Skip to main content

Advertisement

Table 7 Comparison of the results on training- and validation set (when the dataset contains two trends both trends are evaluated separately)

From: Developing a system that can automatically detect health changes using transfer times of older adults

Training scenariosa Validation scenariosb
Detection Rate (DR)
T r SU 100 % V SU 100 %
T r US 100 % V US 100 %
T r SUS   V SUS  
S →U 100 % S →U 100 %
U →S 100 % U →S 100 %
T r USU   V USU  
U →S 100 % S →U 100 %
S →U 100 % U →S 100 %
Average Run Length (ARL)
T r SU 8.05±3.62 V SU 7.70±3.34
T r US 12.95±4.31 V US 10.95±3.59
T r SUS   V SUS  
SU 7.65±4.10 S →U 8.35±3.59
U →S 11±4.30 U →S 12.45±5.25
T r USU   V USU  
U →S 10.15±5.96 U →S 10±3.54
S →U 8±5.58 S →U 8.10±2.88
Average number of false alarms per week (FPR)
T r S 0.15±0.17 V S 0.16±0.07
T r U 0.11±0.07 V U 0.14±0.06
T r SU 0.20±0.10 V SU 0.20±0.07
T r US 0.18±0.07 V US 0.17±0.07
T r SUS 0.23±0.08 V SUS 0.20±0.08
T r USU 0.23±0.07 V USU 0.20±0.08
  1. Notes
  2. aTraining simulation scenarios as described in Table 3
  3. bValidation simulation scenarios as described in Table 4