Fall Prediction in Hypertensive Patients via Short-Term HRV Analysis

Area of Science:

• Gerontology
• Biomedical Engineering
• Cardiology

Background:

• Falls are a significant issue for older adults, impacting quality of life and healthcare costs.
• Existing fall prediction technologies often rely on accelerometers and gyroscopes, with limited success in identifying first-time fallers.
• Hypertensive patients are a suitable population for study due to regular outpatient visits allowing for accessible electrocardiogram (ECG) recordings.

Purpose of the Study:

• To develop and validate a predictive metamodel for identifying first-time fallers.
• To investigate the utility of short-term Heart Rate Variability (HRV) analysis for fall risk assessment.
• To specifically focus on a hypertensive patient cohort for pragmatic clinical application.

Main Methods:

• Collected 55-minute ECG recordings from 170 hypertensive patients (mean age 72 years).
• Extracted linear and nonlinear Heart Rate Variability (HRV) features from 11 consecutive 5-minute ECG excerpts.
• Utilized statistical and data mining techniques, including Multinomial Naïve Bayes, to build a predictive metamodel.

Main Results:

• The best predictive metamodel, based on Multinomial Naïve Bayes, accurately predicted first-time fallers.
• The model achieved a sensitivity of 72%, specificity of 61%, and overall accuracy of 68%.
• Short-term HRV analysis proved effective in distinguishing individuals who would experience a first-time fall within three months.

Conclusions:

• Short-term Heart Rate Variability (HRV) analysis is a promising, non-invasive method for predicting first-time falls in hypertensive individuals.
• This approach can aid in early identification of fall risk, potentially enabling timely interventions.
• The findings suggest integrating HRV analysis into routine cardiovascular assessments for enhanced geriatric fall prevention strategies.