سامانه همایش‌ها TSTA | ثبت‌نام و ارسال مقاله کنفرانس‌ها و کنگره‌های ملی

Home / سی و دومین کنفرانس ملی و دهمین کنفرانس بین المللی مهندسی زیست پزشکی ایران

Investigating the impact of arm swing on lower limb forces using machine learning techniques

Authors :

Mohammad Reza Seidgar¹ Hadi Farahani² Mostafa Rostami³ Elham Naziri⁴ Sadegh Madadi⁵

1- دانشگاه شهید بهشتی 2- دانشگاه شهید بهشتی 3- دانشگاه صنعتی امیرکبیر 4- دانشگاه صنعتی امیرکبیر 5- دانشگاه صنعتی امیرکبیر

Keywords :

Arm Swing،Lower Limb،Force،Machine Learning

Abstract :

Introduction: Arm swing is a key component of gait mechanics, influencing joint stability, muscle forces, and locomotion efficiency. Understanding this relationship involves biomechanical complexity. This study investigates how different arm swing amplitudes affect lower limb muscle forces during walking. Methods: Motion capture data were collected from 20 healthy participants, with arm swings categorized as large (LA), normal (NA), and small (SA). Muscle force data for 41 muscles were analyzed using OpenSim software. Two approaches were used for classification: (1) statistical feature extraction with machine learning (ML) classifiers (logistic regression, SVM, decision tree, random forest, and XGBoost) and (2) time-series analysis using dynamic time warping (DTW) with weighted K-NN. Clustering was also performed using algorithms such as OPTICS, Hebbian learning, DBSCAN, SOM, BIRCH, and agglomerative clustering after applying Fourier Transform. Results: The best classification performance in the (LA-NA)-SA scenario was achieved by SVM and logistic regression, both reaching 98% accuracy. The DTW-based weighted K-NN approach achieved 84% accuracy. Clustering results showed the highest agreement in the (LA-NA)-SA scenario, with the Hebbian learning model achieving a silhouette score of 0.5 and ARI, AMI, and FMI values of 0.44, 0.52, and 0.69, respectively. Discussion: The results suggest that lower limb muscle force patterns are similar in normal and high arm swing ranges, while low arm swing produces distinct patterns, indicating increased muscular effort. Classification and clustering results were consistent and reinforced each other. Conclusion: This study contributes to the understanding of the importance of arm swing in gait mechanics, with implications for rehabilitation and ergonomic design. The integration of motion capture systems with ML models offers a novel and effective approach to human movement analysis.