Shelley Rusincovitch is an informatics architect with CTSI. She spends her time working with researchers who dive deeper in massive amounts of data, and connecting the pieces to enable high-quality research. What motivates her to work with something so big? That fact itself. “I love being a part of something bigger than just me — whether that’s a musical performance or a research project, I know that I have more impact as part of a great team.”
So what is your role here at Duke?
As an informatics architect, I work with data; but just as importantly, I work with people who work with data. My role is all about connecting both sides of the process.
So connecting people to data, is that where the “Applied” part of Applied Informatics comes in?
Yes. Informatics as a discipline has many different specializations. If you think about a field like engineering, you have engineering as a field, but you also have applications like electrical engineering and industrial engineering. Informatics works the same way. You have clinical informatics, clinical research informatics, even imaging informatics. But each of those specializations within informatics is still about connecting information systems with a defining purpose to make things happen.
What motivates you in your work?
I’m very lucky to work with many incredibly smart people, especially the investigators, researchers and clinicians, who have a purpose. They want to accomplish something with data, and that’s not easy. It requires thought and planning, technical logistics and details to make that happen. And that’s what I do.
What kinds of data are you working with now?
A good example of that would be the PCORnet Common Data Model, one of my more recent projects. I’ve been working under the direction of Lesley Curtis (faculty in the Duke Clinical Research Institute) for the past several years — she’s amazing. She leads a large team for data and operations under PCORnet. The PCORnet Common Data Model is a way of organizing electronic health data so that it’s predictable, understandable, and — most importantly — interoperable.
This is so exciting because we’re able to leverage data that already exists. Think about it. Every time anyone visits the doctor’s office, they’re typing data into the computer. Can you imagine how rich a data resource that is?
What is PCORnet doing with this model?
Consider our electronic health record here at Duke Health. It’s called MAESTRO Care, which is based on a software product called Epic. Lots of healthcare providers use Epic, but that’s just one of several systems on the market. The databases are all designed differently, but they’re recording common pieces of data. No matter which doctor you visit, they’re going to record your date of birth, your weight, what time you arrived, and what medicines you’re taking.
So to use all this data together, we need a common way of organizing the data, so that a computer can always predict and organize electronic health data, no matter which system it comes from.
For example, we have a specific way of referencing the values, no matter how the source system references the value. One system may use M/F for sex, while another uses Male/Female, or 1/2. But a computer pulling information from several systems won’t always know that these are all the same. So that makes asking questions of the data much more complicated. You have to build in all the possible variations of one question, and that’s not efficient. Using a common model for the data allows us to ask questions that get accurate answers from all the data, which is much more scalable.
How long have you been at Duke?
I came to Duke as a research assistant thirteen years ago, in 2003. I’ve been really lucky in that I’ve had some key mentors as I’ve progressed at Duke, and they’ve fostered my interest in databases and programming. To see databases in action and be a part of research teams really got me interested in informatics.
Was informatics always a career goal for you?
I actually got involved in clinical research before informatics, rather than the other way around. Early in my career at Duke, I took classes in computer programming at Durham Tech and earned an associate’s degree there; I love their program because they have a strong focus on practicality. I’m continuing my education now as a student in the Duke Master of Management in Clinical Informatics program, and I’ll graduate in August.
I didn’t start out in programming, though. I was a liberal arts major as an undergraduate — I’m a musician, and even now I play the oboe with the Duke Symphony Orchestra under the baton of Maestro Davidson. It’s fun and refreshing to use the creative side of my personality alongside the analytical. People say that musicians make good programmers; I have no idea if that’s true for everyone, but it resonates with me. Many people I know who are quantitatively inclined are great musicians — maybe because we like to be a part of something bigger than us.
To be a part of the orchestra as a whole, to be a part of that performance, something that is bigger than any one person, really appeals to me. Being a part of a performance is a powerful thing. You get to look back on something that may be enjoyed by so many people and think, “I was part of making that happen.”
And what I’m doing at Duke is also part of something bigger than just me; I’m proud to be a part of our impact on the world.
Moving forward, what kind of projects is your team getting involved in?
Predictive analytics is something we’re seeing more and more of: working to get a better idea of which patients are at high risk. The Southeastern Diabetes Initiative (SEDI) is a great example of that. SEDI was using a risk model much earlier than most people realize, under the direction of Robert Califf (faculty in Cardiology).
SEDI had a lot of different pieces to it. At the heart it was the idea that if we’re trying to help large groups of people, how do we know how to most efficiently deploy interventions for people who are at risk? So the SEDI concept was to group patients into pre-defined risk brackets. For example, the high risk algorithm was to identify patients who had diabetes and were at high risk for a catastrophic event in the next year, such as a hospitalization or amputation. The better we could predict their risk, the better we could work to prevent an event like that.
Our clinical leadership was able to deploy teams of home-based healthcare providers to the most at-risk patients. Those teams holistically addressed the patients’ health challenges, and it wasn’t just about prescribing medications, but also about finding out if they can afford healthy meals and helping them find transportation to doctor appointments.
I’m on the data side, so I’m not personally interacting with patients, but I still get to help teams like that go out into the world and improve lives of patients.
What’s the best advice you’ve ever received?
So I never met my great grandmother, but I’m told she had a saying,
“I done the best that I could at the time I done it.”
That’s always resonated with me. You work hard at what you’re doing, and you recognize that you’ll grow, you'll be able to do even more and learn from what you’re doing and apply it. Don’t be scared of jumping in because you’re not sure you know everything.
We’re on the frontier of developing new innovations and methods, and we have this huge responsibility to gain insight from these important health data entrusted to us. We recognize how much there is to learn, and sometimes that can be overwhelming. But it’s important to get started now with what we have, today. It can be scary, but it’s so exciting! We have to be ready to learn as we move forward.