Anthony Sena is an Associate Director of Epidemiology Analytics at Janssen Research and Development where he architects software solutions and data platforms for the analysis and application of observational data sources. Anthony’s areas of expertise include web application development, data modeling, information visualization, technology infrastructure, project management, and informatics. A collaborator on a number of open-source software solutions in OHDSI and one of the co-leads of the ATLAS & WebAPI working group, he has taken a prominent role in the recent CHARYBDIS Project, a characterization study to understand the disease natural history of COVID-19. His focus is on expanding the capabilities of the OHDSI open-source solution architecture to enable transparent and reproducible research using observational data.
Prior to joining Janssen Research and Development, Anthony held many leadership and technical roles of increasing responsibility across a range of business sectors including energy, pharmaceuticals, retail and financial services. Anthony received his undergraduate degree in Computer Science from Drew University.
He recently discussed his journey to OHDSI, his work with open-source tools and some of his most important projects, during the latest Collaborator Spotlight.
Your career has spanned several industries, including energy, retail, and financial services, so how did you find your way to data sciences and informatics?
I’ve been fortunate to gain experience working in different industries during my career across a wide range of information technology projects. Nearly all of these projects included data management, usually in the form of a relational database. So for me, I always thought in terms of data: how is it collected, how is it cleaned/verified and how do we structure it to collect insights. When “big data” and “data science” became the big IT buzz words, I had a hard time separating out the problems people were solving from the technology that was utilized. So, I started reaching out to people in my network that were working in this space.
I was fortunate to work with Frank DeFalco early on in my career and he was my first call when I started to dig into applying data science technology with big data. At the time, we were not working together at Janssen, but when I reached out to him, he was able to share with me the OHDSI mission and the open-sourced tools that were built on the OMOP Common Data Model (CDM) standard. The idea of building open-source solutions that could promote better health decisions and better care for patients was (and still is) very exciting. Five years later, I’m really glad that I reached out to Frank since it led me to be a part of the Janssen team and to the OHDSI community.
What were your earliest challenges when joining OHDSI, and how were you able to navigate them?
I had two sets of challenges. First, I needed to learn about the appropriate use of observational data to answer clinical questions. Thankfully I work with some great colleagues at Janssen who were able to educate me on this topic. Additionally, I’ve had the benefit of being in OHDSI long enough to have attended a number of tutorials that have helped to deepen my understanding in this area. Secondly, I needed to learn about how the OHDSI ecosystem works from a technology perspective. One of the things that I think is remarkable about the OHDSI technology stack is the ability to support so many different database platforms. In my past IT projects, we were able to scope our work to a single database platform and usually a specific operating system to host our code, which makes the development of the application much easier. In OHDSI, we are challenged by the fact that many different organizations use different database platforms based on institutional requirements. When I heard of SqlRender, I thought it was magic — writing one SQL statement that could run on many different database platforms? It seemed too good to be true. So, learning how to write SQL that is compatible with SqlRender was a big challenge but thankfully there are members of the community who were able to support me as I learned this skill.
Side note: go check out SqlRender – it is awesome.
You have collaborated on numerous open-source tools for the community. How important is the open-source nature of OHDSI in building the most efficient analytics tools?
OHDSI’s core values include openness and reproducibility and the open-source nature of OHDSI is critical in supporting those values. Open-source software development has trade-offs but by publishing code used to produce results, we strive to make the work we do as a community as transparent as possible. Another benefit of adopting open-source is that it allows for others to contribute to make tools more performant or to fix bugs. It has been a great experience to work with people in the OHDSI community who have both the ideas and skills to advance the software development forward.
For me personally, the open-source approach adopted by OHDSI allowed me to download and explore the CDM and tools without restriction. Contrast this with exploring commercial software space: you generally have to provide your phone, email, firstborn, etc., just to get a glimpse of how the software works. Then you have to navigate licensing, per-user charges, maintenance fees, and other costs associated with running it in your organization. It was refreshing to be able to see software function without all of that overhead and to have direct access to the code to see how it works.
ATLAS is one of the community’s most-used tools. As Project Manager of the Atlas & WebAPI Working Group, can you explain to newcomers what the tool is, and where it benefits research the most?
ATLAS is a web-based tool used to design of observational studies utilizing the OMOP CDM. You can install ATLAS and configure it to work with your CDM to do things like explore the standardized vocabularies, design and execute cohorts against your patient-level data, and do more advanced operations such as characterization, patient-level prediction, and population-level estimation. One of the benefits of using ATLAS is that it does not require you to directly code against your CDM. Instead, we capture design artifacts as JSON (a machine-readable text format) and translate that into executable code that can run and produce results on your CDM. Keeping the design as JSON allows ATLAS uses to share their designs with other ATLAS users without ever exposing any patient-level data.
Can you discuss the current CHARYBDIS project, both in terms of its potential impact on COVID studies, as well as your role in the development?
CHARYBDIS is a characterization study we are doing to understand the disease natural history of COVID-19. Our objectives are to describe the baseline demographic, clinical characteristics, treatments, and outcomes of interest among individuals tested for SARS-CoV-2 and/or diagnosed with COVID-19 overall and stratified by sex, age, and specific comorbidities. The study provides foundational knowledge about patients that we can use to both understand the patients being treated for COVID-19, as well as the ways in which these patients’ experiences are recorded in each participating site’s database. In terms of the development, I’ve been focused on building out the R package that is used at the participating sites to run the study, collecting diagnostics and results, and providing them back to the study leads. It has been a collective effort by the team and the participating sites to run the package, assemble & review the results, and to work on the manuscripts that are forthcoming. I’m proud to have led the development of the R package and think that this will continue to provide great utility for our understanding of the evolution of COVID-19.
What has inspired you the most about being part of OHDSI, especially in your role in developing so many of the tools within the community?
The OHDSI community is a source of inspiration for me. Take for example the OHDSI COVID-19 Study-a-thon. We had hundreds of people online, across the globe, contributing their talents and expertise to work on a problem that is impacting us all. I’ve attended a number of OHDSI events and interacted with members of the community that are doing amazing work based on the data standards and tools that are made available. OHDSI has helped me view science as a team sport — no one person can do it by themselves. I’m inspired to develop tools and contribute my talents towards OHDSI’s mission.
What are some of your hobbies, and what is one interesting thing that most community members might not know about you?
When I’m not hacking code, I enjoy playing golf, barbecuing, and spending time with family and friends. I’m a big New York Giants football fan and I’m hoping that the NFL can still have games while protecting the players during the pandemic. One interesting thing about me is that I am a father of twin girls. I’m trying to teach them how to golf and I’m pretty sure they’ll be better at it than me in no time!