Skip to main content

Table 1 Performance metrics for the I3D and ResNet-18 CNN models

From: A novel in-bed body posture monitoring for decubitus ulcer prevention using body pressure distribution mapping

 

I3D model

ResNet-18 model

LOSO model

LOEO model

LOSO model

LOEO model

Accuracy (%)

82.22 ± 8.50

77.78 ± 9.76

92.07 ± 5.72

85.37 ± 14.38

F1-Score (%)

84.63 ± 6.75

80.75 ± 8.82

93.26 ± 4.14

87.40 ± 10.54

Sensitivity (%)

84.28 ± 6.79

79.56 ± 11.29

93.33 ± 4.81

87.69 ± 11.47

Specificity (%)

96.30 ± 1.57

95.36 ± 2.46

98.36 ± 1.15

97.01 ± 2.87