Sun, Ruichen
(2019)
Optimizing Donor Human Milk Bank Operations.
Doctoral Dissertation, University of Pittsburgh.
(Unpublished)
Abstract
Human breast milk provides nutritional and medicinal benefits that are important to infants, particularly those who are premature or ill. Donor human milk (DHM) - collected, processed, and dispensed via milk banks - is the standard of care for premature and unhealthy infants whose mothers cannot provide an adequate supply of their own milk. DHM undergoes several processes at milk banks, including the pooling of milk from different donors (to meet macronutrient requirements across different milk types), the batching of pooled milk, and its assignment to different types of pasteurizers, before being dispensed to hospitals and outpatients.
This dissertation concentrates on optimal decision making in non-profit milk banks. First, we formulate two integer programs that optimize the batching-pasteurizing decisions and the integrated pooling-batching-pasteurizing decisions, respectively, and establish valid inequalities for both. Our numerical results suggests a significant improvement in pooling performance and production utility, and a modest reduction in labor. The integrated model is in use at Mothers' Milk Bank of North Texas and has effectively resolved their chronic production imbalance. Second, we reconsider pooling decisions with the contamination uncertainty possibly caused by individual deposits and assume that every deposit can be pooled with others. We formulate a two-stage stochastic integer program to maximize the expected production utility and develop an L-shaped algorithm without solving exponentially many subproblems. The numerical study suggests a possible expected production utility improvement and provides guidelines for the circumstances under which our model outperforms current practice. Finally, motivated by the powerful modeling abilities of multistage mixed-integer stochastic programs, we consider a specially structured class whose integrality restrictions can be relaxed, namely totally umimodular multistage stochastic programs. We establish several sufficient conditions for the totally unimodularity. We also discuss the examples in the literature and the potential applications arising in milk banks (e.g., inventory management, DHM dispensing network).
Share
Citation/Export: |
|
Social Networking: |
|
Details
Item Type: |
University of Pittsburgh ETD
|
Status: |
Unpublished |
Creators/Authors: |
|
ETD Committee: |
Title | Member | Email Address | Pitt Username | ORCID |
---|
Committee Chair | Maillart, Lisa | maillart@pitt.edu | | | Committee CoChair | Schaefer, Andrew | andrew.schaefer@rice.edu | | | Committee Member | Prokopyev, Oleg | | | | Committee Member | Zeng, Bo | | | | Committee Member | Bogen, Debra | | | |
|
Date: |
19 June 2019 |
Date Type: |
Publication |
Defense Date: |
17 January 2019 |
Approval Date: |
19 June 2019 |
Submission Date: |
3 April 2019 |
Access Restriction: |
5 year -- Restrict access to University of Pittsburgh for a period of 5 years. |
Number of Pages: |
118 |
Institution: |
University of Pittsburgh |
Schools and Programs: |
Swanson School of Engineering > Industrial Engineering |
Degree: |
PhD - Doctor of Philosophy |
Thesis Type: |
Doctoral Dissertation |
Refereed: |
Yes |
Uncontrolled Keywords: |
Milk bank, donor human milk processing, integer programming, stochastic mixed integer programming, totally unimodularity, multistage optimization |
Date Deposited: |
19 Jun 2019 15:16 |
Last Modified: |
19 Jun 2024 05:15 |
URI: |
http://d-scholarship.pitt.edu/id/eprint/36038 |
Metrics
Monthly Views for the past 3 years
Plum Analytics
Actions (login required)
|
View Item |