About

Sanjay Mehrotra is a Professor of Industrial Engineering and Management Sciences at Northwestern University, where he also serves as the Director of the Center for Engineering and Health and co-Leads Project Minerva. His research develops methods for decision optimization under uncertainty by studying the geometric and algebraic properties of these problems. He is passionate about applied research problems in Health Systems Engineering, with significant contributions to modeling the US National Transplant System, liver disease, and pandemic modeling. His expertise spans areas such as Energy, Inventory Management, and Optimal Learning, with a focus on balancing mathematical rigor and practical applicability. Professor Mehrotra has established groundbreaking results in the convergence of two-stage stochastic optimization, enabling solutions to previously intractable problems. He has also contributed fundamental insights into quantifying convergence metrics in stochastic systems described as Markov Chains and pioneered efforts in Risk-Adjusted and distributionally robust optimization. His work in healthcare systems engineering includes modeling complex health systems, and he has held leadership roles such as chairing the INFORMS Fellow Committee and serving on the INFORMS board.

Research topics

• Computer Science
• Mathematics
• Artificial Intelligence
• Mathematical optimization
• Marketing
• Computer network
• Business
• Applied mathematics
• Combinatorics
• Operations research
• Medicine
• Virology

Selected publications

• Case—Developing Data-Driven 24/7 Nurse Staffing and Shift Scheduling Plans

  INFORMS Transactions on Education · 2026-04-29

  articleOpen accessSenior author

  Funding: This work was supported by the Division of Civil, Mechanical and Manufacturing Innovation [Grant CMMI-0928936].

  PublisherDOI

• Case Article—Developing Data-Driven 24/7 Nurse Staffing and Shift Scheduling Plans

  INFORMS Transactions on Education · 2026-04-29

  articleOpen accessSenior author

  This case study offers a comprehensive hands-on approach to addressing hospital workforce planning challenges by integrating demand forecasting errors into mathematical programming. Structured around two interconnected projects—refined over a decade of experience—students develop data-driven 24/7 nurse staffing and shift scheduling plans that ensure continuous coverage while balancing cost, service quality, operational constraints, and nurse satisfaction. By employing time-series models to predict hourly patient volumes and integrating these forecasts into integer programming models, the case demonstrates how to construct 24/7 coverage matrices and incorporate auxiliary variables that allow for controlled deviations in staffing levels and service delays due to forecasting errors. Emphasizing multiobjective optimization and Pareto frontier analysis, the case effectively evaluates tradeoffs between overstaffing and understaffing. Designed for both undergraduate and graduate courses in healthcare management science or operations research, this case bridges theoretical concepts with real-world applications, thereby enhancing educators’ ability to deliver effective decision-making training in healthcare operations. Funding: This work was supported by the Division of Civil, Mechanical and Manufacturing Innovation [Grant CMMI-0928936]. Supplemental Material: The Teaching Note and supplemental material are available at https://www.informs.org/Publications/Subscribe/Access-Restricted-Materials .

  PublisherDOI

• A Framework for Stochastic Fairness in Dominant Resource Allocation with Cloud Computing Applications

  ArXiv.org · 2025-01-29

  preprintOpen accessSenior author

  Allocation of limited resources under uncertain requirements often necessitates fairness considerations, with applications in computer systems, health systems, and humanitarian logistics. This paper introduces a distributionally robust (DR) stochastic fairness framework for multi-resource allocation, leveraging rough estimates of the mean and variance of resource requirement distributions. The framework employs a sampled approximation DR (SA-DR) model to develop the concept of stochastic fairness, satisfying key properties such as stochastic Pareto efficiency, stochastic sharing incentive, and stochastic envy-freeness under suitable conditions. We show the convergence of the SA-DR model to the DR model and propose a finitely convergent algorithm to solve the SA-DR model. We empirically evaluate the performance of our moment-based SA-DR model -- which uses only rough estimates of the mean and variance of the resource requirement distribution -- against alternative resource allocation models under varying levels of information availability. We demonstrate that our moment-based partial-information SA-DR model can achieve performance closer to the full-information model than the worst-case information model. Convergence of the sampled approximation model and comparisons across models are illustrated using data from cloud computing applications.

  PublisherOA PDFDOI

• Convergence and bound computation for chance constrained distributionally robust models using sample approximation

  Operations Research Letters · 2025-01-27

  articleSenior authorCorresponding
  PublisherDOI

• Effect of Biomarkers and Patient-Specific Factors on Kidney Rejection Risk

  Journal of the American Society of Nephrology · 2025-10-01

  article
  PublisherDOI

• A New Finite Approximation Method for Evaluating Steady-State Performance of a Continuous-State Markov Chain with an Application to Queues with Customer Abandonment

  Mathematics of Operations Research · 2025-06-12

  articleSenior author

  This paper develops a new method for computing the stationary distribution and steady-state performance measures of stochastic systems that can be described as a continuous-state Markov chain supported on [Formula: see text]. The balance equations are solved by constructing a proxy Markov chain with finite states. We show the consistency of an approximate solution and provide deterministic nonasymptotic error bounds under the supremum norm. Our method is near optimal among all approximation methods using discrete distributions. We apply the developed method to compute the stationary distribution of virtual waiting time and associated performance measures for a GI/GI/1+GI queue in which the large market assumption may not hold and the patience time may follow any bounded distribution. Numerical experiments show that our method outperforms steady-state simulation, phase-type approximation, diffusion approximation, and fluid approximation, particularly for medium or small arrival intensities in overloaded or balanced loaded queues. Funding: S. Li and S. Mehrotra were supported by the National Science Foundation grant CMMI-1763035.

  PublisherDOI

• Using Torque Teno Virus as a Serial Monitoring Tool for the Net State of Immunosuppression of Kidney Transplant Recipients

  Clinical Transplantation · 2025-09-01 · 1 citations

  articleOpen accessSenior author

  Torque Teno Virus (TTV) has emerged as a promising marker reflecting the net state and trajectory of immunosuppression (IS). We analyzed longitudinal TTV data from 252 kidney transplant recipients in a multi-center observational study (average 7.8 visits/patient over 2 years). Patient-specific TTV trajectories, computed as slopes over defined time windows, captured the direction of immune response dynamics. A past 1-year logTTV slope < 0.0066, combined with a low current logTTV (<4.3) and a relatively high historical logTTV average (>5.7), was associated with a 13.88-fold increase in the odds of subclinical AR (95% CI: 5.49-37.42) relative to patients whose slope exceeded 0.0066. In contrast, a past 1-year logTTV slope > 0.076 conferred a 12.15-fold rise in the odds of infection over TX (95% CI: 2.99-81.42). Decision trees incorporating TTV trajectories achieved AUCs of 0.67 for both subclinical AR and infection versus TX-outperforming models using a single TTV measurement. We identified optimal two-sided logTTV thresholds-(4.5,7.8)-stratifying patients into Under-IS, Even-IS, and Over-IS states, where Under-IS status increases subclinical AR odds over TX by 2.39-fold (95% CI: 1.53-3.83), while Over-IS status increases infection odds over subclinical AR by 2.5-fold (95% CI: 1.03-6.42). These findings provide a framework for personalized IS management.

  PublisherDOI

• Cost-Effective Diagnostic Testing Models Using Four Kidney Rejection Biomarkers

  American Journal of Transplantation · 2025-08-01

  articleSenior author
  PublisherDOI

• Using Torque Teno Virus as a Serial Monitoring Tool Representing the Net State of Immunosuppression for Kidney Recipients

  American Journal of Transplantation · 2025-08-01

  articleOpen accessSenior author
  PublisherOA PDFDOI

• Do Patients Think it’s Worth Waiting for a Kidney? Evidence from a Discrete-Choice Experiment

  Patient · 2025-09-04

  article
  PublisherDOI

Recent grants

• Promoting Utilization of Kidneys by Improving Patient Level Decision Making

  NIH · $418k · 2016–2020

• Addressing Geographical Disparities in Transplant Organ Accessibility Across United States

  NSF · $320k · 2011–2016

• Models and Algorithms for Risk Adjusted Optimization with Robust Utilities

  NSF · $200k · 2011–2015

• Methods for Solving Mixed Integer Programs Using Adjoint Lattices

  NSF · $361k · 2005–2010

• Rescuing Kidneys at Risk of Discard

  NIH · $732k · 2019–2023

Frequent coauthors

• Daniela P. Ladner

  Northwestern University

  47 shared

• Vikram Kilambi

  31 shared

• John J. Friedewald

  Northwestern University

  27 shared

• Karolina Schantz

  Vector (United States)

  18 shared

• Ashley E. Davis

  RTI Health Solutions

  17 shared

• Kevin Bui

  University of California, Irvine

  15 shared

• Fengqiao Luo

  14 shared

• Masoud Barah

  Northwestern University

  14 shared

Labs

• Center for Engineering and HealthPI

Awards & honors

• INFORMS Fellow Committee Chair
• INFORMS board member