Common use of Table 3 Clause in Contracts

Table 3. Linear models for predicting SW3 daily step count from wGT3X+ daily step count Method Model for predicting SW3 steps/day R2 RMSE People with PAD wGT3X+/LFE -30.30477 + 0.82201 × steps/day 0.85 1663 wGT3X+/N 1704.7047 + 1.4690 × steps/day 0.89 1392 Older adults wGT3X+/LFE -1897 + 0.9193 × steps/day 0.95 1298 wGT3X+/N 4903 + 1.013 × steps/day 0.93 1574 Note. Steps/day is the mean of the daily step count calculated using the valid days. R2 coefficient and root-mean-squared error (RMSE) were computed using the leave-one-out method. Page 35 of 42 Journal of Aging and Physical Activity 23 FIGURES CAPTIONS 24 Figure 1. Step counts from the SW3 and wGT3X+ monitors related to the outdoor 25 walking session 26 On the scatter plots, the solid line shows the identity, and the dashed lines show the regressions 27 of the wGT3X+/LFE and wGT3X+/N data on the SW3 data respectively. On the equivalence 28 plots, the thick, thin, and very thin black lines show the ±10%, ±15%, ±20% SW3 equivalence 29 zones, respectively; LFE = lower frequency extension filter; N = normal filter; EZs = 30 equivalence zones; 90% CI = 90% confidence interval. 32 Figure 2. Effect of pain occurrence on the agreement between the SW3 and wGT3X+ 33 monitors and on walking speed 34 Panel A shows, for each wGT3X+ method, the individual percent errors (PE) for walking 35 without pain and for walking with pain. Panel B shows, for each wGT3X+ method, the 36 individual differences of absolute percent error (APE) between walking with pain and walking 37 without pain (Change in APE = APE[with pain] minus APE[without pain]), along with the 38 corresponding medians (grey bars) and 95% confidence intervals (black error bars). Panel C 39 shows the individual walking speeds when walking without pain and walking with pain. Panel 40 D shows the differences in walking speed (Change in speed = speed[with pain] minus 41 speed[without pain]), with the mean difference and its corresponding 95% confidence interval. 42 The grey zone depicts non-substantial change in speed. The results are related to 17 participants 43 who indicated the occurrence of pain during the first walking bout. 45 Figure 3. Step-based metrics from the SW3 and wGT3X+ monitors for the 7-day free- 46 living period 31 44 47 On the scatter plots, the solid line shows the identity, and the dashed lines show the regressions 48 of the wGT3X+/LFE and wGT3X+/N data on the SW3 data respectively. On the equivalence 49 plots, the thick, thin, and very thin black lines show the ±10%, ±15%, ±20% SW3 equivalence 50 zones, respectively; LFE = lower frequency extension filter; N = normal filter; EZs = 51 equivalence zones; 90% CI = 90% confidence interval. ACCEPTED MANUSCRIPT - CLEAN COPY Human Kinetics, ▇▇▇▇ ▇ ▇▇▇▇▇▇ ▇▇, ▇▇▇▇▇▇▇▇▇, ▇▇ ▇▇▇▇▇ Journal of Aging and Physical Activity Page 38 of 42 ACCEPTED MANUSCRIPT - CLEAN COPY Human Kinetics, ▇▇▇▇ ▇ ▇▇▇▇▇▇ ▇▇, ▇▇▇▇▇▇▇▇▇, ▇▇ ▇▇▇▇▇ ACCEPTED MANUSCRIPT - CLEAN COPY Human Kinetics, ▇▇▇▇ ▇ ▇▇▇▇▇▇ ▇▇, ▇▇▇▇▇▇▇▇▇, ▇▇ ▇▇▇▇▇ Before conducting the graphical inspection, in-bed time was determined using the “PhysActBedRest” R package (version 1.0, 2016) provided by ▇▇▇▇▇ et al. (▇▇▇▇▇, ▇▇▇▇, ▇▇▇▇, & ▇▇▇▇▇▇▇▇▇, 2018). In-bed time was marked into the merged data file using the ▇▇▇▇▇ et al. (▇▇▇▇▇ et al., 2018) algorithm, the wGT3X+/LFE vector magnitude (VM) counts, and the optimal cut-points (CP0 = 50; CP1 = 210; CP2 = 350) provided by Belletiere et al. (▇▇▇▇▇▇▇▇▇▇▇ et al., 2019). The beginning of the in-bed night-time period was considered as the beginning of the in-bed time period visible at the end of the 24-h day (or as the beginning of the first in-bed time period of the following day if any) that appeared after the last out-of-bed time period of the day. Graphical inspection was conducted in R by plotting both the SW and wGT3X+/LFE steps, wGT3X+/LFE wear time, and in-bed time, against time (see eFigure below), and by using the “plotly” package. In the following three cases, all the concerned “wear” in-bed night-time epochs were reconsidered as “nonwear”: (i) when an initial wear time period began in an out-of-bed time period and was prolonged into an in-bed night time period for ≥30 min (eFigure, panel A); (ii) when an initial wear time period began during an in-bed nighttime period more than 30 min before to be prolonged into the next out-of-bed time period (eFigure, panel B); (iii) when an initial wear time period both began and ended during an in-bed night-time period (eFigure, panel C). In addition, if the first out-of-bed time period of the day included more than 30 min of wGT3X+/LFE wear time without SW steps, the corresponding epochs were reconsidered as “nonwear” (eFigure, panel D). ▇▇▇▇▇▇▇▇▇▇▇, ▇., ▇▇▇▇▇, ▇., ▇▇▇▇▇▇▇, V., Full, K. M., ▇▇▇▇, ▇., ▇▇▇▇▇▇▇, ▇. ▇., . . . Di, C. (2019). Parameterizing and validating existing algorithms for identifying out-of-bed time using hip- worn accelerometer data from older women. Physiol Meas, 40(7), 075008. ▇▇▇▇▇, ▇. ▇., ▇▇▇▇, ▇., ▇▇▇▇, ▇. ▇., & ▇▇▇▇▇▇▇▇▇, M. S. (2018). Identifying bedrest using 24-h waist or wrist accelerometry in adults. PLoS One, 13(3), e0194461.