Breast Cancer Survival Analysis Project

In this semi-personal project, I investigated the role of clinical factors in the long term survival of cancer patients. Statistical techniques I used were Kaplan-Meier survival curves and Cox proportional hazards (PH) models.

Contributions and acknowledgements

I worked on this project as part of the University of Sydney SUDATA (Data Society) Research Mentoring Program--the topic/dataset was chosen out of personal interest. I completed all the analyses in R by myself while receiving guidance from my program mentor throughout.

The final pdf presentation slides (based off this analysis, see presentation/) were prepared and presented in collaboration with two additional people.

File structure

• analysis/ Quarto notebooks
• scripts/
• presentation/ Final presentation PDF

Data

The METABRIC dataset used is from Kaggle: "Breast Cancer Gene Expression Profiles (METABRIC)" by Raghad Alharbi.
https://www.kaggle.com/datasets/raghadalharbi/breast-cancer-gene-expression-profiles-metabric

Running code

0. Download METABRIC_RNA_Mutation.csv as above and put it in data/.

1. Run analysis/01_data_cleaning.qmd once to get the cleaned dataset, data/data_clean.csv

2. You can then render the other two .qmd files in analysis/.

Each notebook sources a shared setup script, source("../scripts/setup.R")

Methods/skills

• Data cleaning, bias considerations
• Survival analysis (using Kaplan-Meier and Cox PH models)
• Analysing effect size and model diagnostics (PH assumption, linearity, interaction, etc)

Extra notes

This project is for demonstration; results should not be interpreted as clinical conclusions.