T his work examines the process of admission to a hospital's intensive care unit (ICU). ICUs currently lack systematic admission criteria, largely because the impact of ICU admission on patient outcomes has not been well quantified. This makes evaluating the performance of candidate admission strategies difficult. Using a large patient-level data set of more than 190,000 hospitalizations across 15 hospitals, we first quantify the cost of denied ICU admission for a number of patient outcomes. We use hospital operational factors as instrumental variables to handle the endogeneity of the admission decisions and identify important specification issues that are required for this approach to be valid. Using the quantified cost estimates, we then provide a simulation framework for evaluating various admission strategies' performance. By simulating a hospital with 21 ICU beds, we find that we could save about $1.9 million per year by using an optimal policy based on observables designed to reduce readmissions and hospital length of stay. We also discuss the role of unobserved patient factors, which physicians may discretionarily account for when making admission decisions, and show that including these unobservables could result in a more than threefold increase in benefits compared to just optimizing the policy over the observable patient factors.

This work examines the impact of discharge decisions under uncertainty in a capacity-constrained high risk setting: the intensive care unit (ICU). New arrivals to an ICU are typically very high priority patients and, should the ICU be full upon their arrival, discharging a patient currently residing in the ICU may be required to accommodate a newly admitted patient. Patients so discharged risk physiologic deterioration which might ultimately require readmission; models of these risks are currently unavailable to providers. These readmissions in turn impose an additional load on the capacity-limited ICU resources.We study the impact of several different ICU discharge strategies on patient mortality and total readmission load. We focus on discharge rules that prioritize patients based on some measure of criticality assuming the availability of a model of readmission risk. We use empirical data from over 5000 actual ICU patient flows to calibrate our model. The empirical study suggests that a predictive model of the readmission risks associated with discharge decisions, in tandem with simple index policies of the type proposed can provide very meaningful throughput gains in actual ICUs while at the same time maintaining, or even improving upon, mortality rates. We explicitly provide a discharge policy that accomplishes this. In addition to our empirical work, we conduct a rigorous performance analysis for the family of discharge policies we consider.We show that our policy is optimal in certain regimes, and is otherwise guaranteed to incur readmission related costs no larger than a factor of (ρ + 1) of an optimal discharge strategy, whereρ is a certain natural measure of system utilization.

Mainstream queueing models are frequently employed in modeling healthcare delivery in a number of settings, and further used in making operational decisions for the same. The vast majority of these queueing models assume that the service requirements of a job are independent of the state of the queue upon its arrival. In a healthcare setting, this assumption is equivalent to ignoring the effects of delay experienced by a patient awaiting care. However, it is only natural to conjecture that long delays may have adverse effects on patient outcomes and can potentially lead to longer lengths of stay (LOS) when the patient ultimately does receive care. At a very coarse level, prior research confirms these natural conjectures. This work sets out to understand these delay issues from an operational perspective. In particular, using data of nearly 6,000 Emergency Department (ED) visits, we use an instrumental variable approach to empirically measure how congestion in the Intensive Care Unit (ICU) can lead to delays in boarding from the ED to the ICU and measure the impact on the patient's ICU LOS.Capturing these empirically observed effects in a queueing model is challenging as the effect introduces potentially long range correlations in service and inter-arrival times. As such, we consider the problem of how to incorporate these measured delay effects into a queueing model and characterize approximations to various quantities of interest when the service time of a job is adversely impacted by the delay experienced by that job. Our findings suggest that this delay effect can be substantial and ignoring it when using queueing models to model healthcare delivery systems may result in significant under-provisioning.

No abstract

In a number of service systems, there can be substantial latitude to vary service rates. However, although speeding up service rate during periods of congestion may address a present congestion issue, it may actually exacerbate the problem by increasing the need for rework. We introduce a state-dependent queuing network where service times and return probabilities depend on the "overloaded" and "underloaded" state of the system. We use a fluid model to examine how different definitions of "overload" affect the long-term behavior of the system and provide insight into the impact of using speedup. We identify scenarios where speedup can be helpful to temporarily alleviate congestion and increase access to service. For such scenarios, we provide approximations for the likelihood of speedup to service. We also identify scenarios where speedup should never be used; moreover, in such a situation, an interesting bi-stability arises, such that the system shifts randomly between two equilibria states. Hence, our analysis sheds light on the potential benefits and pitfalls of using speedup when the subsequent returns may be unavoidable.

Abstract-We study scheduling of multimedia traffic on the downlink of a wireless communication system. We examine a scenario where multimedia packets are associated with strict deadlines and are equivalent to lost packets if they arrive after their associated deadlines. Lost packets result in degradation of playout quality at the receiver, which is quantified in terms of the "distortion cost" associated with each packet. Our goal is to design a scheduler which minimizes the aggregate distortion cost over all receivers. We study the scheduling problem in a dynamic programming (DP) framework. Under well justified modeling reductions, we extensively characterize structural properties of the optimal control associated with the DP problem. We leverage these properties to design a low-complexity Channel, Deadline, and Distortion (CD 2 ) aware heuristic scheduling policy amenable to implementation in real wireless systems. We evaluate the performance of CD 2 via trace-driven simulations using H.264/MPEG-4 AVC coded video. Our experimental results show that CD 2 comfortably outperforms benchmark schedulers like earliest deadline first (EDF) and best channel first (BCF). CD 2 achieves these performance gains by using the knowledge of packet deadlines, wireless channel conditions, and application specific information (per-packet distortion costs) in a systematic and unified way for multimedia scheduling.Index Terms-Wireless networks, video streaming, packet scheduling, dynamic programming.

This paper presents a general class of dynamic stochastic optimization problems we refer to as Stochastic Depletion Problems. A number of challenging dynamic optimization problems of practical interest are stochastic depletion problems. Optimal solutions for such problems are difficult to obtain, both from a pragmatic computational perspective as also from a theoretical perspective. As such, simple heuristics are highly desirable. We isolate two simple properties that, if satisfied by a problem within this class, guarantee that a myopic policy incurs a performance loss of at most 50 % relative to the optimal adaptive control policy for that problem. We are able to verify that these two properties are satisfied for several interesting families of stochastic depletion problems and as a consequence identify efficient near-optimal control policies for a number of interesting dynamic stochastic optimization problems.

In order for a patient to be discharged from a hospital unit, a physician must first perform a physical examination and review the pertinent medical information to determine that the patient is stable enough to be transferred to a lower level of care or be discharged home. Requiring an inspection of a patient's 'readiness for discharge' introduces an interesting dynamic where patients may occupy a bed longer than medically necessary. Motivated by this phenomenon, we introduce a queueing system with time-varying arrival rates in which servers who have completed service cannot be released until an inspection occurs. We examine how such a dynamic impacts common system measures such as stability, expected number of customers in the system, probability of waiting and expected waiting time. Leveraging insights from an infinite-server model, we're able to optimize the timing of inspections and find via theoretical and numerical analysis that 1) optimizing a single inspection time could lead to significant improvements in system performance when the amplitude of the arrival rate function is large, 2) multiple inspections should be uniformly distributed throughout the day, and 3) the marginal improvements of adding additional inspection times is decreasing.