Table 7 Central model performance with different values of ‘\(\:\alpha\:\)’ parameter (TCH study + Ohio T1DM data).

From: A personalized federated learning-based glucose prediction algorithm for high-risk glycemic excursion regions in type 1 diabetes

Alpha

Clark’s EGA Analysis

# of patients HH loss improves over MSE for

Zone: A + B

(True predictions)

Zone: C + D + E

(False predictions)

Hypoglycemia (125 patients)

Hyperglycemia (123 patients)

0

97.58

0.41

64

93

0.1

98.92

1.08

101

70

0.2

99.11

0.89

108

60

0.3

99.18

0.82

107

99

0.4

99.19

0.81

111

90

0.5

99.21

0.79

109

78

0.6

99.28

0.72

115

67

0.7

99.33

0.68

117

62

0.8

99.33

0.67

120

53

0.85

99.34

0.66

121

27

0.9

99.36

0.65

124

18

0.95

99.39

0.61

124

14

1

99.37

0.63

125

0

MSE

97.78

2.21

-

-

Back to article page