Predicting Antimicrobial Resistance (AMR) in Campylobacter, a Foodborne Pathogen, and Cost Burden Analysis Using Machine Learning

Bruno Silvester Lopes; Shatha Ghareeb; Shubham Mishra; The Anh Han; Zia Ush Shamszaman

REVIEW 4 major objections 5 minor 36 references

Predicting Antimicrobial Resistance (AMR) in Campylobacter, a Foodborne Pathogen, and Cost Burden Analysis Using Machine Learning

T0 review · 4 major / 5 minor · reviewed 2026-08-05 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read A random forest model predicts Campylobacter resistance from genomic markers with 74 percent accuracy.

desk verdict Plausible classification, but the 2050 forecasts are not reproducible and the paper oversells them; worth reviewing with major revision. read the letter →

arxiv 2509.03551 v1 pith:APNWAO3K submitted 2025-09-03 q-bio.QM cs.LG

Shubham Mishra , The Anh Han , Bruno Silvester Lopes , Shatha Ghareeb , Zia Ush Shamszaman This is my paper · ORCID

classification q-bio.QMcs.LG

keywords antimicrobialresistanceCampylobacterjejunicolimachinelearningwhole-genomesequencingrandomforesttime-seriesforecastingeconomicburdenofAMR

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper is trying to show that genomic and epidemiological data already collected for UK Campylobacter surveillance can be turned into two useful predictions: which isolates are resistant to frontline antibiotics, and how large the resistance and cost burden could be by 2050 if current trends continue. Using 6,683 C. jejuni and C. coli isolates from 2001–2017, a Random Forest classifier reaches 74% overall accuracy in predicting phenotypic resistance, with higher accuracy for beta-lactams and fluoroquinolones than for tetracyclines. The same historical data, fed into SARIMA, SIR, and Prophet models, project campylobacteriosis rates exceeding 130 cases per 100,000 people and annual costs above £1.9 billion by 2050. If these predictions hold, surveillance could become faster and cheaper, and stewardship policy would have quantified targets. The authors treat the long-run numbers as projections under unchanged conditions, not as fixed destinies.

What carries the argument

The central mechanism is a Random Forest classifier whose features are known resistance determinants — the gyrA T86I mutation, the tet(O) gene, and blaOXA beta-lactamase genes — together with metadata such as host, source, region, and years since 2000. Feature selection is done by recursive feature elimination, and the classifier is validated by temporal splitting and bootstrap confidence intervals. For the forward-looking part, SARIMA, SIR, and Prophet models are fit to 2001–2017 incidence and resistance data and extrapolated to 2050, with costs derived by multiplying forecast cases by a constant per-case cost adjusted for inflation. The Random Forest answers 'what is resistant now'; the ti

What would settle it

Retrain the random forest on UK Campylobacter isolates collected after 2017 (none were used here) and compare its predictions to phenotypic susceptibility tests; if accuracy falls materially below 74%, or if observed 2018–2025 incidence and resistance data fall outside the forecast confidence intervals, the paper's central predictive claims are contradicted.

Watch

Extended reading notes

Core claim

The core claim is that resistance phenotypes for three antibiotic classes in Campylobacter can be predicted from a small set of genomic markers and metadata: the gyrA T86I mutation for fluoroquinolones, tet(O) for tetracyclines, and blaOXA genes for beta-lactams. On 6,683 UK isolates collected from 2001 to 2017, the Random Forest classifier achieves 74% overall accuracy, with class-level accuracies of 95% for beta-lactam resistance, 78% for fluoroquinolone resistance, and 65% for tetracycline resistance, under a temporal train/test split (2001–2011 training, 2012–2017 testing) with bootstrap resampling for uncertainty. The paper further claims that time-series models fit to the same period p

Load-bearing premise

The load-bearing premise is that the 2001–2017 UK trends in resistance, case incidence, and per-case cost will continue unchanged to 2050, with cost per case held constant; any shift in policy, vaccination, antibiotic use, or treatment cost invalidates the long-run projections.

Editorial extensions

If this is right

Routine AMR surveillance could move from slow phenotypic culture tests to genomic-marker screening for these three antibiotic classes, with expected class-level accuracy between 65% and 95%.
If the projections are correct, beta-lactam and fluoroquinolone treatments for Campylobacter will become increasingly ineffective in the UK, strengthening the case for reducing antibiotic use in poultry and livestock.
The projected >130 cases per 100,000 and >£1.9 billion annual cost by 2050 give a concrete economic baseline for judging whether new vaccines, diagnostics, or stewardship programs are worth funding.
Source-specific patterns — higher fluoroquinolone resistance in human and poultry isolates than in cattle isolates — imply that interventions should target particular production sectors rather than agriculture as a whole.
Temporal validation and bootstrap confidence intervals provide a template for reporting uncertainty in long-term resistance forecasts.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

A direct test of the forecasting claim would be to compare the SARIMA/Prophet projections against UK surveillance data from 2018–2025; early divergence would indicate that the flat-trend assumption or the constant per-case cost, not the machine-learning classifier, is the weak link.
Because the classifier was trained only on isolates up to 2017, uncharacterized or newly emerging resistance mechanisms could lower real-world accuracy; a model that continually ingests new genomes and phenotypes would be needed to maintain the reported performance.
The economic projection is best read as a lower-bound scenario under current cost structures; if per-case treatment costs rise with resistance, or if sequelae are undercounted, the true burden could be higher.
The same marker-to-phenotype approach could be transferred to other foodborne pathogens such as Salmonella or Escherichia coli if equivalent paired genomic and phenotypic datasets exist, but accuracy would need to be re-estimated for each pathogen.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Colleague,

Here's my take on arXiv:2509.03551. The paper mixes three things: a Random Forest classifier for AMR phenotypes in Campylobacter, time-series forecasting of resistance and incidence to 2050, and an economic cost projection. The integration on 6,683 UK isolates from 2001–2017 is new as far as I know, and the classification part is the most defensible: temporal train/test split, bootstrap resampling, and feature importances that match known resistance determinants (gyrA T86I, tet(O), blaOXA). That part shows a serious attempt.

The soft spots are in the forecasting and economics, and they are load-bearing. The 74% accuracy is reported without a rule-based baseline, confidence intervals on the accuracy, or reconciliation of the per-class numbers (95% BL, 78% FQ, 65% Tet) with the aggregate. More importantly, the 2050 projections are not reproducible. No equations, parameters, or historical incidence series are given for the SARIMA, SIR, or Prophet models. The reported MAPE=13.48% and RMSE=13.90 are in-sample/short-term fit metrics, not uncertainty bounds on a 33-year extrapolation. The cost model holds per-case cost constant, applies an unspecified inflation adjustment, and the claimed sensitivity analysis is never shown. The limitations section honestly says the models ignore policy changes, vaccines, and stewardship, yet the abstract and conclusion present >130 cases per 100,000 and >£1.9 billion as results. There is also an internal inconsistency: the limitations list the three antibiotic classes as fluoroquinolones, β-lactams, and aminoglycosides, while the study uses tetracyclines.

The circularity concern is real but not fatal. Using known markers to predict phenotype is legitimate for surveillance; the paper just should not frame it as a novel predictive discovery. The GitHub repository is promised but no link or identifier appears, which is trivial to fix.

Net: the classification experiment is worth a referee; the forecasting and cost sections are too underspecified for the claims made. I would send this out for review with clear expectations of major revision, not desk reject, because the dataset and the ML approach are real and the topic is important. But if the forecasts cannot be reconstructed, they should be reframed as conditional scenario analyses or removed.

Referee Report

4 major / 5 minor

Summary. The paper applies a Random Forest classifier to genomic and epidemiological features from 6,683 UK Campylobacter isolates (2001–2017) to predict AMR phenotypes, reporting 74% accuracy for three antibiotic classes. It then uses SARIMA, Prophet, and SIR models to extrapolate resistance trends and campylobacteriosis incidence to 2050, and combines these with a per-case cost estimate to project an annual economic burden exceeding £1.9 billion. An 'enhanced' Random Forest regressor is also claimed to refine long-term resistance forecasts with source-specific trajectories and bootstrap-based uncertainty intervals. The paper's central policy-relevant claims are the 2050 incidence and cost projections, presented as actionable evidence for UK AMR stewardship.

Significance. If the forecasting and economic projections were reproducible and properly uncertainty-quantified, the integrated genomic-surveillance-to-economic-burden workflow would be a useful template for AMR policy planning. The manuscript has strengths: it uses public WGS data, applies a temporal train/test split (2001–2011/2012–2017), performs bootstrap resampling (1,000 samples) for classification confidence, claims code availability, and explicitly acknowledges several limitations. However, the novel contribution—the long-term forecasts and cost burden—is not currently reproducible from the manuscript, and the classification experiment is framed as predictive discovery when its features are the known resistance determinants themselves.

major comments (4)

[Materials and Methods, 'Time-Series Forecasting Models for AMR Trends'; Results, 'Long-Term Trends'] The SARIMA, Prophet, and SIR models are described only by name. No fitted equations, parameter values, order selections (p,d,q,P,D,Q,m), incidence series, or code are provided. The reported MAPE=13.48% and RMSE=13.90 are historical fit/backtest metrics and do not quantify the uncertainty of a 33-year extrapolation. The claim that incidence 'may exceed 130 cases per 100,000 by 2050' is therefore unsupported. Additionally, Fig. 2's caption says historical data from 1990, but the dataset is 2001–2017—an internal inconsistency that must be resolved.
[Materials and Methods, 'Economic Burden Analysis'; Results, 'Economic Impact of AMR in Campylobacter'] No explicit formula links forecasted incidence to economic burden. The text states a 'current total economic cost of £0.90 billion' and 'cost per case was assumed to remain constant over time', but the inflation rate, population growth assumption, and the derivation of £1.9 billion are not shown. The sensitivity analysis is mentioned only as a single sentence ('conservative resistance growth scenarios... £1.2 billion') with no scenario definitions or cost assumptions. Because the cost projection is a linear scaling of incidence, modest changes in per-case cost or trend dramatically change the headline figure; the £1.9 billion claim is not robust as presented.
[Materials and Methods, 'ML Framework for AMR Phenotype Prediction'; Results, 'Predictive Performance'] The Random Forest classifier uses known resistance determinants (gyrA T86I, tet(O), blaOXA) as features, and the feature importance analysis identifies exactly these determinants. The reported 74% accuracy therefore primarily measures how well the phenotype is encoded by known markers, not a novel predictive discovery. The manuscript should reframe this as a validation of known markers and compare against a model using only epidemiological/temporal features to demonstrate added value. Also, the per-class accuracies (95% beta-lactam, 78% fluoroquinolone, 65% tetracycline) are not reconciled with the aggregate 74%; class sizes and a weighted-average calculation are needed.
[Materials and Methods, 'Enhanced Resistance Prediction Model'; Results, 'Enhanced AMR Prediction Model and Uncertainty E] The enhanced Random Forest regressor is described only qualitatively. Evaluation Strategy promises R² and MAE for regression, but no numerical results are reported for this model. Bootstrap confidence intervals are mentioned, but no interval widths are given for any resistance projection (beta-lactam 100%, fluoroquinolone increasing, tetracycline fluctuating). Without these numbers, the source-specific resistance trajectories and their uncertainty cannot be assessed.

minor comments (5)

[Figures 1–3 captions] The figure captions in the PDF text contain garbled '/uni000...' sequences, making the captions unreadable. These must be repaired before publication.
[Limitations section] The text lists 'three high-priority antibiotic classes (fluoroquinolones, β-lactams, and aminoglycosides)', but the study actually analyzes fluoroquinolones, tetracyclines, and β-lactams. Aminoglycosides are not part of the analysis.
[Implementation section] The paper states 'The implementation code is available in a GitHub repository' but provides no repository URL, making the reproducibility claim unverifiable.
[Evaluation Strategy] The evaluation section promises R² and MAE metrics for regression tasks, but no such metrics are reported for the enhanced RF regressor. Either report them or remove the promise.
[References] Several references are incompletely formatted (e.g., 'Richard Smith' not full author list; 'Deborah J Griggs' incomplete). The journal's reference style should be applied consistently.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the ML classification is a standard out-of-sample validation, and the forecasts are explicitly conditional extrapolations.

full rationale

The paper's central derivation chain is not circular. The Random Forest classifier is trained on genomic features (gyrA, tet(O), blaOXA) to predict independently measured phenotypic resistance, with a temporal train/test split (2001–2011 training, 2012–2017 test); this is a supervised validation, not a self-definitional reduction. The fact that known resistance determinants are important features means the model is expected to perform well, but the accuracy is not forced by construction—it is an empirical concordance estimate. The 2050 incidence and cost projections are extrapolations of time-series models fitted to historical data, combined with an assumed constant per-case cost and inflation adjustment; no equation in the paper makes the forecast equal to its input. The limitations section explicitly states the predictions are 'projections based on current epidemiological conditions rather than absolute forecasts' and do not account for policy changes, vaccines, or stewardship, which further confirms the authors are not claiming a derivation from first principles. The only self-citation (reference 10, Lopes et al. 2019) is used for background context on Campylobacter genomics and is not load-bearing for the paper's predictive claims. The paper's real weaknesses—underspecified forecasting equations, unquantified long-horizon uncertainty, and a cost model that assumes constant per-case cost—are matters of reproducibility and statistical validity, not circularity. No step reduces by definition to its own input, so the circularity score is 0.

Assumptions & free parameters 7 free parameters · 5 assumptions · 0 invented entities

The central results rest on (1) the sufficiency of three known genetic determinants to explain resistance phenotypes, (2) the representativeness and quality of the PubMLST/FSA/FSS dataset, (3) the validity of extrapolating 2001-2017 time series to 2050, and (4) the economic assumption of constant per-case cost. The RF classifier is effectively a weighted lookup of known markers, and the cost projection is an accounting identity of forecast cases times assumed unit cost.

free parameters (7)

Random Forest classifier hyperparameters = 100 trees, max depth 10, min leaf 5
Tuned via grid search with 5-fold cross-validation; affect accuracy but are model selection choices.
Random Forest regressor hyperparameters = 200 trees, max depth 12
Tuned via grid search to minimize MSE for the enhanced resistance forecast model.
SARIMA model orders (p,d,q,P,D,Q,m) = Not reported
Fitted to the 2001-2017 incidence series; the specific orders are needed to reproduce the 2050 forecast but are omitted.
SIR model parameters (transmission rate, recovery rate) = Not reported
Estimated or assumed to simulate infection burden; no values or estimation procedure are given.
Cost per case = Implied by £0.90B total annual cost
Assumed constant over time; derived from a literature-based total cost divided by case count, not from a transparent model.
Inflation adjustment = Historical UK inflation rates
Applied to future costs; choice of inflation period affects the £1.9B figure.
Population growth = Linear trend
Assumed linear growth in population to scale case counts; not justified by demographic data.

assumptions (5)

domain assumption The three known genetic determinants (gyrA T86I, tet(O), blaOXA) are sufficient to predict the corresponding resistance phenotypes.
This is the basis for the RF classifier's feature set; if other mechanisms drive resistance, the 74% accuracy would not capture them fully.
domain assumption The PubMLST and FSA/FSS dataset is representative of UK Campylobacter isolates and the phenotypic labels are accurate.
All model training and validation depend on the quality and representativeness of these public surveillance data.
domain assumption UK campylobacteriosis incidence follows a stationary seasonal time series such that SARIMA/Prophet extrapolation to 2050 is valid.
Long-horizon forecasts assume no structural breaks, policy changes, or external shocks.
ad hoc to paper The future economic burden is a linear function of case count with constant cost per case.
The £1.9B projection is obtained by multiplying forecast cases by a fixed unit cost and adjusting for inflation, ignoring treatment complexity, sequelae dynamics, and healthcare system changes.
domain assumption The SIR model can represent AMR dynamics using resistance rates as an input.
No explicit equations or parameter estimation are provided; the model is invoked qualitatively.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Predicting Antimicrobial Resistance (AMR) in Campylobacter, a Foodborne Pathogen, and Cost Burden Analysis Using Machine Learning." pith.science (2026). https://pith.science/paper/APNWAO3K

@misc{pith2026250903551,
  author       = {Pith},
  title        = {Pith review of: Predicting Antimicrobial Resistance (AMR) in Campylobacter, a Foodborne Pathogen, and Cost Burden Analysis Using Machine Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/APNWAO3K}},
  note         = {Machine review of arXiv:2509.03551}
}

read the original abstract

Antimicrobial resistance (AMR) poses a significant public health and economic challenge, increasing treatment costs and reducing antibiotic effectiveness. This study employs machine learning to analyze genomic and epidemiological data from the public databases for molecular typing and microbial genome diversity (PubMLST), incorporating data from UK government-supported AMR surveillance by the Food Standards Agency and Food Standards Scotland. We identify AMR patterns in Campylobacter jejuni and Campylobacter coli isolates collected in the UK from 2001 to 2017. The research integrates whole-genome sequencing (WGS) data, epidemiological metadata, and economic projections to identify key resistance determinants and forecast future resistance trends and healthcare costs. We investigate gyrA mutations for fluoroquinolone resistance and the tet(O) gene for tetracycline resistance, training a Random Forest model validated with bootstrap resampling (1,000 samples, 95% confidence intervals), achieving 74% accuracy in predicting AMR phenotypes. Time-series forecasting models (SARIMA, SIR, and Prophet) predict a rise in campylobacteriosis cases, potentially exceeding 130 cases per 100,000 people by 2050, with an economic burden projected to surpass 1.9 billion GBP annually if left unchecked. An enhanced Random Forest system, analyzing 6,683 isolates, refines predictions by incorporating temporal patterns, uncertainty estimation, and resistance trend modeling, indicating sustained high beta-lactam resistance, increasing fluoroquinolone resistance, and fluctuating tetracycline resistance.

Figures

Figures reproduced from arXiv: 2509.03551 by the authors.

**Figure 1.** Antimicrobial Resistance Trends and Projections (2000-2050) for Fluoroquinolone, Tetracycline, and Beta-Lactam Resistance with Uncertainty Estimates 1990 2000 2010 2020 2030 2040 2050 Year 25 50 75 100 125 150 175 200 Cases per 100,000 SARIMA Model Forecast for UK Campylobacteriosis Cases per 100,000 Population Historical Data SARIMA Forecast 95% Confidence Interval [PITH_FULL_IMAGE:figures/full_fig_p006_1.png] view at source ↗

**Figure 2.** SARIMA Model Forecast for UK Campylobacteriosis Cases per 100,000 Population, Showing Historical Data and Projected Trends from 1990 to 2050 [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗

**Figure 3.** Economic Burden Projection Showing Original and Inflation-Adjusted Costs of Campylobacter Infections from 2022 to 2050 [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

36 extracted references · 36 canonical work pages

[1]

Antimicrobial resistance: Global report on surveillance

World Health Organization. Antimicrobial resistance: Global report on surveillance. WHO, 2019

work page 2019
[2]

Global burden of bacterial antimicrobial resistance in 2019: A systematic analysis

Christopher J L Murray et al. Global burden of bacterial antimicrobial resistance in 2019: A systematic analysis. The Lancet, 2023

work page 2019
[3]

The European Union summary report on antimicrobial resistance in zoonotic and indicator bacteria from humans, animals, and food in 2021/2022

European Food Safety Authority. The European Union summary report on antimicrobial resistance in zoonotic and indicator bacteria from humans, animals, and food in 2021/2022. EFSA Journal , 2024

work page 2021
[4]

A survey of Campylobacter in the UK poultry supply chain

Food Standards Agency. A survey of Campylobacter in the UK poultry supply chain. FSA, 2017

work page 2017
[5]

The economic burden of Campylobacter-associated Guillain-Barr´ e syndrome in the UK

Marie-Jos´ ee J Mangen et al. The economic burden of Campylobacter-associated Guillain-Barr´ e syndrome in the UK. Clinical Infectious Diseases, 2015

work page 2015
[6]

Mechanisms of antimicrobial resistance in Campylobacter

Taradon Luangtongkum et al. Mechanisms of antimicrobial resistance in Campylobacter. Foodborne Pathogens and Disease, 2009

work page 2009
[7]

Computer age statistical inference: Algorithms, evidence, and data science

Bradley Efron. Computer age statistical inference: Algorithms, evidence, and data science. Cambridge University Press, 2016

work page 2016
[8]

Machine learning approaches for predicting antimicrobial resistance in nontyphoidal Salmonella

Marcus Nguyen et al. Machine learning approaches for predicting antimicrobial resistance in nontyphoidal Salmonella. Microbial Genomics, 2019

work page 2019

Show all 36 references

[9]

Prediction of antibiotic resistance in Escherichia coli from whole-genome sequences

Danesh Moradigaravand et al. Prediction of antibiotic resistance in Escherichia coli from whole-genome sequences. Nature Communications, 2018

2018
[10]

Emergence and evolution of multidrug- resistant Campylobacter jejuni sequence type 5136 in the UK

Bruno S Lopes et al. Emergence and evolution of multidrug- resistant Campylobacter jejuni sequence type 5136 in the UK. Journal of Antimicrobial Chemotherapy , 2019

2019
[11]

DeepARG: A deep learning approach for predicting antibiotic resistance genes from metagenomic data

Gustavo Arango-Argoty et al. DeepARG: A deep learning approach for predicting antibiotic resistance genes from metagenomic data. Microbiome, 2018

2018
[12]

Population-level mathematical modeling of antimicrobial resistance: A systematic review

Anna Maria Niewiadomska et al. Population-level mathematical modeling of antimicrobial resistance: A systematic review. BMC Medicine, 2019

2019
[13]

Another look at forecast-accuracy metrics for intermittent demand

Rob J Hyndman. Another look at forecast-accuracy metrics for intermittent demand. Foresight: The International Journal of Applied Forecasting, 2006

2006
[14]

The true cost of antimicrobial resistance

Richard Smith. The true cost of antimicrobial resistance. BMJ, 2013

2013
[15]

Quantifying the economic cost of antibiotic resistance and the impact of related interventions

Mark Jit. Quantifying the economic cost of antibiotic resistance and the impact of related interventions. Clinical Infectious Diseases, 2020

2020
[16]

Tackling drug-resistant infections globally: Final report and recommendations

Jim O’Neill. Tackling drug-resistant infections globally: Final report and recommendations. Review on Antimicrobial Resistance, 2016

2016
[17]

Integrated Campylobacter surveillance

Food Standards Agency. Integrated Campylobacter surveillance. FSA, 2017

2017
[18]

Campylobacter attribution study

Food Standards Scotland. Campylobacter attribution study. FSS, 2017

2017
[19]

Quinolone resistance in Campylobacter: Mechanisms and epidemiology

Sophie Payot et al. Quinolone resistance in Campylobacter: Mechanisms and epidemiology. Veterinary Research, 2006

2006
[20]

Beta-lactam resistance mechanisms in Campylobacter

Deborah J Griggs. Beta-lactam resistance mechanisms in Campylobacter. Journal of Medical Microbiology , 2009

2009
[21]

Feature engineering and selection: A practical approach for predictive models

Max Kuhn. Feature engineering and selection: A practical approach for predictive models. Chapman and Hall/CRC , 2019

2019
[22]

Forecasting: Principles and practice

Rob J Hyndman. Forecasting: Principles and practice. OTexts, 2021

2021
[23]

Forecasting at scale

Sean J Taylor. Forecasting at scale. The American Statistician, 2018

2018
[24]

Scikit-learn: Machine learning in Python

Fabian Pedregosa. Scikit-learn: Machine learning in Python. Journal of Machine Learning Research , 2012

2012
[25]

Fundamentals of data visualization

Claus O Wilke. Fundamentals of data visualization. O’Reilly Media, 2019

2019
[26]

The F AIR Guiding Principles for scientific data management and stewardship

Mark D Wilkinson. The F AIR Guiding Principles for scientific data management and stewardship. Scientific Data, 2016

2016
[27]

Emerging resistance mechanisms in Campylobacter

Yuansha Chen. Emerging resistance mechanisms in Campylobacter. Antimicrobial Agents and Chemotherapy , 2013

2013
[28]

Antimicrobial resistance in the EU/EEA

European Centre for Disease Prevention and Control. Antimicrobial resistance in the EU/EEA. ECDC, 2020

2020
[29]

Global antimicrobial resistance and use surveillance system (GLASS) report

World Health Organization. Global antimicrobial resistance and use surveillance system (GLASS) report. WHO, 2021

2021
[30]

Antibiotic resistance: Are we all doomed? Internal Medicine Journal , 2016

Peter John Collignon. Antibiotic resistance: Are we all doomed? Internal Medicine Journal , 2016

2016
[31]

Global priority list of antibiotic-resistant bacteria to guide research, discovery, and development

Evelina Tacconelli. Global priority list of antibiotic-resistant bacteria to guide research, discovery, and development. The Lancet Infectious Diseases, 2018

2018
[32]

A unified approach to interpreting model predictions

Scott M Lundberg. A unified approach to interpreting model predictions. Advances in Neural Information Processing Systems, 2020

2020
[33]

Common pitfalls and recommendations for using machine learning to detect and prognosticate for COVID-19

Michael Roberts. Common pitfalls and recommendations for using machine learning to detect and prognosticate for COVID-19. Nature Machine Intelligence , 2021

2021
[34]

Restricting the use of antibiotics in food- producing animals and its associations with antibiotic resistance

Karen L Tang. Restricting the use of antibiotics in food- producing animals and its associations with antibiotic resistance. The Lancet Planetary Health , 2017

2017
[35]

Rapid detection of antimicrobial resistance using genomic approaches

M J Ellington. Rapid detection of antimicrobial resistance using genomic approaches. Journal of Antimicrobial Chemotherapy, 2016

2016
[36]

Understanding the mechanisms and drivers of antimicrobial resistance

Alison H Holmes. Understanding the mechanisms and drivers of antimicrobial resistance. The Lancet, 2016

2016

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.

[1] [1]

Antimicrobial resistance: Global report on surveillance

World Health Organization. Antimicrobial resistance: Global report on surveillance. WHO, 2019

work page 2019

[2] [2]

Global burden of bacterial antimicrobial resistance in 2019: A systematic analysis

Christopher J L Murray et al. Global burden of bacterial antimicrobial resistance in 2019: A systematic analysis. The Lancet, 2023

work page 2019

[3] [3]

The European Union summary report on antimicrobial resistance in zoonotic and indicator bacteria from humans, animals, and food in 2021/2022

European Food Safety Authority. The European Union summary report on antimicrobial resistance in zoonotic and indicator bacteria from humans, animals, and food in 2021/2022. EFSA Journal , 2024

work page 2021

[4] [4]

A survey of Campylobacter in the UK poultry supply chain

Food Standards Agency. A survey of Campylobacter in the UK poultry supply chain. FSA, 2017

work page 2017

[5] [5]

The economic burden of Campylobacter-associated Guillain-Barr´ e syndrome in the UK

Marie-Jos´ ee J Mangen et al. The economic burden of Campylobacter-associated Guillain-Barr´ e syndrome in the UK. Clinical Infectious Diseases, 2015

work page 2015

[6] [6]

Mechanisms of antimicrobial resistance in Campylobacter

Taradon Luangtongkum et al. Mechanisms of antimicrobial resistance in Campylobacter. Foodborne Pathogens and Disease, 2009

work page 2009

[7] [7]

Computer age statistical inference: Algorithms, evidence, and data science

Bradley Efron. Computer age statistical inference: Algorithms, evidence, and data science. Cambridge University Press, 2016

work page 2016

[8] [8]

Machine learning approaches for predicting antimicrobial resistance in nontyphoidal Salmonella

Marcus Nguyen et al. Machine learning approaches for predicting antimicrobial resistance in nontyphoidal Salmonella. Microbial Genomics, 2019

work page 2019