Predict the Risk of Cardiovascular Diseases in the Future Using Deep Learning

Date
2018
Authors
Jin, Ruitao
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract

Cardiovascular disease is the leading cause of death worldwide. About 790,000 people in US have heart attacks each year. Of those, 14.4% will die. The estimated annual cost of heart disease in 2013 was $199.6 billion in US. In 2010, the estimated global cost of cardiovascular disease was $863 billion, and it is estimated to rise to $1044 billion by 2030, when 43.9% of the US adult population is projected to have some form of cardiovascular diseases. Prediction the risk factors of cardiovascular has been conducted over the past two decades with validated predictive models including Framingham model in Unites States and QRISK model in England. However, these models are statistical models and the accuracy of the prediction is not good with a high predictive risk over observation ratio, suggesting an overestimated risk for patient. In this study, we applied a deep learning algorithm to establish a neural network model to predict the risk of cardiovascular disease with a training set (clinical data from Cleveland Study with 303 observations) and validate our model with another set of study (617 observations of clinical data from 3 studies: Hungarian, Switzerland, and Long Beach VA datasets). Our model showed a confidence level of 70% comparing against the true diagnostic data, suggesting a powerful prediction tool employing big data and deep learning.

Description
This item is available only to currently enrolled UTSA students, faculty or staff.
Keywords
Citation
Department
Electrical and Computer Engineering