Master's lecture in Applied Statistics - Þórey Heiðarsdóttir | University of Iceland Skip to main content

Master's lecture in Applied Statistics - Þórey Heiðarsdóttir

When 
Tue, 02/06/2020 - 11:00 to 12:00
Where 
Further information 
Free admission

The lecture will be streamed: https://us04web.zoom.us/j/78751676367?pwd=YmIycVFyNXY2bHljbUZNTWRuZXZUZz09

Master's student: Þórey Heiðarsdóttir

Title: Using mixed models to analyse progression of blood pressure and walking distance in a two year longitudinal study

 ___________________________________________

Faculty: Faculty of Physical Sciences

Advisor: Thor Aspelund, Professor at the Centre of Public Health Sciences

Also in the masters committee: Janus Guðlaugsson

Examiner: Jóhanna Jakobsdóttir, Research specialist at the Centre of Public Health Sciences

Abstract

The main purpose of this assignment was to use mixed models to assess the progression of repeated measurements on blood pressure and walking distance for senior citizens, 65-94 years old, in a two-year multi-modal training intervention. Mixed models were compared using linear prediction function and cubic splines. Mixed models with cubic splines gave better estimate of the progression of both blood pressure and walking distance than they did linearly. The spline curve showed that the effect of the training intervention was the largest after the first 6 months of intervention, then stabilized the 12 months thereafter but in the last 6 months of intervention the effect was starting to decline. To gain an accurate description of non-linear relationships it may be a good option to use cubic splines. Another purpose of the assignment was to predict dropout in the study from baseline values of the participants. The results indicate that various physical factors like dynamic balance (8-foot up-and-go test), endurance (6-minute walking test) and strength (grip strength) can indicate weather participants will continue in intervention studies like this one. The relationship wasn’t strong, and it is likely that other factors that the study didn’t measure contributed to the dropout. There can be underlying difference between dropouts and finishers that should be considered before analyzing them together.