Most optimal result for NPM1 mutation prediction using a cohort of 275 cases of AML divided in a training and test set.²¹  The number of probe sets used in this prediction was 22, which represents 18 unique genes. For the 10-fold CV error set, n = 184; for the error validation set, n = 91.

Ten-fold cross-validation (CV) prediction error on training set (n = 184); prediction error on validation set (n = 91). In 10-fold cross-validation, the model is fitted on 90% of the samples and the class of the remaining 10% is predicted. This procedure is repeated 10 times, with each part playing the role of the test samples and the error of all 10 parts added together to compute the overall error. The error within the validation set reflects the number of samples wrongfully predicted in this set.