AVAnkit Vijayvargiya

Ankit Vijayvargiya

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Ankit Vijayvargiya has a total of 28 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Editorial: Enhancing gait therapy with artificial intelligence: current trends and future prospects

    by Ankit Vijayvargiya, Rajesh Kumar, Pilani Nkomozepi, Kieran Moran, Chandra Prakash

    Impaired mobility is among the most common impairments caused by neurologic, neurodegenerative, and orthopaedic disorders. Depending upon the cause being stroke, Parkinson's disease, spinocerebellar ataxia, or osteoarthritis needing arthroplasty, the disruption in gait and mobility causes dependency, increased risk of falls, and decreased quality of life. Conventional methods used for assessing these disruptions have been based on visual inspection, rating scales, and occasional visits to clinics, which are time-consuming, observerdependent, and has limited capacity of providing information on how a patient walks during daily routine activities or on the positioning of interventions and implants.The current Research Topic was compiled to highlight how digital health technologies, such as wearable inertial sensors, markerless motion capture, exergame technology, and artificial intelligence/machine learning (AI/ML), can be employed to make movement assessment more objective, continuous, and personalizable. Five papers included in the current collection present the research that covers a broad spectrum of clinical and technical issues, from a randomized controlled trial of motor-cognitive exergame training in post-stroke rehabilitation to a narrative review of AI-assisted preoperative planning for total hip arthroplasty and three studies demonstrating the validation and application of sensor and machine-learning-based gait analysis in healthy subjects, patients with Parkinson's disease and spinocerebellar ataxia.Personalized motor-cognitive exergaming after stroke Huber et al. [1] present the results of the PEMOCS trial, a single-blind, randomized, controlled trial comparing the effects of conceptually guided, personalized, motor-cognitive exergame training with usual care in 46 community-living stroke survivors with chronic deficits. Although the primary outcome measure (global cognition assessed using the Montreal Cognitive Assessment) was not significantly different between the exergame and usual care groups, it was found that there were more responders in the exergame group. There were some interaction effects between time points in favour of the exergame group, for perceived mobility, intrinsic visual attention, working memory, outdoor gait speed and swing width. It is discussed how the frequency of training, exercise intensity, and a high-functioning baseline sample may have prevented the researchers from finding a bigger effect, and possible directions for future exergame-based interventions in chronic stroke patients are proposed.Wang et al. [2] present a narrative review of artificial intelligence-assisted, three-dimensional preoperative planning of the position and sizing of acetabular cups using CT scans in total hip arthroplasty. This review summarizes findings showing that AI-driven planning based on automated segmentation, deep learning landmark detection, and biomechanical simulation allows for a reduction in the mean angular error of cup inclination and anteversion to less than 3°, 80-85% size-matching accuracy within one implant size, and at least 50% decrease in time needed for planning relative to manual templating. The authors also highlight important limitations of the existing evidence, including a lack of high-quality randomised studies, poor representation of revisions and complicated cases with acetabular defects among patients whose imaging scans were used to train the algorithms, and "black box" algorithms as important challenges impeding clinical adoption of this technology.Cafolla and Chaparro-Rico [3] have performed system characterization of the current version of SANE (eaSy gAit aNalysis systEm), a real-time gait analysis system based on two depth cameras and pose estimation using artificial intelligence. Following the earlier clinical validation of SANE against marker-based methods, the authors concentrate on the robustness of this system in the present study. Over the course of four runs and 80 walks by a healthy individual during one week, with a re-test one week later, including system resets, the relative error for spatiotemporal parameters was mostly under 5%, and the standard deviation was on par with the results from previous studies with marker-based systems. The study illustrates that dual-camera processing could alleviate occlusions and allow for maintaining reliable, highframerate (~80 fps) real-time processing of gait data.Mittal et al. [4] apply Decision Tree, Random Forest, XGBoost, and LightGBM classifiers to ground-reaction-force gait data to detect Parkinson's disease and classify its severity on the Hoehn and Yahr scale, using two open-access benchmark datasets. They have achieved 98.25% accuracy on the PhysioNet gait dataset and 85% accuracy on another independent Figshare dataset, which is better than other classifiers and some earlier published works. Notably, the researchers use techniques of explainable artificial intelligence to show individual decision path that leads to every classification result, an effort aimed at building up clinical trust in the proposed model to be used as a quick, automatic test to screen Parkinson's disease staging, while recognizing the necessity for future validations on wearable devices and prospective datasets.Shah et al. [5] compare the gait parameters derived using body-worn inertial sensors on the two-minute walk test done in a supervised environment, during an in-clinic appointment, and during unsupervised, day-to-day monitoring in a week in 26 patients with spinocerebellar ataxia (SCA types 1, 2, 3, and 6) and 13 age-matched healthy controls. Gait parameters derived during the in-clinic walk test were better at distinguishing the two groups. They had better correlations with clinical and self-reported parameters. Still, the two parameters of gait variability of stride timing, swing time variability and double-support time variability showed robust discrimination power in both cases. This...

  • Open Access

    BAYES-GAIT: A Bayesian Deep Neural Network for Continuous Gait Reference Trajectory Generation Using an Enhanced Loss Function

    by Bharat Singh, Suchit Patel, Ankit Vijayvargiya, Rajesh Kumar, Naveen Gehlot

    Background: Adaptive gait trajectory prediction is essential to achieve natural and stable locomotion in prosthetic limbs and legged robots, particularly under varied conditions such as changing inclines and walking speeds. Although deep neural networks have shown potential for learning complex joint gait trajectory patterns, conventional deep neural networks often suffer from prediction bias and lack mechanisms for uncertainty estimation, which limits their reliability in dynamic or unseen environments. Methods: This study proposes a Bayesian deep neural network framework for predicting continuous joint trajectories in the sagittal plane across diverse ambulation tasks. The approach integrates Bayesian inference to estimate predictive uncertainty and enhance model generalization. A new hybrid loss function is introduced that combines the standard error of the dataset with variational mean absolute error or mean squared error, allowing the model to leverage low-variance gait segments more effectively during training. The framework is trained and evaluated using a dataset that includes varying walking speeds and inclines. Results: The proposed Bayesian model outperforms traditional state-of-the-art methods, particularly deep neural networks and finite state machine baselines, and shows up to 15% and 25% improvements, respectively, in joint trajectory prediction accuracy. Moreover, the Bayesian deep neural network provides 95% prediction intervals, indicating strong reliability and robustness under uncertain conditions. Significance: By incorporating uncertainty quantification and dataset-driven variance into the learning process, this study offers a more reliable and adaptive approach to gait modeling. These improvements support enhanced gait control in assistive technologies and legged robotic systems operating in variable terrain conditions in the real world.

  • Medical Knowledge Paradigms for Enabling the Digital Health Ecosystem

    by Usha Desai, Vivek P. Chavda, Ankit Vijayvargiya, Ravichander Janapati

  • 1D self-organizing feature map technique for diagnosis of COVID-19 infection

    by Naveen Gehlot, Ankit Vijayvargiya, Rajesh Kumar, Akhil Ranjan Garg, Nilanjan Dey

  • Open Access

    S-WD-EEMD: A hybrid framework for imbalanced sEMG signal analysis in diagnosis of human knee abnormality

    by Ankit Vijayvargiya, Aparna Sinha, Naveen Gehlot, Ashutosh Jena, Rajesh Kumar, Kieran Moran

    The diagnosis of human knee abnormalities using the surface electromyography (sEMG) signal obtained from lower limb muscles with machine learning is a major problem due to the noisy nature of the sEMG signal and the imbalance in data corresponding to healthy and knee abnormal subjects. To address this challenge, a combination of wavelet decomposition (WD) with ensemble empirical mode decomposition (EEMD) and the Synthetic Minority Oversampling Technique (S-WD-EEMD) is proposed. In this study, a hybrid WD-EEMD is considered for the minimization of noises produced in the sEMG signal during the collection, while the Synthetic Minority Oversampling Technique (SMOTE) is considered to balance the data by increasing the minority class samples during the training of machine learning techniques. The findings indicate that the hybrid WD-EEMD with SMOTE oversampling technique enhances the efficacy of the examined classifiers when employed on the imbalanced sEMG data. The F-Score of the Extra Tree Classifier, when utilizing WD-EEMD signal processing with SMOTE oversampling, is 98.4%, whereas, without the SMOTE oversampling technique, it is 95.1%.


  • Open Access

    supDQN: Supervised Rewarding Strategy Driven Deep Q-Network for sEMG Signal Decontamination

    by Ashutosh Jena, Naveen Gehlot, Rajesh Kumar, Ankit Vijayvargiya, Mahipal Bukya

    The presence of muscles throughout the active parts of the body, such as the upper and lower limbs, makes electromyography-based human-machine interaction prevalent. However, muscle signals are stochastic and noisy, with noises being both regular and irregular. Irregular noises due to movements or electrical switching require dynamic filtering. Conventionally, filters are stacked, which unnecessarily trims and delays the signal. This study introduces a decontamination technique involving a supervised rewarding strategy to drive a deep Q-network-based agent (supDQN). It applies one of three filters to decontaminate a 1 sec long surface electromyography signal, which is dynamically contaminated. A machine learning agent identifies whether the signal after filtering is clean or noisy, generating a reward accordingly. The identification accuracy is enhanced by using a local interpretable model-agnostic explanation. The deep Q-network is guided by this reward to select the filter optimally while decontaminating a signal. The proposed filtering strategy is tested on four noise levels (−5 dB, −1 dB, +1 dB, +5 dB). supDQN filters the signal desirably when the signal-to-noise ratio (SNR) is between -5 dB to +1 dB but filters less desirably at high SNR (+5 dB). A normalized root mean square ( $\Omega $ ) is formulated to depict the difference of the filtered signal from the ground truth. This is used to compare supDQN and conventional methods, including wavelet denoising with debauchies and symlet wavelet, high-order low-pass filter, notch filter, and high-pass filter. The proposed filtering strategy gives an average $\Omega $ value of 1.1974, which is lower than that of the conventional filters.


  • PC-GNN: Pearson Correlation-Based Graph Neural Network for Recognition of Human Lower Limb Activity Using sEMG Signal

    by Ankit Vijayvargiya, Rajesh Kumar, Parul Sharma

    Artificial intelligence has a plethora of applications in the realm of biomedical sciences, such as pattern recognition, diagnosis of disease, human–machine interaction, medical image processing, robotic limbs, or exoskeletons. Robotic limbs, or exoskeletons, are widely employed to assist with lower limb movement. To increase the exoskeleton's flexibility in the lower extremities, it is critical to recognize the diverse motion intents of the lower limbs of the human body. In this investigation, sEMG signals from lower limb muscles are used for a computer-aided recognition system to correctly identify the lower limb activities because these signals can identify movement ahead of time and enable faster detection of signal fluctuation than other wearable sensors. Several types of noise are introduced into the signal during collection. A multistage classification strategy is proposed to overcome the processing challenges associated with these sEMG signals. Initially, nine time-domain handcrafted features are retrieved using a hybrid of wavelet denoising and ensemble empirical mode decomposition approach with a sliding window of 256 ms and a 25% overlap. Next, a Pearson correlation-based graph is formed from the extracted features and applied to a graph neural network (GNN). GNN not only captures individual information but also makes use of information from other samples to form a graph. The combination of a Pearson correlation-based graph with a GNN is referred to as Pearson correlation-based GNN. The observation states that the approach proposed in the research achieved an accuracy of 99.19%, 99.02%, 96.21% for the walking, sitting, and standing of healthy subjects, while 99.29%, 97.97%, 99.36% for the subjects comprising knee abnormalities, respectively.


  • Data-driven gait model for bipedal locomotion over continuous changing speeds and inclines

    by Bharat Singh, Suchit Patel, Ankit Vijayvargiya, Rajesh Kumar


  • Open Access

    Hybrid Deep Learning Approaches for sEMG Signal-Based Lower Limb Activity Recognition

    by Ankit Vijayvargiya, Bharat Singh, Rajesh Kumar, Usha Desai, Jude Hemanth

    Lower limb activity recognition utilizing body sensor data has attracted researchers due to its practical applications, such as neuromuscular disease detection and kinesiological investigations. The employment of wearable sensors including accelerometers, gyroscopes, and surface electromyography has grown due to their low cost and broad applicability. Electromyography (EMG) sensors are preferable for automated control of a lower limb exoskeleton or prosthesis since they detect the signal beforehand and allow faster movement detection. The study presents hybrid deep learning models for lower limb activity recognition. Noise is suppressed using discrete wavelet transform, and then the signal is segmented using overlapping windowing. Convolutional neural network is used for temporal learning, whereas long short-term memory or gated recurrent unit is used for sequence learning. After that, performance indices of the models such as accuracy, sensitivity, specificity, and F-score are calculated. The findings indicate that the suggested hybrid model outperforms the individual models.


  • Hardware Implementation for Lower Limb Surface EMG Measurement and Analysis Using Explainable AI for Activity Recognition

    by Ankit Vijayvargiya, Puneet Singh, Rajesh Kumar, Nilanjan Dey

    Electromyography (EMG) signals are gaining popularity for several biomedical applications, including pattern recognition, disease detection, human–machine interfaces, medical image processing, and robotic limb or exoskeleton fabrication. In this study, a two-channel data acquisition system for measuring EMG signals is proposed for human lower limb activity recognition. Five leg activities have been accomplished to measure EMG signals from two lower limb muscles to validate the developed hardware. Five subjects (three males and two females) were chosen to acquire EMG signals during these activities. The raw EMG signal was first denoised using a hybrid of Wavelet Decomposition with Ensemble Empirical Mode Decomposition (WD-EEMD) approach to classify the recorded EMG dataset. Then, eight time-domain (TD) features were extracted using the overlapping windowing technique. An investigation into the comparative effectiveness of several classifiers is presented, although it was hard to distinguish how the classifiers predicted the activities. Having a trustworthy explanation for the outcomes of these classifiers would be quite beneficial overall. An approach known as explainable artificial intelligence (XAI) was introduced to produce trustworthy predictive modeling results and applied the XAI technique known as local interpretable model-agnostic explanations (LIME) to a straightforward human interpretation. LIME investigates how extracted features are anticipated and which features are most responsible for each action. The accuracy of the extra tree classifier gives the highest accuracy of the other studied algorithms for identifying different human lower limb activities from sEMG signals.