|
Activity Number:
|
178
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 1, 2016 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Biopharmaceutical Section
|
|
Abstract #318888
|
|
|
Title:
|
A Bayesian Approach for Classification Based on Continuous Glucose Monitoring Data of Diabetic Patients
|
|
Author(s):
|
Hui Zheng*
|
|
Companies:
|
Harvard Medical School
|
|
Keywords:
|
Bayesian ;
Spline ;
classification ;
continuous glucose monitoring ;
diabetes ;
CGM
|
|
Abstract:
|
We propose a Bayesian modeling approach to distinguish subgroups of diabetic patients based on their glucose profiles. The glucose data are collected using continuous glucose monitoring (CGM) devices which measure glucose at a frequency of once 1 to 5 minutes. The classification method is based on a hierarchical model using B-spline basis functions to smooth the glucose measurements while allowing for auto-correlations. In the training phase, the mean and variance-covariance matrices of glucose profiles are obtained for each subgroup. The classification is based the posterior probabilities for a patient to belong to each subgroup. We illustrate the method using CGM data collected in a clinical trial .
|
Authors who are presenting talks have a * after their name.
Back to the full JSM 2016 program
|
