John Silberholz

Assistant Professor of Business Analytics
Ross School of Business, University of Maryland
7699 Mowatt Ln Room 4326
College Park, Maryland 20742

Graduate Education:
Ph.D. from MIT Operations Research Center
Undergraduate Education:
Dual B.S. degrees from University of Maryland (C.S. and Math)

C.V.
Google Scholar Page

I am an Assistant Professor of Business Analytics at the University of Maryland's R.H. Smith School of Business. My scholarship focuses broadly on data-driven decision-making for healthcare problems of significant societal interest. My main research stream focuses on enhancing decision-making around healthcare experimentation. While I have researched more traditional topics around efficient trial designs that provide accurate estimates of treatment effects while minimizing operational objectives such as expected trial length, enrollment, or cost, my focus around healthcare experimentation is broader, also asking questions like "What therapies should be tested in a planned new trial?" and "How can health systems best learn from past trial data?" My work on healthcare experimentation has twice been recognized with the Pierskalla Best Paper Award from the INFORMS Health Applications Society.

Though I consider the healthcare experimentation stream to be my primary research stream, I have examined healthcare decision making and public policy more broadly, studying topics such as improving nurse shift scheduling by taking into account past helping interactions between pairs of nurses and optimizing complex policies using multiple mathematical models. I have also explored socially relevant topics outside of healthcare, such as improving battery economics to better enable renewable integration on the electric grid.

Contact

Email:

josilber@umd.edu

Featured Works

(pdf)	Enhancing Safety Signaling: Integrating Clinical Trials and Post-Marketing Adverse Event Reports Fernanda Bravo, Yunliang (Lawrence) Chen, John Silberholz (2024) Major revision at Manufacturing & Service Operations Management. Post-marketing drug safety surveillance systems like the FDA's FAERS system rely on voluntary reporting of adverse events by patients or their doctors; this reporting suffers both from selection biases (patients do not randomly select the drug they'll take for their condition) and reporting biases (some patients are more motivated to report safety issues). We combine this data with pre-approval clinical trial data to estimate the direction and magnitude of potential biases, enabling us to more accurately raise safety signals.
(pdf)	Can Employees' Past Helping Behavior Be Used to Improve Shift Scheduling? Evidence from ICU Nurses Zhaohui (Zoey) Jiang, John Silberholz, Yixin (Iris) Wang, Deena Costa, Michael Sjoding (In Press) Accepted at Management Science. This empirical work studies a natural question in shift scheduling: do shifts obtain better operational outcomes when scheduling coworkers together who have helped one another in the past? To study this question, we take advantage of the fact that hospital electronic medical records provide a detailed history of instances where one nurse helped another. Our results demonstrate that ICU shifts with more pairs of nurses that have helped each other in the past have shorter average patient lengths of stay; the effect of past helping is significantly stronger than the effect of team familiarity (looking at whether nurses have worked together on past shifts, regardless of whether they helped one another or not). Counterfactual analysis indicates that optimizing schedules to pair nurses who have helped each other in the past could yield managerially relevant boosts in patient length of stay on the order of a 5% improvement.
(pdf)	Cost-saving synergy: Energy stacking in battery energy storage systems Joonho Bae, Roman Kapuscinski, John Silberholz (In Press) Accepted at Management Science. We use analytical modeling to explain the big profitability boosts observed under battery stacking, in which a battery performs multiple electric grid services simultaneously (e.g. simultaneously performing arbitrage - "buy low, sell high" - while also providing frequency regulation - rapidly charging or discharging to help balance the grid over short timescales). A key reason for the profitability boost is a phenomenon we call "cost-saving synergy," in which the wear and tear on a battery when stacking services is often much smaller than if we had performed the services separately. We establish types of services and grid conditions where stacking will be particularly advantageous.
(pdf)	Adaptive Clinical Trial Designs with Surrogates: When Should We Bother? Arielle Anderer, Hamsa Bastani, John Silberholz (2019) Management Science 68(3): 1982-2002. Winner of the 2019 William Pierskalla Best Paper Award. We study clinical trials that simultaneously track a slow-to-measure true outcome (e.g. overall survival in metastatic cancer) and a faster-to-measure surrogate outcome (e.g. time to disease progression in metastatic cancer). We focus on cases with historical data from similar past clinical trials to observe the statistical link between surrogate and true outcomes. We propose a Bayesian trial design that learns about the true outcome both directly (observing time-to-event data for the true outcome) as well as indirectly (observing the time-to-event for the surrogate outcome and contextualizing that information with the historical link between the two outcomes). The result is a trial design that speeds up trials compared to a classical design (which only learns from a single outcome) with no loss of accuracy. We both analytically identify unexpected types of outcome pairs where our trial design is most valuable, in addition to numerically establishing that our design can yield a 16% decrease in trial costs for metastatic breast cancer clinical trials with no decrease in the quality of inference.
(pdf) (supplement) (code)	What Works Best When? A Systematic Evaluation of Heuristics for Max-Cut and QUBO Iain Dunning, Swati Gupta, John Silberholz (2018) INFORMS Journal on Computing 30(3): 608-624. Special recognition for the 2016 INFORMS Computing Society Student Paper Prize. Heuristics papers for combinatorial optimization problems are often irreproducible (no source code published) and evaluate heuristics on small, homogenous sets of problem instances. We performed a large-scale implementation and evaluation effort, publishing an open-source software package with 37 Max-Cut and Quadratic Unconstrained Binary Optimization (QUBO) heuristics along with a paper characterizing the types of problem instances where different heuristics work best.
(pdf)	An Analytics Approach to Designing Combination Chemotherapy Regimens for Cancer Dimitris Bertsimas, Allison O'Hair, Stephen Relyea, John Silberholz (2016) Management Science 62(5): 1511-1531. Winner of the 2013 William Pierskalla Best Paper Award. For many cancers, combination drug therapies are most effective. However, since often dozens of drugs are available for a cancer, there can be thousands or millions of feasible drug combinations that combine these drugs in different dosages and dosing schedules. Clearly, it is impossible to test each feasible regimen, so tools are needed to prioritize regimens. We fit machine learning models to predict the qualities of new combinations and then use optimization to design new combinations. Our work shows promising results, e.g. multiple months of added survival without added toxicity for gastric cancer treatments.
(pdf)	Tenure Analytics: Models for Predicting Research Impact Dimitris Bertsimas, Erik Brynjolfsson, Shachar Reichman, John Silberholz (2015) Operations Research 63(6): 1246-1261. We study how early-career bibliometric data of scholars in the operations community predicts later-career outcomes.

Other Works

	Time to first onset of chest binding-related symptoms in transgender adults Sarah Peitzmeier, John Silberholz, Ivy H. Gardner, Jamie Weinand, Kimberlynn Acevedo (2021) Pediatrics 147(3): e20200728..
(pdf)	Optimal COVID-19 Containment Strategies: Evidence Across Multiple Mathematical Models Hyun-Soo Ahn, John Silberholz, Xueze Song, Xiaoyu Wu (2021) Working Paper.
(pdf)	Measuring Utility and Speculation in Blockchain Tokens John Silberholz, Di (Andy) Wu (2021) Working Paper.
(pdf) (site)	Clinical Benefit, Toxicity and Cost of Metastatic Breast Cancer Therapies: Systematic Review and Meta-analysis John Silberholz, Dimitris Bertsimas, Linda Vahdat (2019) Breast Cancer Research and Treatment 176(3): 535-543..
	Computational Comparison of Metaheuristics John Silberholz, Bruce Golden, Swati Gupta, Xinyu Wang (2019) Handbook of Metaheuristics (M. Gendreau and J. Potvin, eds.), Springer: 581-604.
(pdf)	An Applied Informatics Decision Support Tool for Mortality Predictions in Cancer Patients Dimitris Bertsimas, Jack Dunn, Colin Pawlowski, John Silberholz, Alexander Weinstein, Ying Daisy Zhuo, Eddy Chen, Aymen Elfiky (2018) JCO Clinical Cancer Informatics 2, 1-11.
(pdf)	Optimal healthcare decision making under multiple mathematical models: Application in prostate cancer screening Dimitris Bertsimas, John Silberholz, Tom Trikalinos (2018) Health Care Management Science 21(1): 105-118.
(pdf) (materials)	A Course on Advanced Software Tools for Operations Research and Analytics Iain Dunning, Vishal Gupta, Angela King, Jerry Kung, Miles Lubin, John Silberholz (2015) INFORMS Transactions on Education 15(2): 169-179.
(pdf)	Empirical Analysis of the Effects of Residents on Emergency Department Treatment Times David Anderson, Bruce Golden, John Silberholz, Mike Harrington, Jon Mark Hirshon (2013) IIE Transactions on Healthcare Systems Engineering 3(3): 171-180.
(pdf)	The Impact of the Residency Teaching Model on the Efficacy of the Emergency Department at an Academic Center John Silberholz, David Anderson, Bruce Golden, Mike Harrington, Jon Mark Hirshon (2013) Socio-Economic Planning Sciences 47(3): 183-190.
(pdf) (instances)	Comparison of Heuristics for the Colorful Traveling Salesman Problem John Silberholz, Andrea Raiconi, Raffaele Cerulli, Monica Gentili, Bruce Golden, Si Chen (2013) International Journal of Metaheuristics 2(2): 141-173.
(pdf)	Black Hole Simulations with CUDA Frank Herrmann, John Silberholz, Manuel Tiglio (2011) GPU Computing Gems Emerald Edition (W. Hwu, ed.), Morgan Kaufmann: 103-111.
(pdf)	Comparison of Metaheuristics John Silberholz, Bruce Golden (2010) Handbook of Metaheuristics (M. Gendreau and J. Potvin, eds.), Springer: 625-640.
(pdf)	Statistical constraints on binary black hole inspiral dynamics Chad Galley, Frank Herrmann, John Silberholz, Manuel Tiglio, Gustavo Guerberoff (2010) Classical and Quantum Gravity 27(24): 245007.
(pdf)	Integrating Post-Newtonian Equations on Graphics Processing Units Frank Herrmann, John Silberholz, Matias Bellone, Gustavo Guerberoff, Manuel Tiglio (2010) Classical and Quantum Gravity 27(3): 032001.
(pdf) (instances)	The Effective Application of a New Approach to the Generalized Orienteering Problem John Silberholz, Bruce Golden (2010) Journal of Heuristics 16(3): 393-415. Winner of the 2010 INFORMS Undergraduate Operations Research Prize.
(pdf) (instances)	Comparison of Heuristics for Solving the GMLST Problem Yiwei Chen, Namrata Cornick, Andrew O. Hall, Ritvik Sahajpal, John Silberholz, Inbal Yahav, Bruce Golden (2008) Proceedings of the 9th INFORMS Telecommunications Conference: 191-217.
(pdf) (instances)	The Generalized Traveling Salesman Problem: A New Genetic Algorithm Approach John Silberholz, Bruce Golden (2007) Proceedings of the 10th INFORMS Computing Society Conference: 165-181.

Last modified: June 7, 2025