Capturing Compensatory Reserve in Sarcopenia: A Bioengineering Framework for Multidimensional Temporal Analysis of...

Area of Science:

• Bioengineering
• Gerontology
• Computational Neuroscience

Background:

• Sarcopenia patients exhibit subtle balance deficits often missed by conventional assessments due to compensatory strategies.
• Quantifying compensatory reserve is crucial for understanding and managing balance impairments in older adults.

Purpose of the Study:

• To develop and validate a computational framework for analyzing center-of-pressure (COP) signals to quantify compensatory reserve in sarcopenia patients.
• To differentiate between sarcopenia patients and healthy controls based on subtle balance variations.

Main Methods:

• Collected COP data from 82 older adults (sarcopenia vs. controls) during static standing using a force platform.
• Integrated Dynamic Time Warping (DTW) distances, LSTM embeddings, and statistical metrics for multidimensional temporal analysis.
• Employed feature selection and 5-fold cross-validation with SMOTE to ensure model robustness and mitigate overfitting.

Main Results:

• The developed framework achieved high accuracy (0.84 ± 0.04) and ROC-AUC (0.86 ± 0.05) in discriminating between groups, particularly during semi-tandem stance.
• DTW-based features were identified as primary drivers of classification accuracy, correlating with clinical severity.
• Variability in prediction probabilities indicated a gradient of compensatory reserve within patient groups.

Conclusions:

• The study presents a feasible bioengineering methodology using clinical COP analysis to assess compensatory reserve in sarcopenia.
• This approach offers a promising tool for routine clinical assessment, enhancing the detection of subtle balance impairments.
• Further validation and translation into clinical practice are warranted.