📚 Syllabus & Course Curriculum
Antimicrobial Resistance (AMR)
Module-by-module breakdown of Machine Learning Approaches for Predicting Antimicrobial Resistance (AMR), from foundations to a certified capstone project.
Problem
• Predicting phenotype from genotype: the task and its ceiling
• Binary resistance calls versus MIC regression
• Label quality: phenotypic testing error propagating into training data
Features
• Gene presence-absence, k-mer and SNP-based representations
• Pan-genome construction and reference bias
• Population structure as a confounder that inflates cross-validation scores
Models
• Regularised models and tree ensembles on high-dimensional genomic features
• Phylogeny-aware cross-validation to avoid leakage through relatedness
• Handling severe class imbalance for rare resistance phenotypes
Interpretation
• Feature attribution to known resistance determinants as a sanity check
• Discovering candidate novel determinants and validating them
• Distinguishing mechanism from lineage marker
Translation
• Turnaround time and the clinical decision the prediction must beat
• Regulatory expectations for genomic AST prediction
• Monitoring model performance as resistance mechanisms evolve
Earn government-registered certification in Machine Learning Approaches for Predicting Antimicrobial Resistance (AMR)
e-Certificate and e-Marksheet issued on successful completion.
