: As the proportion of the U.S. population receiving healthcare through value-based care—instead of fee-for-service—increases, incentives will shift. Keeping plan members and patients healthy and at home becomes the goal, and that requires true population health. Knowing where you need to intervene is essential. That’s where HEDIS is invaluable. Incentive programs like CMS Star ratings, as well as state programs, all rely on these kinds of measures. The more we shift to value-based care, the more these measures become the key to improving population health.
Leverage Health: What do you see as the main challenges health plans face in terms of digital quality?
Rebecca Johnson: First, health plans are challenged to make the jump to digital quality, as they’re being pushed from multiple directions. For instance, they need to meet the interoperability requirements set by the Office of the National Coordinator for Health Information Technology (ONC) while also meeting those of the NCQA, which asks that they move from claims to clinical data and from retrospective to prospective review. In addition, they face impending and significant changes to existing HEDIS technology. As I see it, the real goal is for plans to see these different pressures as essentially one overall challenge, which asks that they take a strategic approach to achieving a thoughtful, well-planned transformation.
While not easy, the move to digital quality will result in incredible benefits. Plans that are at the forefront of this change will reap more benefits early on. They’ll be the first to higher quality rates, faster interventions, reduced abrasions, and, ultimately, better quality care and healthier populations.
Leverage Health: Many industry leaders talk about scale when discussing NLP or any machine learning intelligence. Can you speak to the importance of scale when deploying NLP?
Rebecca Jacobson: Scale is a really important factor for success, and often overlooked until it’s too late. For those of us in NLP, scale has two separate and equally important meanings. The first denotes the traditional definition of scale: how much data you can run through your pipeline per unit of time, and what happens when you increase that data input. To do prospective population health using NLP, you must be able to efficiently handle millions of text documents per day. At this scale, the level of experience and sophistication required are far greater, because the problems you encounter in processing at this scale are very different from those you face with lower volumes.
The second meaning of scale is specific to NLP and has to do with how you manage the heterogeneity of data across various note types, facilities, and geographic areas. NLP is notorious for its lack of portability. So, scaling can also refer to how well your solution adapts to that data heterogeneity within one environment as well as across multiple environments. Astrata has met these problems by creating technology and processes that handle very, very high volumes of documents while addressing our customers’ issues of data heterogeneity.
Leverage Health: Being one of the two accredited companies in NCQA’s NLP Working Group is quite an honor. Tell us about Astrata’s relationship with NCQA and what it means for potential clients.
Rebecca Jacobson: Yes, it’s been wonderful being a part of the NLP Working Group. The group finished its work at the end of 2020; I’m looking forward to the 2021 NLP pilot project. NCQA clearly believes that NLP is an important new technology for any plan that uses the HEDIS program. It’s now possible for our clients to do prospective HEDIS across their populations. This will lead to a new degree of automation that will improve their quality-measurement operations and reduce the unit costs of abstraction. This is big!
Leverage Health: What can our readers do to prepare for the shift to data-driven quality measurement?
Rebecca Jacobson: There are three important steps that health plan executives can take right now to help their plans make this transition. First, they need to work to establish a strategy for acquiring clinical EMR (electronic medical records) data. It’s a strategy discussion because it will impact all parts of the business from contracting to networks to IT to the clinical quality workforce. Plans will need a way to ingest, store, and compute both structured and unstructured clinical data.
Second, they’ll need to look carefully at their existing agreements with providers. They’ll also need to create the technical infrastructure to store and compute clinical quality rates directly from this data using technology that most health plans are still just learning about—NLP, Fast Healthcare Interoperability Resources (FHIR), and Clinical Quality Language (CQL).
Next, if their quality teams are not already experimenting with or entirely shifting to a year-round HEDIS strategy, then it’s time for them to initiate this change. A technology-enabled platform can help plans successfully transition without expanding manual abstraction costs. Finally, your readers should start evaluating their existing quality-measurement vendors. This can be a demanding task, and the conversations need to start now. They’ll need to ask what innovations are in the pipelines and how they’ll meet these new requirements.
Leverage Health: What do you see as the five-year outlook for measuring healthcare quality?
Rebecca Jacobson: Not surprisingly, I predict that quality measurement will look very different in five years. Last week at their quality conference, Meaningful Measures 2.0, CMS announced that they were moving their ambitious deadline for the transition to 100% digital measurement of quality from 2030 to 2025. You can learn more about Meaningful Measures 2.0 here: https://www.cms.gov/meaningful-measures-20-moving-measure-reduction-modernization. I think we’ll see a lot more on this topic in the coming months. But, in short, five years from now, I expect health plans will routinely be using clinical EMR data, working on quality measures year-round, intervening much sooner, and having a bigger impact on population health. Technology advances will drive these changes.
Leverage Health: What key areas and pain points in the healthcare experience does Astrata look to address?
Rebecca Jacobson: We aim to make the quality-measurement intervention cycle faster, cheaper, and better. We want healthcare organizations to manage their populations more effectively and to achieve three goals: better care, better experience, and lower costs. For most healthcare organizations, quality measurement is absolutely the Achilles’ heel of value-based care. For years, healthcare has been some combination of claims-based, sample-based, manual, delayed, and expensive—the challenge can be overwhelming. That’s what we’re looking to change for our customers.
Leverage Health: What does the term learning health system mean, and how is it relevant to Astrata?
Rebecca Jacobson: The Institute of Medicine coined the term learning health system, or learning healthcare system, in 2007 to describe the process of translating research and evidence into patient care. The system measures the impact on key variables, deriving new evidence and understandings to help us continually improve care. Essentially, the learning healthcare system is the engine driving value-based care. In regions with growing populations, quality measurement and population health are places where we clearly see the cycle go round and round. Speeding up that cycle, making it better and cheaper across populations instead of across samples, and doing so with clinical data instead of claims data—that’s where Astrata comes in.
Leverage Health: Are there clinical applications other than quality where Astrata’s NLP adds value?
Rebecca Jacobson: For sure, but I still think there’s a huge opportunity within quality—because quality is everywhere in healthcare. While we have focused first on HEDIS, we know there are hundreds of measures that are important to both health plans and providers. For example, Medicare’s Merit-based Incentive Payment System (MIPS), which includes specialty measures, is an area where we’ve worked. Beyond that, I’d say that patient safety and reducing unnecessary care share many features with quality, and our technology could be focused here, too.
Leverage Health: Please elaborate on how Astrata helps a patient on his or her care journey.
Rebecca Jacobson: I’m so glad you asked this, because that journey is our priority. At its core, our technology helps healthcare organizations answer a critical question: Is the right thing being done for a specific patient? For example, Mr. Jackson is diabetic, so providers and plans need to know if he’s receiving the eye care he needs, if his blood pressure is well controlled, if the appropriate steps are being taken to treat or prevent nephropathy. With healthcare as fragmented as it is today, it’s hard to know. But it’s crucial that Mr. Jackson receive consistent quality care, so he can manage his chronic condition and lead a full and healthful life. At the end of the day, Astrata’s technology helps health plans and providers be sure that he is.