Zhanar Kemelbekova
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Zhanar Kemelbekova has a total of 23 co-authors. Here are the top co-authors by number of shared publications:
- A. E. Kozhabekova
M.Auezov South Kazakhstan Research University
3 shared publications - D. U. Ydrysbaev2 shared publications
- E. A. Nysanov
профессор Южно-Казахстанского университета им. М. Ауэзова, Шымкент, Казахстан, доктор физико-математических наук
2 shared publications Zh. Umarova
M.Auezov South Kazakhstan Research University
2 shared publications- A. Baidibekova
M. Auezov South Kazakhstan University
1 shared publication - A. H. Makhatova1 shared publication
Latest Publications
- Open Access
- 30 September 2026
- Published by Astana IT University in Scientific Journal of Astana IT University
- Vol. 27 (3) , 49-65
AN EFFECTIVE METHOD FOR ANALYZING HUMAN MOVEMENT IN REAL TIME BASED ON YOLO-ROI
Real-time human movement analysis has become an essential component of modern computer vision applications, including sports performance assessment, rehabilitation monitoring, intelligent surveillance, and human–computer interaction. Although deep learning–based pose estimation methods provide accurate human body landmark detection, executing person detection on every video frame introduces considerable computational overhead, limiting their applicability on resource-constrained CPU-based systems. This paper proposes an efficient YOLO–ROI framework for real-time human movement analysis that combines scheduled YOLO person detection, region-of-interest (ROI) generation, MediaPipe Pose estimation, and lightweight landmark smoothing. Instead of performing object detection on every frame, the proposed framework periodically executes the YOLO detector while reusing the most recently detected ROI between detection cycles. Consequently, MediaPipe Pose continuously estimates body landmarks within the ROI, significantly reducing redundant object detection operations and improving computational efficiency. The proposed framework was evaluated using a self-collected dataset consisting of 200 videos recorded from 10 participants performing five physical exercises under controlled indoor conditions. Computational performance was analyzed under different ROI sizes, detection intervals, and MediaPipe Pose complexity levels using CPU-based execution. Experimental results demonstrate that the proposed optimization strategy achieves a maximum processing speed of 61.32 FPS with an average processing latency of 16.31 ms, enabling stable real-time operation without GPU acceleration. The obtained results further show that scheduled YOLO detection, ROI reuse, and lightweight landmark smoothing effectively reduce computational workload while preserving continuous pose estimation throughout video processing. The proposed framework provides a simple, computationally efficient, and easily deployable solution for real-time human movement analysis and can be readily integrated into sports analytics, rehabilitation systems, intelligent surveillance applications, and edge-based computer vision platforms.
- 30 September 2026
- Published by Astana IT University in Scientific Journal of Astana IT University
- Vol. 27 (3) , 49-65
Open AccessAN EFFECTIVE METHOD FOR ANALYZING HUMAN MOVEMENT IN REAL TIME BASED ON YOLO-ROI
Real-time human movement analysis has become an essential component of modern computer vision applications, including sports performance assessment, rehabilitation monitoring, intelligent surveillance, and human–computer interaction. Although deep learning–based pose estimation methods provide accurate human body landmark detection, executing person detection on every video frame introduces considerable computational overhead, limiting their applicability on resource-constrained CPU-based systems. This paper proposes an efficient YOLO–ROI framework for real-time human movement analysis that combines scheduled YOLO person detection, region-of-interest (ROI) generation, MediaPipe Pose estimation, and lightweight landmark smoothing. Instead of performing object detection on every frame, the proposed framework periodically executes the YOLO detector while reusing the most recently detected ROI between detection cycles. Consequently, MediaPipe Pose continuously estimates body landmarks within the ROI, significantly reducing redundant object detection operations and improving computational efficiency. The proposed framework was evaluated using a self-collected dataset consisting of 200 videos recorded from 10 participants performing five physical exercises under controlled indoor conditions. Computational performance was analyzed under different ROI sizes, detection intervals, and MediaPipe Pose complexity levels using CPU-based execution. Experimental results demonstrate that the proposed optimization strategy achieves a maximum processing speed of 61.32 FPS with an average processing latency of 16.31 ms, enabling stable real-time operation without GPU acceleration. The obtained results further show that scheduled YOLO detection, ROI reuse, and lightweight landmark smoothing effectively reduce computational workload while preserving continuous pose estimation throughout video processing. The proposed framework provides a simple, computationally efficient, and easily deployable solution for real-time human movement analysis and can be readily integrated into sports analytics, rehabilitation systems, intelligent surveillance applications, and edge-based computer vision platforms.
- 15 September 2026
- Published by National Engineering Academy of the Republic of Kazakhstan in Bulletin of the National Engineering Academy of the Republic of Kazakhstan
- Vol. 101 (3) , 70-81
An explainable AI approach based on XGBoost for leadership index prediction
This article presents a machine learning-based approach to predicting the leadership index based on students’ learning activity, digital skills, and the level of use of artificial intelligence tools. In the course of the study, a synthetic dataset of 110 observations was formed, taking into account the statistical characteristics typical of educational analytics. Based on this data, an XGBoost regression model based on the gradient boosting algorithm was developed. The effectiveness of the model was assessed using the coefficient of determination (R²), root mean square error (RMSE), and mean absolute error (MAE). The results obtained showed that the proposed model works with high accuracy and can effectively identify nonlinear and complex dependencies in the data. In the course of the study, factor importance, permutation analysis, and partial dependence methods were used to interpret the model results. The analysis results showed that the performance indicator is the main determinant in the formation of the leadership index, and the level of use of artificial intelligence tools was identified as a factor with an important but heterogeneous effect. The scientific novelty of the study is characterized by the provision of an integrated approach to predicting students’ complex competencies by combining educational data analysis and explanatory artificial intelligence methods. The proposed method can serve as a basis for developing decision-making support systems in the field of educational analytics, identifying risk groups, and forming personalized learning strategies.
- 15 September 2026
- Published by National Engineering Academy of the Republic of Kazakhstan in Bulletin of the National Engineering Academy of the Republic of Kazakhstan
- Vol. 101 (3) , 70-81
An explainable AI approach based on XGBoost for leadership index prediction
This article presents a machine learning-based approach to predicting the leadership index based on students’ learning activity, digital skills, and the level of use of artificial intelligence tools. In the course of the study, a synthetic dataset of 110 observations was formed, taking into account the statistical characteristics typical of educational analytics. Based on this data, an XGBoost regression model based on the gradient boosting algorithm was developed. The effectiveness of the model was assessed using the coefficient of determination (R²), root mean square error (RMSE), and mean absolute error (MAE). The results obtained showed that the proposed model works with high accuracy and can effectively identify nonlinear and complex dependencies in the data. In the course of the study, factor importance, permutation analysis, and partial dependence methods were used to interpret the model results. The analysis results showed that the performance indicator is the main determinant in the formation of the leadership index, and the level of use of artificial intelligence tools was identified as a factor with an important but heterogeneous effect. The scientific novelty of the study is characterized by the provision of an integrated approach to predicting students’ complex competencies by combining educational data analysis and explanatory artificial intelligence methods. The proposed method can serve as a basis for developing decision-making support systems in the field of educational analytics, identifying risk groups, and forming personalized learning strategies.
- 30 March 2026
- Published by National Engineering Academy of the Republic of Kazakhstan in Bulletin of the National Engineering Academy of the Republic of Kazakhstan
- Vol. 99 (1) , 85-86
Integration of LSTM neural networks and hydrological models for flood forecasting
This paper proposes a hybrid forecasting approach that integrates LSTM neural networks with classical hydrological models to improve the accuracy of early flood prediction. In the course of the study, the Saint-Venant equations were employed to describe the dynamics of open-channel flow, while the Boussinesq equations were used to model groundwater movement. The results of physical-based modeling, together with time-series data, were utilized to train the LSTM neural network. The performance of the proposed hybrid model was evaluated using the RMSE and MAE metrics and compared with the results of standalone classical and data-driven models. The results demonstrate that the hybrid approach provides higher accuracy and robustness in flood forecasting. This study can serve as a basis for early flood risk detection and its application in intelligent management systems.
- 30 March 2026
- Published by National Engineering Academy of the Republic of Kazakhstan in Bulletin of the National Engineering Academy of the Republic of Kazakhstan
- Vol. 99 (1) , 85-86
Integration of LSTM neural networks and hydrological models for flood forecasting
This paper proposes a hybrid forecasting approach that integrates LSTM neural networks with classical hydrological models to improve the accuracy of early flood prediction. In the course of the study, the Saint-Venant equations were employed to describe the dynamics of open-channel flow, while the Boussinesq equations were used to model groundwater movement. The results of physical-based modeling, together with time-series data, were utilized to train the LSTM neural network. The performance of the proposed hybrid model was evaluated using the RMSE and MAE metrics and compared with the results of standalone classical and data-driven models. The results demonstrate that the hybrid approach provides higher accuracy and robustness in flood forecasting. This study can serve as a basis for early flood risk detection and its application in intelligent management systems.
- 30 December 2025
- Published by National Engineering Academy of the Republic of Kazakhstan in Bulletin of the National Engineering Academy of the Republic of Kazakhstan
- Vol. 98 (4) , 9-20
Modeling metabolic networks in systems biology with artificial intelligence tools
This paper discusses the experience of applying artificial intelligence methods for modeling metabolic networks in systems biology. Based on data from the KEGG database, a network of interactions between key metabolites and enzymes of human metabolism was constructed and analyzed using the Python NetworkX library. Topological properties of the network nodes and edges – indices of degree, intermediate and centrality of proximity – were calculated, and the role of essential and non-essential metabolites in the network was quantified. Statistical analysis confirmed the biological validity of the model and showed that essential metabolites play the role of central nodes. This study provides effective methods for a comprehensive and systematic study of metabolic processes at the interface of artificial intelligence and systems biology and opens up the possibility of predicting dynamic changes in in metabolism using graph neural networks.
- 30 December 2025
- Published by National Engineering Academy of the Republic of Kazakhstan in Bulletin of the National Engineering Academy of the Republic of Kazakhstan
- Vol. 98 (4) , 9-20
Modeling metabolic networks in systems biology with artificial intelligence tools
This paper discusses the experience of applying artificial intelligence methods for modeling metabolic networks in systems biology. Based on data from the KEGG database, a network of interactions between key metabolites and enzymes of human metabolism was constructed and analyzed using the Python NetworkX library. Topological properties of the network nodes and edges – indices of degree, intermediate and centrality of proximity – were calculated, and the role of essential and non-essential metabolites in the network was quantified. Statistical analysis confirmed the biological validity of the model and showed that essential metabolites play the role of central nodes. This study provides effective methods for a comprehensive and systematic study of metabolic processes at the interface of artificial intelligence and systems biology and opens up the possibility of predicting dynamic changes in in metabolism using graph neural networks.
- 30 September 2025
- Published by National Engineering Academy of the Republic of Kazakhstan in Bulletin of the National Engineering Academy of the Republic of Kazakhstan
- Vol. 97 (3) , 203-211
Modeling and calculation of moisture distribution taking into account soil characteristic
In this article, a mathematical model and a method for calculating moisture distribution have been developed, taking into account soil characteristics. When modeling the flow of water in a porous medium, the latter is divided into two zones: the zone of full and incomplete saturation. The incomplete saturation zone is an aeration zone, i.e. a moisture content zone. The motion of moisture is described by a nonlinear partial differential equation, the exact solution of which cannot be written explicitly. The most effective way to solve moisture transfer problems is numerical methods based on the use of a computer mathematics system. Finite difference methods are widely used in the practice of numerical calculations of moisture transfer. The paper considers the change in humidity over time and in soil depth. The problem is solved with the appropriate initial and boundary conditions for the moisture transfer equation. At the initial moment of time, we consider the humidity in a porous medium to be known. On the surface of the soil, we consider Darcy’s generalized law to be true. At the end of the depth of moisture spreading downwards, the initial humidity is maintained. In this case, a two-layer six-point implicit scheme of the finite difference method and the Mathcad computer mathematics system are used.
- 30 September 2025
- Published by National Engineering Academy of the Republic of Kazakhstan in Bulletin of the National Engineering Academy of the Republic of Kazakhstan
- Vol. 97 (3) , 203-211
Modeling and calculation of moisture distribution taking into account soil characteristic
In this article, a mathematical model and a method for calculating moisture distribution have been developed, taking into account soil characteristics. When modeling the flow of water in a porous medium, the latter is divided into two zones: the zone of full and incomplete saturation. The incomplete saturation zone is an aeration zone, i.e. a moisture content zone. The motion of moisture is described by a nonlinear partial differential equation, the exact solution of which cannot be written explicitly. The most effective way to solve moisture transfer problems is numerical methods based on the use of a computer mathematics system. Finite difference methods are widely used in the practice of numerical calculations of moisture transfer. The paper considers the change in humidity over time and in soil depth. The problem is solved with the appropriate initial and boundary conditions for the moisture transfer equation. At the initial moment of time, we consider the humidity in a porous medium to be known. On the surface of the soil, we consider Darcy’s generalized law to be true. At the end of the depth of moisture spreading downwards, the initial humidity is maintained. In this case, a two-layer six-point implicit scheme of the finite difference method and the Mathcad computer mathematics system are used.
- 15 December 2020
- Published by National Academy of Sciences of the Republic of Kazakshtan in Series of Geology and Technical Sciences
- Vol. 6 (444) , 203-212
CALCULATION OF TWO-SPEED FLOW OF TWO-PHASE OPEN FLOW
In this article the mathematical model of unsteady flow the two-phase open stream taking into account the redistribution of the particulate concentration, the depth of flow and water filtration on the bottom of the channel, and also created an efficient method of calculation. In this case, the two-speed flow is considered, i.e. the presence of the longitudinal and vertical components of the phase velocities is taken into account, and we also believe that the flow parameters along the flow do not change. Initial and boundary conditions are established based on theoretical and empirical formulas, which are widely used in practice. The flow in open channels is non-pressurized, occurs under the influence of gravity and is characterized by the fact that the flow has a free surface. At the initial moment of time, we consider the flow to be uniform in the longitudinal direction and all parameters are set by known theoretical and empirical formulas. At the bottom of the channel for longitudinal velocity component of the water use condition of adhesion, and for the longitudinal velocity component of solid phase condition for the shift and believe the known concentrations of solid particles, and vertical components of velocity the phases of the filtering conditions (for water), and hydraulic size (for solid particles). On the free surface, we consider that there are no solid particles, and for the longitudinal components of the phase velocities we neglect the force of air friction, and for the vertical components of the phase velocities we use the condition of non-uniformity of the free surface in time. On the basis of the developed mathematical model and the created method of calculation, the changes of the main parameters in the depth of the flow and in time are determined.
- 15 December 2020
- Published by National Academy of Sciences of the Republic of Kazakshtan in Series of Geology and Technical Sciences
- Vol. 6 (444) , 203-212
CALCULATION OF TWO-SPEED FLOW OF TWO-PHASE OPEN FLOW
In this article the mathematical model of unsteady flow the two-phase open stream taking into account the redistribution of the particulate concentration, the depth of flow and water filtration on the bottom of the channel, and also created an efficient method of calculation. In this case, the two-speed flow is considered, i.e. the presence of the longitudinal and vertical components of the phase velocities is taken into account, and we also believe that the flow parameters along the flow do not change. Initial and boundary conditions are established based on theoretical and empirical formulas, which are widely used in practice. The flow in open channels is non-pressurized, occurs under the influence of gravity and is characterized by the fact that the flow has a free surface. At the initial moment of time, we consider the flow to be uniform in the longitudinal direction and all parameters are set by known theoretical and empirical formulas. At the bottom of the channel for longitudinal velocity component of the water use condition of adhesion, and for the longitudinal velocity component of solid phase condition for the shift and believe the known concentrations of solid particles, and vertical components of velocity the phases of the filtering conditions (for water), and hydraulic size (for solid particles). On the free surface, we consider that there are no solid particles, and for the longitudinal components of the phase velocities we neglect the force of air friction, and for the vertical components of the phase velocities we use the condition of non-uniformity of the free surface in time. On the basis of the developed mathematical model and the created method of calculation, the changes of the main parameters in the depth of the flow and in time are determined.
- 15 December 2019
- Published by National Academy of Sciences of the Republic of Kazakshtan in Series of Geology and Technical Sciences
- Vol. 6 (438) , 246-255
CALCULATIONS OF EXCESS LOAD ON THE NETWORK
- 15 December 2019
- Published by National Academy of Sciences of the Republic of Kazakshtan in Series of Geology and Technical Sciences
- Vol. 6 (438) , 246-255
CALCULATIONS OF EXCESS LOAD ON THE NETWORK