A potential predictive model based on machine learning and CPD parameters in elderly patients with aplastic anemia and myelodysplastic neoplasms

Table 3 Comparative evaluation of six machine learning models for ten-fold resampling-validation

Classifier	Cohorts	AUC	Cut off	Accuracy	Sensitivity	Specificity	Positive predictive value	Negative predictive value	F1
XGBoost	Training cohort	1.000	0.794	0.992	1.000	1.000	1.000	0.986	1.000
XGBoost	Validation cohort	0.759	0.794	0.705	0.679	0.842	0.765	0.629	0.712
Logistic Regression	Training cohort	0.842	0.555	0.778	0.672	0.878	0.813	0.759	0.735
Logistic Regression	Validation cohort	0.827	0.555	0.762	0.750	0.866	0.773	0.770	0.748
LightGBM	Training cohort	0.984	0.495	0.937	0.938	0.952	0.939	0.937	0.938
LightGBM	Validation cohort	0.770	0.495	0.697	0.728	0.780	0.708	0.714	0.712
Random Forest	Training cohort	1.000	0.545	0.988	1.000	0.997	0.998	0.981	0.999
Random Forest	Validation cohort	0.787	0.545	0.678	0.750	0.747	0.808	0.620	0.769
AdaBoost	Training cohort	0.994	0.499	0.958	0.978	0.956	0.948	0.967	0.962
AdaBoost	Validation cohort	0.705	0.499	0.676	0.705	0.718	0.618	0.734	0.644
SVM	Training cohort	0.819	0.428	0.766	0.780	0.770	0.723	0.806	0.749
SVM	Validation cohort	0.780	0.428	0.719	0.716	0.825	0.698	0.747	0.700

XGBoost eXtreme gradient boosting, LightGBM light gradient boosting machine, AdaBoost Adaptive boosting, SVM Support vector machines

ISSN: 1472-6947