Insulin Dose Prediction

Description

Developing a machine learning model to predict insulin dose based on patient data such as glucose levels, calories, heart rate, steps, and other parameters for type 1 diabetes.

Setup

To run a specific model, execute the following command in the main directory:

Random Forest:

python -m models.rf_predict_dose

CatBoost:

python -m models.cb_predict_dose

Decision Tree:

python -m models.dt_predict_dose

K-Nearest Neighbors:

python -m models.knn_predict_dose

Linear Regression:

python -m models.lr_predict_dose

Support Vector Machine:

python -m models.svm_predict_dose

XGBoost:

python -m models.xgb_predict_dose

LightGBM:

python -m models.lgb_predict_dose

All models at once:

python -m plot_metrics.plot_metrics

Main dataset

Hidalgo, J. Ignacio; Alvarado, Jorge; Botella, Marta; Aramendi, Aranzazu; Velasco, J. Manuel; Garnica, Oscar (2024), “HUPA-UCM Diabetes Dataset”, Mendeley Data, V1, doi: 10.17632/3hbcscwz44.1

Additional datasets

Khan, Murtaza (2024). Diabetes Dataset. figshare. Dataset. 10.6084/m9.figshare.27959529.v1