Skip to main content

Table 6 Comparison of the performance of Artificial Neural Networks (ANN) classifier with gradient descent backpropagation using hidden units {1, 2, 4, 8, 32} and the momentum {0,0.2,0.5,0.9} using sampling

From: Comparison of machine learning techniques to predict all-cause mortality using fitness data: the Henry ford exercIse testing (FIT) project

  H = 1 H = 2 H = 4 H = 8 H = 32
  M = 0 M = 0.2 M = 0.5 M = 0.9 M = 0 M = 0.2 M = 0.5 M = 0.9 M = 0 M = 0.2 M = 0.5 M = 0.9 M = 0 M = 0.2 M = 0.5 M = 0.9 M = 0 M = 0.2 M = 0.5 M = 0.9
Sensitivity 54.27% 45.83% 57.63% 56.81% 56.15% 50.86% 55.79% 55.61% 55.89% 47.38% 54.87% 51.87% 50.94% 51.02% 43.82% 45.96% 44.17% 58.70%
Specificity 88.01% 88.35% 88.01% 89.10% 90.51% 90.61% 90.78% 90.30% 89.82% 90.13% 90.43% 89.67% 90.31% 90.29% 90.39% 89.77% 90.43% 90.60% 90.19% 88.09%
Precision 0 3.60% 0 11.90% 24.93% 26.00% 27.70% 23.60% 18.47% 21.47% 24.37% 17.77% 23.30% 23.57% 24.50% 18.33% 25.87% 27.13% 23.37% 0.90%
F-score 0 6.75% 0 18.89% 34.81% 35.67% 37.10% 32.24% 27.75% 30.98% 33.94% 25.84% 32.71% 32.41% 33.09% 26.97% 32.53% 34.12% 30.56% 1.77%
RMSE 0.30 0.30 0.30 0.30 0.29 0.29 0.29 0.30 0.29 0.29 0.29 0.30 0.30 0.30 0.30 0.32 0.32 0.32 0.32 0.35
AUC 0.77 0.76 0.74 0.72 0.8 0.79 0.77 0.72 0.8 0.81 0.82 0.78 0.81 0.81 0.81 0.68 0.77 0.77 0.78 0.52