2026-09-20T07:56:06Zhttps://ecommons.cornell.edu/server/oai/request

oai:ecommons.cornell.edu:1813/704142026-05-15T19:47:20Zcom_1813_35col_1813_47

Cazer, Casey Lu Grohn, Yrjo Booth, James Ivanek Miojevic, Renata Marquis, Helene 2020-08-10T20:24:11Z 2021-06-08T06:00:13Z 2020-05 ProQuest Submission ID: 11946 ProQuest Publication ID: 27957704 https://hdl.handle.net/1813/70414 https://doi.org/10.7298/gktf-2s14 13254297 244 pages Supplemental file(s) description: Supplementary Table 5.3. Antimicrobial resistant bacteria existed before the medical use of antimicrobials but increasing antimicrobial use has escalated the development and spread of resistant bacteria in human and animal populations. Resistant bacteria and resistance genes spread between individuals through direct and indirect pathways, such that antimicrobial use in one population can result in increased morbidity and mortality from resistant bacteria in another population. In particular, there is a risk of resistant foodborne bacteria due to antimicrobial use in livestock, but this risk is challenging to quantify and likely varies with bacterial species, antimicrobial, and host species. Data of varying quality have been collected on antimicrobial use, pharmacokinetics, pharmacodynamics, resistance genes, and resistance phenotypes over several decades. Novel and innovative methodologies are required to maximize the utility of these data and improve our understanding of antimicrobial resistance ecology. In this dissertation, I first gather all available information on the impact of tylosin on beef cattle enteric bacteria with a systematic review. These data are used to inform and validate a meta-population mathematical model of tylosin pharmacokinetics and pharmacodynamics. The model uses previously published data to estimate the change in macrolide-resistant enterococci within cattle and in the feedlot environment during and after tylosin administration. The mathematical model and systematic review reveal a limited impact of tylosin on macrolide-resistant enterococci and other potential foodborne pathogens. Further along the foodborne transmission chain, the National Antimicrobial Resistance Monitoring System collects data on resistant bacteria at slaughterhouses, on retail meats, and from human infections. Multidrug resistant bacteria are associated with more severe health outcomes yet are more challenging to analyze because the number of multidrug resistance phenotypes grows exponentially with the number of antimicrobials considered. I present the first application of association rule mining, a machine learning methodology, to analyze multidrug resistance in chicken-associated Escherichia coli. Network analytics and the association rules demonstrated that E. coli from chicken carcasses and chicken meat had similar multidrug resistance phenotypes, despite the high variability in the dataset. Finally, I refine and expand the association mining techniques and apply them to hospital surveillance data, demonstrating that association mining can identify and quantify clinically important multidrug resistance in small, surveillance datasets. These studies establish the importance of using complementary analytic techniques to maximize the value of existing and future antimicrobial resistance data. Machine learning promises to substantially change the approach to antimicrobial resistance data, and may become necessary as the amount of antimicrobial resistance data collected expands, but my work on association mining indicates that efforts to validate and increase the rigor of these methods are essential. The merging of systematic reviews with mathematical models, and the union of machine learning with traditional statistics and visualization, will improve our understanding of antimicrobial resistance ecology. en Attribution-NonCommercial-NoDerivatives 4.0 International https://creativecommons.org/licenses/by-nc-nd/4.0/ antimicrobial resistance machine learning mathematical model multi-drug resistance one health MODELING AND MINING ANTIMICROBIAL RESISTANCE IN HUMAN AND ANIMAL POPULATIONS dissertation or thesis https://catalog.library.cornell.edu/catalog/13254297 application/pdf application/vnd.ms-excel Biomedical and Biological Sciences Cornell University Doctor of Philosophy Ph. D., Biomedical and Biological Sciences https://hdl.handle.net/1813/59810 Publication Cornell Theses and Dissertations Cazer, Casey Lu 5893a6ea-7af3-41d7-abc6-04bcd26ab5df open.access open.access <Publication xmlns="https://www.openaire.eu/cerif-profile/1.1/" id="03b1ae05-cca8-46a5-841d-ff149c7cea8c"> <Type xmlns="https://www.openaire.eu/cerif-profile/vocab/COAR_Publication_Types">http://purl.org/coar/resource_type/c_1843</Type> <Language>en</Language> <Title>MODELING AND MINING ANTIMICROBIAL RESISTANCE IN HUMAN AND ANIMAL POPULATIONS</Title> <PublishedIn> <Publication> </Publication> </PublishedIn> <PublicationDate>2020-05</PublicationDate> <DOI>https://doi.org/10.7298/gktf-2s14</DOI> <Authors> <Author> <DisplayName>Cazer, Casey Lu</DisplayName> <Affiliation> <OrgUnit> </OrgUnit> </Affiliation> </Author> </Authors> <Editors> </Editors> <Publishers> <Publisher> <OrgUnit /> </Publisher> </Publishers> <License>https://creativecommons.org/licenses/by-nc-nd/4.0/</License> <Keyword>antimicrobial resistance</Keyword> <Keyword>machine learning</Keyword> <Keyword>mathematical model</Keyword> <Keyword>multi-drug resistance</Keyword> <Keyword>one health</Keyword> <Abstract>Antimicrobial resistant bacteria existed before the medical use of antimicrobials but increasing antimicrobial use has escalated the development and spread of resistant bacteria in human and animal populations. Resistant bacteria and resistance genes spread between individuals through direct and indirect pathways, such that antimicrobial use in one population can result in increased morbidity and mortality from resistant bacteria in another population. In particular, there is a risk of resistant foodborne bacteria due to antimicrobial use in livestock, but this risk is challenging to quantify and likely varies with bacterial species, antimicrobial, and host species. Data of varying quality have been collected on antimicrobial use, pharmacokinetics, pharmacodynamics, resistance genes, and resistance phenotypes over several decades. Novel and innovative methodologies are required to maximize the utility of these data and improve our understanding of antimicrobial resistance ecology. In this dissertation, I first gather all available information on the impact of tylosin on beef cattle enteric bacteria with a systematic review. These data are used to inform and validate a meta-population mathematical model of tylosin pharmacokinetics and pharmacodynamics. The model uses previously published data to estimate the change in macrolide-resistant enterococci within cattle and in the feedlot environment during and after tylosin administration. The mathematical model and systematic review reveal a limited impact of tylosin on macrolide-resistant enterococci and other potential foodborne pathogens. Further along the foodborne transmission chain, the National Antimicrobial Resistance Monitoring System collects data on resistant bacteria at slaughterhouses, on retail meats, and from human infections. Multidrug resistant bacteria are associated with more severe health outcomes yet are more challenging to analyze because the number of multidrug resistance phenotypes grows exponentially with the number of antimicrobials considered. I present the first application of association rule mining, a machine learning methodology, to analyze multidrug resistance in chicken-associated Escherichia coli. Network analytics and the association rules demonstrated that E. coli from chicken carcasses and chicken meat had similar multidrug resistance phenotypes, despite the high variability in the dataset. Finally, I refine and expand the association mining techniques and apply them to hospital surveillance data, demonstrating that association mining can identify and quantify clinically important multidrug resistance in small, surveillance datasets. These studies establish the importance of using complementary analytic techniques to maximize the value of existing and future antimicrobial resistance data. Machine learning promises to substantially change the approach to antimicrobial resistance data, and may become necessary as the amount of antimicrobial resistance data collected expands, but my work on association mining indicates that efforts to validate and increase the rigor of these methods are essential. The merging of systematic reviews with mathematical models, and the union of machine learning with traditional statistics and visualization, will improve our understanding of antimicrobial resistance ecology.</Abstract> <Access xmlns="http://purl.org/coar/access_right" > </Access> </Publication>