|
Activity Number:
|
286
- Missing Data Methods
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 11, 2021 : 1:30 PM to 3:20 PM
|
|
Sponsor:
|
Biometrics Section
|
|
Abstract #318572
|
|
|
Title:
|
Impact of Degree of Missingness and Sample Size on the Performance of Imputation Methods for Mass Spectrometry Data
|
|
Author(s):
|
Sandra L Taylor and Matthew Dominic Ponzini and Machelle D Wilson and Kyoungmi Kim*
|
|
Companies:
|
University of California, Davis and University of California, Davis and University of California, Davis and University of California, Davis
|
|
Keywords:
|
metabolomics;
mass spectrometry;
missing data;
imputation;
sample size
|
|
Abstract:
|
Missing values are common in high-throughput mass spectrometry (MS) data. A common analytical strategy is to remove compounds with more than a pre-specified level of missingness, typically 30-50%, and analyze the remaining compounds after imputing missing values predominantly to reduce bias due to imputation. The most appropriate threshold could depend on sample size with higher levels of missingness tolerated for larger sample sizes. This study evaluated the effect of sample size and percentage of missing values on statitsical inference of imputation methods and methods that specifcally account for missing values. We compared statistical inference based on selected imputation and observed data-based methods over a wide range of missingness and sample sizes. Using several simulation studies motivated by real data, we developed empirical recommendations for analyzing mass spectrometry data with missing values.
|
Authors who are presenting talks have a * after their name.
Back to the full JSM 2021 program
|
