As an epidemiologist by training, Michelle Winkler has extensive experience with research design and biostatistics, having focused her doctoral and clinical research on ovarian cancer. She spent part of her career working in public health, which is when she began to see the limitations and opportunities in large population-based data sets.
“While health departments have access to incredibly large datasets, the majority of this data has a very limited scope of information,” she said, “and in working closely with local health care systems and insurance companies, it became clear that population health initiatives would greatly benefit from the ability to scale clinical information to a population level.”
Winkler further developed her skills as a data scientist at an AI supply chain management startup spun out of a large health system. Her role there focused on developing solutions to manage pharmacy spend and high-cost resources like MRI and CT scanners within healthcare systems. While there, she created predictive models focused on pharmacy cost containment — specifically demand planning and inventory management — as well as several other initiatives to aggregate siloed data and provide decision support.
Winkler, who has a PhD in epidemiology from the University of Pittsburgh, is fluent in a number of programming languages, analytic platforms and data visualization tools. She joined Agilum in August 2020 as a senior data scientist but was named director earlier this year.
Winkler lives near Pittsburgh with her husband and 1-year-old son. In her free time, she enjoys camping and doing “Top Chef”-style cooking challenges at home.
How did you learn about the opportunity at Agilum and what drew you to it?
A recruiter reached out to me and wanted to discuss my background in clinical and population health research. When I started talking to the recruiter and Dr. Kirsh (Agilum’s Chief Medical Information Officer), I was impressed with the breadth and depth of Agilum’s datasets, and that’s what really drew me to the position. It’s rare to have access to longitudinal data that includes detailed clinical information for such a large population of patients. As an epidemiologist, I was intrigued by the possibility of designing studies using real-world data that would have the power to study relatively rare diseases and treatments.
Epidemiologic research is traditionally designed with homogenous study populations to minimize biases that could influence specific research questions. However, this isn’t reflective of real populations that include patients such as those that are elderly, pediatric, or have been diagnosed with multiple diseases. It was very exciting when I realized that Agilum’s real-world data provides the ability to assess treatments and outcomes in actual healthcare patient populations. I also recognized that such a large, multifaceted dataset was particularly well suited for cognitive analytics. The opportunity to detect patterns using machine learning methodologies, particularly in situations such as COVID-19, where so much about the disease etiology and treatment was unknown, was also very interesting to me. Ultimately, what convinced me that I needed to be a part of the Agilum team was the impact I could have by applying my epidemiologic and data science skillsets to such a unique and robust dataset.
I have had access to large health outcomes datasets with very few datapoints and I’d also worked with smaller-scale clinical and supply chain data sets. Each dataset was valuable, but only provided insight into specific aspects of healthcare. With the resources at Agilum I’m able to connect these datapoints and build a more comprehensive picture of how supply chain decisions impact operational and clinical outcomes. Instead of efforts to reduce pharmacy costs that only utilize supply chain data, Agilum can provide intelligent decision support that builds on purchase data while accounting for factors such as payer remittance, readmission rates, and length of stay to help customers overcome the financial pressures that they face.
I was going to ask, given your background in epidemiology, whether you had plans to do some analysis of the vaccines or maybe some kind of follow up on therapeutic treatments or whatever. Any plans for that?
There’s a lot of talk about the possibilities. I think right now our biggest task is prioritizing which projects we want to go after first. The entire team is excited to build on the COVID-19 research that was published last summer. We now have an even larger population of COVID-19 patients as well as a longer follow-up period for survivors, which will allow us to investigate not just treatment effectiveness but also any long-term effects. I think there’s also the opportunity to study some of the effects of the pandemic on healthcare systems, such as the impact of changes to outpatient care delivery.
What are you primarily focusing on in your role as director of data science and quality?
Our biggest priority is adding enhancements to CRCA that will benefit our customers by providing data visibility. Healthcare systems generate a tremendous amount of data that is typically very siloed, making it difficult for them to have clarity into the true cost of care. My team is focused on connecting our customers’ data feeds and translating those into actionable insights.
One example of this is our drug remittance dashboard. Many healthcare organizations don’t have the ability to reconcile the purchase amount for a drug and how much they received as payment. We are building a feature that will link purchase and claims data to remittance data. This will enable them to determine whether they are losing money every time they dispense a specific drug or whether they should consider switching to an equivalent or biosimilar drug.