Abstract
This article is motivated by some longitudinal clinical data of kidney transplant recipients, where kidney function progression is recorded as the estimated glomerular filtration rates at multiple time points post kidney transplantation. We propose to use the functional principal component analysis method to explore the major source of variations of glomerular filtration rate curves. We find that the estimated functional principal component scores can be used to cluster glomerular filtration rate curves. Ordering functional principal component scores can detect abnormal glomerular filtration rate curves. Finally, functional principal component analysis can effectively estimate missing glomerular filtration rate values and predict future glomerular filtration rate values.
Original language | English (US) |
---|---|
Pages (from-to) | 3785-3796 |
Number of pages | 12 |
Journal | Statistical Methods in Medical Research |
Volume | 27 |
Issue number | 12 |
DOIs | |
State | Published - Dec 1 2018 |
Externally published | Yes |
Keywords
- functional data analysis
- Functional principal component analysis
- missing data
- outlier
- renal disease
ASJC Scopus subject areas
- Epidemiology
- Statistics and Probability
- Health Information Management