Meet the Expert: Rebecca Jacobson, vice president of analytics

UPMC Enterprises’ expert in natural language processing on why the technology is essential to improving health care

Rebecca Jacobson joined UPMC Enterprises in June 2017 as vice president of analytics to lead the organization’s efforts around natural language processing (NLP), a technology that’s key to many projects that are harnessing large amounts of health data.

Jacobson, who spent 16 years as a professor of biomedical informatics at the University of Pittsburgh, was well suited to transition from academics to an organization focused on innovation and commercialization. She is a co-founder of Nexi Inc., a Pittsburgh company that uses NLP and other analytics to produce insights across very large networks of clinical data. She remains a scientific advisor to the company, which licenses clinical NLP technology produced in Jacobson’s lab at Pitt.

For more than 10 years, she also was project director of the TIES Cancer Research Network, an NIH-funded program at Pitt that supports data sharing of NLP-processed notes, structured data, and images across a network of six academic health centers: Pitt/UPMC, University of Pennsylvania, Roswell Park Cancer Institute, Georgia Regents University, Thomas Jefferson University Medical Center, and Stonybrook Health.

With nearly a year under her belt at UPMC Enterprises, Jacobson talked about what she and her team have been doing and where she sees her work with NLP headed.

What made you want to join UPMC Enterprises?

After many years in academics, I was getting restless and ready for change. I saw that health care leaders such as UPMC were building their own NLP, predictive analytics, and machine learning teams. With so much exciting innovation emerging in the industry and health care sectors, I realized that we were reaching a tipping point. The next 10 years will be all about what organizations like ours can do with these technologies to transform health care. And our strength is really in combining the technologies with our two most important assets – our data and our health care acumen. I wanted to be a part of that transformation.

What have you and your team been doing in your first 10 months here?

During our first six months, we had a great first assignment with medCPU, a UPMC portfolio company that is developing real-time clinical decision support applications. Bringing our expertise in NLP to the company, we were able to help them get their latest product into production at UPMC within a few months. We are excited to see the results emerging within the next few months from this first trial at UPMC. A second project we’ve started with the UPMC Wolff Center will produce a new NLP-based system for rapidly abstracting clinical quality metrics. And we have many more projects queued up for the rest of the year.

Why is NLP a focus at UPMC Enterprises and the digital health industry?

So much of the data that is needed for improving health care quality, reducing health care costs, and managing the health of populations has been historically locked away in the free text of our clinical notes. Efforts to make providers codify all of the data they collect has limitations, because it takes so much more time to enter data into discrete fields when compared with dictating a note. Natural language is also the most expressive way for clinicians to communicate. It just seems inevitable that we will always want to return to it.

What is a problem in health care that could be solved by NLP but hasn’t been yet?

I think one of the most interesting “unsolved problems” for NLP is the identification of Social Determinants of Health within the patient record. It’s been argued that 60 percent to 70 percent of health outcomes are driven by factors such as your social support network, access to stable housing, food, and transportation, and the degree to which you are exposed to violence or substance abuse. But this information is hardly ever captured by electronic health records within explicitly structured fields. Identifying and extracting this information from the clinical notes could help to improve predictive analytics and eventually help us to intervene more effectively.

What was your first job and what was the most important lesson you took from it?

For three years, between college and medical school, I was a case worker and then case manager in a community mental health center in North Philadelphia. I worked with a dual-diagnosed population of clients – with both psychiatric and substance abuse problems. I learned that what really matters to me is having impact: using my skills and expertise to make a positive difference. That’s what I look for in every job, and in everything I do.