Kerek Ádám, Szabó Ábel, Jerzsele Ákos
Department of Pharmacology and Toxicology, University of Veterinary Medicine, István utca 2, HU-1078 Budapest, Hungary.
National Laboratory of Infectious Animal Diseases, Antimicrobial Resistance, Veterinary Public Health and Food Chain Safety, University of Veterinary Medicine, István utca 2, HU-1078 Budapest, Hungary.
Antibiotics (Basel). 2025 May 10;14(5):491. doi: 10.3390/antibiotics14050491.
: Antimicrobial resistance (AMR) poses a growing threat to veterinary medicine and food safety. This study examines antibiotic resistance patterns in ducks, focusing on multidrug-resistant (MDR) strains. Understanding resistance patterns and predicting MDR occurrence are critical for effective intervention strategies. : isolates were collected from duck samples across multiple regions. Descriptive statistics and resistance frequency analyses were conducted. A decision tree classifier and a neural network were trained to predict MDR status. Cross-resistance relationships were visualized using graph-based models, and Monte Carlo simulations estimated MDR prevalence variations. : Monte Carlo simulations estimated an average MDR prevalence of 79.6% (95% CI: 73.1-86.1%). Key predictors in MDR classification models were enrofloxacin, neomycin, amoxicillin, and florfenicol. Strong cross-resistance associations were detected between neomycin and spectinomycin, as well as amoxicillin and doxycycline. : The high prevalence of MDR strains underscores the urgent need to revise antibiotic usage guidelines in veterinary settings. The effectiveness of predictive models suggests that machine learning tools can aid in the early detection of MDR, contributing to the optimization of treatment strategies and the mitigation of resistance spread. The alarming MDR prevalence in isolates from ducks reinforces the importance of targeted surveillance and antimicrobial stewardship. Predictive models, including decision trees and neural networks, provide valuable insights into resistance trends, while Monte Carlo simulations further validate these findings, emphasizing the need for proactive antimicrobial management.
抗菌药物耐药性（AMR）对兽医学和食品安全构成了日益严重的威胁。本研究调查了鸭的抗生素耐药模式，重点关注多重耐药（MDR）菌株。了解耐药模式并预测MDR的出现对于有效的干预策略至关重要。从多个地区的鸭样本中收集了分离株。进行了描述性统计和耐药频率分析。训练了决策树分类器和神经网络来预测MDR状态。使用基于图的模型可视化交叉耐药关系，并通过蒙特卡洛模拟估计MDR流行率的变化。蒙特卡洛模拟估计MDR的平均流行率为79.6%（95%置信区间：73.1 - 86.1%）。MDR分类模型中的关键预测因素是恩诺沙星、新霉素、阿莫西林和氟苯尼考。在新霉素与壮观霉素之间以及阿莫西林与强力霉素之间检测到强烈的交叉耐药关联。MDR菌株的高流行率凸显了在兽医环境中修订抗生素使用指南的迫切需要。预测模型的有效性表明，机器学习工具可有助于早期检测MDR，有助于优化治疗策略并减轻耐药性传播。鸭分离株中令人担忧的MDR流行率强化了针对性监测和抗菌药物管理的重要性。包括决策树和神经网络在内的预测模型提供了有关耐药趋势的宝贵见解，而蒙特卡洛模拟进一步验证了这些发现，强调了积极进行抗菌药物管理的必要性。
