Survival analysis on subchondral bone length for total knee replacement

This work is published in Skeletal Radiology (https://doi.org/10.1007/s00256-024-04627-1).

Prerequisites

The tool was developed based on the following dependencies:

1. NumPy
2. Scipy
3. Pandas
4. Lifelines
5. Statistics
6. Matplotlib
7. Sksurv
8. Sklearn
9. Jupyter Notebooks

Data files

All data files for this project are stored within the folder 'OAI_SBL_Analysis_Data'. Within this folder all information for patients regarding demographics, TKR, BML size, and SBL can be found.

Notice: You must update the path to this folder within each script before conducting Cox Proportional Hazards or Kaplan-Meier Analysis.

loc_data = "~/OAI_SBL_Analysis_Data/" --> path to data
loc_module = "~/OAI_Github_scripts/" --> path to analysis scripts

Scripts for Kaplan Meier Analysis

These scripts can be found under the 'Kaplan_Meier_Analysis' folder. By running these you will be able to generate Figures 2A and 2B from the paper.

Notice: When saving figures from Kaplan-Meier analysis, be sure to change the path to your suited location.

plt.savefig('~/KMF_Curves/kmf_BML.pdf', format='pdf', bbox_inches = 'tight')

Scripts for Cox Proportional Hazards Analysis

These scripts can be found under the 'CPH_Analysis' folder. By running these you will be able to generate the other figures from the paper.

Notice: When saving figures from CPH analysis, be sure to change the path to your suited location.

plt.savefig('~AUC_Curves/AUC_ALL_Knees_SBL_BML_figure_all_SBL.pdf', format='pdf', bbox_inches = 'tight')

Stastistics

Time-Dependent Area Under the Curve (AUC)

Concordance Index

Brier Score

Log-Rank Test

Results

Each CPH Analysis script will produce a time-dependent AUC, Concordance Index, and Brier Score. Kaplan-Meier Analysis scripts will generate Kaplan Meier curves and the Log-Rank test associated with each curve. Generated figures are located in ~/AUC_Curves/ and ~/KMF_Curves/.