Machine learning approaches for cardiovascular hypertension stage estimation using photoplethysmography and clinical...

Area of Science:

• Biomedical Engineering
• Cardiology
• Machine Learning in Healthcare

Background:

• Cardiovascular diseases (CVDs) are a major global health concern, with hypertension as a primary risk factor.
• Early hypertension detection and management are crucial for preventing CVDs and complications.

Purpose of the Study:

• To develop and evaluate a novel approach for estimating hypertension stages using acceleration photoplethysmography (APG) waveform features and clinical parameters.
• To assess the efficacy of different machine learning algorithms in classifying hypertension stages based on APG data.

Main Methods:

• Extracted novel features from the acceleration photoplethysmography (APG) waveform using a specialized algorithm on a public dataset.
• Employed three supervised machine learning algorithms: Decision Tree (DT), Linear Discriminant Analysis (LDA), and Linear Support Vector Machine (LSVM) for classification.
• Validated model performance using cross-validation and a separate test dataset.

Main Results:

• The Decision Tree (DT) model achieved 100% training accuracy and 96.87% test accuracy.
• Linear Support Vector Machine (LSVM) demonstrated strong performance with 88.77% cross-validation and 93.75% test accuracy.
• Linear Discriminant Analysis (LDA) yielded 85.02% cross-validation and 84.37% test accuracy.

Conclusions:

• APG waveform analysis, combined with machine learning, shows significant potential for accurate hypertension staging.
• This approach can serve as a valuable non-invasive tool for clinicians, supporting early detection and intervention strategies for cardiovascular health.
• The findings advocate for the integration of APG analysis into routine clinical practice to improve cardiovascular healthcare outcomes.