Hospital Stockpiling Problems with Inventory Sharing
Authors: Eric Lofgren, Anil Vullikanti
AAAI 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | 6 Experimental Results We study the HSTOCKPILE formulation on a hospital network formed by the ten largest hospitals in the state of North Carolina, as shown in Figure 3. All of these hospitals are Level I or Level II trauma centers that would reasonably be expected to provide patient care during a public health emergency. We assume a completely connected undirected network between these ten hospitals, with the cost of sharing supplies between hospitals dictated solely by the distance between them. This, in effect, creates a network with several clusters of hospitals that can share easily those within the Triangle made up of the cities of Raleigh, Durham and Chapel Hill, as well as those in the cities of Winston-Salem and Greensboro as well as more distant hospitals who will have a more difficult time obtaining supplies from other hospitals. We choose the cost Cij to be proportional to the distance d(i, j) between the hospitals, and choose Ci = mini,j d(i, j). We vary the penalty p. Epidemics are inherently stochastic and unpredictable, making the allocation of supplies a difficult problem. Stockpiling strategies must be robust to both unexpectedly severe outbreaks, such as the 2009 H1N1 influenza epidemic, as well as less severe outbreaks that may not necessitate extensive stockpiles. To simulate this varying demand, we simulate 1000 stochastic epidemics of a flu-like illness, using a variation of a Susceptible-Exposed-Infected-Recovered (SEIR) epidemic model, where infected patients are subdivided into ten compartments (Ij), indicating that the demand for their care is the responsibility of Hospital j, allocated proportionally based on a weight (κj). In one experiment, this weight was proportional to the bed-size of the hospital, and in another, the weight was assigned randomly. |
| Researcher Affiliation | Academia | Eric Lofgren and Anil Vullikanti , Biocomplexity Institute of Virginia Tech, Virginia Tech Department of Computer Science, Virginia Tech Email: {lofgrene, akumar}@vbi.vt.edu |
| Pseudocode | No | The paper presents mathematical formulations and descriptions of its model and proofs, including a linear programming objective and constraints, but it does not contain any structured pseudocode or algorithm blocks. |
| Open Source Code | No | The paper does not provide any statement or link regarding the open-sourcing of the code for the methodology described. |
| Open Datasets | No | The paper describes using a hospital network from North Carolina and simulating 1000 stochastic epidemics using an SEIR model. However, it does not provide access information (link, citation, repository) for the specific data generated or used in their experiments, nor does it refer to a standard, publicly accessible dataset. |
| Dataset Splits | No | The paper does not provide specific details on training, validation, or test dataset splits (e.g., percentages, sample counts, or predefined standard splits) needed to reproduce the data partitioning. |
| Hardware Specification | No | The paper does not provide any specific hardware details such as GPU/CPU models, processor types, or memory used for running the experiments. |
| Software Dependencies | No | The paper describes the use of an SEIR epidemic model but does not list any specific software dependencies or libraries with version numbers (e.g., 'Python 3.x', 'CPLEX 12.x') that would be required for reproducibility. |
| Experiment Setup | Yes | We choose the cost Cij to be proportional to the distance d(i, j) between the hospitals, and choose Ci = mini,j d(i, j). We vary the penalty p. ... We choose the parameters so that they correspond to pandemic-grade flu, with the reproductive number R0 (which corresponds to the expected number of secondary infections caused by any individual) between 1.7 and 2.0 (Halloran et al. 2008). Figure 4 shows the stockpile levels for the social optimum solution computed using the algorithm from Section 5 for the ratio C/p varying from 0.5 to 1. |