Tuesday, December 10 2019
3:30pm - 5:30pm
Masters Presentation
Linear Mixed-Effects Models for Analyzing the Impact of Maternal Nutrition Supplement on Hemoglobin Levels

Applied research often focuses on how the mean response differs by treatment group or over levels of a predictor. This can include how mean response varies over time, which introduces correlation between observations that must be taken into account. Linear mixed-effects models (LMMs) explores different approaches to analyzing the above question by properly modeling correlation. In this paper, we describe LMMs, their assumptions, and validation steps. Then, we apply the model to a specific data set: the Women First trial. A sample of 1701 women is from a rural site of the Democratic Republic of Congo. Our application focuses on the impact of a maternal nutrition supplement on hemoglobin levels (Hb) using LMMs. Predictors that were considered are the nutrition supplement commencing at least 3 months prior pregnancy, the same nutrition supplement starting at 12-14 weeks of gestation, the parity, the body mass index at registration, and the timepoint. The results of our study demonstrate that the nutrition supplement commencing at least 3 months prior pregnancy, the same nutrition supplement starting at 12-14 weeks of gestation, the parity, and the body mass index at registration do not have a significant impact on hemoglobin levels with p-values of 0.67, 0.86, 1.00, and 0.60, respectively; however, the timepoint is significant with p-values less than 10-6.
Keywords: Hemoglobin level, Linear Mixed-Effects Models, Model fitting, Nutrition supplement, Women First trial
Speaker:Ptshou Nzazi Duki
Affiliation:
Location:CU 320A


Email Address:  
   Sign up to be notified by email before this event takes place.

Download as iCalendar

Done