PLT Predicting

This dataset accompanies a retrospective study on predicting platelet recovery in patients with cancer treatment-induced thrombocytopenia (CTIT) receiving recombinant human thrombopoietin (rh-TPO). Data from 400 patients treated in 2023 were used to build a stacking ensemble machine learning model. Key features were selected via LASSO regression, and model performance was evaluated using an independent test set with SHAP analysis for interpretation. The dataset can be used to reproduce the predictive model and support research on individualized management of CTIT.