Translational Science Highlight
- Through its Small Business Innovation Research program, NCATS supported the development of a technology to use electronic health records data to advance value-based and enhanced patient care.
Electronic health records (EHRs) hold a wealth of information that can be used to look at how well health care providers are treating their patients and point to ways to improve care. The data can also help researchers conduct population health studies. But much of this useful clinical information is entered in the notes section of EHRs rather than into a specific field.
Referred to as “unstructured data,” this information is hard to extract and use in health analyses. This bothered Dan Riskin, M.D., M.B.A., FACS, adjunct professor of surgery and biomedical informatics research at Stanford University, clinical informaticist and CEO of Vanguard Medical Technologies (VMT). He created a spin-off company from VMT, Health Fidelity, to develop technology to use this unstructured clinical data to analyze patient risk and quality from the EHR. In 2012, NCATS awarded Riskin and Health Fidelity a Small Business Innovation Research (SBIR) Phase 1 grant to develop and evaluate the technology and to help bring it to market.
NCATS’ SBIR and Small Business Technology Transfer (STTR) programs speed innovation in translational science. Often, small startup companies lack the early-stage funding to carry an idea through for further investment and into commercial markets to benefit patients.
“NCATS helps bridge this translational gap by awarding grants, contracts and technical assistance to small businesses and research organizations focused on advancing translational research and technologies that will improve disease prevention, detection and treatment,” said Lili Portilla, M.P.A., director of NCATS’ SBIR and STTR programs.
“The NCATS grant came at a defining time for a defining technology,” Riskin said. “Based on work from that grant, we were able to push the boundaries of unstructured data use, exploring automated assessment of cohorts, quality and risk.”
Technology Can Make the Process More Cost-Effective
The value-based care model provides rewards or penalties based on treatment outcomes, rather than just the volume of treatments used in the fee-for-service model. The goal is to improve the quality of care and the patient experience. But this requires analyzing patient risk and outcomes by using the right data in the right format. Manually getting the data — by having someone review each patient’s chart — is too time-consuming and cost-prohibitive.
That’s where natural language processing comes in. It is a way for computers to extract meaning from human language and is used to do everything from automatically translating between languages to correcting grammar and summarizing blocks of text.
Built in collaboration with Columbia University, Riskin’s technology was unique in that it was developed specifically for use with clinical language, which is highly specialized. Doctors use a lot of abbreviations and acronyms in their notes, which can make the meaning ambiguous, and this can make them especially difficult to automate. For example, “DM” in an EHR notes section could either mean “diabetes mellitus” or refer to someone’s initials. The new technology pulls out meaningful data based on the context, such as whether a medication used to treat diabetes is found elsewhere in the EHR. The software can also look for risk factors that directly affect outcomes, such as whether the patient is regularly taking their medication.
Spurring High-Risk Early Development
With the initial SBIR grant, Health Fidelity could program the software for natural language processing of physicians’ notes and then evaluate the precision of the platform.
At the time, Health Fidelity already had undergone a first round of venture capital funding.
“Still, we had quite a bit of research to do,” Riskin said. “Some of the high-risk basic science efforts funded by NCATS became foundational to the company’s first commercial product.”
Riskin added that having validation from NIH’s peer review that the Health Fidelity team’s work was scientifically sound was invaluable and that value-based care is an important area of scientific research. Because of this funding, Health Fidelity was able to make great strides before many others in the field were even thinking about technologies to analyze narrative data for value-based care. Health Fidelity was therefore in a prime position when recent health care and related legislation helped accelerate the move to this model.
In 2013, Health Fidelity established collaborations with University of Pittsburgh Medical Center (UPMC) and several other large institutions to pilot-test the technology. A year later, the company signed a deal with UPMC and raised $29 million in investment capital. The company has since flourished, generating revenue and giving hospitals and health systems a better way to look at patient risk and pave the way to improved care through value-based models.
“The NCATS grant was crucial to the company’s success,” said Riskin. “And the work is now supporting value-based care for hundreds of thousands of patients.”
Since bringing this technology to market, Riskin has handed off his leadership role at Health Fidelity and returned to VMT, where he is working to solve the next critical problem in health care through innovative technologies.
“Dan’s story is just one example of how leveraging an SBIR award to obtain additional investment from the venture capital community can lead to further success,” Portilla said. “By providing early-stage funding for developing new translational bioinformatics tools, NCATS is improving health through smarter science.”
Posted July 2017