XGBoost
Overview
XGBoost (eXtreme Gradient Boosting) is a gradient-boosted decision tree machine learning framework known for high performance on structured data. It is widely used for classification and regression tasks in cancer genomics PMID:27634761.
Used by
- PMID:27634761 — XGBoost was used as the core machine learning framework for the ATLAS (AI Tumor Lineage and Site) classifier; trained on 8,249 RNA-seq samples from TCGA and CCLE to classify 22 cancer sites of origin and 8 cancer lineages; achieved 91.4% site-of-origin accuracy and 97.1% lineage accuracy on a validation set of 10,376 samples; high-confidence predictions (score >= 0.99) achieved 98–99% accuracy PMID:27634761.
- Gradient boosting (XGBoost-class) model trained to predict ICI response (ROC-AUC=0.77 training, 0.78 JAVELIN validation) and TKI response (ROC-AUC=0.74 validation, n=822) in ccRCC, outperforming all published single-biomarker signatures PMID:22138691
Notes
- XGBoost outperformed random forests and SVMs in the ATLAS cancer classifier benchmark; RNA expression alone sufficed without requiring DNA mutation or copy number features PMID:27634761.
- Corpus-grown slug; not present in canonical ontology.
Sources
This page was processed by entity-page-writer on 2026-04-11. - PMID:22138691
This page was processed by wiki-cli on 2026-05-06.