Skip to main content
. 2020 Jul 9;8(7):e17216. doi: 10.2196/17216

Table 1.

Accuracy of the convolutional neural network model according to exercise.

Exercise Personal data (%) Total data (%)
Aa Gb A+Gc A G A+G
No exercise 100
(1224/1224)d 		95.8
(1172/1224)		 100
(1224/1224)		 97.4
(1192/1224)		 97.8
(1197/1224)		 100
(1224/1224)
Bilateral flexion 98.5
(1103/1120)		 97.8
(1095/1120)		 99.0
(1109/1120)		 96.1
(1076/1120)		 98.1
(1099/1120)		 97.2
(1089/1120)
Wall push 99.0
(1014/1024)		 93.7
(959/1024)		 100
(1024/1024)		 93.8
(960/1024)		 86.5
(886/1024)		 93.2
(954/1024)
Active scapula 93.0
(1030/1108)		 97.6
(1081/1108)		 100
(1108/1108)		 92.2
(1022/1108)		 87.0
(964/1108)		 93.0
(1030/1108)
Towel slide 100
(1125/1125)		 95.5
(1074/1125)		 100
(1125/1125)		 94.2
(1060/1125)		 88.7
(998/1125)		 95.5
(1074/1125)
Total 98.1
(5496/5601)		 96.0
(5381/5601)		 99.9
(5590/5601)		 94.8
(5310/5601)		 91.8
(5144/5601)		 95.8
(5371/5601)

aAccelerometer data.

bGyroscope data.

cAccelerometer combined with gyroscope data.

dIndicates the value of correct samples divided by total samples.