Yang, Yuwen
(2020)
Optimal Design and Operation of WHO-EPI Vaccine Distribution Chains.
Doctoral Dissertation, University of Pittsburgh.
(Unpublished)
Abstract
Vaccination has been proven to be the most effective method to prevent infectious diseases and in 1974 the World Health Organization (WHO) established the Expanded Programme on Immunization (EPI) to provide universal access to all important vaccines for all children, with a special focus on underserved low- and middle-income countries. However, there are still roughly 20 million infants worldwide who lack access to routine immunization services and remain at risk, and millions of additional deaths could be avoided if global vaccination coverage could improve. The broad goal of this research is to optimize the design and operation of the WHO-EPI vaccine distribution chain in these underserved low- and middle-income countries. We first present a network design problem for a general WHO-EPI vaccine distribution network by developing a mathematical model that formulates the network design problem as a mixed integer program (MIP). We then present three algorithms for typical problems that are too large to be solved using commercial MIP software. We test the algorithms using data derived from four different countries in sub-Saharan Africa and show that with our final algorithm, high-quality solutions are obtained for even the largest problems within a few minutes. We then discuss the problem of outreach to remote population centers when resources are limited and direct clinic service is unavailable. A set of these remote population centers is chosen, and over an appropriate planning period, teams of clinicians and support personnel are sent from a depot to set up mobile clinics at these locations to vaccinate people there and in the immediate surrounding area. We formulate the problem of designing outreach efforts as an MIP that is a combination of a set covering problem and a vehicle routing problem. We then incorporate uncertainty to study the robustness of the worst-case solutions and the related issue of the value of information. Finally, we study a variation of the outreach problem that combines Set Covering and the Traveling Salesmen Problem and provides an MIP formulation to solve the problem. Motivated by applications where the optimal policy needs to be updated on a regular basis and where repetitively solving this via MIP can be computationally expensive, we propose a machine learning approach to effectively deal with this problem by providing an opportunity to learn from historical optimal solutions that are derived from the MIP formulation. We also present a case study on outreach operations and provide numerical results. Our results show that while the novel machine learning based mechanism generates high quality solution repeatedly for problems that resemble instances in the training set, it does not generalize as well on a different set of optimization problems. These mixed results indicate that there are promising research opportunities to use machine learning to achieve tractability and scalability.
Share
| Citation/Export: |
|
| Social Networking: |
|
Details
| Item Type: |
University of Pittsburgh ETD
|
| Status: |
Unpublished |
| Creators/Authors: |
|
| ETD Committee: |
|
| Date: |
28 September 2020 |
| Date Type: |
Publication |
| Defense Date: |
15 July 2020 |
| Approval Date: |
28 September 2020 |
| Submission Date: |
20 July 2020 |
| Access Restriction: |
No restriction; Release the ETD for access worldwide immediately. |
| Number of Pages: |
170 |
| 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: |
Vaccines; Supply Chain; Mixed integer programming; Machine Learning |
| Date Deposited: |
28 Sep 2020 20:12 |
| Last Modified: |
28 Sep 2020 20:12 |
| URI: |
http://d-scholarship.pitt.edu/id/eprint/39390 |
Metrics
Monthly Views for the past 3 years
Plum Analytics
Actions (login required)
 |
View Item |
![[feed]](/images/feed-icon-32x32.png){kind=icon}