About

Retsef Levi is the J. Spencer Standish (1945) Professor of Operations Management at the MIT Sloan School of Management. He is a member of the Operations Management Group at MIT Sloan and affiliated with the MIT Operations Research Center. Levi also serves as the faculty leader for Food Chain Supply Analytics. His research focuses on designing analytical data-driven decision support models and tools that address complex business and system design decisions under uncertainty, particularly in health and healthcare management, supply chain, procurement, inventory management, revenue management, pricing optimization, and logistics. Levi has led several industry-based collaborative research efforts with major academic hospitals and organizations across the U.S., including the FDA, Walmart Foundation, and various healthcare and food safety institutions. He has received numerous awards for his contributions, including the NSF Faculty Early Career Development award, the INFORMS Optimization Prize for Young Researchers, the Daniel H. Wagner Prize, and the Harold W. Kuhn Award. Levi teaches courses on operations management, analytics, risk management, system thinking, and healthcare, engaging students from various programs and industry partners. He has graduated multiple PhD, Master, and postdoctoral students and has been recognized for his teaching excellence.

Research topics

• Computer Science
• Business
• Medicine
• Econometrics
• Economics
• Machine Learning
• Political Science
• Artificial Intelligence
• Operations research
• Mathematics
• Geography
• Actuarial science
• Engineering
• Mathematical optimization
• Environmental health
• Meteorology
• World Wide Web
• Statistics
• Finance
• Microeconomics
• Nursing
• Algorithm
• Data science

Selected publications

• Author Correction: Increased emergency cardiovascular events among under-40 population in Israel during vaccine rollout and third COVID-19 wave

  Scientific Reports · 2026-03-12

  articleOpen accessSenior authorCorresponding

  The original version of this Article contained a typographical error in the Results section in which the percentage of patients aged 16-39 receiving their second vaccination dose was reported as 32.2% rather than 42.2%.This has now been updated and the Results section, "Among the 5,506,398 patients receiving their 1st vaccination dose and 5,152,417 patients receiving their 2nd vaccination dose, 2,382,864 (43.3%) and 2,176,172 (32.2%) patients were of age 16-39, respectively." now reads: "Among the 5,506,398 patients receiving their 1st vaccination dose and 5,152,417 patients receiving their 2nd vaccination dose, 2,382,864 (43.3%) and 2,176,172 (42.2%) patients were of age 16-39, respectively."

  PublisherOA PDFDOI

• Reducing Food Waste through a Reservation Scheme

  SSRN Electronic Journal · 2025-01-01

  preprintOpen access1st authorCorresponding
  PublisherDOI

• Designing Inclusive Offerings

  Management Science · 2025-11-26

  article1st authorCorresponding

  The trend toward inclusivity in product and service design—ensuring equal access to, or benefit from, products and services—has gained significant attention in industries such as beauty, wellness, and fashion. However, to our knowledge, there is currently no rigorous definition of inclusivity in the product line design or assortment optimization literature. This paper addresses the need for rigorous models, frameworks, and methods to create inclusive product lines and to measure the inclusivity of existing ones. We introduce the first formal definition of inclusive offerings and propose two continuous measures of inclusivity. We then formulate the inclusive offering design problem, which aims to select a minimum-cardinality set of offerings from a continuous feature space to ensure suitability for all users. Despite its NP-hardness, we present an optimal algorithm for realistic settings and one-dimensional feature spaces. Additionally, we explore the inclusivity resulting from offerings designed to maximize profit, revealing a nonmonotone relationship between inclusivity and the diversity of the user population. Our methods and metrics are applied to the foundation lines of eleven leading makeup brands, demonstrating their practical relevance and impact to the inclusive beauty movement. This paper was accepted by Victor Martínez-de-Albéniz, operations management. Supplemental Material: The online appendices and data files are available at https://doi.org/10.1287/mnsc.2024.06742 .

  PublisherDOI

• Matters Arising: Safety of SARS-CoV-2 vaccination during pregnancy - obstetric outcomes from a large cohort study: methodological biases in study design with potential impact on the study’s interpretation

  BMC Pregnancy and Childbirth · 2025-07-01

  letterOpen access1st authorCorresponding
  PublisherOA PDFDOI

• Improved intrahospital transport time via proximity-based staff assignments

  Journal of the American Medical Informatics Association · 2025-05-09 · 1 citations

  articleOpen accessSenior author

  BACKGROUND: Intrahospital patient transport is pivotal in enabling hospital operations and facilitating safe and efficient patient movement. However, transport delays are common in hospitals, signaling a need for improvement. This study develops, implements, and evaluates a proximity-based transporter-to-request assignment system aimed at improving transport service system efficiency. MATERIALS AND METHODS: In this observational study, we used discrete-event simulation to design and optimize an enhancement to an electronic medical record's original first-in, first-out transporter-to-request assignment system, and we implemented it at a quaternary care academic medical center. Our enhancement prioritizes requests based on the proximity of available transporters within pre-specified areas. We compared transport request completion time (primary outcome) and the percentage of transports exceeding 45 minutes (secondary outcome) during control (01/2021-02/2022) and intervention (02/2022-03/2023) periods and estimated their differences using multivariate generalized linear models to adjust for confounding factors including variable workforce levels and workload. RESULTS: A total of 136 414 transport requests were included in the study. The intervention was associated with an adjusted 5.0% (95% confidence interval 1.8%-8.5%) reduction in completion times and a 16.0% (7.4%-23.9%) relative reduction in the percentage of trips exceeding the 45-minute completion time target. DISCUSSION: The intervention's improvements stem from reductions in unnecessary travel time between transport requests, common to first-in first-out assignment systems. The intervention was designed to be natively integrated into existing electronic health record systems, reducing barriers to real-world adoption. CONCLUSION: Implementing a proximity-based assignment system designed based on simulation-optimization modeling improved intrahospital patient transport efficiency without requiring additional staff.

  PublisherOA PDFDOI

• Observed-to-Expected Fetal Losses Following mRNA COVID-19 Vaccination in Early Pregnancy

  medRxiv · 2025-06-20

  preprintOpen accessSenior author

  ABSTRACT Background The clinical trials used to approve COVID-19 vaccines excluded pregnant women, and existing safety assessments of COVID-19 vaccination, particularly during early stages of pregnancy, are limited to observational studies prone to various types of potential bias, including healthy vaccinee bias. Methods The study includes pregnancies in Israel with last menstruation period (LMP) between March 1, 2016 and February 28, 2022. The main analysis presents observed-to-expected comparisons of the number of eventual fetal losses among pregnant women exposed to mRNA COVID-19 vaccination (almost all Pfizer) during gestational weeks 8-13 and 14-27, respectively. Women vaccinated for influenza during gestational weeks 8-27, as well as women vaccinated prior to pregnancy for COVID-19 or influenza, were used as comparative controls. Cohort-specific expected number of fetal losses are established based on estimates from a regression model trained on historical data from 2016-2018 that incorporates individual-level risk factors and gestational week of each pregnant woman included in the cohort. Results Analysis of 226,395 singleton pregnancies in Israel from 2016 to 2022 indicates that COVID-19 vaccination with dose 1 during weeks 8-13 was associated with higher-than-expected observed number of fetal losses of approximately 13 versus 9 expected for every 100 exposed pregnancies, i.e., nearly 3.9 (95% CI: [2.55-5.14]) additional fetal losses above expected per 100 pregnancies Most of the excess fetal losses occurred after gestational week 20 and nearly half occurred after gestational week 25. Similarly, women vaccinated with dose 3 during weeks 8-13 exhibited a higher-than-expected number of fetal losses with nearly 1.9 (95% CI: 0.39-3.42]) additional fetal losses above expected per 100 pregnancies. In contrast, pregnant women vaccinated for influenza during weeks 8-27 exhibited a consistently lower-than-expected observed number of fetal losses, likely the result of healthy vaccinee bias. Women vaccinated for COVID-19 or influenza prior to pregnancy exhibited according-to-expected or lower-than-expected numbers of fetal losses. Conclusion The results provide evidence for a substantially higher-than-expected number of eventual fetal losses associated with COVID-19 vaccination during gestational weeks 8-13.

  PublisherOA PDFDOI

• Supply Chain Characteristics as Predictors of Cyber Risk: A Machine-Learning Assessment

  IEEE Transactions on Dependable and Secure Computing · 2025-05-19 · 1 citations

  article

  This paper provides the first large-scale data-driven analysis to evaluate the predictive power of digital supply chain attributes for assessing risk of cyberattack data breaches. Motivated by rapid increase in the complexity of digital supply chains and related third-party enabled cyberattacks, the paper provides the first quantitative empirical evidence that digital supply-chain attributes are significant predictors of enterprise cyber risk. The analysis leverages externally observable cybersecurity ratings that aim to capture the quality of the enterprise internal cybersecurity management, but augments these with original supply chain features that are inspired by observed third-party cyberattack scenarios, as well as concepts from network science research. The main quantitative result of the paper is to show that these supply chain network features add significant detection power to predicting enterprise cyber risk, relative to merely using enterprise-only attributes. In particular, compared to a base model that relies only on internal enterprise features, the supply chain network features improve the out-of-sample AUC by 2.3%. Given that a cyber data breach on a specific enterprise is a low probability high impact risk event, these improvements in the prediction power have significant value. Additionally, the model highlights several cybersecurity risk drivers related to third-party cyberattack and breach mechanisms and provides important insights as to what key metrics should be monitored and what interventions might be effective to mitigate these risks.

  PublisherDOI

• Optimal Interventions for Increasing Healthy Food Consumption Among Low-Income Populations

  Management Science · 2025-09-30

  article1st authorCorresponding

  More than $60 billion per year in the United States is spent on policies aimed to increase fruit and vegetable (FV) consumption among low-income households. Many of these policy interventions are either monetary (e.g., financial incentives) or education related. The goal of this paper is to improve the performance of these interventions through a more strategic and personalized allocation of funds. This paper introduces a consumer behavioral model for grocery shopping decisions, which is nested into the policymaker’s upper-level optimization problem. The policymaker’s goal is to ensure that the FV spending of all consumers in a given population exceeds a specified threshold by utilizing a small strategic set of different intervention bundles—combinations of monetary and education-related interventions. Although an exact solution to the upper-level problem is intractable, we provide an analytical upper bound on the number of intervention bundles needed to achieve the policymaker’s goal, as well as a method for constructing these intervention bundles and assigning them to individuals based on their characteristics. We demonstrate the practicality of the model and approach using the low-income households in the U.S. Department of Agriculture’s FoodAPS data set. This paper was accepted by Jayashankar Swaminathan, operations management. Funding: This material is based upon work supported by the National Science Foundation Graduate Research Fellowship under [Grant 1745302]. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2022.02324 .

  PublisherDOI

• Twelve-Month All-Cause Mortality after Initial COVID-19 Vaccination with Pfizer-BioNTech or mRNA-1273 among Adults Living in Florida

  medRxiv · 2025-04-29

  preprintOpen access1st authorCorresponding

  ABSTRACT Comparative effects of the mRNA COVID-19 vaccines on mortality and other health outcomes are uncertain. Matched cohort analysis of 1,962,822 adults living in Florida who received a first dose of either BNT162b2 (Pfizer) or mRNA-1273 (Moderna) between December 18, 2020, and August 31, 2021, was used to examine 12-month risk of all-cause, cardiovascular, COVID-19, and non-COVID-19 mortality. Matching was performed on seven criteria, including census tract. Compared with mRNA-1273 recipients, BNT162b2 recipients had significantly higher risk for all-cause mortality (783.6 vs. 562.4 deaths per 100,000; odds ratio, OR [95% CI]: 1.407 [1.358, 1.457]), cardiovascular mortality, COVID-19 mortality and non-COVID-19 mortality. Negative control outcomes and robustness analyses did not show any indication of meaningful unobserved residual confounding that is likely to alter the results. The results of this study are consistent with cumulative evidence from prior literature showing worse outcomes in recipients of BNT162b2 compared to mRNA-1273. Primary Funding Source The study did not receive any external funding.

  PublisherOA PDFDOI

• The Limits to Learning a Diffusion Model

  Management Science · 2025-04-18 · 1 citations

  article

  This paper provides the first sample complexity lower bounds for the estimation of simple diffusion models, including the Bass model (used in modeling consumer adoption) and the Susceptible-Infected-Recovered (SIR) model (used in modeling epidemics). We show that one cannot hope to learn such models until quite late in the diffusion. Specifically, we show that the time required to collect a number of observations that exceeds our sample complexity lower bounds is large. For the Bass model, our results imply that when new adopters are predominantly driven by imitation, one cannot hope to predict the eventual number of adopting customers until one is at least two-thirds of the way to the time at which the rate of new adopters is at its peak. In a similar vein, our results imply that in the case of an SIR model, one cannot hope to predict the eventual number of infections until one is approximately two-thirds of the way to the time at which the infection rate has peaked. This lower bound in estimation further translates into a lower bound in regret for decision making in epidemic interventions. Our results formalize the challenge of accurate forecasting and highlight the importance of incorporating additional data sources. To this end, we analyze the benefit of a seroprevalence study in an epidemic, where we characterize the size of the study needed to improve SIR model estimation. Extensive empirical analyses on product adoption and epidemic data support our theoretical findings. This paper was accepted by David Simchi-Levi, data science. Funding: This work was supported by the Division of Civil, Mechanical and Manufacturing Innovation [Grant 1727239]. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2022.02953 .

  PublisherDOI

Recent grants

• An Innovative Optimization and Computational Framework for Assortment Problems Under Consider-Then-Rank Choice Models

  NSF · $300k · 2015–2019

• CAREER: New Algorithmic Approaches to Computationally Challenging Stochastic Supply Chain and Revenue Management Models

  NSF · $400k · 2009–2015

• MSPA-MCS: Collaborative Research: Algorithms for Near-Optimal Multistage Decision-Making under Uncertainty: Online Learning from Historical Samples

  NSF · $172k · 2007–2010

Frequent coauthors

• Peter F. Dunn

  Harvard University

  35 shared

• David B. Shmoys

  Cornell University

  18 shared

• Georgia Perakis

  18 shared

• Jonathan P. Wanderer

  Vanderbilt University Medical Center

  16 shared

• J Standish

  Vanderbilt University

  16 shared

• Brett A. Simon

  16 shared

• Bethany Daily

  Massachusetts General Hospital

  16 shared

• Yanchong Zheng

  14 shared

Labs

• System DynamicsPI

Awards & honors

• NSF Faculty Early Career Development award
• INFORMS Optimization Prize for Young Researchers (2008)
• Daniel H. Wagner Prize (2013)
• Harold W. Kuhn Award (2016)
• MSOM Responsible Research in Operations Management award (20…