Bias in Surface Electromyography Features across a Demographically Diverse Cohort

2026-04-15 · Source: Machine Learning · Field: Science & Research — Artificial Intelligence & Machine Learning, Health & Medical Research, Engineering & Applied Sciences · Depth: Expert, quick

Summary

A study analyzed surface electromyography (sEMG) features from 81 demographically diverse individuals performing discrete hand gestures to identify associations with demographic characteristics. The research utilized a dataset and derived 147 common sEMG features, employing mixed-effects linear models and partial least squares (PLS) analysis. Key demographic variables considered included age, sex, height, weight, skin properties, subcutaneous fat, and hair density. The findings revealed that 33% (49 of 147) of the commonly used sEMG features exhibited significant associations with these demographic factors. This variability in sEMG characteristics often necessitates extensive personalization for reliable performance in human-machine interfaces and assistive devices.

Key takeaway

For Machine Learning Engineers developing sEMG-based neural interfaces, understanding and mitigating demographic bias is crucial. Your models must account for individual differences like age, sex, and body composition, as 33% of common sEMG features are significantly affected. Prioritize robust personalization strategies or bias-aware feature engineering to ensure fair and consistent performance across diverse user populations.

Key insights

Demographic differences significantly bias sEMG features, impacting human-machine interface performance and fairness.

Principles

• sEMG features are highly idiosyncratic.
• Individual differences alter sEMG signal quality.

Method

The study used mixed-effects linear models and PLS analysis on 147 sEMG features from 81 individuals to identify associations with demographic variables.

In practice

• Consider demographic factors in sEMG device design.
• Account for individual variability in sEMG feature extraction.

Topics

• Surface Electromyography
• Demographic Bias
• Human-Machine Interfaces
• Gesture Decoding
• Machine Learning

Best for: AI Scientist, Machine Learning Engineer, Research Scientist

Related on AIssential

• Fair Cognitive Impairment Detection Through Unlearning — Multimodal fusion and gradient reversal unlearning can significantly reduce demographic bias in MCI detection models while improving overall performance.
• A Practical Upper Bound on Selection Bias Effects in Medical Prediction Models — A novel upper bound estimates worst-case model performance under selection bias with partially observed data.
• Auditing Multimodal LLM Raters: Central Tendency Bias in Clinical Ordinal Scoring — Multimodal LLMs exhibit a central tendency bias in clinical ordinal scoring, compressing predictions towards the scale's middle.
• Effect of Demographic Bias on Skin Lesion Classification — Demographic biases in skin lesion classification models stem from distinct mechanisms requiring targeted mitigation strategies.
• Comparison of sEMG Encoding Accuracy Across Speech Modes Using Articulatory and Phoneme Features — SPARC features predict sEMG better than phonemes across speech modes, supporting silent-speech modeling.
• Gender Disambiguation in Machine Translation: Diagnostic Evaluation in Decoder-Only Architectures — Decoder-only MT models exhibit gender bias, but post-training improves contextual awareness and reduces masculine "Prior Bias."

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Machine Learning.