Skip to Main Content
Frequently Asked Questions
Submit an ETD
Global Search Box
Need Help?
Keyword Search
Participating Institutions
Advanced Search
School Logo
Files
File List
39308.pdf (3.64 MB)
ETD Abstract Container
Abstract Header
Managing a Hybrid Oral Medication Distribution System in a Pediatric Hospital: A Machine Learning Approach
Author Info
Thaibah, Hilal
ORCID® Identifier
http://orcid.org/0000-0002-0793-5445
Permalink:
http://rave.ohiolink.edu/etdc/view?acc_num=ucin1626356839363113
Abstract Details
Year and Degree
2021, PhD, University of Cincinnati, Pharmacy: Pharmaceutical Sciences.
Abstract
Background: The efficient and safe delivery of medications represent a challenge, particularly within in-patient hospital pharmacies. Different medication distribution systems have evolved and were deployed to meet this challenge. Centralized, decentralized, and hybrid medication distribution systems comprise the main medication distribution systems. Moreover, the optimization of these systems is an ongoing process, and there is a need for innovative managing tools that align with these distribution systems. Objectives: This dissertation aimed at managing the oral medication distribution system within a Hybrid Medication Distribution System (HMDS) in a pediatric hospital using Artificial Neural Network (ANN) modeling. This aim was sought in two specific aims: to develop and validate an ANN model to determine the assignment of oral medications to either centralized or decentralized distribution system within the HMDS, and to evaluate the expandability of the developed ANN model in managing these assignments in another high-throughput nursing unit. Methods: Retrospective data analyses were performed using a one-year dispensing data from the Cincinnati Children’s Hospital Medical Center between January 1st – December 31st of 2018. A subset of the oral medication dispense transactions was obtained, and two nursing units were selected to carry out the analyses for the two aims. The ANN model was developed and validated, and the model’s quality metrics were obtained. The expandability of the developed ANN model, as well as the retraining of the model, were evaluated. Results: There was a total of 82,961 oral medication dispense transactions in aim 1 dispensing unit. The centralized distribution accounted for 54.18% of the oral medication dispense transactions, and 45.82% pertained to the decentralized distribution. The ANN model was developed, and cross-validated using 75% training (n= 62,002) and 25% testing (n= 20,667) data. The ANN training model had a 98% accuracy, 2% misclassification, 98% precision, and 97% recall rates. The developed ANN model was then applied to the testing data and resulted in 97% accuracy, 3% misclassification, 98% precision, and 97% recall rates. A total of 63,829 oral dispense transactions were found in the aim 2 dispensing unit. The utilization of the HMDS was found at 53.09% for the centralized distribution and 45.91% for the decentralized distribution. The developed ANN model in aim 1 was applied on the dispensing data from aim 2 and had an accuracy rate of 85%, misclassification rate of 15%, and precision rate of 95%. Furthermore, the retrained ANN model produced better overall quality metrics. It showed 91% accuracy, 9% misclassification, 88% precision, and 93% recall rates on training and testing datasets. Conclusion: The ANN modeling was applicable in determining the assignment of the oral medication dispenses to either a centralized or decentralized distribution system within the HMDS in a pediatric hospital. The expandability of the developed ANN model could be achieved with caution, and retraining the model on the data pertaining to a different nursing unit was warranted. Using a systematic approach in developing and validating the ANN model, including the assessment of relevant variables, could help in obtaining a model of high quality.
Committee
Alex Lin, Ph.D. (Committee Chair)
Jianfei (Jeff) Guo, Ph.D. (Committee Member)
Ana Hincapie, Ph.D. (Committee Member)
Marepalli Rao, Ph.D. (Committee Member)
Bingfang Yan, D.V.M. Ph.D. (Committee Member)
Pages
106 p.
Subject Headings
Pharmaceuticals
Keywords
Medication Distribution Systems
;
Pharmacy Operation
;
Hospital Pharmacy Management
;
Machine Learning
;
Artificial Neural Network
Recommended Citations
Refworks
EndNote
RIS
Mendeley
Citations
Thaibah, H. (2021).
Managing a Hybrid Oral Medication Distribution System in a Pediatric Hospital: A Machine Learning Approach
[Doctoral dissertation, University of Cincinnati]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=ucin1626356839363113
APA Style (7th edition)
Thaibah, Hilal.
Managing a Hybrid Oral Medication Distribution System in a Pediatric Hospital: A Machine Learning Approach.
2021. University of Cincinnati, Doctoral dissertation.
OhioLINK Electronic Theses and Dissertations Center
, http://rave.ohiolink.edu/etdc/view?acc_num=ucin1626356839363113.
MLA Style (8th edition)
Thaibah, Hilal. "Managing a Hybrid Oral Medication Distribution System in a Pediatric Hospital: A Machine Learning Approach." Doctoral dissertation, University of Cincinnati, 2021. http://rave.ohiolink.edu/etdc/view?acc_num=ucin1626356839363113
Chicago Manual of Style (17th edition)
Abstract Footer
Document number:
ucin1626356839363113
Download Count:
215
Copyright Info
© 2021, all rights reserved.
This open access ETD is published by University of Cincinnati and OhioLINK.