Logo image
Assessing Trend Changes in Functional and Structural Characteristics: Combining Principal Components Methods and Functional Data Analysis
Journal article   Open access

Assessing Trend Changes in Functional and Structural Characteristics: Combining Principal Components Methods and Functional Data Analysis

Johannes Ledolter
Journal of Ophthalmology & Clinical Research, Vol.6(3), 057
09/24/2019
DOI: 10.24966/OCR-8887/100057
url
https://doi.org/10.24966/ocr-8887/100057View
Published (Version of record) Open Access

Abstract

We classify subjects into one of several groups based on their time progression. Medical time series are available for each subject, but time series are short and observations are collected at different time periods. Instead of characterizing each subject by its average and slope, we use principal components analysis to determine the appropriate summary features and classify the subjects on their most important principal components scores. Since observations on subjects are collected at different time periods, we first use functional data analysis to transform the irregularly-spaced data into a complete data array with rows representing subjects and columns representing time. We illustrate the technique - functional data analysis to create a complete data matrix and principal components analysis to determine the most important features for classification - on the average thickness of the retinal nerve fiber layer that come from two groups of patients.

Details

Metrics

16 Record Views
Logo image