Fig. 3

Optimization of the MDS diagnostic work-up. A Decision tree based on a threshold equal to 0.23 to guide PLT-F reflex test and an additional cut-off at 3% for IPF to guide slide review. According to this decision tree, slide review was not indicated in 375 patients (including 349 non-MDS and 26 MDS patients) and suggested in 150 patients (including 8 non-MDS and 142 MDS patients). B Performance of this decision tree in real-life considering all slide review criteria. 18 of 26 misclassified patients with MDS had additional criteria for slide review. Combining the decision tree and the other criteria for slide review, 349 of non-MDS (97.8%) and 160 of MDS patients (95.2%) were correctly classified. C Histogram showing the frequency of MDS and non-MDS patients according to the MDS-CBC score. D Decision tree based on a threshold equal to 0.23 and inferior to 0.6 to guide PLT-F reflex test and an additional cut-off at 3% for IPF to guide slide review. Slide review was not indicated in 361 patients (including 342 non-MDS and 19 MDS patients) and suggested in 164 patients (including 15 non-MDS and 149 MDS patients). E Performance of this decision tree in real-life considering all slide review criteria. 13 of 19 misclassified patients with MDS had additional criteria for slide review. Combining the decision tree and the other criteria for slide review, 342 of non-MDS (95.8%) and 162 of MDS patients (96.4%) were correctly classified