Paul Nagy, PhD, FSIIM is Associate Professor in the Johns Hopkins University School of Medicine Department of Radiology with a joint appointments in Medicine, Public Health, and the Department of Biomedical Engineering in the School of Engineering. He received his BS from Carnegie Mellon University and his PhD at the Medical College of Wisconsin. His research focus is developing biomarkers from medical imaging to enable real world reproducible evidence from observational research.
He is the director of education for the training programs in the Biomedical Informatics and Data Science section of the Department of Medicine. He leads the Observational Health and Data Science Informatics (OHDSI) efforts at Johns Hopkins as part of the Precision Medicine initiative. He is a leader in the OHDSI community, especially within the developer community. He began the Kheiron Cohort in 2022, which serves to welcome and mentor developers, and he joined Adam Black to start the open-source community workgroup this year as well.
Nagy, who is also active in several other workgroups and co-leads the Medical Imaging WG, helped lead the DevCon 2022 event, and has developed tools to track OHDSI impact in several areas. He shared some thoughts about his career, his many contributions to the OHDSI community, and plenty more in the latest Collaborator Spotlight.
What is your background, and how did you develop an interest in data science?
I learned to program Fortran on punch cards before college and remember my first Numerical Recipes in Fortran book very fondly where they actually shared the code! I’ve been hooked on open science and open source ever since. I was trained as an experimental physicist at CMU and got my PhD in Diagnostics Medical Physics at the Medical College of Wisconsin. My role as a hospital-based physicist was to help Radiologists and Radiation Oncologists with using technology in patient care. As Radiology moved filmless in the early 2000s, I found I could better aid clinicians by enabling information technology and data science delivered at the point of care.
What is your role with Johns Hopkins, and how did you first connect with OHDSI?
My mission is to partner with clinical faculty to empower their clinic as an informatics laboratory. I have a few roles that help me in this mission as well as appointments across the schools of engineering, public health, and medicine to allow me to work across our system.
My principal role is as the program director for the Biomedical Informatics and Data Science graduate training program in School of Medicine where we have 60 graduate students. 40 of our students are professionals seeking a masters degree online while 20 are full time masters or doctoral students. http://dhsi.med.jh.edu/
In the health system I serve as the deputy director of the JHM Technology Innovation Center (http://tic.jh.edu). The JHTIC is a team of 55 software developers, designers, and data scientists working with clinical inventors to build, deploy, and evaluate digital health and clinical decision support solutions within the Johns Hopkins Medical System.
Johns Hopkins launched the precision medicine initiative in 2017 to work with clinics across JHM and empower them with a next generation data science platform as well as deliver AI models back into clinical care with a standards based delivery platform. In April of 2020 as part of this work, I was commissioned to convert our Epic EMR data into OMOP to help our clinical researchers study the progression of disease and evaluate treatments for COVID.
You have taken a leadership role with our open-source community. Why do you see OHDSI as the right place for open-source developers to thrive, and what do you think they can bring to the community?
OHDSI is an open-science community where we are developing the leading methods to work with observational medical records and claims data. There are over 400 contributors in the 180 github repositories in the OHDSI github organization. Open source development is one of the best ways for a software developer to learn professional development practices and get mentorship while making a real impact. While we already have 400 contributors, I see room in OHDSI for many times that number as the field of computational observational research is relatively new and the entire healthcare IT industry is still quite young in bringing computational evidence back into clinical care.
You and Adam Black developed the Khieron Contributor Cohort, which will help onboard and mentor open-source developers in the community. Now that you are a few months in, can you discuss its progress, and is it something you hope continues each year?
Adam is a tremendous leader in our open source community. Together, we lit a spark and made an open invitation to new software contributors to join the OHDSI development community.We were happily surprised that 25 talented individuals committed 10% of their time for an entire year to join us on our journey. The Kheiron project to catalyze our conversation about how to produce open source software that is stable, feature rich, and extensible. We plan on running this program again and again where previous year’s participants will become the following year’s faculty. This is a great career development program for software developers and also a lot of fun. We have held many workshops on standing up the R environment, the database, as well as Atlas and WebApi. Several of our Khierons are already contributing to other projects across OHDSI.
You helped lead a panel on building organizational support for using OMOP/OHDSI, but can you highlight a few key ways that Johns Hopkins has ‘joined the journey’ for people with similar goals?
Johns Hopkins was founded to train physicians as scientists and to have them use their clinic as a laboratory. What excites me the most is that we are using OHDSI and OMOP to combine research with clinical care in our quest to exemplify a learning health system. Our goal is to enable physicians who formulate questions by observing the patient’s progression of disease and response to treatment. OHDSI can help them formulate those questions across many health systems to create reproducible real world evidence. We have a precision medicine initiative with the purpose of bringing those machine learning prediction models back into clinical care through clinical decision support applications.
From your work with DevCon and the Khieron Cohort to your leadership with the open-source community, you seem to be inspiring many community members, but what is it about OHDSI that inspires you to take such an active role?
The values, goals, and people of OHDSI I find incredibly inspiring. The people in OHDSI are highly talented, practical, and collegial. They are always welcoming new people on the journey. OHDSI is a multi-disciplinary community pushing the boundaries of computational observational research so there is always something new to learn. The values of being transparent in our methods and in our software resonate with me in trying to make a lasting impact. I get energized from going to OHDSI meetings because I always meet new people, learn new things, and am part of making a difference in improving healthcare.
What are some of your hobbies, and what is one interesting thing that most community members might not know about you?
My pandemic hobby has been woodworking and I have built out a workshop next to my office which is one of the reasons I enjoy working from home. Most people don’t know that I am often covered in sawdust 10 seconds before I hop on my OHDSI calls.