Date of Award

January 2024

Document Type

Open Access Thesis

Degree Name

Master of Public Health (MPH)

Department

School of Public Health

First Advisor

Leying Guan

Abstract

Background Cardiovascular disease (CVD) is one of the leading causes of morbidity and mortality worldwide. There are several studies researching the risk factor of cardiovascular diseases based on BRFSS data in recent years, but there is still a lack of comprehensive research and further discussion is needed for the prediction of CVD. This study aims to provide a systematic CVD prediction model based on BRFSS survey data

Method In this study, the subjects’ characteristics factors will be summarized in tables separated by different aspects and CVD groups. Two-thirds of the records will be randomly selected as the training set to build prediction models, and the remaining records will be used as the validation set to further evaluate the performance of the models. The ROC curves will be used to evaluate the predictive performance of different models

Results The AUCs for training set of logistic regression, random forest, decision tree and SVM were 0.784, 1.000, 0.844 and 0.983. While for validation set were 0.774, 0.921, 0.855 and 0.875. The results of DeLong test showed no significant differences (p=0.376) in decision tree and SVM based on radial basis function in validation set, but the p values were all less than 0.05 for other combinations.

Discussion Random forest prediction model performed best in both training set and validation set. The most important feature of random forest was the status of a subject ever had CHD or MI, and subjects that ever had CHD or MI were more likely to have a CVD. Other subjects’ characteristics like employment status, currently taken medicine for high cholesterol status, and lonely status also served as important parts, and further discussion is needed.

Key words CVD, BRFSS, Logistic Regression, Random Forest, Decision Tree, SVM

Comments

This is an Open Access Thesis.

Download
Open Access

This Article is Open Access

Share

COinS