Skip to main content

Table 10 FOG Prediction: One-freezer-held-out cross validation with the 2-layer LSTM model after 4 training epochs

From: Prediction and detection of freezing of gait in Parkinson’s disease from plantar pressure data using long short-term memory neural-networks

Participant held out

Sensitivity (%)

Specificity (%)

Precision (%)

F1 score

1

82.5

60.2

15.3

0.26

2

62.7

83.9

36.0

0.46

3

68.1

71.8

4.8

0.09

6

78.8

88.6

18.4

0.30

7

84.1

71.6

54.6

0.66

8

90.8

74.4

17.7

0.30

9

92.6

83.4

5.2

0.10

Mean ± SD

79.9 ± 10.3

76.3 ± 9.0

21.7 ± 16.5

0.31 ± 0.19