Table 7 Baseline results using data preprocessed following the approach proposed by Carreiro et al.¹² learned with 4 classifiers: Naive Bayes (NB), Support Vector Machine (SVM), Random Forests (RF) and XGB (eXtreme Gradient Boosting) to predict the Evolution for each of the target endpoints, C4 and C5, within the considered time windows (90, 180 and 365 days), respectively.

C4—need for a caregiver

 90 days

  NB

76.85 ± 3.44

64.00 ± 9.05

72.77 ± 2.31

  SVM

72.58 ± 3.71

63.89 ± 6.38

68.66 ± 3.29

  RF

75.35 ± 4.08

57.16 ± 8.70

77.13 ± 1.78

  XGB

76.10 ± 3.34

57.37 ± 8.36

76.89 ± 2.06

 180 days

  NB

79.45 ± 2.06

64.21 ± 3.74

75.93 ± 1.99

  SVM

78.63 ± 2.56

72.05 ± 4.84

70.42 ± 2.23

  RF

78.89 ± 1.63

64.46 ± 3.70

76.51 ± 2.32

  XGB

78.61 ± 1.75

64.81 ± 3.50

76.28 ± 2.25

 365 days

  NB

77.61 ± 2.05

58.76 ± 4.18

77.33 ± 2.64

  SVM

77.58 ± 2.10

65.22 ± 2.72

74.74 ± 2.81

  RF

83.33 ± 1.57

75.05 ± 3.20

76.55 ± 2.41

  XGB

80.83 ± 1.43

73.30 ± 2.85

74.07 ± 2.46

C5—need for a wheelchair

 90 days

  NB

80.83 ± 2.92

77.44 ± 8.53

72.16 ± 1.75

  SVM

79.32 ± 2.58

73.60 ± 6.60

68.66 ± 2.16

  RF

79.65 ± 3.23

64.16 ± 8.04

78.78 ± 2.02

  XGB

81.85 ± 2.75

68.48 ± 6.72

77.95 ± 1.96

 180 days

  NB

82.19 ± 1.79

73.14 ± 4.57

74.48 ± 1.60

  SVM

83.90 ± 1.87

81.80 ± 3.91

71.51 ± 1.76

  RF

81.31 ± 1.79

66.55 ± 4.68

79.53 ± 1.53

  XGB

82.13 ± 1.75

68.39 ± 4.73

79.56 ± 1.62

 365 days

  NB

78.53 ± 1.71

66.26 ± 2.98

74.47 ± 1.45

  SVM

81.13 ± 1.97

78.13 ± 3.93

69.66 ± 1.81

  RF

82.54 ± 1.64

68.46 ± 3.18

80.41 ± 1.63

  XGB

80.87 ± 1.30

66.43 ± 3.53

80.06 ± 1.79