Using functional principal component analysis (FPCA) to quantify sitting patterns derived from wearable sensors

Table 2 Association of blood pressure with MFPCA/POPAI variables from MLR

Outcome	Estimate	MFPCA		POPAI
Outcome	Estimate	Level 1 PC1	Level 1 PC2	Average daily inactive sitting time	Average daily active sitting time
SBP	\(\hat{\beta }\) (s.e)\(^1\)	0.014 (0.017)	-0.026 (0.027)	0.025 (0.011)	-0.018 (0.021)
	p\(^2\)	0.42	0.34	0.03	0.40
	p-FDR\(^3\)	0.42	0.42	0.08	0.42
DBP	\(\hat{\beta }\) (s.e)	0.040 (0.012)\(^4\)	0.034 (0.019)	0.021 (0.008)\(^5\)	-0.030 (0.015)
	p	0.0009	0.07	0.008	0.05
	p-FDR	0.007	0.11	0.03	0.10

¹ Parameter estimates and their standard errors (s.e) from the MLR between independent variables (MFPCA or POPAI variables) and outcomes (SBP or DBP) controlling for other baseline characteristics, including age, race (ref: non-white), education (ref: below college), employment status (ref: not working), BMI, hypertension status (ref: no), as well as concurrent wearing days of devices and average daily non-sitting time
² p : p values from MLRs
³ p-FDR : adjusted p values based on Benjamini and Hochberg false discovery rate (BH FDR) [34] accounted for 8 pairwise comparisons
⁴ Standardized \(\hat{\beta }\) (s.e) = 2.041 (0.607) between DBP and level 1 PC1
⁵ Standardized \(\hat{\beta }\) (s.e) = 1.636 (0.614) between DBP and inactive sitting time

ISSN: 1479-5868