Activity Number:
|
106
- New Statistical Models for Functional Data/ Longitudinal Data
|
Type:
|
Topic Contributed
|
Date/Time:
|
Monday, August 8, 2022 : 8:30 AM to 10:20 AM
|
Sponsor:
|
Section on Nonparametric Statistics
|
Abstract #320983
|
|
Title:
|
A Nonparametric Mixed-Effects Mixture Model for Patterns of Clinical Measurements Associated with COVID-19
|
Author(s):
|
Xiaoran Ma* and Wensheng Guo and Mengyang Gu and Len Usvyat and Peter Kotanko and Yuedong Wang
|
Companies:
|
University of California, Santa Barbara and University of Pennsylvania and University of California, Santa Barbara and Fresenius Medical Care and Renal Research Institute and University of California, Santa Barbara
|
Keywords:
|
clustering;
mixture model;
mixed-effects model;
spline;
COVID-19;
EM algorithm
|
Abstract:
|
Some, but not all, COVID-19 patients had changes in biological/clinical variables such as temperature and oxygen saturation days before symptoms occur. We propose a flexible nonparametric mixed-effects mixture model (NMEM) that simultaneously identifies risk factors and classifies patients with biological change. We model the latent biological change probability using a logistic regression model with L1 penalty and trajectories in each latent class using splines. We apply the EM algorithm and penalized likelihood to estimate all parameters and mean functions. Simulation studies indicate the proposed method performs well. We apply the NMEM model to investigate changes in temperature in COVID-19 patients receiving hemodialysis.
|
Authors who are presenting talks have a * after their name.