Mathivanan / Mallik / Sangeetha Advanced Explorations in Machine Learning, Computer Vision, and IoT

1. Auflage 2026
ISBN: 979-8-89881-258-4
Verlag: Bentham Science Publishers
Format: EPUB
Kopierschutz: 0 - No protection

E-Book, Englisch, 340 Seiten

Advanced Explorations in Machine Learning, Computer Vision, and IoT focuses on the convergence of machine learning algorithms, computer vision techniques, and Internet of Things (IoT) infrastructures to enable scalable, adaptive, and real-time intelligent applications.
Balancing strong theoretical foundations with system-level design considerations, the book serves as a structured guide for readers interested in how advanced mathematical models and learning paradigms drive modern AI-enabled IoT ecosystems.
The book begins with the mathematical, probabilistic, and computational principles underlying machine learning and visual intelligence, with subsequent chapters exploring linear and nonlinear models, kernel methods, neural networks, deep learning architectures, and optimisation techniques.

Autoren/Hrsg.

Mathivanan, Sandeep Kumar

Mallik, Saurav

Sangeetha, S. K. B.

Leseproben

AI-Driven Gait Analysis using Wearable Assistive Devices for Personalized Healthcare

Alfred Daniel J.1, *

1 Department of Computer Science and Engineering, Karpagam Academy of Higher Education, Coimbatore, India

Abstract

This chapter addresses the limitations of traditional gait analysis methods by introducing wearable assistive devices integrated with machine learning for a more sophisticated approach. The primary problem is the need for advanced and accurate gait analysis, especially in healthcare and rehabilitation. The approach involves utilizing wearable devices equipped with sensors to collect gait data and applying machine learning algorithms for analysis. The key findings showcase the effectiveness of the proposed integrated approach in providing precise insights into gait patterns. The machine learning model plays a pivotal role in enhancing the accuracy of gait analysis, allowing for more nuanced and personalized assessments. The proposed model uses the LSTM networks framework for AI-driven gait analysis. The system model is evaluated based on metrics such as joint angle time series, Gait Phase Probability Distribution, Gait Recognition for Abnormal Gait Detection, LSTM-based Gait Abnormality Detection, and LSTM-based Gait Prediction.

Keywords: Gait analysis, Healthcare, Machine learning, Rehabilitation, Wearable assistive devices.

* Corresponding author Alfred Daniel J.: Department of Computer Science and Engineering, Karpagam Academy of Higher Education, Coimbatore, India; E-mail: [e-mail]

INTRODUCTION

Gait analysis refers to the systematic study of human walking patterns, encompassing the movement of the limbs, body, and associated biomechanical aspects during locomotion. This analysis provides valuable insights into the functioning of the musculoskeletal system, helping to understand and diagnose various gait-related disorders. Gait analysis is a crucial diagnostic tool in identifying and assessing various neurological, musculoskeletal, and orthopaedic conditions [1]. It aids in detecting abnormalities in walking patterns that may indicate underlying health issues.

Healthcare professionals use gait analysis to formulate personalized treatment plans for patients with conditions such as cerebral palsy, stroke, or orthopaedic injuries. It helps in tailoring interventions to address specific gait abnormalities. Gait analysis serves to monitor the progress of patients undergoing rehabilitation. It allows healthcare practitioners to track changes in gait parameters over time and adjust treatment plans accordingly [2].

Gait analysis is instrumental in orthopaedic rehabilitation after surgeries or injuries. It guides therapists in designing exercises and interventions that promote optimal gait mechanics and reduce the risk of secondary complications. Recent advancements in technology, including wearable devices and machine learning, have further enhanced the precision and accessibility of gait analysis [3]. Wearable assistive devices equipped with sensors can continuously monitor gait patterns outside the laboratory setting, offering valuable real-world data. The gait analysis is a multidisciplinary tool with significant implications for healthcare and rehabilitation. By understanding and quantifying human walking patterns, healthcare professionals can better diagnose, treat, and monitor individuals with various conditions affecting mobility. In cases of neurological disorders such as Parkinson's disease or spinal cord injuries, a gait analysis helps in understanding the impact of these conditions on walking patterns. It aids in developing targeted rehabilitation strategies to enhance mobility. Gait analysis is crucial for individuals using prosthetic limbs or orthotic devices. It ensures these devices properly fit and function, improving mobility and quality of life [4]. The problem or gap in current gait analysis methods that wearable assistive devices aim to address lies in the limitations of traditional laboratory-based assessments. Conventional gait analysis often relies on expensive and immobile equipment, such as motion capture systems, force plates, and cameras, which restrict the assessment to controlled environments like gait laboratories. This approach poses several challenges and shortcomings:

Traditionally, gait analyses are typically conducted in laboratory settings, which may not fully represent diverse and dynamic conditions in real-world scenarios. Walking patterns in everyday life can differ significantly from those observed in a controlled environment. Laboratory-based assessments only offer snapshots of a person's Gait within a limited timeframe. This may not capture the variability and nuances of Gait over an extended period or under different conditions, potentially missing relevant information about walking abnormalities [5]. Gait laboratories are not easily accessible to everyone, especially those living in remote areas or with mobility constraints. This limits the inclusivity of gait analysis, hindering its application to a broader population. The equipment used in traditional gait analysis can be intrusive and may induce changes in natural walking patterns. This can result in an altered gait, as proposed by the Hawthorne effect, which undermines the accuracy and reliability of the assessments. Conducting gait analysis in a laboratory setting requires dedicated time and resources, making it less feasible for routine monitoring and long-term assessments.

Wearable devices equipped with sensors allow for continuous, real-time monitoring of gait patterns in naturalistic settings, providing a more comprehensive understanding of an individual's walking behaviour over time, and allowing gait analysis to be performed outside the confines of a laboratory [6]. This mainly benefits individuals who cannot easily access traditional gait analysis facilities. Wearable devices are less intrusive, minimizing the impact on natural walking patterns and reducing the likelihood of the Hawthorne effect. This facilitates more accurate and ecologically valid assessments. Wearable assistive devices enable longitudinal studies by providing data over extended periods, allowing researchers and healthcare professionals to track changes in gait patterns and assess the effectiveness of interventions [7]. By addressing these limitations, wearable assistive devices contribute to a more holistic and practical approach to gait analysis, enhancing their applicability in various healthcare and rehabilitation contexts [8-11].

The primary goals of this chapter are to investigate and demonstrate the effectiveness of integrating machine learning into gait analysis using wearable assistive devices. The emphasis lies in leveraging machine learning techniques to enhance gait analysis's precision, efficiency, and interpretability in real-world scenarios [12-15]. Assess the performance and feasibility of wearable assistive devices equipped with sensors for collecting gait data in naturalistic environments. Integrate machine learning algorithms into analyzing gait data obtained from wearable devices [16]. This involves applying advanced computational techniques to extract meaningful patterns, features, and abnormalities in the gait signal. Enhance the accuracy and reliability of gait analysis by leveraging the capabilities of machine learning models [17]. This includes the ability to discern subtle variations in gait patterns that may indicate neurological, musculoskeletal, or other health-related conditions. A substantial body of literature discusses traditional methods of gait analysis using laboratory-based equipment, including motion capture systems, force plates, and electromyography. These studies emphasize the importance of understanding temporal, spatial, and kinematic parameters in evaluating gait abnormalities associated with various medical conditions [18]. Existing literature often explores the clinical applications of gait analysis in fields such as orthopaedics, neurology, and rehabilitation. Researchers highlight the diagnostic value of gait parameters in identifying and monitoring conditions like cerebral palsy, Parkinson's disease, and musculoskeletal disorders [19]. Numerous studies delve into developing and applying wearable devices for gait analysis. These devices, equipped with accelerometers, gyroscopes, and other sensors, provide an alternative to traditional methods by enabling continuous and unobtrusive monitoring of gait patterns in real-world settings [20]. Literature includes validation studies comparing the accuracy and reliability of data obtained from wearable devices with that from traditional laboratory setups. Researchers investigate the feasibility of using wearables in diverse populations and conditions, addressing sensor placement, data synchronization, and signal quality concerns.

Longitudinal studies discuss the potential of wearable devices for continuous gait monitoring. Researchers explore the benefits of capturing day-to-day variations in Gait, assessing the impact of interventions, and providing insights into the progression of chronic...