Research Article

Diagnostic Model of In-Hospital Mortality in Patients with Acute ST-Segment Elevation Myocardial Infarction Used Artificial Intelligence Methods

Table 3

Confusion matrix and AUC.

 Original to originalDownsampling to original
TPaTNbFPcFNdAccuracyRecallSpecificityPrecisionF1 scoreF2 scoreAUCTPaTNbFPcFNdAccuracyRecallSpecificityPrecisionF1 scoreF2 scoreAUC

Logistic regressionTraining10393907475341300.890.200.980.580.300.230.594243377514039260.770.820.730.750.780.810.77
Test9873792875839080.890.200.980.570.300.230.5939792894497429160.760.810.750.290.430.600.78
Validation9943576473330070.910.250.980.580.350.280.6132942809484037070.780.820.770.280.420.590.80

Decision treeTraining13613927455338080.900.260.990.710.380.300.624316402211568530.810.830.780.790.810.830.81
Test9973779289438980.890.200.980.530.290.230.593735286471003911600.740.760.740.270.400.560.75
Validation8923557092731090.900.220.970.490.310.250.6030942776587329070.760.770.760.260.390.560.77

XgboostTraining13393925157638300.900.260.990.700.380.300.624496390412746730.810.870.750.780.820.850.81
Test10413788879838540.890.210.980.570.310.240.60401128110105768840.740.820.730.270.410.590.77
Validation8923557092731090.900.220.970.490.310.250.6033062725392446950.750.830.750.260.400.580.79

K nearest neighbourTraining3703968913847990.890.071.000.730.130.090.5336994128105014700.760.720.800.780.750.730.76
Test2803853914746150.890.061.000.660.110.070.53324631054763216490.790.660.800.300.410.530.73
Validation2903633216537110.900.071.000.640.130.090.53267829935656213230.810.670.820.290.400.530.74

Multilayer perceptronTraining10823906276540870.890.210.980.590.310.240.604398363415447710.780.850.700.740.790.830.78
Test10433792975738520.890.210.980.580.310.240.60412928018106687660.740.840.720.280.420.600.74
Validation8043599650131970.910.200.990.620.300.230.59347726476100215240.740.870.730.260.400.590.74