Hypertension data set

Description

**Dataset Description** The dataset contains **14,352 anonymized patient records** and **12 variables** used for hypertension risk prediction. The variables include **age, sex, body mass index (BMI), residence, systolic blood pressure (SBP), diastolic blood pressure (DBP), smoking status, comorbid disease status, serum creatinine, blood urea nitrogen (BUN), number of medications, and the hypertension class label (level)**. The dataset was preprocessed through data cleaning, missing-value handling, and feature transformation before training and evaluating the machine learning models. A labeled version of the dataset is provided as supplementary material to support transparency and reproducibility. 1. 14,352 records 2. 12 variables (11 predictors + 1 target label) 3. Target variable: level (hypertension class) 4. Predictors: age, sex, BMI, residence, SBP, DBP, smoking, comorbid disease, creatinine, BUN, number of medications.

Steps to reproduce

1. Download the dataset and import it into a data analysis environment such as Python (version 3.9 or later). 2. Load the dataset using the provided CSV file. 3. Perform data preprocessing, including data cleaning, handling missing values, feature transformation, and encoding categorical variables where applicable. 4. Split the dataset into training and testing subsets using the same experimental protocol described in the manuscript. 5. Train the Random Forest, Extra Trees, and Bagged Decision Tree models using the specified hyperparameters. 6. Evaluate model performance using Accuracy, Precision, Recall, F1-score, and Area Under the Receiver Operating Characteristic Curve (AUC-ROC). 7. Compare the predictive performance of the models and analyze feature importance to identify the most influential clinical and demographic risk factors associated with hypertension. 8. The preprocessing workflow, model configuration, and evaluation methodology are fully described in the accompanying manuscript.

Institutions

• University of Gondar

  Amhara, Gonder

Categories

Licence