A Transfer Learning Approach for Toe Walking Recognition Using Surface Electromyography on Leg Muscles

Sensors, Год журнала: 2025, Номер 25(5), С. 1305 - 1305

Опубликована: Фев. 20, 2025

Gait is a complex motor process that involves the coordination and synchronization of various body parts through continuous interaction with environment. Monitoring gait crucial for early detection abnormalities, such as toe walking, which characterized by limited or absent heel contact floor during walking. Persistent walking can cause severe foot, ankle, musculature conditions; poor balance; increased risk falling tripping; affect overall quality life, making it difficult, example, to participate in sports social activities. This study proposes new approach detect using surface Electromyography (sEMG) on lower limbs. sEMG sensors, measuring electrical activity muscles, see signals before movement corresponding muscle activation, contributing an possible problem. The signal presents significant complexity due its noisy nature challenge extracting meaningful features classification. To address this issue enhance model’s robustness across different devices configurations, Transfer Learning (TL) introduced. method leverages pre-trained models effectively handle variability data improve classification accuracy. In particular, Continuous Wavelet Transform (CWT) applied sEMG-filtered (with time windows 1 s) convert them into 2D images (scalograms). Preliminary tests were performed public dataset some most well-known architectures, obtaining accuracy about 95% InceptionResNetV2.

Язык: Английский

---

Gender-Based Differences in Biomechanical Walking Patterns of Athletes Using Inertial Sensors

Journal of Functional Morphology and Kinesiology, Год журнала: 2025, Номер 10(1), С. 82 - 82

Опубликована: Фев. 27, 2025

Background: Wearable inertial sensors are essential tools in biomechanics and sports science for assessing gait real-world conditions. This study explored gender-based differences biomechanical walking patterns among healthy Greek athletes using the BTS G-Walk system, focusing on key parameters to inform gender-specific training rehabilitation strategies. Methods: Ninety-five (55 men, 40 women), aged 18 30 years, participated this study. Each athlete performed a standardized 14 m walk while 17 were recorded sensor. Statistical analyses conducted SPSS assess gender left–right foot symmetry. Results: No significant asymmetry was found between left right feet most parameters. Men exhibited longer stride lengths (left: p = 0.005, Cohen’s d 0.61; right: 0.009, 0.53) cycle durations 0.025, 0.52; 0.53). Women showed higher cadence (p 0.022, −0.52) greater propulsion index 0.001, −0.71; −0.73), as well percentage of first double support 0.030, −0.44). Conclusions: These findings highlight impact biological patterns, emphasizing need rehabilitation. The system proved reliable analysis, with potential optimizing performance, injury prevention, athletes. Future research should explore larger, more diverse populations multi-sensor setups.

Язык: Английский

---

Readiness, recovery, and strain: an evaluation of composite health scores in consumer wearables

Deleted Journal, Год журнала: 2025, Номер unknown

Опубликована: Апрель 9, 2025

Abstract Introduction Consumer wearables increasingly provide users with Composite Health Scores (CHS) – integrated biometric indices that claim to quantify readiness, recovery, stress, or overall well-being. Despite their growing adoption, the validity, transparency, and physiological relevance of these scores remain unclear. This study systematically evaluates CHS from leading wearable manufacturers assess underlying methodologies, contributors, scientific basis. Content Information was synthesised publicly available company documentation, including technical white papers, user manuals, app interfaces, research literature where available. We identified 14 across 10 major manufacturers, Fitbit (Daily Readiness), Garmin (Body Battery™ Training Oura (Readiness Resilience), WHOOP (Strain, Recovery, Stress Monitor), Polar (Nightly Recharge™), Samsung (Energy Score), Suunto Resources), Ultrahuman (Dynamic Recovery), Coros Stress), Withings (Health Improvement Score). The most frequently incorporated contributors in this catalogue were heart rate variability (86 %), resting (79 physical activity (71 sleep duration %). However, significant discrepancies data collection timeframes, metric weighting, proprietary scoring methodologies. None disclosed exact algorithmic formulas, few provided empirical validation peer-reviewed evidence supporting accuracy clinical scores. Summary outlook While concept represent a promising innovation digital health, applicability uncertain. Future should focus on establishing standardized sensor fusion frameworks, improving evaluating diverse populations. Greater collaboration between industry, researchers, clinicians is essential ensure serve as meaningful health metrics rather than opaque consumer tools.

Язык: Английский

---

Advancements in Wearable Sensor Technologies for Health Monitoring: A Systematic Review of Clinical Applications, Rehabilitation, and Disease Risk Assessment (Preprint)

Опубликована: Апрель 17, 2025

Wearable sensor technologies, such as inertial measurement units (IMUs), smartwatches, and multi-sensor systems, have emerged valuable tools in clinical real-world health monitoring. These devices allow continuous, non-invasive tracking of gait, mobility, functional across a variety populations. However, significant challenges remain, including variability placement, data processing methodologies, insufficient validation settings. This systematic review aims to evaluate recent literature on the research applications wearable sensors. Specifically, it investigates how these technologies are used assess predict disease risk, support rehabilitation. It also identifies limitations proposes future directions. The was conducted according PRISMA guidelines. A comprehensive search PubMed, Scopus, Web Science databases performed for studies published past ten years. Inclusion criteria focused using sensors or environments. total 30 eligible were identified qualitative synthesis. Data extracted included study design, population characteristics, type machine learning algorithms, outcomes. Among reviewed studies, observational designs most common (43.3%), followed by experimental (26.7%) randomized controlled trials (10%). IMU-based 66.7% with wrist-worn being placement (43.3%). Machine techniques frequently applied, random forest (20%) deep (16.7%) models predominating. Clinical spanned Parkinson’s disease, stroke, multiple sclerosis, frailty, several reporting high predictive accuracy fall risk mobility decline (AUROC up 0.919, p < 0.05). demonstrate strong potential enhancing monitoring, assessment, rehabilitation both remain standardizing protocols analysis. Future should focus large-scale, longitudinal harmonized pipelines, integration cloud-based systems improve scalability translation.

Язык: Английский

---