The Pittsburgh Health Data Alliance announced its collaboration with Amazon Web Services Aug. 7, which aims to advance innovation in cancer diagnostics, precision medicine, voice-enabled technologies and medical imaging through machine learning research.
Rob Hartman, PhD, director of translational science at Pittsburgh-based UPMC Enterprises, and Zariel Johnson, PhD, program manager at UPMC Enterprises, were two major players in the partnership and within the PHDA.
The alliance was formed by UPMC, University of Pittsburgh and Carnegie Mellon University to use data to transform how diseases are treated and prevented as well as develop strategies to better engage with patients.
Below, Drs. Hartman and Johnson describe how the recent partnership with AWS will further the PHDA.
Editor's note: Responses are lightly edited for clarity and length.
Question: What are some of the research projects you are exploring with AWS?
Dr. Rob Hartman: There's certainly a methodical view at UPMC. We can provide a view of how healthcare is now and where we think it is headed. UPMC also has a strong sense of general trends in the healthcare marketplace, where the unmet needs are and where the industry is going. We considered which types of projects would be data-intensive or would rely on the kinds of tools and resources that AWS or the cloud has. Other criteria we had when choosing which projects would get funding was the impact it would achieve. We wanted to build solutions that would move science and clinical care forward, whether it be a big disruptive finding or something that could be scaled broadly and rapidly.
Dr. Zariel Johnson: Some of the projects can be broadly classified as improving personalized medicine or precision medicine. Other projects use genomic data and imaging data in innovative ways. There are four projects that are looking to use voice-enabled technology and improving operational efficiencies.
One specific project is being led David Vorp, PhD, out of the University of Pittsburgh. His team is tracking problems related to abdominal aortic aneurysm. Currently, there aren't good objective standards for identifying the middle range of aneurysms and how they should be treated. His team is using imaging data and other information to predict in an objective fashion how the aneurysms should be treated. Some of the AWS resources are helpful for the type of model training and testing that Dr. Vorp is doing.
Another project is pulling information from an EHR and using that information to better understand what medical codes are associated with that information. That is a process today that requires a lot of manual labor and a lot of searching through diagnostic codes, which is time intensive and can lead to errors. This team is looking at exciting innovation, both from a scientific standpoint and the applications in healthcare, to automate the processes. The tool would then provide diagnostics codes that could be reviewed by the appropriate department.
Q: What excites you the most about this partnership? Looking 5 years ahead, how do you see this partnership expanding?
RH: The opportunity to work on the AWS platform and to build things that can be easily integrated and scaled beyond the research setting is quite exciting. Additionally, it's exciting about where this could go. I think as we achieve great results, it will draw more attention from AWS and others to the Pittsburgh Health Data Alliance. What makes the alliance strong is the collaboration of an innovative academic medical center, the world's number one computer science and AI school and one of the highest image-funded research universities working together to advance healthcare. By bringing in the external support of AWS will take the alliance to new heights.
In an ideal world, we would like to see a closer partnership with AWS — a partnership where we are able to leverage more expertise from both sides on identifying projects and supporting projects to achieve the ability to scale and transition out of the academic world. There is complementary expertise between technology companies and progressive academic medical centers that have only begun to be leveraged. As we achieve real results, we will be able to further our partnership where we leverage the complementary strengths. What we hope to build is the infrastructure to support both academic research in healthcare as well as help solutions grow and expand across the healthcare industry.
ZJ: One really exciting part about this collaboration is bringing in the training and culture that AWS has to help some of these academic researchers really understand what it takes to get products to the next level in healthcare implementation.
This is a huge opportunity. In the future, based on the strength of the type of researchers that are involved in the Pittsburgh Health Data Alliance, we expect to see more and more projects that would lend themselves to this type of collaboration with AWS.
Q: Why Amazon Web Services?
RH: The Pittsburgh Health Data Alliance as it is now is UPMC funding research programs and investigators at Carnegie Mellon University and the University of Pittsburgh. The goal is to take the researcher and turn it into scientific advancement as well as real world translation or commercial opportunity.
We've been active now for four years, guiding the research to positively affect healthcare. Our discussion with AWS began with the motivation of finding a way to accelerate and advance what is going on in the university realm, where research is being done to increase the impact on what is happening in the academic realm. We were also motivated by seeing the translation of the research and development to clinical care. AWS provides cloud-based tools and resources that can allow for the acceleration of advancement.
We work with AWS to identify projects that are of high value in healthcare. We then served as a facilitator of AWS' support. In a sense, it is a unique partnership because we were facilitating funding of high-value research projects that we identified. Collectively, we worked with them to determine which projects were best to fund.