Dmitry krass | University of Toronto (original) (raw)

Papers by Dmitry krass

Health Care Management Science, Jan 12, 2023

Developing rapid tools for early detection of viral infection is crucial for pandemic containment... more Developing rapid tools for early detection of viral infection is crucial for pandemic containment. This is particularly crucial when testing resources are constrained and/or there are significant delays until the test results are available-as was quite common in the early days of Covid-19 pandemic. We show how predictive analytics methods using machine learning algorithms can be combined with optimal pre-test screening mechanisms, greatly increasing test efficiency (i.e., rate of true positives identified per test), as well as to allow doctors to initiate treatment before the test results are available. Our optimal test admission policies account for imperfect accuracy of both the medical test and the model prediction mechanism. We derive the accuracy required for the optimized admission policies to be effective. We also show how our policies can be extended to re-testing high-risk patients, as well as combined with pool testing approaches. We illustrate our techniques by applying them to a large data reported by the Israeli Ministry of Health for RT-PCR tests from March to September 2020. Our results demonstrate that in the context of the Covid-19 pandemic a pre-test probability screening tool with conventional RT-PCR testing could have potentially increased efficiency by several times, compared to random admission control. Keywords Data analytics • Machine learning • Predictive analytics • Optimal test admission policies Ramy Elitzur, Dmitry Krass and Eyal Zimlichman contributed equally to this work.

Annals of Operations Research, Dec 1, 1995

The problem of locating flow-intercepting facilities on a network with probabilistic customer flo... more The problem of locating flow-intercepting facilities on a network with probabilistic customer flows and with facility set-up costs is studied in this paper. Two types of models are investigated, namely the double-counting model and the no-double-counting model (double-counting refers to ...

Management Science, Apr 1, 2019

This paper shows that operational flexibility interacting with informational uncertainty may lead... more This paper shows that operational flexibility interacting with informational uncertainty may lead to truthful information exchange in equilibrium even when the communication is nonbinding and unverifiable, i.e., “cheap talk.” We consider a model consisting of a manufacturer releasing a new product with uncertain release date and demand, and a retailer who must determine the allocation of limited capacity between a preexisting third-party product and the manufacturer’s new product that may or may not be released on time. The manufacturer has a private forecast about the likelihood of the product release and/or about the demand, which he shares (either truthfully or not) with the retailer. We show that under the “traditional” supply chain structure (one-time opportunity to order) no truthful equilibrium can emerge. However, if (1) the supply chain structure allows for postponement, i.e., the ability to delay orders at a certain cost by the retailer, and (2) the manufacturer has informational uncertainty about the retailer’s capacity, then truthful information exchange may emerge in equilibrium, where the manufacturer transmits his true forecast and the retailer treats the transmission as truthful. The genesis of this effect is preference reversal, where the manufacturer is not sure which way to distort the forecast to best motivate the retailer to wait for the new product. Thus, we show that a truth-revealing mechanism can emerge from a relatively rich setup featuring two-sided information asymmetry interacting with postponement. This paper was accepted by Gad Allon, operations management.

International Regional Science Review, Mar 21, 2014

Spatial planners often make “comprehensive” decisions on the location of public service facilitie... more Spatial planners often make “comprehensive” decisions on the location of public service facilities by using the concept of urban hierarchy: population centers at the upper level of the hierarchy (typically large cities) get the highest level facilities, such as specialized hospitals and universities, while the centers at the lower levels of hierarchy get lower-level facilities. Intuitively, this suggests that there should be a link between urban hierarchy designations and population dynamics in future periods. This link must be taken into account in planning decisions, as it suggests that today’s decisions affect tomorrow’s demand for services. Indeed, this link was assumed in a previously published planning model. Yet, no direct evidence of such a link appears in the literature. The primary goal of this article is to fill this gap by using the census data for Portugal for the period 1991–2001 and the changes in the urban hierarchy that were implemented during 1980s and early 1990s. While our results support the link between urban hierarchy designations and population dynamics that has been assumed in previously published work, the mechanism describing this link appears to be somewhat different from the one postulated previously. Several extensions and directions for future work are also discussed.

Iie Transactions, Mar 3, 2009

A location problem with stochastic demand and congestion where mobile servers respond to service ... more A location problem with stochastic demand and congestion where mobile servers respond to service calls originating from nodes is considered. The problem is of the set-covering type: only servers within the coverage radius of the demand-generating node may respond to a call. The service level constraint requires that at least one server must be available to respond to an arriving call, with some prespecified probability. The objective is to minimize the total number of servers. It is shown that earlier models quite often overestimate servers' availability and thus may lead to infeasible solutions (i.e., solutions that fail to satisfy the service level constraint). System stability conditions and lower bounds on system availability are developed by analyzing the underlying partially accessible queueing system. These lead to the development of two new models for which feasibility is guaranteed. Simulation-based computational experiments show that the proposed models achieve feasibility without significantly increasing the total number of servers. [Supplementary materials are available for this article. Go to the publisher's online edition of IIE Transactions for the following free supplemental resource: Appendix of Tables of Computational Results for Section 7.

Communications in statistics, 1997

Flow-intercepting facilities are facilities such as gas stations and advertising billboards, whic... more Flow-intercepting facilities are facilities such as gas stations and advertising billboards, which provide “service” to passing-by customer flows. Customers that pass by such facilities are said to be intercepted. A typical objective in locating these facilities is the maximization of the total customer flow intercepted. We consider two new models for locating flow-intercepting facilities on a network. Both models use only information on the fraction of customers travelling from any node to any adjacent nodes and the fraction of customers that start their travel at any given node. This information is more practical to obtain than the complete path flow information required by existing models. The trade-off for using this reduced data set is the loss of equivalence with the path-flow based models. However, a set of computational experiments shows that the proposed models can be regarded as an approximation of the existing ones. In the first model, repeat visits by a customer to the facilities (e.g., billbo...

Manufacturing & Service Operations Management, Jul 1, 2008

W e analyze the problem of optimal location of a set of facilities in the presence of stochastic ... more W e analyze the problem of optimal location of a set of facilities in the presence of stochastic demand and congestion. Customers travel to the closest facility to obtain service; the problem is to determine the number, locations, and capacity of the facilities. Under rather general assumptions (spatially distributed continuous demand, general arrival and service processes, and nonlinear location and capacity costs) we show that the problem can be decomposed, and construct an efficient optimization algorithm. The analysis yields several insights, including the importance of equitable facility configurations (EFCs), the behavior of optimal and near-optimal capacities, and robust class of solutions that can be constructed for this problem.

arXiv (Cornell University), Mar 30, 2023

Business Process Simulation (BPS) is an approach to analyze the performance of business processes... more Business Process Simulation (BPS) is an approach to analyze the performance of business processes under different scenarios. For example, BPS allows us to estimate what would be the cycle time of a process if one or more resources became unavailable. The starting point of BPS is a process model annotated with simulation parameters (a BPS model). BPS models may be manually designed, based on information collected from stakeholders and empirical observations, or automatically discovered from execution data. Regardless of its origin, a key question when using a BPS model is how to assess its quality. In this paper, we propose a collection of measures to evaluate the quality of a BPS model w.r.t. its ability to replicate the observed behavior of the process. We advocate an approach whereby different measures tackle different process perspectives. We evaluate the ability of the proposed measures to discern the impact of modifications to a BPS model, and their ability to uncover the relative strengths and weaknesses of two approaches for automated discovery of BPS models. The evaluation shows that the measures not only capture how close a BPS model is to the observed behavior, but they also help us to identify sources of discrepancies.

Annals of Operations Research, Apr 1, 2005

ABSTRACT

IEEE Transactions on Automatic Control, 1995

In this paper we address the following basic feasibility problem for infinite-horizon Markov deci... more In this paper we address the following basic feasibility problem for infinite-horizon Markov decision processes (MDP's): can a policy be found that achieves a specified value (target) of the long-run limiting average reward at a specified probability level (percentile)? Related optimization problems of maximizing the target for a specified percentile and vice versa are also considered. We present a complete (and discrete) classification of both the maximal achievable target levels and of their corresponding percentiles. We also provide an algorithm for computing a deterministic policy corresponding to any feasible target-percentile pair. Next we consider similar problems for an MDP with multiple rewards and/or constraints. This case presents some difficulties and leads to several open problems. An LP-based formulation provides constructive solutions for most cases.

2022 Winter Simulation Conference (WSC)

Ensuring feasibility in location problems with stochastic

In this paper, we analyze how well a machine can solve a general problem in queueing theory. To a... more In this paper, we analyze how well a machine can solve a general problem in queueing theory. To answer this question, we use a deep learning model to predict the stationary queue-length distribution of an M/G/1M/G/1M/G/1 queue (Poisson arrivals, general service times, one server). To the best of our knowledge, this is the first time a machine learning model is applied to a general queueing theory problem. We chose M/G/1M/G/1M/G/1 queue for this paper because it lies "on the cusp" of the analytical frontier: on the one hand exact solution for this model is available, which is both computationally and mathematically complex. On the other hand, the problem (specifically the service time distribution) is general. This allows us to compare the accuracy and efficiency of the deep learning approach to the analytical solutions. The two key challenges in applying machine learning to this problem are (1) generating a diverse set of training examples that provide a good representation of a "gen...

Production and Operations Management, 2022

Discrete Applied Mathematics, 2021

SSRN Electronic Journal, 2020

Assessment of the anterior chamber angle (ACA) is an essential part of the ophthalmological exami... more Assessment of the anterior chamber angle (ACA) is an essential part of the ophthalmological examination. It is intrinsically related to the diagnosis and treatment of glaucoma and has a role in its prevention. Although slit-lamp gonioscopy is considered the gold-standard technique for ACA evaluation, its poor reproducibility and the long learning curve are well-known shortcomings. Several new imaging techniques for angle evaluation have been developed in the recent years. However, whether these instruments may replace or not gonioscopy in everyday clinical practice remains unclear. This review summarizes the last findings in ACA evaluation, focusing on new instruments and their application to the clinical practice. Special attention will be given to the comparison between these new techniques and traditional slit-lamp gonioscopy. Whereas ultrasound biomicroscopy and anterior segment optical coherence tomography provide quantitative measurements of the anterior segment's structures, new gonio-photographic systems allow for a qualitative assessment of angle findings, similarly to gonioscopy. Recently developed deep learning algorithms provide an automated classification of angle images, aiding physicians in taking faster and more efficient decisions. Despite new imaging techniques made analysis of the ACA more objective and practical, the ideal method for ACA evaluation has still to be determined.