The goal of this project was to build a prediction model for 5-year mortality in participants of the Cardiovascular Health Study (a government sponsored cohort study of adults aged 65 years and older in four US communities) using Deep Learning. Written report is available upon request. This project was part of a graduate-level biostatistics course.
Challenge: modelling using ~700 observations
-
Develop a predictive model for 5-year mortality.
-
Is self rated health predictive of mortality beyond the predictive capabilities of the available data?
-
Comment on how useful some of the harder to obtain measurements (MRI, lab variables) are in predicting mortality relative to the other variables.
- mri_dl.ipynb: notebook with code for results