Michał Jan Burdukiewicz

This study evaluates class-balancing algorithms on metabolomics datasets from brain tumor patients and controls, finding that undersampling improves model performance while oversampling risks overfitting.

The paper identifies influential data points in linear regression analyses of 100 scientific papers, finding that 29 relied on a single influential point for their conclusions.

This study shows that AMP prediction benchmarks are biased due to negative data selection and introduces a web server for fair benchmarking.

PCRedux analyzes qPCR amplification curves to calculate interpretable features for machine learning, achieving high sensitivity and specificity in classification.

MegaGO assesses functional similarity between meta-omics datasets using semantic GO term similarity, offering fast and scalable analysis.