Academic project · 2025–2026
Genomic AMR prediction
Evaluating genomic predictors of resistance, with grouped validation and an explicit account of what internal results can establish.
Python / Pandas / scikit-learn
- Methods
- Metadata matching; Profile-grouped cross-validation; Fold-specific feature selection and bootstrap intervals.
- Results
- Internal balanced accuracy of 0.949 for azithromycin and 0.969 for ciprofloxacin; independent validation is still needed.
- Limitations
- Internal cross-validation only. Supplied features were already GWAS-filtered, so upstream selection leakage cannot be excluded. These results do not establish clinical utility.