Performance and Characteristics of Wearable Sensor Systems Discriminating and Classifying Older Adults According to...

Area of Science:

• Gerontology
• Biomedical Engineering
• Rehabilitation Medicine

Background:

• Sensor-based fall risk assessment (SFRA) uses wearable sensors to monitor motion for fall risk evaluation.
• Existing reviews highlight the need for methodological enhancements to integrate SFRA into clinical practice.
• This review addresses the evidence base and methodological quality of SFRA studies.

Purpose of the Study:

• To systematically review the discriminative capability and classification performance of SFRA.
• To identify methodological factors contributing to the risk of bias in SFRA studies.
• To provide recommendations for improving SFRA methodology.

Main Methods:

• Systematic review of scientific literature following recommended guidelines.
• Inclusion of 33 studies from 389 screened records.
• Analysis of study design, sample characteristics, sensor features, classification models, and validation methods.

Main Results:

• Significant differences in sensor features and classification models were found between fall risk groups (fallers/non-fallers).
• Classification performance metrics (AUC ≥ 0.74, accuracy ≥ 84%) were achieved in several studies.
• Methodological limitations identified include insufficient prospective designs, small sample sizes, and limited use of recommended validation methods.

Conclusions:

• SFRA demonstrates evidence of effectiveness in differentiating fall risk.
• Methodological rigor must be enhanced to reduce bias and improve clinical utility.
• Future research should prioritize prospective designs, larger and more representative samples, and standardized validation techniques.