Loading
Leakage is when information from the test data reaches the model during training, through a scaler fitted on all samples, a patient split across folds or a benchmark that appeared in the pretraining set, making the model look better than it is.
In statistics and machine learning, leakage is the use of information during training that would not be available at prediction time, which produces overly optimistic performance estimates (Wikipedia). Common cancer-data forms are normalising or selecting genes on the full cohort before splitting, duplicated patients across TCGA and cBioPortal studies, and benchmark contamination, where a foundation model's pretraining corpus contained the public test set (TCGA slides or profiles), so that its downstream score is partly recall.
Shares Benchmarks, leaderboards and contamination, Drug-response data splits: leave-cell-line-out, leave-drug-out, leave-tissue-out, Cancer AI vocabulary (CanSim terms map) and the tag cansim-terms.
Shares Cross-validation and stratified k-fold, Drug-response data splits: leave-cell-line-out, leave-drug-out, leave-tissue-out, Train, validation and test split discipline, Cancer AI vocabulary (CanSim terms map) and the tag cansim-terms.
Shares Train, validation and test split discipline, Cancer AI vocabulary (CanSim terms map) and the tag cansim-terms.
Shares Batch effects and harmonisation, Cancer AI vocabulary (CanSim terms map) and the tag cansim-terms.
Shares Drug-response data splits: leave-cell-line-out, leave-drug-out, leave-tissue-out, Cancer AI vocabulary (CanSim terms map) and the tag cansim-terms.
Shares Cross-validation and stratified k-fold, Cancer AI vocabulary (CanSim terms map) and the tag cansim-terms.
Shares Batch effects and harmonisation, Cancer AI vocabulary (CanSim terms map) and the tag cansim-terms.
Shares Principal component analysis (PCA) as a feature compressor, Cancer AI vocabulary (CanSim terms map) and the tag cansim-terms.