Barriers to population level AMR research in UK livestock and opportunities for data science

  • Jon Massey

Student thesis: Doctoral ThesisDoctor of Philosophy (PhD)


Antimicrobial resistance (AMR) presents a significant threat to animal and human health the world over. The role that animal agriculture plays in this threat is not yet fully knownbut greater knowledge may be found through analysis of routinelycollected data sets. Examples of such routine data regarding UK livestock are veterinary treatment records, clinical diagnoses, animal movements, and production records. Work towards building an open access research databank for One Health AMR research revealed barriers to using such data for research purposes, manifesting from interactions of social, legal, and technological issues. Detailed examination of three problem areas –identification and quantification of veterinary medicines usage, mismatches in conceptualisation and representation of key entities across domains, and sharing of privacy-constrained data –revealed opportunities for data science to overcome these barriers. To do sorequires consideration of the whole data lifecycle and the social, legal, and technological contexts of its activities.Drawing on existing work in the fields of natural language processing, privacy-preserving record linkage, and open data standards, novel solutionsare provided for the highlighted problem areas. Proposedsolutions are imperfect yetreadily-implementable, or more sustainable and complete yet requiring of significant engagement from key stakeholders and not fully realised during this study. A real-world example of what analysis of routine UK livestock data may provide is demonstrated througha highly predictive machine learned model of milk antibiotic residue test results, based on results from a UK-wide cohort of dairy farmers over a 3-yearperiod. Implementation of analytical tools based on this modelling work and adoption of data standards proposed in this thesis are examples of high impact outcomes from this thesis. Continuation of this impact for population-level livestock AMR research requires further study into both its data challenges and the complex social and legal domains in which they exist.
Date of Award28 Sept 2021
Original languageEnglish
Awarding Institution
  • The University of Bristol
SupervisorKristen K Reyher (Supervisor), Andrew Dowsey (Supervisor) & Seth Bullock (Supervisor)

Cite this