Table 3:

Actual surrogate measures identified for use in prediction models. All standard deviations are day-to-day standard deviations calculated for each participant.

Result	Quantity Measured	Calculation
(−) association: mi1	Average total log activity count (TLAC)	$\frac{1}{J_i}\sum _{j=1}^{J_i}\sum _{t=1}^{1440}{X}_{ij}(t)$
(+) association: si1	Standard deviation of total log activity counts	$\text{sd}\left(\sum _{t=1}^{1440}{X}_{ij}(t)\right)$
(−) association: si5	Standard deviation of difference in average log transformed activity counts comparing 10AM-3PM to 4PM-7PM	$\text{sd}\left(\frac{{\sum }_{t=661}^{900}{X}_{ij}\left(t\right)}{240}-\frac{{\sum }_{t=961}^{1140}{X}_{ij}\left(t\right)}{180}\right)$
(−) association: si6	Standard deviation of difference in average log transformed activity counts comparing {8AM-10AM, 3PM-5PM, 10PM-12AM} to {5AM-7AM, 11AM-1PM, 6PM-8PM }	$\text{sd}\left(\frac{{\sum }_{t\in t_a}{X}_{ij}\left(t\right)}{\left|t_a\right|}-\frac{{\sum }_{t\in t_b}{X}_{ij}\left(t\right)}{\left|t_b\right|}\right)$ ta = {481,…, 600, 901,…, 1020, 1321,…, 1440} tb = {301,…, 420, 661,…, 780, 1081,…, 1200}