A periodical of the Faculty of Natural and Applied Sciences, UMYU, Katsina
ISSN: 2955 – 1145 (print); 2955 – 1153 (online)
ORIGINAL RESEARCH ARTICLE
Rodiat Olabisi Omotoso1*, Ismail Ayoade Odetokun2, Toyeeb Olamilekan Abubakar1 and Adebowale Olusola Adejumo1
1Department of Statistics, Faculty of Physical Sciences, University of Ilorin, Ilorin, Nigeria
2Department of Veterinary Public Health and Preventive Medicine, University of Ilorin, Ilorin, Nigeria
*Corresponding Author: Rodiat Olabisi Omotoso 
omotosorodiatolabisi@gmail.com
Antimicrobial resistance (AMR) in livestock has emerged as a major global threat, with direct implications for public health, food security, and sustainable development under the One Health framework. Livestock production systems significantly contribute to global antimicrobial use, exerting selective pressures that drive the establishment and rapid spread of resistant infections across the animal, human, and environmental interfaces. This review examines existing studies on the spatial distribution of AMR in livestock, with emphasis on surveillance databases, spatial statistical techniques, and modeling methods. This underscores the importance of global AMR surveillance resources, such as the ResistanceBank, an open-access livestock AMR database. This systematic review, conducted using the PRISMA guidelines, searched three databases (PubMed, Scopus, and Google Scholar) for studies published between 2016 and 2025. After duplicates were removed and the remaining articles were screened by two independent reviewers based on predefined inclusion and exclusion criteria, 13 studies were included. Study quality was appraised using the McMaster critical evaluation framework, and data on study characteristics, together with the key findings, were extracted. Some studies employed hybrid modeling techniques; most applied spatial modeling (84.62%); over half used Machine Learning (ML) techniques; and some included statistical modeling approaches (53.85%). Spatial analyses revealed clustering, hotspots, and spillover patterns of AMR. ML models exhibited strong predictive performance for AMR features. Hybrid modeling approaches enhanced robustness and interpretability by capturing epidemiological processes and integrating heterogeneous datasets. In conclusion, this review showed that integrating spatial analysis with predictive modeling provides a robust framework for advancing livestock antimicrobial resistance surveillance, improving risk identification, and supporting targeted antimicrobial stewardship and policy interventions.
Keywords: Antimicrobial resistance, spatial analysis, predictive modeling, beta regression, machine learning
- Livestock has a significant impact on the spread of AMR across the globe, thereby calling for a global understanding of the state of AMR across countries
- Spatial heterogeneity is a key feature of AMR, due to differences in AMU practices, livestock production systems, environmental conditions, and socio-economic factors.
- Beyond spatial description, predictive models (statistical and machine learning) provide significant insight into the factors contributing to AMR.
- A shift towards spatially informed predictive modeling approaches will strengthen livestock AMR research.
Antimicrobial resistance (AMR) emerged as a major threat to global health, food security, and sustainable development in the twenty-first century. AMR occurs when microorganisms form resistance that renders antimicrobial agents ineffective, causing persistent infections, elevating morbidity and mortality, and increasing healthcare costs (Prestinaci et al., 2015; Tang et al., 2023; Andrew et al., 2024; Hamisu & Salisu, 2025; Rabiu et al., 2022; Saheed et al., 2025; Salisu et al., 2019, 2017; Usman et al., 2025). AMR is a substantial global burden, with 4.95 million deaths associated with resistant infections alone in 2019 (Murray et al., 2022; Naghavi et al., 2024). Recently, AMR continued to be associated with approximately 5 million deaths every year, making it one of the leading causes of mortality globally (WHO 2024). Projections estimated AMR-related mortality could substantially progress to a cumulative total of more than 39 million deaths between 2025 and 2050 if effective intervention is not provided globally.
Historically, AMR has been examined within clinical and human contexts; however, there is increasing recognition of its complexity and interconnectedness across human, animal, and environmental systems. The interconnected nature of AMR underpins the One Health framework, which highlights the role of livestock production systems in the progress, amplification, and dissemination of AMR pathogens (Chokshi et al., 2019; Abbas et al., 2024; Mohammed et al., 2017). In livestock production systems, especially in intensive farming, antimicrobials are widely used for disease treatment, prevention, and growth promotion, exerting selective pressure that enhances the development of resistant pathogens (Hosain et al., 2021; Matheou et al., 2025).
AMR in livestock has profound implications for public health. Resistant pathogens can be transmitted to humans through multiple channels, including the food chain, direct animal contact, and environmental pathways such as water and soil contamination (Panicker et al., 2025; Meyer et al., 2024). AMR in livestock is associated with a substantial economic burden, affecting animal productivity, trade, and livelihoods, particularly in resource-limited regions (Babo Martins et al., 2024). Livestock constitute a significant source of human protein, and the inappropriate use of antimicrobials in animals can also facilitate the transfer of AMR pathogens (Babo Martins et al., 2024; Trinchera et al., 2025). AMU in livestock is estimated to have surpassed human consumption, intensified AMR, and explained the central role of food-producing animal systems in shaping the increased AMR burden (Van Boeckel et al., 2015; Salisu et al., 2017; Hosain et al., 2021; Babo Martins et al., 2024). Factors contributing to inappropriate antimicrobial use (AMU) and poor monitoring of resistance patterns include weak regulatory enforcement, limited access to veterinary services, informal drug markets, and inadequate surveillance systems (Grace, 2015; Iskandar et al., 2021; Davis et al., 2025). These factors pose AMR challenges in livestock.
Spatial heterogeneity is a key feature of AMR, driven by differences in AMU practices, livestock production systems, environmental conditions, and socio-economic factors. Spatial analysis employs powerful tools to quantify geographical dependence, identify resistance clusters, and detect high-risk areas that may require targeted interventions (Spets et al., 2023; Legenza et al., 2023). Studies on infectious disease and AMR have applied measures of spatial autocorrelation, such as Global and Local Moran’s I, to reveal non-random spatial patterns (Kou et al., 2025). However, their application to livestock AMR at a global scale remains limited.
Beyond spatial description, predictive models (statistical and machine learning) provide significant insight into the factors contributing to AMR. Regression methods, such as beta regression, are suitable for modeling resistance proportions bounded between 0 and 1, while machine learning (ML) approaches can capture complex, non-linear relationships and feature interactions (Mulchandani et al., 2024). Integration of spatial analysis with predictive modeling provides a comprehensive framework for understanding the distribution and determinants of AMR in livestock. This review examines existing studies on the spatial distribution of AMR in livestock, with emphasis on surveillance databases, spatial statistical techniques, and modeling methods.
AMR rapidly increases disease progression, co-infection with pathogens such as salmonellosis, and other transmissible diseases, with the potential to accelerate the spread of the resistant pathogen across the nation or globe. Furthermore, systematic analysis shows that mortality due to six pathogens (Escherichia coli, Staphylococcus aureus, Klebsiella pneumoniae, Streptococcus pneumoniae, Acinetobacter baumannii, and Pseudomonas aeruginosa) has an estimation of 929000 (660000-1270000 95% UI), with each pathogen accounting for over 250000 deaths due to AMR (Naghavi et al., 2024). The analysis provided insight into the global burden of AMR and confirmed that it is a major global challenge. Studies on resistome show that livestock gastrointestinal systems are substantial hosts of diverse AMR pathogens transmissible to environmental or zoonotic pathogens (Ma et al., 2021).
AMR burden is disproportionate in LMICs due to structural vulnerabilities, including limited veterinary facilities, ineffective regulation, high-density farming, lack of biosecurity, and unrestricted access to antimicrobials (Iskandar et al., 2021; Odey et al., 2024). Additionally, ecological pollution resulting from improper waste management, unmanaged wastewater flow, and livestock effluent runoff accelerates the dissemination of resistance genes in LIMCs (Azabo et al., 2022; Meyer et al., 2024). The close proximity of livestock, humans, and water systems in these nations further exacerbates One Health transmission mechanisms (Davis et al., 2025). However, the lack of spatially focused AMR data limits the identification of hotspots and the effective allocation of resources by policymakers, underscoring a major gap.
An adequate surveillance system is important for understanding AMR patterns, identifying threats, and implementing targeted interventions. Several databases exist, but they have different coverage, resolution, and quality.
ResistanceBank.org is an open-access database of AMR in livestock and one of the most comprehensive data sources for livestock AMR surveillance (Criscuolo, Pires, & Van Boeckel, 2021). This database compiles study-based AMR data from several countries, pathogens, and livestock species, providing a large-scale dataset on resistance proportions across antimicrobial classes.
The database collated AMR data obtained mainly from prevalence surveys published between 2000 and 2021, compiling over 1200 surveys and 33186 resistance estimates across livestock species, including poultry, cattle, pigs, sheep, ducks, horses, and buffalo. The reports include key public health pathogens: Escherichia coli, Salmonella, Staphylococcus aureus, and Campylobacter spp. Countries represented in the Resistance Bank database span across Africa, Asia, Latin America, and parts of Europe and North America.
Livestock AMR surveillance encounters structural challenges. Strengths of existing databases include organized laboratory testing protocols, improved global coordination of AMR data reporting, and inclusive datasets covering multiple livestock species, pathogens, and antimicrobial classes.
A major strength that underpins the credibility of AMR data for statistical analysis is its reliance on organized laboratory testing protocols. ResistantBank.org collates AMR data obtained through standard antimicrobial susceptibility testing (AST) procedures carried out in accredited laboratories, with most contributing studies following internationally recognized guidelines such as those of the Clinical and Laboratory Standards Institute (CLSI) and the European Committee on Antimicrobial Susceptibility Testing (EUCAST). This harmonized testing methodology ensured consistency in pathogen isolation, antimicrobial panels, and the interpretation of AMR across diverse data sources.
Improvements in global coordination on AMR data reporting enhance comparability and provide high-quality evidence for informed decision-making, surveillance, and efficient intervention strategies. Particularly in livestock, where AMR patterns are influenced by diverse animal species, production systems, and AMU, ResistanceBank significantly contributes to the coordination of AMR data reporting by serving as a centralized, standardized repository, thereby reducing fragmentation across individual studies.
Furthermore, the availability of multidimensional data enables comparisons across species, pathogens, and antimicrobial classifications, thereby enhancing the identification of shared AMR patterns and transmission pathways within animal populations. Moreover, inclusivity in data collation reduces bias associated with single-species and single-pathogen surveillance approaches and supports robust epidemiological and spatial analyses.
However, its limitations remain significant. This includes sparse geographic coverage in low-income countries (Criscuolo et al., 2021; Iskandar et al., 2021), a lack of continuous AMR reporting, sampling biases toward outbreaks or commercial farms, limited ability to model predictors or spatial dependencies, and despite the awareness of transmission pathways, there is low environmental and One Health integration. (Bengtsson-Palme et al., 2023; Venkateswaran et al., 2024).
Despite the strengths, livestock AMR surveillance, such as ResistanceBank, is disproportionately populated with data from countries where laboratory infrastructure, routine surveillance systems, and reporting capacity are more established. Figure 1 shows the disproportionate representation of low-income and low-middle-income countries in Resistance Bank data, resulting in substantial data gaps across many countries, where AMU in food processing animals is often less regulated, and surveillance systems are absent.
Figure 1: Contribution of Countries' Economic Classification in the ResistanceBank Database According to the World Bank
LIC – Low Income Country, LMI – Low Middle Income, HIC – High Income country, UMI – Upper Middle Income
Another notable setback of existing livestock AMR databases, particularly ResistanceBank, is the lack of continuous, longitudinal AMR data collection. Much of the collated data in the ResistanceBank was obtained from cross-sectional studies, periodic surveys, or short-term surveillance rather than sustained, routine monitoring systems. This results in the absence of continuous reporting, with substantial gaps between reporting periods for many countries and livestock species. This limited the ability to accurately assess temporal trends, seasonal variations, and the long-term impact of antimicrobial stewardship interventions in the livestock production system. Discontinuity in AMR data reduces statistical power and may introduce bias in spatial and temporal epidemiological analyses, thereby limiting the utility of ResistanceBank for trend modeling and predictive risk assessment, and delaying early detection of emerging resistance patterns and timely public health and veterinary intervention.
Additionally, Figure 2 shows that livestock AMR data, such as those from ResistanceBank, are often subject to sampling biases, with disproportionate contributions from China, India, and Brazil, and with data often obtained from research-driven studies rather than routine sampling across subsistence production systems. The disproportionate sampling may overestimate resistance prevalence and limit the comparability and generalisability of findings to broader livestock populations. Moreover, despite growing awareness of AMR transmission pathways at the human-animal-environment interface, One Health integration remains constrained within existing livestock datasets, such as ResistanceBank, which primarily focus on isolates from animals with little linkage to environmental reservoirs, such as water, soil, and farm waste, or to human clinical data.
Figure 2: Spatial distribution of countries' publications reported in ResistanceBank
International travel and trade make the transmission of resistant pathogens across nations inevitable, creating the need for intensified global interventions to address AMR (Prestinaci et al. 2015; Chhokshi et al. 2019). With the availability of environmental data, the use of spatial techniques helps in understanding the spread of AMR across countries (Legenza et al., 2023).
The robust framework offered by spatial epidemiology techniques for analyzing the geographic spread of diseases and resistance patterns enables researchers to identify clusters and hotspots and to investigate the influence of environmental and socioeconomic factors (Spets et al., 2023). With regards to AMR in livestock, spatial techniques are important because resistance does not occur at random; it reflects connections between AMU practices, livestock production systems, environmental contamination, and regional health and policies (Kou et al., 2025).
Spatial autocorrelation estimates the degree of similarity in observations at nearby locations. Positive spatial autocorrelation underscores clustering of similar values (e.g., high AMR in neighboring nations), while negative autocorrelation underlines spatial heterogeneity (Spet et al., 2023). Global Moran’s I is a measure of spatial autocorrelation that examines the overall spatial pattern across regions (Moran, 1950; Anselin, 1995). It is useful for identifying AMR clusters, dispersion, or randomness. Local Moran’s I (LISA) is also useful for identifying specific clusters (hotspots and coldspots), enabling regionalized policymaking and interventions. These spatial techniques have been widely applied in disease mapping but remain underused in livestock AMR research, particularly in Low-income countries.
Resistance is often measured in proportion (percentage of resistant isolates). Values bounded between 0 and 1 violate the assumptions of ordinary least squares (OLS) regression. Beta regression is therefore preferred because it accommodates boundness, heteroscedasticity, and non-normality (Matheou et al., 2025).
Beta regression is commonly employed in ecological and environmental health studies but is rarely used in livestock AMR modeling, despite its suitability for modeling resistance proportions across species, pathogens, and antimicrobial classes. It also permits flexible link functions and precision parameters, enhancing improved model fit compared with linear models, and produces accurate and interpretable estimates of resistance levels across livestock species, pathogens, and regions. Beta regression enhances the ability to capture spatial dependency, and when integrated with spatial analysis, it supports more reliable identification of AMR hotspots linked to animal trade, strengthening spatial epidemiological insight. The use of explanatory modeling techniques enables the clear identification of the driver of AMR emergence in livestock populations.
Machine learning (ML) techniques – such as gradient boosting, random forests, and neural networks – give powerful alternatives for identifying nonlinear relationships among AMR predictors (Lewnard et al., 2024). ML approaches are mainly valuable when independent variables exhibit complex interactions, relationships are not linear, and data possess high-dimensional features such as pathogen, drug class, species, and environmental metadata.
Beyond improved prediction accuracy, the use of ML models in livestock AMR surveillance can enhance risk stratification and decision-making in data-limited settings. When integrated with spatial analysis, ML techniques can provide early warning systems, identify high-risk livestock populations, and inform targeted intervention. Consequently, a combination of ML models with explainable models and domain-informed features represents a major pathway for advancing livestock AMR modeling, because the deployment of ML in this context relies on careful model validation, transparency, and interpretability to ensure AMR predictions are epidemiologically meaningful.
Figure 3 highlights the contributions of spatial techniques, regression modeling, ML modeling techniques, and the integration of three analytical methods in understanding AMR trends in Livestock.
Figure 3: Contribution of spatial analysis, regression modeling, and machine learning modeling in understanding AMR patterns
The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement was used in conducting this systematic review (Moher et al., 2015). A search protocol was designed and registered on Rayyan.
Three databases (PubMed, Scopus, and Google Scholar) were systematically searched to obtain relevant articles published between January 1, 2016, and December 30, 2025. The search strategy combined keywords and related vocabulary for AMR, Spatial Modeling, and ML modeling. The core search string for the database search is ("antimicrobial resistance" OR "antibiotic resistance" OR "drug resistance") AND ("spatial analysis" OR "spatial modeling" OR "geospatial" OR GIS OR "spatial epidemiology") AND ("predictive modeling" OR "machine learning" OR "statistical modelling"). Filters were applied to the publication year range and language. The final search was conducted; a total of 162 articles were identified (Figure 4) and uploaded to Rayyan. Sixteen duplicate articles were deleted, and 146 articles were subjected to primary screening. In addition to this review’s comprehensive search strategy, a blind search was conducted across the primary databases, leading to the inclusion of one additional article and further strengthening the review and confirming a detailed assessment of the relevant literature.
To determine the article to included in the review, these inclusion criteria were developed – studies focusing on AMR, studies using spatial analysis or predictive modelling techniques, original research articles, studies published in English, studies that focus on livestock or zoonotic diseases, and studies published between 2016 and 2025 – together with is exclusion criteria – reviews, editorials, opinion papers, studies with modelling components, human focused articles, and non-English publications. Two independent reviewers conducted the primary (title and abstract screening) and secondary (full-text screening), and disputes were resolved through discussion. The reporting quality and bias selection in the included studies were appraised by two independent reviewers using the McMaster critical evaluation methods for both quantitative and qualitative study design, study location, animal type, sample type, and the method of detection (Ducat and Kumar, 2015).
A data extraction form was developed to capture the following information: author names, year, aim, modeling techniques, modeling type, study region, pathogen, and key results, which are considered crucial for this study.
Thirteen articles were found eligible for this study (Figure 4), including publications from 2018 to 2025. These studies covered diverse geographical regions, including global analyses (23.08%), continental level studies (15.38%) (Africa and Europe), and country-specific studies (61.54%), particularly in China, Italy, Spain, Australia, the United States of America (USA), and Peru (Table 1).
The majority of the studies (84.62%) employed spatial modeling approaches to identify geographical patterns, clustering, and hotspots of AMR. The techniques employed include geospatial mapping, spatial econometric models, Bayesian spatial regression, Moran’s I and spatial scan statistics. Studies by Zhao et al. (2024) and Muchandani et al. (2024) identified global and regional AMR hotspots, while Teng et al. (2020) and Kou et al. (2025) demonstrated significant spatial clustering and spillover effects. A key finding from Warren et al. (2018) showed that multidrug-resistant tuberculosis spillover extended approximately 5.47km, highlighting the role of localized transmission forces (Table 1).
Predictive modeling techniques were applied in many studies (69.23%), focusing on identifying AMR drivers. ML models were widely employed, including Random Forests (RF), Gradient Boosting Machines (GBM), LightGBM, and Neural networks. RF was identified as a top-performing algorithm, and LightGBM estimated high prediction accuracy. Overall, ML approaches demonstrated strong predictive performance and efficiently identified some key determinants of AMR. For example, Feng et al. reported a fourfold increase in antimicrobial resistance gene (ARG) abundance using RF alongside genomic and geospatial analyses to predict global trends of Streptococcus agalactiae, while Adedeji et al. (2025) employed multiple ML algorithms, including Random Forest and Light Gradient Boosting Machine (LightGBM), to analyze surveillance data in Africa. Their findings highlighted the robustness of ML techniques for determining key AMR drivers.
Several studies (53.85%) also employed statistical modeling techniques, including logistic growth models, Bayesian semi-mechanistic models, and regression-based models. These models were used to capture underlying epidemiological processes, account for data heterogeneity across sources, and improve the robustness of AMR trend predictions. For instance, models trained on heterogeneous datasets showed improved predictive accuracy and reliability. Hybrid approaches integrated spatial, statistical, and ML techniques to improve both predictability and interpretability.
Warren et al. (2018) demonstrated the strength of hierarchical Bayesian modeling in capturing spatial spillover effects of multidrug-resistant tuberculosis (MDR-TB) in Peru, estimating a spillover risk extending approximately 5.47km and providing evidence of community-level transmission. Likewise, Teng et al. (2020) applied Bayesian spatial regression (BYM2), along with Moran’s I and spatial scan statistics, to identify significant spatial clusters and a west-to-east gradient in Salmonella infection risk in Spain. Ou et al. (2025) employed logistic regression in combination with spatial and genomic analyses to investigate transmission patterns of Mycobacterium tuberculosis in China, revealing a high level of recent transmission (~50%) and identifying key hotspots.
Figure 4: PRISMA flow diagram of study identification, screening, and inclusion for antimicrobial resistance in livestock studies
Table 1: Overview of modeling techniques and key findings in antimicrobial resistance studies
| Authors | Aim | Modelling Techniques | Modelling Type | Study Region | Pathogen | Key Result |
|---|---|---|---|---|---|---|
| Fang et al. 2025 | To analyze global temporal trends and predict future antibiotic resistance in GBS | Genomic analysis, geospatial analysis, Random Forest prediction | Machine learning + Spatial analysis | Global | Streptococcus agalactiae | 4-fold increase in ARG abundance; Asia identified as a hotspot; 187% predicted |
| Sobkowich et al. 2024 | To assess the prevalence and trends of CRE in animals | Space-time cluster analysis, temporal trend analysis | Spatial + Statistical modeling | USA | Enterobacter ales | Low prevalence (98.86% susceptibility); identified regional clusters |
| Adedeji et al. 2025 | To identify predictors of AMR and develop predictive models using surveillance data | Random Forest, LightGBM, linear regression, chi-square tests | Machine learning + Statistical modeling | Africa | Multiple bacteria | RF showed high classification performance; LightGBM achieved ~81% accuracy; and identified key AMR drivers. |
| Garcia-Vozmediano et al. 2025 | To develop a data-driven surveillance framework for Salmonella | Statistical models + machine learning integration | Hybrid (Statistical + ML) | Italy | Salmonella enterica | Improved surveillance through the One Health framework (data-driven integration) |
| Mulchandani et al. 2024 | To map AMR prevalence and identify hotspots | Geospatial modeling, predictive mapping | Spatial modelling | Europe | E. coli, Salmonella, Campylobacter | Identified AMR hotspots and geographic variation; supports targeted interventions |
| Zhao et al. 2024 | To map AMR prevalence globally using survey data | Geospatial modeling using point-prevalence surveys | Spatial modelling | Global | E. coli, Salmonella | Identified global AMR hotspots; predicted future resistance thresholds geographically |
| Smit et al. 2025 | To analyze small-scale spatial and temporal variation in AMR | Geospatial mapping, temporal trend analysis | Spatial + Statistical modeling | Australia | E. coli | Resistance varied by region and time; identified local hotspots |
| Teng et al. 2020 | To analyze the spatial distribution and risk factors of Salmonella infection | Bayesian spatial regression (BYM2), Moran’s I, spatial scan statistics | Spatial + Statistical modeling | Spain | Salmonella | Identified spatial clusters and a west-to-east risk gradient |
| Ou et al. 2025 | To study TB transmission using genomic and spatial data | Logistic regression, spatial analysis, genomic clustering | Spatial + Statistical modeling | China | Mycobacterium tuberculosis | High recent transmission (~50%); spatial hotspots identified |
| Zhelyazkova et al. 2021 | To predict AMR risk and sample origin using spatial modeling | Bayesian spatial convolution model, ML (RF, GBM, NN) | Hybrid (Spatial + ML) | Global | Multiple bacteria | RF performed best; spatial modeling improves AMR risk estimation |
| Shang et al. 2022 | To analyze TB distribution and forecast trends | Spatial analysis, exponential smoothing model | Spatial + Time-series modeling | China | Mycobacterium tuberculosis | Identified clusters and seasonal peaks; the model is effective for forecasting |
| Warren et al. 2018 | To investigate spatial spillover of MDR-TB | Hierarchical Bayesian modelling, spatial analysis | Spatial + Statistical modelling | Peru | Mycobacterium tuberculosis | Spillover risk extends ~5.47 km; evidence of community transmission |
| Kou et al. 2025 | To assess spatial patterns and drivers of E. coli antimicrobial resistance | Spatial panel modelling, spatial econometric analysis | Spatial modelling | China | Escherichia coli | Significant spatial clustering; clear interregional spillover effects; AMR drivers identified |
GBS – Group B Streptococcus, ARG – Antimicrobial Gene, CRE – Carbapenem-resistant Enterobacterales, ML – Machine Learning, AMR – Antimicrobial Resistance, TB – Tuberculosis, MDR-TB – Multidrug-Resistant Tuberculosis
This systematic analysis reviewed literatures on the application of spatial and predictive modeling methods in AMR research. The findings demonstrate growing integration of these techniques in understanding the distribution, drivers, and future trajectories of AMR across diverse settings. The complementary roles of spatial and predictive modeling approaches are a notable finding of this review: spatial models were used to characterize geographic heterogeneity, clustering, and hotspot identification, while predictive models focused on forecasting and identifying determinants of AMR. The increased use of hybrid frameworks suggests a methodological shift toward integrated modeling, aligning with the need for multidimensional modeling in AMR studies.
This review provides strong evidence of significant spatial heterogeneity in AMR distribution, along with the identification of hotspots at both global and local levels. Evidence of spatial spillover effects and short-range transmission underscores the role of geographic proximity and connectivity in the spread of AMR. Machine learning models, including RF, GBM, and neural networks, are able to capture nonlinear relationships and high-dimensional interactions, making them particularly suitable for AMR prediction and consistently showing strong predictive performance across studies. However, despite their predictive strength, many machine learning models lacked interpretability, which is essential for public health decision-making. Conversely, statistical models such as Bayesian and logistic models provide better insight into underlying epidemiological processes, though often at the cost of predictive accuracy.
The existing literature demonstrates the growing potential of spatial AMR analysis, but it remains limited. A systematic review by Spet et al. (2023) found that AMR in environmental hosts exhibited significant spatial clustering driven by anthropogenic contamination. Geospatial modeling conducted by Legenza et al. (2023) examined neighborhood-level antibiotic susceptibility in the United States and demonstrated how spatial heterogeneity can reveal local antibiotic-resistant hotspots. Kou et al. (2025) employed spatial panel data analysis to assess AMR trends in E. coli across China, exhibiting substantial spatial spillover effects. Bengtsson-Palme et al. (2023) stressed that environmental AMR surveillance requires spatially explicit frameworks given the uneven distribution of antimicrobial resistance genes (ARGs) worldwide.
In these examples, most spatial AMR studies focus on environmental or human health data. Very few studies on AMR in low-income livestock countries exist, even though these nations face the most severe AMR challenges (Iskandar et al., 2021). Thus, employing spatial techniques to analyze livestock resistance data bridges a major methodological and geographic gap. The incorporation of spatial analysis enables a clearer understanding of how AMR spreads across nations – whether through livestock trade, shared water sources, or farm management practices. This abides with One Health principles, highlighting the interconnectedness of humans, animals, and the environment (Trinchera et al., 2025; Panicker et al., 2025). The application of ML to livestock AMR in LMICs is underexplored, even though it has been widely adopted in clinical AMR modeling. The few studies that compared livestock AMR models show that ML can outperform classical methods in predictive accuracy.
Although substantial progress has been made in understanding AMR patterns globally, this literature identifies persistent gaps in the utilization of spatial and predictive analyses for AMR in livestock.
Despite improvements in recognition of the spatial determinants of AMR, few studies have employed spatial autocorrelation techniques, such as Global Moran’s I or LISA, to analyze livestock datasets in LMICs. Tang et al. (2023) note that most AMR studies focus on clinical pathogens in human hospitals, with little emphasis on geospatial epidemiology in animal production systems. Similarly, Venkateswaran et al. (2024) emphasized the scarcity of spatially disaggregated livestock AMR data, due to incomplete laboratory coverage and fragmented reporting structures. As a result, the lack of spatial maps highlighting clusters, hotspots, or emerging threats within livestock systems affects policymaking.
The availability of a global AMR surveillance repository represents a significant advance in consolidating livestock AMR data, yet these data are underused in the scientific literature. Studies show that many researchers rely on small-scale or local datasets, mainly because global AMR repositories are relatively new and require advanced analytical skills. However, this resulted in a lack of comprehensive multi-country LMIC analyses that employed standard comparable livestock AMR data for predictive and spatial modeling. These gaps support the relevance and novelty of this study.
Researchers (Prestinaci et al., 2015; Chokshi et al., 2019) emphasized the dynamic complexity of AMR, yet most studies rely on classical statistical models, despite ML's capacity to model nonlinear relationships between species, pathogens, antimicrobial classifications, and environmental conditions. Machine learning models such as XGBoost, decision trees, and random forests are underused in AMR surveillance, especially in LMIC livestock settings, where data may be incomplete or imbalanced.
Some studies explain AMR patterns, others explore predictive modeling, and very few incorporate spatial statistics with ML. WHO (2021) 's Global strategy highlights the importance of system-level surveillance; however, current research rarely combines autocorrelation detection with regression or ML-based predictors. This limits the ability to understand not only where AMR occurs, but also why it emerges in a specific nation and production system.
In conclusion, this review shows that AMR in livestock exhibits pronounced spatial heterogeneity, influenced by AMU, production patterns, environmental contamination, and socio-economic factors. Global databases such as ResistanceBank provide a substantial foundation for robust analyses, but are limited by uneven geographical coverage and weak One Health integration. Beyond descriptive analysis, integrating spatial statistics with predictive modeling offers a robust framework for advancing livestock AMR surveillance towards risk identification. Overall, a shift towards spatially informed predictive modeling approaches will strengthen livestock AMR research, enable the identification of resistance hotspots and elucidation of resistance drivers, and support targeted antimicrobial stewardship and interventions.
Abbas, A., Barkhouse, A., Hackenberger, D., & Wright, G. D. (2024). Antibiotic resistance: A key microbial survival mechanism that threatens public health. Cell Host & Microbe, 32(6), 837–851. [Crossref]
Adedeji, R., Agboeze, T., Akomolafe, A., & Daramola, O. I. (2025). Supervised learning model systems to predict and identify drivers of AMR in Africa. Wellcome Open Research, 10, Article 410. [Crossref]
Andrew, O. E., Umeh, E. U., Elabor, I., Zakari, S., Samuel, E., & Florence Anna, S. (2024). Antibacterial Potential and Time-Kill Kinetics of Calotropis procera Extracts. UMYU Scientifica, 3(4), 459–468. [Crossref]
Anselin, L. (1995). Local indicators of spatial association—LISA. Geographical Analysis, 27(2), 93–115. [Link]
Azabo, R., Dulle, F., Mshana, S. E., Matee, M., & Kimera, S. (2022). Antimicrobial use in cattle and poultry production and occurrence of multidrug-resistant Escherichia coli: A systematic review with focus on sub-Saharan Africa. Frontiers in Veterinary Science, 9, 1000457. [Crossref]
Babo Martins, S., Sucena Afonso, J., Fastl, C., Huntington, B., & Rushton, J. (2024). The burden of antimicrobial resistance in livestock: A framework to estimate its impact within the Global Burden of Animal Diseases programme. One Health, 19, 100917. [Crossref]
Bengtsson-Palme, J., Abramova, A., Berendonk, T. U., Coelho, L. P., Forslund, S. K., Gschwind, R., & Zahra, R. (2023). Towards monitoring of antimicrobial resistance in the environment: For what reasons, how to implement it, and what are the data needs? Environment International, 178, 108089. [Crossref]
Chokshi, A., Sifri, Z., Cennimo, D., & Horng, H. (2019). Global contributors to antibiotic resistance. Journal of Global Infectious Diseases, 11(1), 36–42. [Crossref]
Criscuolo, N. G., Pires, J., Van Boeckel, T. P., & Resistance Bank Team. (2021). ResistanceBank.org: An open-access repository for surveys of antimicrobial resistance in animals. Scientific Data, 8, 187. [Crossref]
Davis, K., Limato, R., Monga, M., Egid, B., Paul, S., Okioma, S., & Steege, R. (2025). Antimicrobial resistance, equity and justice in low- and middle-income countries: An intersectional critical interpretive synthesis. Nature Communications, 16, 9078. [Crossref]
Ducat, W. H., & Kumar, S. (2015). A systematic review of professional supervision experiences and effects for allied health practitioners working in non-metropolitan health care settings. Journal of Multidisciplinary Healthcare, 8, 397–407
Fang, H., Li, D., Wei, X., Wang, R., Su, Y., Yu, Y., Bai, H., Gao, H., Hou, J., Mao, D., & Luo, Y. (2025). Global dynamics of Streptococcus agalactiae antibiotic resistance: Past, present, and future. Journal of Hazardous Materials, 499, Article 140199. [Crossref]
Food and Agriculture Organization of the United Nations. (2017). The future of food and agriculture: Trends and challenges. FAO. [Link]
Garcia Vozmediano, A., Romano, A., Begovoeva, M., Pitti, M., Crescio, E., Brenda, A., Di Roberto, M., Gioia, A., Giraldo, A., Massone, E., Nobile Lanzarini, M., Raggio, A., De Vita, E., Ru, G., & Maurella, C. (2025). Integrating statistical and machine learning approaches for Salmonella enterica surveillance in northwestern Italy: A One Health data driven framework. Microorganisms, 13(12), 2773. [Crossref]
Grace, D. (2015). Review of evidence on antimicrobial resistance and animal agriculture in developing countries. International Livestock Research Institute. [Link]
Hamisu, S., & Salisu, B. (2025). GC-MS analysis and Synergistic Inhibition of Staphylococcus aureus , Streptococcus pyogenes and Dermatophytes by Novel Plant Oil Blends Developed for Skin and Hair Therapy. UMYU Journal of Microbiology Research (UJMR), 10(1), 284–295. [Crossref]
Hosain, M. Z., Kabir, S. M. L., & Kamal, M. M. (2021). Antimicrobial uses for livestock production in developing countries. Veterinary World, 14(1), 210–221. [Crossref]
Iskandar, K., Molinier, L., Hallit, S., Sartelli, M., Hardcastle, T. C., Haque, M., & Roques, C. (2021). Surveillance of antimicrobial resistance in low- and middle-income countries: A scattered picture. Antimicrobial Resistance & Infection Control, 10, 63. [Crossref]
Kou, R., Wang, H., Zou, D., Hu, J., Wu, Y., Xu, Q., & Zhang, X. (2025). Spatial panel data analysis of antimicrobial resistance in Escherichia coli in China. Scientific Reports, 15, 23753. [Crossref]
Legenza, L., McNair, K., Gao, S., Lacy, J. P., Olson, B. J., Fritsche, T. R., & Rose, W. E. (2023). A geospatial approach to identify patterns of antibiotic susceptibility at a neighborhood level in Wisconsin, United States. Scientific Reports, 13, 7122. [Crossref]
Lewnard, J. A., Charani, E., Gleason, A., Hsu, L. Y., Khan, W. A., Karkey, A., & Laxminarayan, R. (2024). Burden of bacterial antimicrobial resistance in low-income and middle-income countries avertible by existing interventions. The Lancet, 403(10442), 2439–2454. [Crossref]
Ma, T., McAllister, T. A., & Guan, L. L. (2021). A review of the resistome within the digestive tract of livestock. Frontiers in Microbiology, 12, 761773. [Crossref]
Matheou, A., Abousetta, A., Pascoe, A. P., Papakostopoulos, D., Charalambous, L., Panagi, S., & Johnson, E. O. (2025). Antibiotic use in livestock farming: A driver of multidrug resistance? Microorganisms, 13(4), 779. [Crossref]
Meyer, C., Price, S., & Ercumen, A. (2024). Do animal husbandry operations contaminate groundwater sources with antimicrobial resistance: A systematic review. Environmental Science and Pollution Research, 31(11), 16164–16176. [Crossref]
Mohammed, Na ’inna, S. Z., Yusha ’u, M., Salisu, B., Adamu, U., & Garba, S. A. (2017). In vitro Antibacterial Activity of Psidium guajava Leaves Extracts against Clinical Isolates of Salmonella specie. UMYU Journal of Microbiology Research (UJMR), 2(1), 1–5. [Link]
Moher, D., Shamseer, L., Clarke, M., Ghersi, D., Liberati, A., Petticrew, M., Shekelle, P., Stewart, L. A., & PRISMA-P Group. (2015). Preferred reporting items for systematic review and meta-analysis protocols (PRISMA-P) 2015 statement. Systematic Reviews, 4, 1.
Moran, P. A. P. (1950). Notes on continuous stochastic phenomena. Biometrika, 37(1–2), 17–23.
Mulchandani, R., Zhao, C., Tiseo, K., Pires, J., & Van Boeckel, T. P. (2024). Predictive mapping of antimicrobial resistance for Escherichia coli, Salmonella, and Campylobacter in food-producing animals, Europe, 2000–2021. Emerging Infectious Diseases, 30(1), 96–104. [Crossref]
Murray, C. J. L., Ikuta, K. S., Sharara, F., Swetschinski, L., Robles Aguilar, G., Gray, A., & Naghavi, M. (2022). Global burden of bacterial antimicrobial resistance in 2019: A systematic analysis. The Lancet, 399(10325), 629–655. [Crossref]
Naghavi, M., Vollset, S. E., Ikuta, K. S., Swetschinski, L. R., Gray, A. P., Wool, E. E., & GBD 2021 AMR Collaborators. (2024). Global burden of bacterial antimicrobial resistance 1990–2021: A systematic analysis with forecasts to 2050. The Lancet, 404(10459), 1199–1226. [Crossref]
Odey, T. O. J., Tanimowo, W. O., Afolabi, K. O., Jahid, I. K., & Reuben, R. C. (2024). Antimicrobial use and resistance in food animal production: Food safety and associated concerns in sub-Saharan Africa. International Microbiology, 27(1), 1–23. [Crossref]
Ou, X., Li, X., Pei, S., Zhao, B., Feng, L., Teng, C., Lu, Y., Zhu, H., Zhou, Y., Xia, H., Liu, Z., Wang, X., Wang, Y., Anthony, R., & Zhao, Y. (2025). High proportion of tuberculosis recent transmission in rural areas of Northeastern China: A 3-year prospective population-based genotypic and spatial analysis in Hinggan League, China. Microbiology Spectrum, 13(8), e00169-25. [Crossref]
Panicker, A., Changaroth, A., Gangadharan, S. V., Yamamoto, T., Noothalapati, H., Nambudiri, M. K., & Poornachandran, P. (2025). From farms to homes: Navigating antimicrobial resistance landscapes from livestock to humans. One Health Advances, 3, 20. [Crossref]
Prestinaci, F., Pezzotti, P., & Pantosti, A. (2015). Antimicrobial resistance: A global multifaceted phenomenon. Pathogens and Global Health, 109(7), 309–318. [Crossref]
Rabiu, M., Abdulkadir, B., Aliyu, I. A., & Kumurya, A. S. (2022). Systematic Review on the Antibacterial Resistance of Vibrio Cholerae. UMYU Scientifica, 1(1), 60–66. [Crossref]
Saheed, Y., Titus, I., Iliyasu, M. Y., & Umar, A. F. (2025). Molecular Detection of BlaSHV and Sul2 Genes in Multidrug-Resistant Shigella flexneri from Clinical Specimens in North-East Nigeria. UMYU Scientifica, 4(3), 182–188. [Crossref]
Salisu, B. D., Magaji, A. M., & Abdulkadir, B. (2017). Phytochemical Determination and In Vitro Antimicrobial Activity of Crude Ethanolic Extract of Stem Bark of Boswellia dalzielii. International Journal of Science and Research (IJSR), 6(12), 1484–1492. [Crossref]
Salisu, Dandashire, B., Magashi, A. M., Abdulkadir, B., Abbas, M. A., Dauda Goni, M., & Yakubu, A. (2019). Toxicological studies and bioactivity-guided identification of antimicrobially active compounds from crude aqueous stem bark extract of Boswellia dalzielii. Journal of Advanced Veterinary and Animal Research Research, 6(2), 183–192. [Crossref]
Shang, Y., Zhang, T. T., Wang, Z. F., Ma, B. Z., Yang, N., Qiu, Y. T., Li, B., Zhang, Q., Zhang, Q. L., Huang, Q. L., & Liu, K. Y. (2022). Spatial epidemiological characteristics and exponential smoothing model application of tuberculosis in Qinghai Plateau, China. Epidemiology and Infection, 150, e37. [Crossref]
Smit, C. C. H., Keighley, C. L., Rogers, K., Miyakis, S., Taxis, K., Sanderson Smith, M., Nicholas, N., Robertson, H. A., & Pont, L. G. (2025). Geotemporal variation in the antimicrobial resistance of Escherichia coli in the community. Antibiotics, 14(3), Article 233. [Crossref]
Sobkowich, K., Poljak, Z., Weese, J. S., Plum, A., Szlosek, D., & Bernardo, T. M. (2024). Prevalence and distribution of carbapenem‑resistant Enterobacterales in companion animals: A nationwide study in the United States using commercial laboratory data. Journal of Veterinary Internal Medicine, 38(5), 2642–2653. [Crossref]
Spets, P., Ebert, K., & Dinnétz, P. (2023). Spatial analysis of antimicrobial resistance in the environment: A systematic review. Geospatial Health, 18(1), 1168. [Crossref]
Tang, K. W. K., Millar, B. C., & Moore, J. E. (2023). Antimicrobial resistance (AMR). British Journal of Biomedical Science, 80, 11387. [Crossref]
Teng, K. T. Y., Martínez Avilés, M., Ugarte Ruiz, M., Bárcena, C., de la Torre, A., López, G., Moreno, M. A., Domínguez, L., & Álvarez, J. (2020). Spatial trends in Salmonella infection in pigs in Spain. Frontiers in Veterinary Science, 7, Article 345. [Crossref]
Trinchera, M., De Gaetano, S., Sole, E., Midiri, A., Silvestro, S., Mancuso, G., & Catalano, T. (2025). Antimicrobials in livestock farming and resistance: Public health implications. Antibiotics, 14(6), 606. [Crossref]
Usman, Z., Mukhtar, F., Salisu, B., & Saleh, A. D. (2025). Integrated Phytochemical Profiling (GC-MS/FTIR), Molecular Docking, and Bioevaluation of Vernonia amygdalina; and &Psidium guajava Against Multidrug-Resistant &Salmonella typhi; UMYU Scientifica, 4(4), 88–111. [Crossref]
Van Boeckel, T. P., Brower, C., Gilbert, M., Grenfell, B. T., Levin, S. A., Robinson, T. P., & Laxminarayan, R. (2015). Global trends in antimicrobial use in food animals. Proceedings of the National Academy of Sciences, 112(18), 5649–5654. [Crossref]
Venkateswaran, N., Swetschinski, L. R., Fastl, C., … collaborators. (2024). Using priorities between human and livestock bacterial antimicrobial resistance (AMR) to identify data gaps in livestock AMR surveillance. BMC Infectious Diseases, 24, 1027. [Crossref]
Warren, J. L., Grandjean, L., Moore, D. A. J., Lithgow, A., Coronel, J., Sheen, P., Zelner, J. L., Andrews, J. R., & Cohen, T. (2018). Investigating spillover of multidrug-resistant tuberculosis from a prison: A spatial and molecular epidemiological analysis. BMC Medicine, 16(1), 122. [Crossref]
World Health Organization, Food and Agriculture Organization of the United Nations, & World Organisation for Animal Health. (2019). Monitoring and evaluation of the global action plan on antimicrobial resistance. WHO. [Link]
World Health Organization. (2001). WHO global strategy for containment of antimicrobial resistance (WHO/CDS/CSR/DRS/2001.2). WHO. [Link]
World Health Organization. (2025). Global antibiotic resistance surveillance report 2025: GLASS summary. WHO. [Link]
World Health organization. (2024). Antimicrobial resistance. [Link]
World Health Organization. (2021). Global antimicrobial resistance and use surveillance system (GLASS) report 2021. [Link]
Zhao, C., Wang, Y., Mulchandani, R., Van Boeckel, T. P., Muloi, D., Tiseo, K., & GBD AMR Collaborators. (2024). Global surveillance of antimicrobial resistance in food animals using priority-drugs maps. Nature Communications, 15, 763. [Crossref]
Zhelyazkova, M., Yordanova, R., Mihaylov, I., Kirov, S., Tsonev, S., Danko, D., Mason, C., & Vassilev, D. (2021). Origin sample prediction and spatial modeling of antimicrobial resistance in metagenomic sequencing data. Frontiers in Genetics, 12, 642991. [Crossref]