Survival Curves for Hierarchical Data

Principal Investigator

Name
Ali Parsaee

Degrees
Masters

Institution
University of Alberta

Position Title
Student

Email
parsaee@ualberta.ca

About this CDAS Project

Study
PLCO (Learn more about this study)

Project ID
PLCO-1241

Initial CDAS Request Approval
May 30, 2023

Title
Survival Curves for Hierarchical Data

Summary
In the field of survival analysis we wish to generate individualized survival curves to give a probability of an event (such as death) occurring for any one patient at a given time. There are many strong machine learning methods that build effective models for this task. However assume your data is on Lung cancer and one of the features in the dataset describes the state of cancer the patient is in. If a patient comes in with stage 4 lung cancer, do you build the survival curve based only on the subset of the dataset dealing with stage 4 lung cancer? The whole dataset? A mix of the two?

Aims

-Generate the most effective individualized survival curve for a coming patient
-Describe with theory and experiment what subset of the dataset to take

Collaborators

Professor Russ Greiner, University of Alberta