memaef_prediction

Overview

This repository contains code for the development and evaluation of a prediction model for endometriosis in girls and young women with menstrual pain. This repository is part of the MeMäF Project (visit website).

Structure

The repository contains one code area:

• develop_prediction_model

In it, there are several subdirectories:

• prediction_src for functionalities
• data for local data storage
• notebooks for data preprocessing, descriptive analysis, model development, evaluation and application to the data of the new care form within MeMäF.

In the params.py file, several mapping dictionaries between EBM and ICD-10 codes and predictor names for different models and different predictor sets are stored, as well as lists of computable calibration and evaluation metrics.

memaf_prediction
└── README.md
└── .gitignore
└── params └── develop_prediction_model
├── prediction_src
│ ├── application_prediction
│ ├── calibration
│ ├── descriptive_analysis_kk_daten
│ ├── evaluate_prediction_model
│ ├── fit_prediction_model
│ ├── join_kk_data
│ ├── load_data
│ ├── prediction_pipeline
│ └── utils
├── notebooks
│ ├── 00_Join_KK_Daten
│ ├── 01_Descriptive_Analysis
│ ├── 02_Fit_DAG
│ ├── 03_Calibration
│ ├── 04_Evaluation
│ └── 05_Application_to_NVF
├── models
└── data
├── raw
└── processed
└── results

How to run

The data subdirectory is empty - make sure to store your data here in tabular format. Also, the models folder is empty - here the models should be saved.

The notebooks are numbered and can be run in sequential order. The final model from our work will be "DAG_model_youden_final".