Siti Nurulain Mohd Rum
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
Research Subjects
Siti Nurulain Mohd Rum has a total of 40 co-authors. Here are the top co-authors by number of shared publications:
- Mudathir Ahmed Mohamud
Department of Computer Science, Faculty of Computing, SIMAD University, Mogadishu, Somalia
6 shared publications - Fatimah Sidi
Universiti Putra Malaysia
5 shared publications - Hamidah Ibrahim
Department of Computer Science, Faculty of Computer Science and Information Technology, Universiti Putra Malaysia (UPM), Serdang, Selangor, Malaysia
5 shared publications - Zhang Xiaowei
Department of Computer Science, Faculty of Computer Science and Information Technology, Universiti Putra Malaysia, Serdang, Selangor, Malaysia
3 shared publications - Auzi Asfarian
Software Engineering and Information Science Division, School of Data Science, Mathematics, and Informatics, IPB University, Bogor, Indonesia
2 shared publications - Raihani Mohamed
Fakulti Sains Komputer dan Teknologi Maklumat, Universiti Putra Malaysia, Malaysia
2 shared publications
Latest Publications
- 9 September 2026
- Published by Universiti Putra Malaysia in Journal of Intelligent Media Computing
- Vol. 1  (1) , 135
An Efficient Framework for Processing Range Skyline Queries over Incomplete Database
Skyline queries retrieve objects that are not dominated by any other object in the database, playing a key role in preference analysis and multi-criteria decision-making system. As an important variant, range skyline queries retrieve skyline results within a database defined by the range value across one or more dimensions. The presence of incomplete data is inevitable in large-scale, multidimensional databases, introducing new challenges in processing range skyline queries. To address this challenge, an efficient framework named Optimized Incomplete Range Skyline (OIRS) is proposed to optimize range skyline query execution by incorporating a range pruning strategy and a bucket metadata reuse technique. The OIRS framework takes precomputed buckets and their metadata as input and consists of three phases: pruning objects with range query, computing bucket skylines, and deriving final skyline results. First, a range pruning algorithm prunes irrelevant objects based on bucket metadata. Next, a bucket skyline computation algorithm efficiently computes bucket skylines for the remaining objects by leveraging a bucket skyline reuse technique. Finally, the final skyline results are derived from the candidate objects across the bucket skylines. Extensive experiments conducted on both synthetic and real-world NBA datasets demonstrate that OIRS significantly outperforms the naïve scan-and-compute approach across three key metrics: the number of object scans in the range pruning phase, the number of pairwise comparisons in computing bucket skylines, and the number of pairwise comparisons for deriving final incomplete skylines.
- 9 September 2026
- Published by Universiti Putra Malaysia in Journal of Intelligent Media Computing
- Vol. 1  (1) , 135
An Efficient Framework for Processing Range Skyline Queries over Incomplete Database
Skyline queries retrieve objects that are not dominated by any other object in the database, playing a key role in preference analysis and multi-criteria decision-making system. As an important variant, range skyline queries retrieve skyline results within a database defined by the range value across one or more dimensions. The presence of incomplete data is inevitable in large-scale, multidimensional databases, introducing new challenges in processing range skyline queries. To address this challenge, an efficient framework named Optimized Incomplete Range Skyline (OIRS) is proposed to optimize range skyline query execution by incorporating a range pruning strategy and a bucket metadata reuse technique. The OIRS framework takes precomputed buckets and their metadata as input and consists of three phases: pruning objects with range query, computing bucket skylines, and deriving final skyline results. First, a range pruning algorithm prunes irrelevant objects based on bucket metadata. Next, a bucket skyline computation algorithm efficiently computes bucket skylines for the remaining objects by leveraging a bucket skyline reuse technique. Finally, the final skyline results are derived from the candidate objects across the bucket skylines. Extensive experiments conducted on both synthetic and real-world NBA datasets demonstrate that OIRS significantly outperforms the naïve scan-and-compute approach across three key metrics: the number of object scans in the range pruning phase, the number of pairwise comparisons in computing bucket skylines, and the number of pairwise comparisons for deriving final incomplete skylines.
- Open Access
- 5 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 4675-4687
TB-FusionNet: A Multi-Scale Feature Fusion Algorithm With Spatial and Channel Cross-Attention for Tuberculosis Detection
The lesions of tuberculosis (TB) in X-ray images are highly complex, exhibiting a variety of sizes, shapes, and structural variations. Single-scale features are insufficient to fully represent this diversity and complexity, thereby limiting the effectiveness of TB detection. As a result, multi-scale feature fusion has become a widely explored approach in the field of TB detection. However, current multi-scale feature fusion methods still have several limitations. First, the weight allocation in existing methods typically remains at the feature level without extending to local features and channels. This limitation prevents the model from precisely controlling the significance of local features and channels, resulting in coarse feature representations. Second, current methods neglect the contextual information between features at different levels, which further undermines the consistency of the fused features and leads to a lack of semantic coherence. To address the aforementioned issues, this study proposes TB-FusionNet, a multi-scale feature fusion algorithm based on channel and spatial cross-attention mechanisms, for tuberculosis classification. The algorithm first calculates the similarity between local features at different levels to precisely select low-level detail features that are closely related to high-level semantic features, thereby generating a more hierarchical feature representation. Next, the algorithm computes the dependencies between feature channels at different levels, allowing low-level channel features to be appropriately enhanced or suppressed under the guidance of high-level channels, effectively improving the semantic consistency between cross-level features. Through these operations, the model can dynamically adjust the weight distribution of local features and channels according to task requirements, thereby more flexibly adapting to the challenges of complex tasks. Experiments were conducted on the Shenzhen, Montgomery and HSAAS datasets in this study. The results demonstrate that the proposed method outperforms current state-of-the-art approaches, validating its effectiveness and robustness.
- 5 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 4675-4687
Open AccessTB-FusionNet: A Multi-Scale Feature Fusion Algorithm With Spatial and Channel Cross-Attention for Tuberculosis Detection
The lesions of tuberculosis (TB) in X-ray images are highly complex, exhibiting a variety of sizes, shapes, and structural variations. Single-scale features are insufficient to fully represent this diversity and complexity, thereby limiting the effectiveness of TB detection. As a result, multi-scale feature fusion has become a widely explored approach in the field of TB detection. However, current multi-scale feature fusion methods still have several limitations. First, the weight allocation in existing methods typically remains at the feature level without extending to local features and channels. This limitation prevents the model from precisely controlling the significance of local features and channels, resulting in coarse feature representations. Second, current methods neglect the contextual information between features at different levels, which further undermines the consistency of the fused features and leads to a lack of semantic coherence. To address the aforementioned issues, this study proposes TB-FusionNet, a multi-scale feature fusion algorithm based on channel and spatial cross-attention mechanisms, for tuberculosis classification. The algorithm first calculates the similarity between local features at different levels to precisely select low-level detail features that are closely related to high-level semantic features, thereby generating a more hierarchical feature representation. Next, the algorithm computes the dependencies between feature channels at different levels, allowing low-level channel features to be appropriately enhanced or suppressed under the guidance of high-level channels, effectively improving the semantic consistency between cross-level features. Through these operations, the model can dynamically adjust the weight distribution of local features and channels according to task requirements, thereby more flexibly adapting to the challenges of complex tasks. Experiments were conducted on the Shenzhen, Montgomery and HSAAS datasets in this study. The results demonstrate that the proposed method outperforms current state-of-the-art approaches, validating its effectiveness and robustness.
- Open Access
- 5 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 5054-5081
An Efficient Skyline Model for Processing Multiple Continuous Skyline Queries Over Uncertain Data Stream
Skyline queries based on pareto dominance have gained significant attention for their ability to effectively identify interesting objects from large multi-dimensional datasets. They are particularly useful in applications that involve multi-criteria decision support. Recently, several techniques have been proposed for processing continuous skyline queries over uncertain data streams, however, they mainly focus on uncertainty that results from objects having many instances. Uncertainty owing to objects having range values wherein the exact values of the objects are not known at the point of processing has not been extensively explored. Moreover, identical objects may recur in the stream at different times while continuous skyline queries submitted by different users can overlap within the same time frame. Thus, it would be inefficient to repeatedly process the same segment of the data stream and the same objects in processing these multiple queries. We seek to overcome the aforementioned problems by introducing an efficient skyline model named MCSQ-UDS, for computing multiple continuous skyline queries over uncertain data stream. Several extensive experiments were conducted on synthetic and real datasets with different parameter settings. The results show that the proposed model outperforms the baseline techniques with regard to number of pairwise comparisons and execution time.
- 5 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 5054-5081
Open AccessAn Efficient Skyline Model for Processing Multiple Continuous Skyline Queries Over Uncertain Data Stream
Skyline queries based on pareto dominance have gained significant attention for their ability to effectively identify interesting objects from large multi-dimensional datasets. They are particularly useful in applications that involve multi-criteria decision support. Recently, several techniques have been proposed for processing continuous skyline queries over uncertain data streams, however, they mainly focus on uncertainty that results from objects having many instances. Uncertainty owing to objects having range values wherein the exact values of the objects are not known at the point of processing has not been extensively explored. Moreover, identical objects may recur in the stream at different times while continuous skyline queries submitted by different users can overlap within the same time frame. Thus, it would be inefficient to repeatedly process the same segment of the data stream and the same objects in processing these multiple queries. We seek to overcome the aforementioned problems by introducing an efficient skyline model named MCSQ-UDS, for computing multiple continuous skyline queries over uncertain data stream. Several extensive experiments were conducted on synthetic and real datasets with different parameter settings. The results show that the proposed model outperforms the baseline techniques with regard to number of pairwise comparisons and execution time.
- Open Access
- 24 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 216802-216831
Finding Informative Skyline Results Over Incomplete Data With Threshold-Based Bucket Skyline Algorithm
Skyline queries which return objects that are not dominated by any other objects, have been integrated into various real-world database applications. However, the presence of incomplete data in databases, may lead into cyclic dominance relation that none of the objects are considered as skylines, while the transitive dominance property may no longer hold. Although, these issues have received great attention, most of the existing skyline algorithms typically ignore the quality of the returned skyline results, thereby limiting the insights they offer. In this paper, a Threshold-based Bucket Skyline Algorithm (TBSA) is proposed with the aim to derive informative skyline results, i.e. results that are not too few and with low missing rate. TBSA organises the objects in the database into a two-layer structure, namely: bucket and cluster, to ensure only dominant objects are retained for further skyline analysis. Meanwhile, to minimise the tuning iterations, a threshold prediction model is constructed that identifies in advanced the subspaces to be excluded from the skyline computation. Extensive experiments have been carried out on the synthetic and real datasets and the results show that TBSA outperforms the previous approach with respect to the number of iterations, pairwise comparisons, skyline results, informative skyline results, and processing time.
- 24 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 216802-216831
Open AccessFinding Informative Skyline Results Over Incomplete Data With Threshold-Based Bucket Skyline Algorithm
Skyline queries which return objects that are not dominated by any other objects, have been integrated into various real-world database applications. However, the presence of incomplete data in databases, may lead into cyclic dominance relation that none of the objects are considered as skylines, while the transitive dominance property may no longer hold. Although, these issues have received great attention, most of the existing skyline algorithms typically ignore the quality of the returned skyline results, thereby limiting the insights they offer. In this paper, a Threshold-based Bucket Skyline Algorithm (TBSA) is proposed with the aim to derive informative skyline results, i.e. results that are not too few and with low missing rate. TBSA organises the objects in the database into a two-layer structure, namely: bucket and cluster, to ensure only dominant objects are retained for further skyline analysis. Meanwhile, to minimise the tuning iterations, a threshold prediction model is constructed that identifies in advanced the subspaces to be excluded from the skyline computation. Extensive experiments have been carried out on the synthetic and real datasets and the results show that TBSA outperforms the previous approach with respect to the number of iterations, pairwise comparisons, skyline results, informative skyline results, and processing time.
- 29 July 2025
- Published by Springer Nature
- p. 116-127
Skyline Query Processing Over Data Stream with Dominance Analyses Reduction and Hashing Technique
- 29 July 2025
- Published by Springer Nature
- p. 116-127
Skyline Query Processing Over Data Stream with Dominance Analyses Reduction and Hashing Technique
- Open Access
- 6 May 2025
- Published by Akademia Baru Publishing in Journal of Advanced Research Design
- Vol. 129  (1) , 46-59
Uncovering Depression on Social Media using BERT Model
The lack of early detection and effective treatment programs for depression has left millions struggling with mental illnesses, including anxiety, sleep disorders and, in severe cases, self-harm or suicide. Adolescents and young adults, who are among the most active users of social media platforms like Twitter, represent a particularly vulnerable demographic. With Twitter often serving as a digital diary where individuals share thoughts and emotions, its data offers a unique opportunity for mental health research. This study explores the potential of leveraging Twitter data to detect depression and identify at-risk individuals. Using advanced machine learning techniques, including Natural Language Processing (NLP) with the BERT (Bidirectional Encoder Representations from Transformers) model, this research builds a predictive system that analyses linguistic cues, sentiment and user interactions. The methodology involves integrating and preprocessing data from multiple sources, ensuring a comprehensive and reliable approach to model development. The findings demonstrate the BERT model’s superior accuracy, outperforming traditional machine learning methods like Logistic Regression and Support Vector Machines. This study underscores the effectiveness of social media as a tool for early detection and intervention, offering an innovative approach to mitigating the global burden of mental health disorders.
- 6 May 2025
- Published by Akademia Baru Publishing in Journal of Advanced Research Design
- Vol. 129  (1) , 46-59
Open AccessUncovering Depression on Social Media using BERT Model
The lack of early detection and effective treatment programs for depression has left millions struggling with mental illnesses, including anxiety, sleep disorders and, in severe cases, self-harm or suicide. Adolescents and young adults, who are among the most active users of social media platforms like Twitter, represent a particularly vulnerable demographic. With Twitter often serving as a digital diary where individuals share thoughts and emotions, its data offers a unique opportunity for mental health research. This study explores the potential of leveraging Twitter data to detect depression and identify at-risk individuals. Using advanced machine learning techniques, including Natural Language Processing (NLP) with the BERT (Bidirectional Encoder Representations from Transformers) model, this research builds a predictive system that analyses linguistic cues, sentiment and user interactions. The methodology involves integrating and preprocessing data from multiple sources, ensuring a comprehensive and reliable approach to model development. The findings demonstrate the BERT model’s superior accuracy, outperforming traditional machine learning methods like Logistic Regression and Support Vector Machines. This study underscores the effectiveness of social media as a tool for early detection and intervention, offering an innovative approach to mitigating the global burden of mental health disorders.
- Open Access
- 6 November 2024
- Published by Association for Computing Machinery (ACM)
- p. 29-35
Dominance Analyses Reduction in Skyline Query Processing over Data Stream with Data Mining Technique
The database community has observed in the past two decades, the growth of research interest in skyline queries, which aim to report to users interesting objects—commonly known as skylines—based on their preferences. The identification of skyline objects becomes more challenging when skylines are to be identified from a collection of continuously generated input data streams. In this paper, we proposed the Dominance Analyses Reduction (DAR) framework, which aims at addressing the issues of redundant dominance analyses that arise while determining skylines over data stream. Dominance analyses are repeated for objects that are in the overlapped frames of two windows and for pairs of objects that later reappear in the stream. DAR employs the Apriori algorithm, one of the most prevalent data mining algorithms, to identify the frequently occurring dominance analyses. Instead of conducting the dominance analyses again, their results are stored and utilised in the subsequent derivation of skylines. The DAR framework has been validated through several experiments. Its results exhibit significant reduction in the number of pairwise comparisons at both object and dimension levels and execution time.
- 6 November 2024
- Published by Association for Computing Machinery (ACM)
- p. 29-35
Open AccessDominance Analyses Reduction in Skyline Query Processing over Data Stream with Data Mining Technique
The database community has observed in the past two decades, the growth of research interest in skyline queries, which aim to report to users interesting objects—commonly known as skylines—based on their preferences. The identification of skyline objects becomes more challenging when skylines are to be identified from a collection of continuously generated input data streams. In this paper, we proposed the Dominance Analyses Reduction (DAR) framework, which aims at addressing the issues of redundant dominance analyses that arise while determining skylines over data stream. Dominance analyses are repeated for objects that are in the overlapped frames of two windows and for pairs of objects that later reappear in the stream. DAR employs the Apriori algorithm, one of the most prevalent data mining algorithms, to identify the frequently occurring dominance analyses. Instead of conducting the dominance analyses again, their results are stored and utilised in the subsequent derivation of skylines. The DAR framework has been validated through several experiments. Its results exhibit significant reduction in the number of pairwise comparisons at both object and dimension levels and execution time.
- Open Access
- 7 October 2024
- Published by Akademia Baru Publishing in Journal of Advanced Research in Applied Sciences and Engineering Technology
- p. 168-177
Public Transport Delay Prediction using Deep Learning
A system of trains, buses, subways, ferries, and other vehicles that are accessible to the public is referred to as public transportation, also known as public transit or mass transit. These networks are usually run by private or public organisations are made to efficiently transport large numbers of people in urban and suburban areas. It usually operates on set schedules and routes and has a listed fare for each journey. Long wait periods and lengthy travel times resulting from delays are two problems that frequently plague public transportation services. There are several things that can cause a delay in public transport, such as an excess of passengers, heavy traffic, accidents, and other unforeseen circumstances. The availability of a more accurate delay prediction for public transportation might increase users’ confidence and their willingness to pay more for transit services. Over the past two decades, a number of studies on prediction algorithms for transportation data have been proposed. Most of the work is on machine learning model development, focusing on delay prediction and taking into account several factors such as weather conditions and infrastructure issues. This paper proposes a deep learning model to predict public transportation delays using data from public transportation and the weather. The results obtained from this research work are compared with several other existing works. Our experiment has demonstrated that the deep neural network (DNN) is the best model to predict transit delay compared to several other machine learning and deep learning models.
- 7 October 2024
- Published by Akademia Baru Publishing in Journal of Advanced Research in Applied Sciences and Engineering Technology
- p. 168-177
Open AccessPublic Transport Delay Prediction using Deep Learning
A system of trains, buses, subways, ferries, and other vehicles that are accessible to the public is referred to as public transportation, also known as public transit or mass transit. These networks are usually run by private or public organisations are made to efficiently transport large numbers of people in urban and suburban areas. It usually operates on set schedules and routes and has a listed fare for each journey. Long wait periods and lengthy travel times resulting from delays are two problems that frequently plague public transportation services. There are several things that can cause a delay in public transport, such as an excess of passengers, heavy traffic, accidents, and other unforeseen circumstances. The availability of a more accurate delay prediction for public transportation might increase users’ confidence and their willingness to pay more for transit services. Over the past two decades, a number of studies on prediction algorithms for transportation data have been proposed. Most of the work is on machine learning model development, focusing on delay prediction and taking into account several factors such as weather conditions and infrastructure issues. This paper proposes a deep learning model to predict public transportation delays using data from public transportation and the weather. The results obtained from this research work are compared with several other existing works. Our experiment has demonstrated that the deep neural network (DNN) is the best model to predict transit delay compared to several other machine learning and deep learning models.
- Open Access
- 28 August 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 12, 120877-120898
A Performance Analysis of Prediction Techniques in Handling High-Dimensional Uncertain Data for the Application of Skyline Query Over Data Stream
The proliferation of high-dimensional data in many advanced database applications is a result of today’s technological advancements. These data points that correspond to objects are often without a precise description, which make their representation uncertain. While the concept of data streaming is not new, its practical uses are only recently emerging. This research focuses on continuous range data—a type of uncertain data common in database applications—that do not have explicit representations of their exact values. Furthermore, the identification of skyline objects—one of the popular database applications—becomes more challenging when skylines are to be identified from a collection of continuously generated input data streams where objects might have imprecise values. This makes it imperative to determine which approach has the optimal accuracy for estimating or predicting the uncertain values and at the same time able to handle a massive streams of data that are continuously generated and analyze them almost instantly to provide accurate and timely responses. Given this, the following techniques are selected— Linear Regression (LR) , k-Nearest Neighbour (k-NN) , Random Forest (RF) , Decision Trees (DT) , and Centre and Range Method (CRM) and their effectiveness is evaluated in terms of execution time, precision, recall, F1-score, and root mean square error (RMSE). Additionally, in order to verify the accuracy of each prediction technique, the predicted data derived from its model is used to derive skyline objects, which are subsequently compared to the actual skyline results. An inaccurate prediction of a continuous range value would result in incorrect set of skyline objects.
- 28 August 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 12, 120877-120898
Open AccessA Performance Analysis of Prediction Techniques in Handling High-Dimensional Uncertain Data for the Application of Skyline Query Over Data Stream
The proliferation of high-dimensional data in many advanced database applications is a result of today’s technological advancements. These data points that correspond to objects are often without a precise description, which make their representation uncertain. While the concept of data streaming is not new, its practical uses are only recently emerging. This research focuses on continuous range data—a type of uncertain data common in database applications—that do not have explicit representations of their exact values. Furthermore, the identification of skyline objects—one of the popular database applications—becomes more challenging when skylines are to be identified from a collection of continuously generated input data streams where objects might have imprecise values. This makes it imperative to determine which approach has the optimal accuracy for estimating or predicting the uncertain values and at the same time able to handle a massive streams of data that are continuously generated and analyze them almost instantly to provide accurate and timely responses. Given this, the following techniques are selected— Linear Regression (LR) , k-Nearest Neighbour (k-NN) , Random Forest (RF) , Decision Trees (DT) , and Centre and Range Method (CRM) and their effectiveness is evaluated in terms of execution time, precision, recall, F1-score, and root mean square error (RMSE). Additionally, in order to verify the accuracy of each prediction technique, the predicted data derived from its model is used to derive skyline objects, which are subsequently compared to the actual skyline results. An inaccurate prediction of a continuous range value would result in incorrect set of skyline objects.
- Open Access
- 5 March 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 12, 37540-37556
Road Crash Injury Severity Prediction Using a Graph Neural Network Framework
Crash severity prediction is a challenging research area, where the objective is to accurately assess the extent of severity of an injury resulting from road traffic accidents. The main aim of existing studies is to precisely assess the potential severity of crashes under diverse circumstances, such as weather conditions, vehicle attributes, road characteristics and layout, and traffic control factors. This effort aids authorities in establishing effective emergency response systems. The novelty and objective of our work involve contributing to this research area by employing a graph architecture to capture relationships among various crash records to uncover any hidden patterns that traditional ML models might overlook. The current study extends existing knowledge by leveraging Graph Neural Networks (GNN) and comparing their performance to popular ensemble-based models, which include Extreme Gradient Boosting (XGBoost) and Random Forest (RF) and Artificial Neural Networks (ANNs). Real data from the United Kingdom (UK) was employed to achieve our goal. All models underwent training using the training dataset, followed by performance evaluation using diverse metrics such as the accuracy, precision, recall, f1-score, Matthews Correlation Coefficient (MCC), confusion matrix, and computational cost on the test dataset. Overall, our proposed GNN-based model demonstrated better performance when compared to other models. Specifically, the GNN model outperformed all other models across all metrics. For instance, the accuracy stood at 85.55% for GNN as compared to 83.36%, 83.18%, and 83.27% for XGBoost, RF, and ANNs models, respectively. The GNN model assisted in identifying hidden patterns by considering non-linear relationships among crash records. Thus, the model had the potential to improve its ability to predict severe accidents, which could in turn significantly improve emergency response efforts and reduce the likelihood of severe accidents resulting in fatalities.
- 5 March 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 12, 37540-37556
Open AccessRoad Crash Injury Severity Prediction Using a Graph Neural Network Framework
Crash severity prediction is a challenging research area, where the objective is to accurately assess the extent of severity of an injury resulting from road traffic accidents. The main aim of existing studies is to precisely assess the potential severity of crashes under diverse circumstances, such as weather conditions, vehicle attributes, road characteristics and layout, and traffic control factors. This effort aids authorities in establishing effective emergency response systems. The novelty and objective of our work involve contributing to this research area by employing a graph architecture to capture relationships among various crash records to uncover any hidden patterns that traditional ML models might overlook. The current study extends existing knowledge by leveraging Graph Neural Networks (GNN) and comparing their performance to popular ensemble-based models, which include Extreme Gradient Boosting (XGBoost) and Random Forest (RF) and Artificial Neural Networks (ANNs). Real data from the United Kingdom (UK) was employed to achieve our goal. All models underwent training using the training dataset, followed by performance evaluation using diverse metrics such as the accuracy, precision, recall, f1-score, Matthews Correlation Coefficient (MCC), confusion matrix, and computational cost on the test dataset. Overall, our proposed GNN-based model demonstrated better performance when compared to other models. Specifically, the GNN model outperformed all other models across all metrics. For instance, the accuracy stood at 85.55% for GNN as compared to 83.36%, 83.18%, and 83.27% for XGBoost, RF, and ANNs models, respectively. The GNN model assisted in identifying hidden patterns by considering non-linear relationships among crash records. Thus, the model had the potential to improve its ability to predict severe accidents, which could in turn significantly improve emergency response efforts and reduce the likelihood of severe accidents resulting in fatalities.