Predicting medical outcomes using deep learning with CT chest images
Principal Investigator
Name
Lyle Palmer
Degrees
PhD
Institution
University of Adelaide
Position Title
Professor of Genetic Epidemiology
Email
About this CDAS Project
Study
NLST
(Learn more about this study)
Project ID
NLST-311
Initial CDAS Request Approval
Jun 8, 2017
Title
Predicting medical outcomes using deep learning with CT chest images
Summary
We intend to build on our previous work in predicting important medical outcomes from routinely acquired CT chest images. We have previously used deep learning to predict all cause mortality in healthy older patients, and are currently working with a much larger dataset to extend this to other medical outcomes, including clinical covariates.
We intend to investigate outcomes including all-cause mortality, disease specific mortality, incidence of chronic and malignant diseases, and complications following procedures. These outcomes will inform predictive models of disease for precision medicine purposes.
NLST will be used in conjunction with a large local dataset.
We intend to investigate outcomes including all-cause mortality, disease specific mortality, incidence of chronic and malignant diseases, and complications following procedures. These outcomes will inform predictive models of disease for precision medicine purposes.
NLST will be used in conjunction with a large local dataset.
Aims
1. Development of a deep learning system which incorporates clinical covariates to predict medical outcomes in the study group, including all cause mortality within 5 years.
2. Use the NLST dataset to validate our model trained with local data.
Collaborators
Gustavo Carneiro, University of Adelaide
Lyle Palmer, University of Adelaide