Predictive Epidemiology: ML Frameworks for Global AMR Tracking and Surveillance

The primary aim of this 3-day workshop is to bridge the gap between data science and public health by equipping participants with the theoretical knowledge and hands-on technical skills required to leverage Machine Learning (ML) and data analytics for global Antimicrobial Resistance (AMR) tracking, risk classification, and surveillance intelligence.

To achieve this aim, the workshop will focus on the following key operational objectives:

• Demystify AMR Data Infrastructure: Introduce participants to global surveillance frameworks (like WHO GLASS) and specialized genomic databases (CARD, ResFinder, NCBI Pathogen Detection).

• Build Data Proficiency: Provide practical training in Python-based data workflows to clean, manipulate, and visualize highly complex, multi-dimensional epidemiological and microbiological datasets.

• Deploy Predictive Frameworks: Guide participants through framing AMR challenges as ML problems, engineering relevant biological/geographical features, and building supervised models (such as Random Forests and XGBoost) to predict resistance risks.

• Translate Insights into Action: Teach participants how to design basic early-warning systems and interactive dashboards (using Streamlit and Plotly) that translate raw data into actionable public health intelligence.

📅 Day 1: Foundations of Predictive Epidemiology and Global AMR Surveillance

• Antimicrobial resistance: global burden, drivers, and public health impact
• Principles of infectious disease surveillance and predictive epidemiology
• AMR data sources: clinical, microbiological, genomic, hospital, environmental, and public health datasets
• Global AMR surveillance systems and reporting frameworks
• Data preparation for AMR analytics: missing values, categorical variables, resistance labels, and temporal trends
• Role of machine learning in AMR detection, trend analysis, and surveillance intelligence
• Tools covered: Google Colab, Python, Pandas, NumPy, Matplotlib, WHO GLASS resources, open AMR datasets

Hands-on Activity:

• AMR Surveillance Dataset Exploration in Google Colab: Clean sample AMR records, explore pathogen-antibiotic resistance patterns, visualize resistance trends, and prepare data for predictive modeling

📅 Day 2: Machine Learning Models for AMR Prediction and Risk Classification

• ML problem framing for AMR: classification, risk scoring, trend prediction, and hotspot identification
• Feature engineering using pathogen, antibiotic, geography, time, patient, and sample-level variables
• Supervised learning models: Logistic Regression, Random Forest, Gradient Boosting, and XGBoost concepts
• Model evaluation for AMR prediction: accuracy, precision, recall, F1-score, ROC-AUC, and confusion matrix
• Interpreting model outputs for epidemiological and public health decision-making
• Bias, uncertainty, data quality, and responsible AI considerations in healthcare surveillance
• Tools covered: Google Colab, Python, Scikit-learn, Statsmodels, Plotly, Seaborn

Hands-on Activity:

• Building an AMR Risk Classification Model: Develop a machine learning model to classify resistance risk, evaluate performance, identify important predictors, and interpret results for surveillance use cases

📅 Day 3: Global AMR Tracking, Genomic Surveillance and Early-Warning Dashboards

• Global AMR tracking: geographic trends, temporal patterns, and hotspot analysis
• Integrating clinical, epidemiological, genomic, and environmental AMR signals
• Introduction to genomic surveillance for AMR: resistance genes, pathogen lineages, and mutation tracking
• AMR databases and platforms: CARD, ResFinder, NCBI Pathogen Detection, Microreact, and Nextstrain
• Early-warning frameworks for emerging resistance threats
• Designing interactive dashboards for AMR surveillance and public health reporting
• Tools covered: Streamlit, Plotly, CARD, ResFinder, NCBI Pathogen Detection, Microreact, Nextstrain, GitHub-based open datasets

Hands-on Activity:

• AMR Surveillance Dashboard and Early-Warning Risk Tracker: Create a simple dashboard concept to visualize resistance trends, identify high-risk pathogen-antibiotic combinations, highlight geographic or temporal patterns, and generate basic early-warning insights

• Public Health & Medical Professionals: Epidemiologists, public health officers, healthcare data analysts, and hospital infection control specialists looking to adopt data-driven predictive tools.

• Life Sciences & Microbiology Researchers: Microbiologists, bioinformatics learners, and antibiotic stewardship teams who want to transition from traditional lab data analysis to advanced computational modeling.

• Data Scientists & Computational Chemists: Computational biologists and industry professionals in pharmaceuticals or global health policy who want to apply ML architectures to urgent, real-world biosecurity and infectious disease threats.

,

• Public Health & Medical Professionals: Epidemiologists, public health officers, healthcare data analysts, and hospital infection control specialists looking to adopt data-driven predictive tools.

• Life Sciences & Microbiology Researchers: Microbiologists, bioinformatics learners, and antibiotic stewardship teams who want to transition from traditional lab data analysis to advanced computational modeling.

• Data Scientists & Computational Chemists: Computational biologists and industry professionals in pharmaceuticals or global health policy who want to apply ML architectures to urgent, real-world biosecurity and infectious disease threats.

• Understand the global significance of antimicrobial resistance and predictive epidemiology
• Identify key AMR data sources used in public health, microbiology, and genomic surveillance
• Clean, analyze, and visualize AMR surveillance datasets using Python workflows
• Apply machine learning models for AMR prediction and risk classification
• Evaluate model performance using healthcare-relevant metrics
• Interpret AMR trends across pathogens, antibiotics, regions, and time periods
• Understand the role of genomic surveillance in resistance tracking
• Explore open AMR platforms and databases for research and surveillance
• Design simple early-warning and dashboard frameworks for AMR monitoring
• Connect ML-based AMR analytics with public health decision-making and global surveillance priorities