Comparison of Human Interpretation and a Rule-Based Algorithm for Instrumented Sit-to-Stand Test

Area of Science:

• Gerontology
• Biomedical Engineering
• Physical Therapy

Background:

• The five times sit-to-stand test (5STS) is a widely used clinical tool for evaluating lower extremity physical function.
• Accurate assessment of 5STS is crucial for diagnosing and monitoring conditions like sarcopenia.
• Current 5STS interpretation relies on human observation, which can be subjective.

Purpose of the Study:

• To assess the correlation between human interpretation of the 5STS (5STShuman) and a novel rule-based algorithm (5STSrule).
• To evaluate the performance of an instrumented 5STS using two sensors for automated data collection.
• To compare the ability of both methods in classifying sarcopenia.

Main Methods:

• Analysis of clinical records from 148 geriatric outpatients undergoing the electronic Short Physical Performance Battery (eSPPB) protocol.
• Instrumented 5STS using a load cell and LiDAR to generate time-weight and time-distance curves.
• Calculation of 5STShuman via manual assessment of curves and 5STSrule via an empirical rule-based algorithm.

Main Results:

• No significant difference was found between mean 5STShuman (12.2±0.4s) and 5STSrule (11.4±0.4s) times (p=0.232).
• A strong positive correlation was observed between 5STShuman and 5STSrule (β=0.99, R2=0.99).
• Both methods demonstrated comparable ability in classifying sarcopenia (C-indices of 0.826 for 5STShuman and 0.820 for 5STSrule).

Conclusions:

• An empirical rule-based algorithm for instrumented 5STS demonstrates strong correlation with human interpretation.
• The automated algorithm exhibits comparable diagnostic capability for sarcopenia as traditional human assessment.
• This suggests a reliable and objective automated approach for 5STS physical performance evaluation.