Table 4 Performance and computational time comparison of different models on PE detection.

From: Harnessing hybrid stacking ensemble learning for accurate pulmonary embolism diagnosis using tabular clinical data

Model

Acc (%)

Pre (%)

Rec (%)

F1-score (%)

Training time/Fold (Avg.)

MLP

76.5

74.1

72.5

73.2

6.1 min

LGBM

78.3

75.5

74.3

72.6

2.4 min

XGB

85.7

82.1

80.8

81.6

3.6 min

SAINT

90.4

87.1

85.2

86.6

20.3 min

MPA-optimized-HSE (Proposed)

92.3

88.2

87.4

89.5

41.7 min

Back to article page