I had the pleasure of interviewing Crystal Son, an applied data science manager at Civis Analytics’. She is a trained epidemiologist and worked at the NYC Department of Health and Memorial Sloan-Kettering Cancer Center before turning her focus to data and analytics consulting. She graduated from Williams College with a B.A. in History and Columbia University with a Masters in Public Health.
Thank you so much for doing this with us! What is your “backstory”?
I initially wanted to go to medical school, but the more I learned about public health, the more I became interested in creating change in healthcare at a macro-level. I’ve also always had a weird fascination with learning about diseases, so epidemiology was a natural choice. What I didn’t anticipate was becoming so interested in working with data. You have to become really comfortable with all aspects of working with data as an epidemiologist — getting it, transforming it, analyzing it, and interpreting it. I did that for a while, working first in diabetes prevention and control for the city of New York, then as a hospital epidemiologist at a cancer hospital. What I learned was that there’s so much data in healthcare, but not enough information. The underlying infrastructure and processes make it so hard to get the data out. I went into data analytics consulting for a while because I wanted to learn how other industries might do it better. Now, I’m in a role where I can combine my knowledge of the healthcare space with my experience organizing teams, building data products, and delivering solutions to an industry that’s calling for it.
What do you think makes your company stand out? Can you share a story?
A lot of firms that service clients focus on strategy (they tell you what to do and how to do it) or they focus on implementation (they do what you want once you tell them what you want). Civis is uniquely able to do both. If you know you have a specific problem X that you want the answer to, we have the ability to answer that for you. But as we work with you and get to know your business and where you want to go in the future, we can provide thought leadership and strategic advice on how to get there.
Are you working on any new or exciting projects now?
We recently worked on a few projects where we’ve adapted some of our matching algorithms to resolving entities like businesses, vendors, distributors, and even products across various datasets. Normally, our identity resolution methods are used for matching people, because we often don’t have a unique identifier we can use, so we have to come up with our own way of being able to trace a person across different areas of the data. In healthcare, especially for hospitals and other providers, there’s usually a medical record number that can serve as a unique identifier, so person matching is less of a need. But there are plenty of other things that need matching! There’s so much siloed data and so many legacy systems with different naming mechanisms in healthcare, compounded by a lot of parties who actually benefit from data obscurity, especially around costs. In order to achieve more transparency, identity resolution is one of the first unavoidable hurdles. Back in 2014, Civis built a model for the Chicago Department of Public Health that was used to do better outreach to women at risk of breast cancer. They used our uninsured model to encourage women to sign up for free mammograms, which was especially important for African-American women because their mortality rate from breast cancer was so much larger than Non-Hispanic White Women (in fact, this disparity in Chicago was the largest in the country). In 2016, some exciting reports were published with data showing that Chicago now has the lowest disparity because of efforts by the city like this mammogram campaign.
Can you share the top ways that technology is changing the experience of going to the doctor?
– We’re already seeing signs that healthcare consumerism is on the rise, and while shopping for a doctor really isn’t quite the same as shopping for a pair of pants, patients will continue to do more comparison-shopping and demand greater cost transparency from providers. Hopefully, the frustration over the tools that do this today (and the data that backs them up) will reach an unbearable point and better customer-facing platforms for estimating costs and comparing providers will surface. Organizations that encourage data sharing across different payers like the Healthcare Cost Institute (http://www.healthcostinstitute.org/) are helping bring visibility to what some things cost and how wide the ranges can be.
– I think we’ll see even more companies scrambling, trying to leverage all of the data that we leave behind as a footprint — like purchasing behavior and data from wearables. This, motivated by the regulatory and cost pressures to move to value-based care, will allow providers to have conversations with patients about lifestyle and behavior instead of only focusing on symptoms. Applying analytics and data science to this expanded dataset might also allow providers to proactively identify patients at risk before negative outcomes occur. Fitbit and Google Health just announced a collaboration: (https://www.businesswire.com/news/home/20180430005464/en/Fitbit-Google-Announce-Collaboration-Accelerate-Innovation-Digital). And Apple is collaborating with Stanford Health to monitor participants in their heart study (https://www.apple.com/watch/apple-heart-study/).
Originally published at medium.com