AMR Data Challenge 2025: AMR Data Challenge 2025: Identifying patterns and drivers of AMR among pathogens in India

Our data analytics project will leverage datasets from the GASAR and PLEA studies, encompassing over 3,876 non-duplicate gram-negative bacterial isolates collected across India between 2011 and 2023. The data includes detailed phenotypic susceptibility profiles and genotypic information on key resistance genes such as blaTEM, blaSHV, blaCTX-M, blaNDM, blaVIM, and blaOXA.
The project will focus on identifying patterns and drivers of antimicrobial resistance among pathogens including E. coli, Klebsiella pneumoniae, Pseudomonas aeruginosa, Escherichia coli, Pseudomonas aeruginosa, and Acinetobacter baumannii. The sources of the samples are from blood, skin, urine, respiratory channels. Using statistical modeling and machine learning approaches, the project will:
• Analyze temporal and spatial trends in resistance profiles across the three study phases.
• Identify correlations between resistance phenotypes and the presence of specific resistance genes.
• Use clustering and dimensionality reduction techniques to group isolates by resistance mechanisms and antibiotic susceptibility profiles.
• Evaluate the effectiveness of key antibiotics over time, particularly carbapenems and polymyxins.
• Explore the potential emergence and spread of multidrug-resistant strains by source and region.
The outputs will include heatmaps, resistance trend plots, gene-phenotype correlation matrices, and interactive dashboards to visualize and interpret key findings. These insights can support the optimization of empirical therapy, guide infection control strategies, and contribute to national and regional antibiotic stewardship efforts.
The project will be structured in phases: data cleaning and preprocessing (Week 1–2), exploratory and statistical analysis (Week 3–5), advanced modeling (Week 6–7), and final visualization/reporting (Week 8).

本数据分析项目将依托GASAR与PLEA研究的数据集，该数据集涵盖2011年至2023年间在印度境内采集的3876余株非重复革兰氏阴性细菌分离株（gram-negative bacterial isolates）。数据集包含详细的表型药敏谱（phenotypic susceptibility profiles）信息，以及针对blaTEM、blaSHV、blaCTX-M、blaNDM、blaVIM、blaOXA等关键耐药基因的基因型数据。
本项目将聚焦于明确大肠杆菌（E. coli）、肺炎克雷伯菌（Klebsiella pneumoniae）、铜绿假单胞菌（Pseudomonas aeruginosa）、大肠埃希菌（Escherichia coli）、铜绿假单胞菌（Pseudomonas aeruginosa）以及鲍曼不动杆菌（Acinetobacter baumannii）等病原菌的抗菌药物耐药（antimicrobial resistance）模式与驱动因素。样本来源涵盖血液、皮肤、尿液及呼吸道。本项目将采用统计建模与机器学习方法开展如下工作：
• 分析三个研究阶段内耐药谱的时间与空间分布趋势；
• 明确耐药表型与特定耐药基因携带状态之间的关联；
• 采用聚类与降维技术，依据耐药机制与抗生素药敏谱对分离株进行分组；
• 评估关键抗生素（尤其是碳青霉烯类（carbapenems）与多粘菌素类（polymyxins））随时间推移的抗菌有效性；
• 按样本来源与地域，探究多重耐药菌株（multidrug-resistant strains）的潜在出现与传播态势。
项目产出将包括热图、耐药趋势图、基因-表型关联矩阵，以及用于可视化解读核心研究结果的交互式仪表盘（interactive dashboards）。这些研究结果可辅助优化经验性治疗（empirical therapy）方案、指导感染防控策略制定，并助力国家及区域层面的抗生素管理（antibiotic stewardship）工作。
本项目将按阶段推进：第1-2周完成数据清洗与预处理，第3-5周开展探索性与统计分析，第6-7周进行高级建模，第8周完成最终可视化与报告撰写。