The optimized XGBoost algorithm demonstrated good discrimination ability with an area under the curve (AUC) of

/XGBoost model achieves AUC 0.766 (0.875 with reliable cell lines) with significant per-drug score separation (p=0.003)