NYU Langone Medical Center Knee Alignment Research

@author Jack Bosco

---

What is the Goal?

Using machine learning techniques such as feature selection and data clustering, we hope to develop our understanding of knee alignment morphologies.

Where is the data?

Due to compliance reasons, I cannot upload the datasheet to GitHub. However, if you have the mako_data.xlsx file, drop that in raw.

How do I run the project?

1. cd to the project directory
2. Make sure you have the right dependencies by running the command below. You only need to do this once.

   pip3 install -r requirements.txt
   

3. To treat the data, run the command below. This creates the treated spreadsheet treated/morphologies.csv. You only need to run this once.

   python3 treat_data.py
   

4. Visualize the treated data by running

   python3 data_viz.py
   

5. Create and visualize a regression model for planning postop aHKA alignments

   python3 regression.py
   

Config options

Configure the date file locations in config.py:

• raw_path is the path to the raw data
• treated_path is the path to the treated data
• norm_path is the path to the input normalizer
• de_norm_path is the path to the output normalizer
• model_path is the path to the pre-trained model

regression.py is the only file that will run if you don't supply the de-anonymized patient data yourself

I cannot privide the files in this repo due to compliance reasons, but please reach out to me if you would like to run this on your own dataset.

linkedin: linkedin.com/in/JackBosco.