I am currently a Staff Fellow (Regulatory Scientist) at the U.S. Food and Drug Administration in the Division of Imaging Diagnostics and Software Reliability at the Center for Devices and Radiological Health.

I recently completed my PhD in computer science at Johns Hopkins University, where I was advised by Prof. Suchi Saria. My research focuses on improving the reliability of machine learning models (see our tutorial here) and expanding the technical toolkit available for answering the question: "Is a machine learning model safe and effective?" This includes developing new learning paradigms, methods for monitoring and evaluating performance, and model updating strategies. I most often use tools from causal inference, graphical models, and robust optimization to develop solutions.

I'm particularly interested in applications in healthcare—a safety-critical domain in which the cost of failure is high. I am motivated by the observation that, as a field, we currently lack tools to determine the safety and efficacy of machine learning models developed for healthcare. Over the course of my PhD, I have worked to bridge this gap with respect to the problem of dataset shift: systematic differences between the conditions in which a model was trained and the conditions in which it is deployed or tested that can lead to failures. Some useful starter resources include:

• Our NEJM letter containing a checklist that overviews different ways in which dataset shifts manifest in healthcare.
• Our Biostatistics article presenting technical challenges in this space related to dataset shift.

Previously, I received B.S. degrees in computer science and mathematics from Vanderbilt University, where I worked on ensuring privacy through de-identification in electronic medical records with Professor Brad Malin.

Preprints

• A hierarchical decomposition for explaining ML performance discrepancies.
  Jean Feng, Harvineet Singh, Fan Xia, Adarsh Subbaswamy, and Alexej Gossmann.
  

• I-SPEC: An End-to-End Framework for Learning Transportable, Shift-Stable Models.
  Adarsh Subbaswamy and Suchi Saria.
  

Publications

• Designing monitoring strategies for deployed machine learning algorithms: navigating performativity through a causal lens.
  Jean Feng, Adarsh Subbaswamy, Alexej Gossmann, Harvineet Singh, Berkman Sahiner, Mi-Ok Kim, Gene Pennello, Nicholas Petrick, Romain Pirrachio, and Fan Xia.
  Causal Learning and Reasoning, CLeaR 2024.
  

• A bias evaluation checklist for predictive models and its pilot application for 30-day hospital readmission models.
  H Echo Wang, Matthew Landers, Roy Adams, Adarsh Subbaswamy, Hadi Kharrazi, Darrell J Gaskin, and Suchi Saria
  Journal of the American Medical Informatics Association (JAMIA), 2022.
  

• A unifying causal framework for analyzing dataset shift-stable learning algorithms.
  Adarsh Subbaswamy, Bryant Chen, and Suchi Saria.
  Journal of Causal Inference, 2022.
  

• The Clinician and Dataset Shift in Artificial Intelligence.
  Sam G Finalyson*, Adarsh Subbaswamy*, Karandeep Singh, John Bowers, Annabel Kupke, Jonathan Zittrain, Isaac S Kohane, Suchi Saria. (*Equal Contribution)
  New England Journal of Medicine, 2021.
  

• Evaluating Model Robustness to Dataset Shift.
  Adarsh Subbaswamy*, Roy Adams*, and Suchi Saria. (*Equal Contribution)
  Artificial Intelligence and Statistics, AISTATS 2021.
  

• From Development to Deployment: Dataset Shift, Causality, and Shift-Stable Models in Health AI
  Adarsh Subbaswamy and Suchi Saria.
  Biostatistics, 2020.
  

• Preventing Failures Due to Dataset Shift: Learning Predictive Models That Transport.
  Adarsh Subbaswamy, Peter Schulam, and Suchi Saria.
  Artificial Intelligence and Statistics, AISTATS 2019.
  (Previously an oral presentation at NeurIPS 2018 Causal Learning Workshop)

• Counterfactual Normalization: Proactively Addressing Dataset Shift and Improving Reliability Using Causal Mechanisms.
  Adarsh Subbaswamy and Suchi Saria.
  Uncertainty in Artificial Intelligence, UAI 2018.

• Treatment-Response Models for Counterfactual Reasoning with Continuous-time, Continuous-valued Interventions.
  Hossein Soleimani*, Adarsh Subbaswamy*, and Suchi Saria. (*Equal Contribution)
  Uncertainty in Artificial Intelligence, UAI 2017.

• Non-intrusive occupancy monitoring using smart meters.
  Dong Chen, Sean Barker, Adarsh Subbaswamy, David Irwin, and Prashant Shenoy.
  Proceedings of the 5th ACM Workshop on Embedded Systems For Energy-Efficient Buildings. 2013.