Table 8 Summary of the best AUC results obtained with the triclustering-based classification approach for each of the target endpoints according to each of the considered time windows.

From: Triclustering-based classification of longitudinal data for prognostic prediction: targeting relevant clinical endpoints in amyotrophic lateral sclerosis

 

90 days

180 days

365 days

C1

86.24 ± 4.03

83.33 ± 2.71

86.63 ± 3.39

(XGB; D; 5 CS)

(RF; D; 5 CS)

(RF; C; 5 CS)

C2

94.12 ± 3.14

94.14 ± 1.84

93.63 ± 3.23

(RF; D; 5 CS)

(RF; D; 4 CS)

(RF; D; 5 CS)

C3

91.53 ± 5.28

93.23 ± 2.87

89.92 ± 5.38

(XGB; D; 4 CS)

(XGB; D; 3 CS)

(XGB; D; 5 CS)

C4

85.52 ± 4.10

86.35 ± 2.43

91.58 ± 2.36

(RF; D; 3 CS)

(RF; D; 5 CS)

(RF; D; 5 CS)

C5

85.18 ± 5.60

81.23 ± 3.34

81.45 ± 4.92

(SVM; C; 4 CS)

(RF; D; 5 CS)

(RF; D; 5 CS)

  1. D stands for distance matrices as learning examples, while C stands for correlation matrices. C1, need for NIV; C2, need for an auxiliary communication device; C3, need for PEG; C4, need for a caregiver, and C5, need for a wheelchair.
Back to article page