TABLE 4.
Choosing the threshold for moderate-to-vigorous intensity physical activity by examining the false positives and false negatives based for approach 2 (eight activities)
| Accelerometer (counts·30 s-1) | True Positive (N) | True Negative (N) | False Positive (N)* | False Negative (N)** | Error Rate (%) | Validation Error Rate (%)*** (5th, 95th Percentile) |
|---|---|---|---|---|---|---|
| 1100 | 239 | 273 | 43 | 6 | 8.7 | 6.4,11.4 |
| 1150 | 242 | 273 | 40 | 6 | 8.2 | 5.9,10.3 |
| 1200 | 252 | 270 | 30 | 9 | 7.0 | 4.6, 8.9 |
| 1250 | 259 | 270 | 23 | 9 | 5.7 | 3.5, 7.5 |
| 1300 | 263 | 269 | 19 | 10 | 5.2 | 3.4, 7.0 |
| 1350 | 268 | 267 | 14 | 12 | 4.6 | 2.9, 6.7 |
| 1400 | 269 | 266 | 13 | 13 | 4.6 | 3.0, 6.7 |
| 1450 | 274 | 261 | 8 | 18 | 4.6 | 3.0, 6.6 |
| 1500 | 276 | 258 | 6 | 21 | 4.8 | 2.8, 6.5 |
| 1550 | 277 | 250 | 5 | 29 | 6.1 | 4.1, 7.9 |
| 1600 | 279 | 247 | 3 | 32 | 6.2 | 3.9, 8.2 |
Counts measured under the protocols for watching TV, playing a computer game, sweeping the floor, and slow walking that were above the threshold for moderate intensity activity were considered as false positives.
Counts measured under the protocols for brisk walk, stair walk, shoot baskets, and run that were below the threshold for moderate intensity activity were considered as false negative (exclude step aerobics and bicycling).
Based on 100 bootstrap samples, each half the size as the original sample.