Vardhan is using Argonne supercomputing and AI resources to advance biomedical research.
The Argonne Leadership Computing Facility (ALCF), a U.S. Department of Energy (DOE) user facility located at DOE’s Argonne National Laboratory, named postdoctoral researcher Madhurima Vardhan the latest recipient of its Margaret Butler Fellowship in Computational Science.
Vardhan, a computational scientist who specializes in developing high performance computing (HPC) and artificial intelligence (AI) algorithms for biomedical and clinical applications, earned a PhD in Biomedical Engineering at Duke University. While at Duke, she won the Student Design and Research award at the Annual Biomedical-Engineering Society Meeting for her project “Deriving Fractional Flow Reserve (FFR) from Routine Angiogram-Based Fluid Simulations.” Madhurima was also awarded the American Heart Association Predoctoral Fellowship in 2019 for her work “Novel Anatomy-Physiology Guided Diagnostic Metric for Complex Coronary Lesions.”
The Margaret Butler Fellowship in Computational Science honors the lifetime achievements of Margaret Butler, a pioneering researcher in both computer science and nuclear energy. Among her many notable accomplishments, Butler served as the director of Argonne’s National Energy Software Center and was the first female Fellow of the American Nuclear Society.
In this Q&A, Vardhan discusses her journey into the world of scientific computing, as well as some of the unique biomedical projects she plans to work on at the ALCF.
What drew you to apply for the Margaret Butler Fellowship?
I learned about the Margaret Butler Fellowship when I attended the Argonne Training Program on Extreme-Scale Computing (ATPESC) in 2019. With an incredible line-up of speakers and several hands-on sessions on leadership-class computing systems, ATPESC proved to be beyond informational. I was very excited to learn about different dimensions of developing novel scientific applications for HPC, and the Margaret Butler Fellowship offered just the same. What I also liked about the fellowship was the opportunity to engage in multidisciplinary collaborations with different science project teams across Argonne. Overall, it really resonated with my future aspirations of developing computational frameworks for solving complex science problems utilizing HPC.
What initially sparked your interest in the field of computer science?
I have always wanted to help people lead healthier lives, and I believe computer science can revolutionize medical advancements and enable system-wide improvements to healthcare. Integrating computer science and medicine can save time, money, and lives. As such, combining physician expertise and computational algorithms in medical practice has the potential to reduce errors in the diagnosis and treatments of patients. Recent advancements in AI can even improve communication between patients and doctors, making it easier for patients to receive accurate medical treatments. Beyond clinical informatics and diagnostics, the applications of computer science in biomedicine are endless and span across drug development, automated healthcare assistants, public health, and creating a digital twin. The fact that computer science can impact so many different areas in healthcare is what truly piqued my interest to dive deeper into this field.
What do you plan to do during your time in the fellowship?
I am excited about developing scalable computational tools that enable AI to tackle healthcare challenges beyond human intuition. The last decade of developments in AI have demonstrated transformative progress in fields such as computer vision (ImageNet) and natural language understanding (WordNet). This was possible due to the rapid increase in computational capabilities and availability of open access datasets. However, the lack of such open access and high-quality health datasets has impeded AI-driven healthcare applications, due to the complex and sensitive nature and privacy concerns. Addressing this challenge is important for developing and training scalable AI algorithms that rely on ethically sourced, trustworthy, well-defined, and accessible health datasets. I believe a viable and attractive solution is generating synthetic data from scratch such that it is representative of real data. Therefore, during my time in the fellowship, I would like to develop AI algorithms trained on synthetic data which can be used by biomedical and clinical scientists interested in a specific healthcare problem, ultimately leading to real medical innovations.
Can you tell us about your current research project(s)?
We are working to demonstrate the usefulness of the pretrained AI models on synthetic and real-patient datasets to predict functional biomarkers of long COVID. To pursue this work, we would use data from the Million Veteran Program and the UK Biobank, and subsequently train models in the Argonne Biomedical Learning Enclave (ABLE). For this work, ABLE is the ideal compute facility with not just state-of-art HPC hardware for conducting AI-driven research, but it also has the necessary privacy controls required for handling sample patient data.
Can you tell us about your research and use of HPC before coming to the ALCF?
During my PhD, I developed a massively parallel fluid model to simulate blood flow on traditional HPC and cloud infrastructures. I validated this model in a multicenter clinical trial and derived promising novel biomarkers that could predict adverse events. Taking my PhD work forward, I worked in Google Research’s AI and Health group and developed advanced AI models that could help people lead healthier lives. To this end, I worked on large language models that could function as an automated and personalized health coach. With my research experience so far, I learned that achieving breakthrough science requires not only an in-depth understanding of complex software and hardware but also domain knowledge. While doing research I have focused on both aspects, first identifying the research problems and unmet needs in health, and then solving them by developing novel computational algorithms using HPC and AI.
What are you most looking forward to while working at the ALCF?
First off, the ALCF is a very strong, integrated HPC facility that provides ample engagement between researchers from diverse scientific backgrounds. I would like to advance multimodal AI models in biomedicine using HPC and work towards realizing the same pivotal role which AI played in the fields of computer vision and natural language understanding. Due to the technical expertise, cross functionality, and computational resources required for this endeavor, I believe ALCF will provide an ideal research environment.
Outside of the professional sphere, what can you tell us about yourself – unique hobbies, favorite places, etc.? Is there anything about you your colleagues might be surprised to learn?
I picked up rock climbing, hiking, and playing flute, all during my time at graduate school. I have a liking towards mountains and was thrilled to go ziplining over Swiss Alps. Also, this past summer I was able to visit the Himalayas.
The Argonne Leadership Computing Facility provides supercomputing capabilities to the scientific and engineering community to advance fundamental discovery and understanding in a broad range of disciplines. Supported by the U.S. Department of Energy’s (DOE’s) Office of Science, Advanced Scientific Computing Research (ASCR) program, the ALCF is one of two DOE Leadership Computing Facilities in the nation dedicated to open science.
Argonne National Laboratory seeks solutions to pressing national problems in science and technology. The nation's first national laboratory, Argonne conducts leading-edge basic and applied scientific research in virtually every scientific discipline. Argonne researchers work closely with researchers from hundreds of companies, universities, and federal, state and municipal agencies to help them solve their specific problems, advance America's scientific leadership and prepare the nation for a better future. With employees from more than 60 nations, Argonne is managed by UChicago Argonne, LLC for the U.S. Department of Energy's Office of Science.
The U.S. Department of Energy's Office of Science is the single largest supporter of basic research in the physical sciences in the United States and is working to address some of the most pressing challenges of our time. For more information, visit https://energy.gov/science