Assessing elderly persons' fall risk using spectral analysis on accelerometric data--a clinical evaluation study.

Area of Science:

• Gerontology
• Biomedical Engineering
• Clinical Biomechanics

Background:

• Falls are a major cause of morbidity, mortality, and disability in the elderly.
• Accelerometry shows promise for detecting fall risk, but studies often lack standardized settings and clear reference measures.
• Existing methods may not be suitable for unsupervised or real-world clinical application.

Purpose of the Study:

• To develop a simple, unsupervised method for assessing elderly fall risk using accelerometry.
• To evaluate the method's accuracy against established clinical fall risk assessment scores.
• To provide a practical tool for identifying individuals at high risk of falls.

Main Methods:

• Utilized spectral analysis of triaxial accelerometer data recorded in a clinical setting.
• Employed simple logistic regression classifier models for risk assessment.
• Validated the method against three established clinical reference scores for fall risk.

Main Results:

• Achieved overall prediction accuracy ranging from 65.5% to 89.1%.
• Demonstrated sensitivity between 78.5% and 99%, and specificity between 15.4% and 60.4%.
• Indicated fair to good predictive accuracy when compared to common clinical reference scores.

Conclusions:

• The proposed simple, unsupervised method effectively detects high fall risk in the elderly.
• Parameters derived from spectral analysis are procedure-independent, suitable for unsupervised use.
• Future research will involve prospective validation in a larger cohort study.