Saied Alshahrani
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Saied Alshahrani has a total of 18 co-authors. Here are the top co-authors by number of shared publications:
- Shah Khalid
School of Electrical Engineering and Computer Science, National University of Sciences and Technology (NUST), Islamabad, Pakistan
3 shared publications Hafiz Syed Muhammad Bilal
School of Electrical Engineering and Computer Science, National University of Sciences and Technology (NUST), Islamabad, Pakistan
2 shared publicationsAmani K. Samha
1 shared publicationAreeba Tanveer
School of Electrical Engineering and Computer Science, National University of Sciences and Technology (NUST), Islamabad, Pakistan
1 shared publicationFahd Alzahrane
1 shared publication- Haifa Fahad Alhasson
Department of Information Technology, College of Computer, Qassim University, Buraydah 51452, Saudi Arabia
1 shared publication
Latest Publications
- Open Access
- 21 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 18491-18508
Toward Early Alzheimer’s Diagnosis: A Fusion of Cognitive and Genetic Data via Ensemble Learning
Early detection of Alzheimer’s disease (AD) remains a critical challenge in neurology, with significant implications for patient care and therapeutic efficacy. Although gold-standard diagnostic techniques, such as neuroimaging and cerebrospinal fluid biomarkers, offer high accuracy, their invasiveness and resource intensity limit their routine clinical application. Current non-invasive methods are more accessible but often lack sufficient predictive power for reliable early-stage diagnosis. To address this gap, we propose an integrative diagnostic framework that combines cognitive assessments and genetic risk profiling using an ensemble machine learning approach. Our method fuses Mini-Mental State Examination (MMSE) scores, Clinical Dementia Rating (CDR) metrics, APOE ε4 genotype information, and patient age within a three-model ensemble: a temporal progression tracker, a patient similarity mapper, and a structured data interpreter. Evaluated on a dataset of over 2,500 participants from the Alzheimer’s Disease Neuroimaging Initiative (ADNI), the ensemble achieved an accuracy of 0.897, a sensitivity of 0.897, and a specificity of 0.948, substantially outperforming conventional models. The inclusion of genetic data notably enhanced diagnostic precision, affirming its value in early risk stratification. This accessible and non-invasive solution offers a scalable tool for early AD detection with the potential to significantly improve clinical outcomes through earlier intervention.
- 21 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 18491-18508
Open AccessToward Early Alzheimer’s Diagnosis: A Fusion of Cognitive and Genetic Data via Ensemble Learning
Early detection of Alzheimer’s disease (AD) remains a critical challenge in neurology, with significant implications for patient care and therapeutic efficacy. Although gold-standard diagnostic techniques, such as neuroimaging and cerebrospinal fluid biomarkers, offer high accuracy, their invasiveness and resource intensity limit their routine clinical application. Current non-invasive methods are more accessible but often lack sufficient predictive power for reliable early-stage diagnosis. To address this gap, we propose an integrative diagnostic framework that combines cognitive assessments and genetic risk profiling using an ensemble machine learning approach. Our method fuses Mini-Mental State Examination (MMSE) scores, Clinical Dementia Rating (CDR) metrics, APOE ε4 genotype information, and patient age within a three-model ensemble: a temporal progression tracker, a patient similarity mapper, and a structured data interpreter. Evaluated on a dataset of over 2,500 participants from the Alzheimer’s Disease Neuroimaging Initiative (ADNI), the ensemble achieved an accuracy of 0.897, a sensitivity of 0.897, and a specificity of 0.948, substantially outperforming conventional models. The inclusion of genetic data notably enhanced diagnostic precision, affirming its value in early risk stratification. This accessible and non-invasive solution offers a scalable tool for early AD detection with the potential to significantly improve clinical outcomes through earlier intervention.
- Open Access
- 12 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 10991-11018
MCQs Generation With Large Language Models: A Survey of Methodologies, Evolution, and Open Research Issues
In recent years, there has been growing interest in the automatic generation of Multiple Choice Questions (MCQs), mainly to reduce the effort and time involved in manual question construction. Many researchers and practitioners have tried different methods to produce accurate and good-quality MCQs automatically. The progress in this area has been steady but also diverse, which makes it difficult to identify which approaches work best in practice. The research trend has moved gradually from early rule-based methods to more advanced systems based on Large Language Models (LLMs). Although several review papers and surveys have summarized earlier work, most of them give limited discussion on how the field has evolved over time or what key problems remain to be solved. This paper aims to fill that gap by reviewing important studies on MCQs generation and providing a structured overview of the main question types, their essential components, and the processes involved in automated generation. It also discusses the essential stages in a general system architecture, including preprocessing, key selection, and distractor generation. In addition, the review summarizes the most frequently used datasets and evaluation measures for assessing question quality and highlights important applications and research challenges. In doing so, it outlines possible future directions to advance the use of automated MCQs generation in education, training, and professional assessment.
- 12 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 10991-11018
Open AccessMCQs Generation With Large Language Models: A Survey of Methodologies, Evolution, and Open Research Issues
In recent years, there has been growing interest in the automatic generation of Multiple Choice Questions (MCQs), mainly to reduce the effort and time involved in manual question construction. Many researchers and practitioners have tried different methods to produce accurate and good-quality MCQs automatically. The progress in this area has been steady but also diverse, which makes it difficult to identify which approaches work best in practice. The research trend has moved gradually from early rule-based methods to more advanced systems based on Large Language Models (LLMs). Although several review papers and surveys have summarized earlier work, most of them give limited discussion on how the field has evolved over time or what key problems remain to be solved. This paper aims to fill that gap by reviewing important studies on MCQs generation and providing a structured overview of the main question types, their essential components, and the processes involved in automated generation. It also discusses the essential stages in a general system architecture, including preprocessing, key selection, and distractor generation. In addition, the review summarizes the most frequently used datasets and evaluation measures for assessing question quality and highlights important applications and research challenges. In doing so, it outlines possible future directions to advance the use of automated MCQs generation in education, training, and professional assessment.
- 27 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Biomedical and Health Informatics
- Vol. PP (21682194) , 1-12
ContextVecNet: A Context-Driven Multimodal Learning Framework for Depression Detection
Depression is a major global health issue, and social media offers a rich source of data for early detection. However, existing multimodal approaches often fail to effectively capture the contextual relationships between text and images and tend to ignore temporal information, thereby limiting their predictive accuracy and reliability. To address these limitations, we propose ContextVecNet, a multimodal deep learning framework. It is built on a CLIP-based architecture enhanced by learnable context vectors, which are integrated into the text and image encoding branches. This design allows the model to dynamically learn and adapt to the specific visual and linguistic markers of depression present in social media data. The architecture further incorporates a cross-modal transformer with time-aware embeddings, which jointly models temporal dynamics and cross-modal interactions across user posts. Evaluated on a publicly available multimodal Twitter dataset, ContextVecNet achieves state-of-the-art performance, with an AUC of 0.9922 and an F1-score of 0.9619, significantly outperforming existing methods. An ablation study confirms that freezing the context vectors results in notable performance degradation, emphasizing their importance in depression detection.
- 27 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Biomedical and Health Informatics
- Vol. PP (21682194) , 1-12
ContextVecNet: A Context-Driven Multimodal Learning Framework for Depression Detection
Depression is a major global health issue, and social media offers a rich source of data for early detection. However, existing multimodal approaches often fail to effectively capture the contextual relationships between text and images and tend to ignore temporal information, thereby limiting their predictive accuracy and reliability. To address these limitations, we propose ContextVecNet, a multimodal deep learning framework. It is built on a CLIP-based architecture enhanced by learnable context vectors, which are integrated into the text and image encoding branches. This design allows the model to dynamically learn and adapt to the specific visual and linguistic markers of depression present in social media data. The architecture further incorporates a cross-modal transformer with time-aware embeddings, which jointly models temporal dynamics and cross-modal interactions across user posts. Evaluated on a publicly available multimodal Twitter dataset, ContextVecNet achieves state-of-the-art performance, with an AUC of 0.9922 and an F1-score of 0.9619, significantly outperforming existing methods. An ablation study confirms that freezing the context vectors results in notable performance degradation, emphasizing their importance in depression detection.