Classification and Prediction on Hypertension with Blood Pressure Determinants in a Deep Learning Algorithm

Area of Science:

• Cardiology
• Biomedical Informatics
• Machine Learning

Background:

• Hypertension prediction often lacks robust classification methods.
• Deep learning applications for hypertension diagnosis are underexplored.
• Blood pressure determinants are key indicators for cardiovascular health.

Purpose of the Study:

• To develop and evaluate a deep learning algorithm for hypertension classification and prediction.
• To assess the impact of energy intake adjustment on deep learning model performance.
• To compare deep neural network (DNN) efficacy against traditional machine learning models.

Main Methods:

• A deep neural network (DNN) was constructed and trained using the Korean Genome and Epidemiology Study dataset.
• Hypertension diagnosis was performed using blood pressure-related factors.
• Model performance was evaluated with varying energy intake adjustment methods, compared against a decision tree.
• Hyperparameter tuning (hidden layers, nodes) was experimentally determined for the DNN.

Main Results:

• The DNN demonstrated superior performance over the decision tree in sensitivity, specificity, F1-score, and accuracy.
• Deep learning models achieved optimal results without energy intake adjustment.
• Energy intake adjustment was found to be unnecessary for DNN-based hypertension prediction.

Conclusions:

• Deep learning algorithms, specifically DNNs, offer a powerful tool for hypertension classification and prediction.
• Optimal performance of DNNs for hypertension prediction is achieved without energy intake adjustment.
• The findings suggest a novel approach to leveraging big data for cardiovascular disease risk assessment.