Automated Detection of Hypertension Using Physiological Signals: A Review

Area of Science:

• Biomedical Engineering
• Cardiovascular Research
• Artificial Intelligence in Medicine

Background:

• Arterial hypertension (HT) is a significant risk factor for cardiovascular disease, stroke, kidney failure, and mortality.
• Early diagnosis and effective treatment of HT are crucial for preventing adverse outcomes.
• Current methods for HT detection using physiological signals like ECG, PPG, HRV, and BCG are often manual, time-consuming, and error-prone.

Purpose of the Study:

• To systematically review studies on the automated detection of hypertension (HT) using physiological signals.
• To evaluate the performance of machine learning (ML) and deep learning (DL) methods for HT detection.
• To identify potential for developing wearable devices for continuous, cuffless blood pressure monitoring.

Main Methods:

• Systematic review of 250 screened papers, identifying 23 eligible studies.
• Analysis of studies employing electrocardiography (ECG), heart rate variability (HRV), photoplethysmography (PPG), and ballistocardiography (BCG) signals.
• Discussion of study methodologies, databases, nonlinear techniques, feature extraction, and diagnostic performance.

Main Results:

• Machine learning and deep learning methods, particularly those utilizing ECG and HRV signals, demonstrated the highest diagnostic performance.
• The review identified 23 relevant studies meeting the eligibility criteria.
• Various nonlinear techniques and feature extraction methods were employed across the reviewed studies.

Conclusions:

• Automated detection of HT using ML/DL algorithms based on ECG and HRV signals is a viable approach.
• These findings support the development of computer-aided diagnosis systems for hypertension.
• The research provides insights for creating wearable devices for continuous, cuffless remote blood pressure monitoring.