Hypertension Screening Using Acoustic Analysis and Machine Learning of Random Speech Samples: A Feasibility Study

Area of Science:

• Cardiology
• Biomedical Engineering
• Speech Processing

Background:

• Hypertension is a major risk factor for cardiovascular diseases.
• Early detection and treatment are crucial for managing hypertension.
• Cuff-based blood pressure measurement, while accurate, faces accessibility challenges in remote areas.

Purpose of the Study:

• To investigate the feasibility of using overt speech for hypertension screening.
• To evaluate speech analysis as an accessible alternative to traditional blood pressure monitoring.

Main Methods:

• Collected speech recordings from 573 participants.
• Extracted acoustic features (temporal, spectral, nonlinear) using machine learning models.
• Evaluated model performance using leave-one-subject-out cross-validation.

Main Results:

• Achieved balanced accuracy (BACC) of 61% (males) and 70% (females) under relaxed hypertension criteria (SBP ≥135 OR DBP ≥85).
• Achieved BACC of 71% (males) and 78% (females) under stricter ESH guidelines (SBP ≥140 OR DBP ≥90).

Conclusions:

• Overt speech analysis shows potential for hypertension screening.
• This acoustic approach could improve accessibility to hypertension detection.