Fast Multilevel Functional Principal Component Analysis

• Activity Number: 402 - Advances in Statistical Methods for Wearable and Mobile Health Data Analysis
• Type: Invited
• Date/Time: Wednesday, August 10, 2022 : 10:30 AM to 12:20 PM
• Sponsor: Biometrics Section
• Author(s): Erjia Cui* and Ruonan Li and Ciprian M. Crainiceanua and Luo Xiao
• Companies: Johns Hopkins University and North Carolina State University and Johns Hopkins University and North Carolina State University
• Keywords: multilevel models; functional principal component analysis; mixed model equations

Abstract

We introduce fast multilevel functional principal component analysis (fast MFPCA), which scales up to high dimensional functional data measured at multiple visits. The new approach is orders of magnitude faster than and achieves comparable estimation accuracy with the original MFPCA (Di et al., 2009). Methods are motivated by the National Health and Nutritional Examination Survey (NHANES), which contains minute-level physical activity information of more than 10000 participants over multiple days and 1440 observations per day. While MFPCA takes more than five days to analyze these data, fast MFPCA takes less than five minutes. The associated function "mfpca.face()" is available and will be incorporated in the "refund" package in R software.