Master of Science
Epidemiology and Biostatistics
Introduction: Multimorbidity, the presence of two or more chronic diseases in an individual, is a pressing medical condition. Novel prevention methods are required to reduce the incidence of multimorbidity. Prognostic predictive models estimate a patient’s risk of developing chronic disease. This thesis developed a single predictive model for three diseases associated with multimorbidity: diabetes, hypertension, and osteoarthritis.
Methods: Univariate logistic regression models were constructed, followed by an analysis of the dependence that existed using copulas. All analyses were based on data from the Canadian Primary Care Sentinel Surveillance Network.
Results: All univariate models were highly predictive, as demonstrated by their discrimination and calibration. Copula models revealed the dependence between each disease pair.
Discussion: By estimating the risk of multiple chronic diseases, prognostic predictive models may enable the prevention of chronic disease through identification of high-risk individuals or delivery of individualized risk assessments to inform patient and health care provider decision-making.
Black, Jason E., "Prognostic Predictive Model to Estimate the Risk of Multiple Chronic Diseases: Constructing Copulas Using Electronic Medical Record Data" (2018). Electronic Thesis and Dissertation Repository. 5926.
Available for download on Sunday, September 01, 2019