RBRoza Beisembekova

Roza Beisembekova

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

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Roza Beisembekova has a total of 24 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    ROI-SSIM guided framework for 3D knee MRI reconstruction under artifact degradation

    by Amanali Bekbolat, Almira Askhatova, Aibek Niyetkaliyev, Prashant Jamwal, Kassymbek Ozhikenov, Zhadyra Alimbayeva, Roza Beisembekova, Yerbolat Igembay

  • Open Access

    DEEP REINFORCEMENT LEARNING CONTROL OF A TENSION-CONSTRAINED TENDON-DRIVEN UNDERACTUATED ROBOTIC FINGER: A COMPARATIVE STUDY OF DDPG AND SAC

    by Kanat Suleimenov, Roza Beisembekova, Beibit Abdikenov, Aiman Ozhikenova, Assylbek Ozhiken, Akim Kapsalyamov

    Tendon-driven robotic fingers are underactuated and strongly nonlinear, and their unidirectional, friction-attenuated tendon tension imposes actuation constraints that any practical controller must respect. A previous study established that a deterministic learning controller (DDPG) can track a grasp trajectory for such a finger and compared it against model-based schemes. The present work extends that line of investigation by asking a narrower but unresolved question: for this same single-actuator, three-link tendon-driven architecture, how does a deterministic actor-critic policy (DDPG) compare with a stochastic, maximum-entropy policy (SAC) once a tension-limit penalty is built directly into the learning objective? Using the previously validated physics-based simulation model, both agents are trained under identical conditions with a reward that jointly penalizes tracking error and tension-constraint violation, and are evaluated across random, sinusoidal, and step disturbances using steady-state error, RMS, ISE, and IAE. The results show a consistent and interpretable trade-off: the stochastic SAC policy attains lower tracking error and better disturbance rejection, especially under time-varying and abrupt loads, at the cost of noisier control and longer training, whereas the deterministic DDPG policy converges faster and produces smoother torque and tension profiles but adapts less well to coupled disturbances. The study isolates the role of the exploration mechanism and the tension penalty in shaping these behaviors and offers practical guidance on selecting between deterministic and stochastic actor-critic methods for physically constrained, underactuated tendon-driven systems.

  • Open Access

    ROI-SSIM Guided Framework for 3D Knee MRI Reconstruction under Artifact Degradation

    by Amanali Bekbolat, Almira Askhatova, Aibek Niyetkaliyev, Prashant Jamwal, Aibek Ozhiken, Zhadyra Alimbayeva, Roza Beisembekova, Yerbolat Igembay

    Magnetic resonance imaging (MRI) reconstruction is highly sensitive to undersampling and acquisition artifacts. Although deep learning-based approaches have significantly improved reconstruction quality, they are typically optimized using global fidelity metrics that may not fully reflect reconstruction quality within clinically important structures such as the anterior cruciate ligament (ACL), meniscus, and articular cartilage. This study proposes aa Region-of-Interest aware MRI reconstruction framework based on a Vision Transformer autoencoder (ViT-AE) guided by a ROI Structural Similarity (ROI-SSIM) loss. The method leverages fastMRI+ bounding-box annotations and incorporates a hybrid artifact simulation framework spanning k-space and image-domain degradations. Experiments on the fastMRI knee single-coil validation set (PD contrast, R=4) show that the proposed ViT-AE achieves competitive reconstruction fidelity to established baselines (SSIM 0.819, Peak Signal-to-Noise Ratio (PSNR) ), while ROI-SSIM supervision improves preservation of relevant anatomy. Under mixed artifact conditions, region-specific structural similarity decreased by only , compared with a decrease in global SSIM. These results indicate that region-aware optimization improves robustness and anatomical preservation under realistic acquisition conditions.Overall, the method presented in this paper:• Introduces ROI-SSIM supervision for region-aware MRI reconstruction.• Develops a hybrid artifact simulation framework spanning k-space and image-domain degradations.• Demonstrates improved anatomical preservation under artifact conditions.

  • Open Access

    Advancing Orthopaedic Imaging: Self-Supervised Vision Transformers for 3D Reconstruction of Knee Magnetic Resonance Imaging

    by Amanali Bekbolat, Kassymbek Ozhikenov, Roza Beisembekova, Damira Mussina, Chingiz Alimbayev, Nariman Imashev

    The main aim of this research is to develop and evaluate a self -supervised learning model based on Vision Transformers for three-dimensional reconstruction of knee Magnetic Resonance Imaging. Three-dimensional reconstruction of hard tissues is an essential procedure for visualization and structural analysis of bones, both in clinical and research settings. Traditional compressed sensing approaches rely on handcrafted priors and iterative solvers, while supervised convolutional neural networks require large datasets of fully sampled Magnetic Resonance Images, which are costly to obtain. To address these limitations, this paper proposes the self-supervised framework based on Vision Transformers architecture along with the integrated convolutional neural network decoder for reconstructing volumetric information from an under-sampled Magnetic Resonance Imaging data. We propose a self-supervised learning framework using Vision Transformers for three – dimensional knee magnetic resonance image reconstruction. Our method leverages masked patch pre-training on a custom dataset of proton density fat-saturated knee magnetic resonance imaging scans, followed by volumetric decoding with a lightweight three- dimensional convolutional neural network. Results demonstrate competitive reconstruction quality with Dice score of 0.87, IoU of 0.83, and recall of 0.93, showing that self-supervised vision transformers effectively capture long-range dependencies and volumetric continuity. This study highlights the potential of Vision transformer-based self ‒ supervised learning frameworks to overcome the limitations of traditional supervised deep learning approaches. The proposed method demonstrated that it can reduce dependence on large fully sampled datasets and support accurate three -dimensional reconstructions for clinical and research applications.

  • Open Access

    Machine learning-based classification of local muscle fatigue using electromyography signals for enhanced rehabilitation outcomes

    by Zhanel Baigarayeva, Assiya Boltaboyeva, Baglan Imanbek, Kassymbek Ozhikenov, Nurgul Karymssakova, Roza Beisembekova

    Muscle fatigue is a key factor affecting rehabilitation progress, safety, and patient engagement. Accurate detection of fatigue during physical activity remains a challenge, particularly in clinical and remote settings. This study presents the development of an Internet of things-based system for classifying local muscle fatigue using surface electromyography (EMG) signals and machine learning. A wearable device was used to collect real-time EMG data and subjective fatigue ratings from 10 healthy participants during sustained isometric grip exercises. Feature extraction was performed on-device, and the data were transmitted wirelessly for analysis. Machine learning models including logistic regression, decision tree (DT), random forest, and extreme gradient boosting (XGBoost) were trained to classify fatigue states. The DT model achieved the highest accuracy of 90.7%, with a precision of 90.7% and a recall of 90.9%. SHAP analysis revealed time under load, smoking, and alcohol use as the most influential factors in fatigue classification. These results show that wearable EMG devices combined with smart algorithms are effective for real-time fatigue monitoring during rehabilitation.


  • Open Access

    AI-Driven Smart Wearable Systems for Personalized Rehabilitation Monitoring

    by Roza Beisembekova, Tanzhuldyz Aidarova, Assiya Boltaboeva, Baglan Imanbek, Kassymbek Ozhikenov, Nurgul Karymsakova

    The growing demand for personalized and effective rehabilitation strategies has driven the development of AI-driven smart wearable systems. These systems provide real-time health monitoring, prediction, and adaptive feedback, which helps to improve clinical outcomes in neuro-musculoskeletal and postoperative rehabilitation. This scoping review analyzes 14 peer-reviewed articles published between 2018 and 2025, comprising clinical trials, cohort studies, and engineering applications. The selected studies were identified through structured searches in IEEE, MDPI, and other scholarly databases, based on relevance to AI-enhanced wearable rehabilitation devices. Commonly used AI algorithms include support vector machines (SVM), convolutional neural networks (CNN), and reinforcement learning (RL), enabling functions such as gait analysis, joint movement recognition, muscle activation tracking, and postural control. In addition, integration with IoT sensor networks, cloud-based platforms, and telemedicine interfaces was widely reported. The review finds that AI-enabled wearables significantly improve patient adherence, monitoring accuracy, and personalized therapy delivery. Nonetheless, challenges remain in data security, sensor calibration, interoperability, and long-term user retention. These results confirm that smart wearables play an important role in supporting personalized, data-driven rehabilitation.