Fig. 5: Boxplots of intra-person correlations between actigraphy measures and self-reported symptoms.

Boxplots of intra-person Pearson pairwise (a, b) and partial (c, d) correlations between a range of actigraphy measures (x-axis) and (a, c) global pain scores, or (b,d) global BFI scores. In (c) and (d), the partial correlations indicate that all the other variables presented are controlled for when computing each individual correlation, as opposed to (a) and (b) where pairwise correlations do not control for any variables. Each point represents the correlation within a specific participant’s own data, and red-dashed lines indicate where |R | =0.3. We refer to Fig. 9 and Supplementary Data 1 for definitions of variable names. Only participants with at least 20 non-missing value pairs were included, and smartwatch cycles in which participants received surgery for endometriosis (either as part of the surgical sub-study or otherwise) were excluded, resulting in n = 54 participants included in the figure, with a total number of days included ranging from 3,591-3,907 for subfigures (a) and (b) and 3,327 total days in subfigures (c) and (d). We refer readers to Supplementary Table 1 where we explored linear mixed-effect models for fatigue and pain (complementing findings presented in subplots (c) and (d)).