MC3 -- Prediction Model for Myocardial Infarction Incidence
Start Date
8-4-2022 10:30 AM
End Date
8-4-2022 12:15 PM
Location
URC Greatroom
Document Type
Event
Abstract
Heart disease is the leading cause of death in the United States. Myocardial infarction (heart attack) is one of the various forms of heart disease. Determining the factors associated with the risk of heart attack is important to the health professionals and the public. We investigated the prevalence of myocardial infarction incidence (1 = if myocardial infarction occurred, 0 = no myocardial infarction occurred) from NHANES III data and used a logistic regression model for the analysis of binary data. The model indicated that sex, age, diabetes, high cholesterol, congestive heart failure, and chest pain are the significant risk factors for heart attack with sex, congestive heart failure, and chest pain as the three most significant ones. The classification accuracy of the fitted logistic regression model was 92.18%. Because the prevalence of the incidence was fairly low, we also used Firth logistic regression model and compared it with the logistic regression model. The results were almost similar but comparing the AIC values, it was found that the firth logistic regression was slightly better. The estimated logistic regression equation is useful in predicting whether an individual is prone to the risk of heart attack based on his/her personal information, diet, smoking habits, and health.
Keywords
Math, Computer Science, Informatics
MC3 -- Prediction Model for Myocardial Infarction Incidence
URC Greatroom
Heart disease is the leading cause of death in the United States. Myocardial infarction (heart attack) is one of the various forms of heart disease. Determining the factors associated with the risk of heart attack is important to the health professionals and the public. We investigated the prevalence of myocardial infarction incidence (1 = if myocardial infarction occurred, 0 = no myocardial infarction occurred) from NHANES III data and used a logistic regression model for the analysis of binary data. The model indicated that sex, age, diabetes, high cholesterol, congestive heart failure, and chest pain are the significant risk factors for heart attack with sex, congestive heart failure, and chest pain as the three most significant ones. The classification accuracy of the fitted logistic regression model was 92.18%. Because the prevalence of the incidence was fairly low, we also used Firth logistic regression model and compared it with the logistic regression model. The results were almost similar but comparing the AIC values, it was found that the firth logistic regression was slightly better. The estimated logistic regression equation is useful in predicting whether an individual is prone to the risk of heart attack based on his/her personal information, diet, smoking habits, and health.