Tharmakulasingam M, Gardner B, La Ragione R, Fernando A. Rectified Classifier Chains for Prediction of Antibiotic Resistance From Multi-Labelled Data With Missing Labels. IEEE/ACM Trans Comput Biol Bioinform. 2022;20(1):625-636. doi:10.1109/TCBB.2022.3148577.
This study presents the Rectified Classifier Chain (RCC) method for predicting antimicrobial resistance (AMR) from genomic data, particularly in multi-drug resistance scenarios with missing labels. The RCC model, utilizing an XGBoost base, outperformed other multi-label classification methods, achieving a 3.3% improvement in Hamming accuracy and a 7.8% increase in F1-score compared to the binary relevance model. The study also demonstrated that RCC could identify biomarkers contributing to AMR prediction. This model may aid genome annotation and help discover new biomarkers linked to AMR.
Comments