Quantifying Accelerometer-Based Tremor Features of Neuromuscular Fatigue in Healthy and Diabetic Adults

Neuromuscular fatigue affects workers' productivity and health, which is further deteriorated with chronic conditions such as type 1 diabetes (T1D). Enhanced physiological tremor, a key indicator of neuromuscular fatigue, shows great potential in detecting the onset of neuromuscular fatigue. This study aims to determine the feasibility of using a cost-effective wearable accelerometer-based microelectromechanical sensor to convey neuromuscular fatigue-related tremor information in healthy and T1D adults. 42 adults (22 healthy, 20 T1D), equipped with a finger and a wrist accelerometer, performed intermittent submaximal isometric handgrip fatigue exercises using a grip dynamometer. Motor variability feature, namely, Coefficient of Variation (CV), and motor complexity feature, namely approximate entropy (ApEn), were extracted from the force signal of dynamometer and from the finger and wrist tremor accelerometry signals and subjected to statistical analysis. First, significant positive correlations were found between tremor accelerometry and force signal in terms of motor variability and complexity features. Second, a three-way (fatigue phase: early, middle, late; gender: male, female; condition: healthy, T1D) analysis of variance resulted in a significant fatigue effect on both accelerometry and force measurements in terms of motor variability and complexity features. Apart from finger CV, no other features showed any gender or condition effects. These findings indicate that finger and wrist tremors measured by accelerometer-based sensors can retain the robustness of fatigue-related motor variability and complexity. Wrist tremor features were found to capture fatigue development across both healthy and diabetic males and females, thereby offering comparable fatigue detection and management in adults with chronic conditions.