Ontology-based venous thromboembolism risk assessment model developing from medical records

Table 4 Comparison of AUC scores of four ML models using word vectors with different dimensions

Dimension	10	20	40	80	100	120	150	200
AUC (GBDT)	0.863 ± 0.024	0.891 ± 0.023	0.916 ± 0.026	0.916 ± 0.017	0.929 ± 0.014	0.929 ± 0.015	0.927 ± 0.020	0.927 ± 0.020
AUC (RF)	0.852 ± 0.022	0.871 ± 0.030	0.883 ± 0.023	0.881 ± 0.029	0.884 ± 0.021	0.893 ± 0.020	0.897 ± 0.018	0.884 ± 0.019
AUC (LR)	0.851 ± 0.023	0.884 ± 0.019	0.912 ± 0.019	0.921 ± 0.020	0.926 ± 0.022	0.923 ± 0.022	0.920 ± 0.026	0.923 ± 0.014
AUC (SVM)	0.862 ± 0.019	0.873 ± 0.034	0.861 ± 0.031	0.879 ± 0.021	0.869 ± 0.025	0.880 ± 0.031	0.867 ± 0.023	0.830 ± 0.034

The value of AUC score is formatted with ‘Mean value ± Standard deviation’. All terms are used to train models

ISSN: 1472-6947