Hala Alshamlan

Background: Automated brain-MRI analysis is usually judged by accuracy at a single operating point, yet a model that is accurate on average may still fail without signalling that it is about to fail. Selective prediction addresses this by allowing a model to abstain, but the resulting trade-off between coverage and risk is rarely quantified in medical imaging. Objective: This study asks whether confidence- and entropy-based selective prediction produces a measurable reduction in selective risk in a biomedical vision–language classifier by concentrating errors among withheld cases, and at what coverage cost. Three further components of a brain-MRI-analysis framework—volumetric segmentation, glioma grading and structured report generation—are evaluated as complementary evidence about the reliability profiles of different modules, not as independent contributions. Methods: A LoRA-fine-tuned LLaVA-Med classifier was evaluated on the 1000-image BRISC2025 test partition, with a disjunctive rejection rule based on softmax confidence and predictive entropy. Volumetric segmentation used nnU-Net v2 (3D full-resolution) on BraTS2020 with a three-fold ensemble evaluated on 74 held-out cases; glioma grading (high- versus low-grade) used DenseNet121 on multi-slice representations; report generation used Meta Llama 3.2 3B Instruct fine-tuned with QLoRA on structured findings. All proportions are reported with 95% Wilson score intervals under a single analysis plan documented before the reported analyses were finalised. Results: Withholding 1.80% of predictions (95% CI 1.14–2.83) raised selective accuracy to 97.76% (95% CI 96.63–98.52) and reduced selective risk from 3.40% to 2.24%, a relative reduction of 34.1%. Withheld predictions were strongly enriched for errors (66.7% versus 2.24%; odds ratio 87.3, Fisher exact p = 4.6 × 10⁻¹⁵), capturing 35.3% of all errors while issuing 99.38% of the predictions that were correct. Considered as a detector of its own errors, the rule has a balanced accuracy of 67.34%; this describes the binary rejection decision at the realised threshold and does not identify the AUROC of the underlying continuous score, which the single operating point leaves anywhere in [0.351, 0.996]. Because the per-sample probability vectors were not retained, the full risk–coverage curve, the rule ablations and calibration error could not be computed, and the thresholds (τ_c = 0.6, τ_H = 1.0) were selected empirically on an undocumented partition; the selective-prediction finding is therefore reported as a single-operating-point case study and is exploratory. Among the complementary components, three-fold ensembling improved held-out mean Dice from 0.7948 to 0.8150 (+0.0202), reported descriptively because the per-case values needed for a paired test were not preserved. Glioma grading did not achieve better-than-chance discrimination: balanced accuracy 49.91%, Cohen’s κ −0.0023, Matthews correlation coefficient −0.0026, low-grade recall 8.96% (95% CI 5.81–13.57), and accuracy (74.03%) below a constant majority-class classifier (79.46%); the associated McNemar statistic is reported descriptively because slice-level observations are not independent. Report generation produced high automatic similarity scores (BERTScore F1 0.9513) that are not interpretable without baselines. Conclusions: Confidence- and entropy-based abstention produced a measurable reduction in selective risk at low rejection rates, and it did so because errors concentrated among the cases were withheld by the confidence/entropy rule rather than because predictions were discarded indiscriminately. At the realised coverage the rule captured 65.5% (95% bootstrap CI 41.9–87.8) of the reduction achievable by an oracle that rejects only true errors. The component-wise comparison shows that this reliability behaviour cannot be assumed for every module: high aggregate accuracy and useful discrimination came apart in the grading task. These findings are component-wise and do not constitute an end-to-end, externally validated or clinically validated pipeline. The vision–language evaluation is at image level: BRISC2025 carries no patient identifiers, so patient-level independence cannot be established and the reported intervals should be read with that constraint in mind.

Alzheimer’s disease and related dementias are projected to affect more than 150 million people worldwide by 2050. Early staging with validated instruments such as the Clinical Dementia Rating (CDR) scale is essential for timely intervention, yet access to clinician-administered CDR assessment remains constrained by workforce, time, and geographic barriers. This study complements a previously published machine learning pipeline for Alzheimer’s disease prediction by addressing the downstream task of dementia staging. Because the global CDR score is already derived from the six sub-domain ratings through an established rule-based procedure, the contribution reported here lies not in discovering that mapping but in encoding it in a transparent, deployable form: an explainable decision tree classifier embedded in DiAbot, a large-language-model-fronted conversational system that supports self-administered CDR-style assessment. We extracted 13,453 CDR records from the Alzheimer’s Disease Neuroimaging Initiative (ADNI), removed administrative variables, invalid entries, and missing rows (final n = 13,290), and trained decision tree classifiers under two impurity criteria, Information Gain and Gini Index, using a 70/30 stratified record-level hold-out and ten-fold stratified record-level cross-validation. This classifier-level evaluation uses the six domain scores as recorded during ADNI’s clinician-administered assessment, not scores elicited by the DiAbot chatbot; the trained classifier was separately embedded in a web application in which a prompt-engineered large language model conducts a CDR-style interview and normalizes responses to ordinal domain scores, but the end-to-end accuracy of that full conversational pipeline (chatbot elicitation through to final CDGLOBAL) has not yet been measured, and is not what the headline accuracy figures below report. The Information Gain Decision Tree reproduced the established mapping from the six CDR sub-domain scores to the CDGLOBAL with 99.86% accuracy under the record-level hold-out protocol (matching macro-averaged precision, recall, and F1-score), with a ten-fold record-level cross-validated mean of 99.81% (SD 0.07); this result represents fidelity to the established CDR scoring rule rather than independent dementia-diagnosis accuracy. Gini-based trees performed almost identically (99.79% hold-out, 99.74% cross-validated). Memory dominated feature importance, consistent with its role as the primary domain in the official CDR scoring algorithm. Residual misclassifications were confined to adjacent CDR stages. Because the CDGLOBAL is deterministically derived from the six sub-domain scores, these figures should be read throughout as evidence of high-fidelity reconstruction of the established CDR scoring relationship, not as general dementia-diagnosis accuracy comparable to imaging- or biomarker-based classifiers; further, participant-independent generalization remains unverified under the record-level protocol evaluated here. An interpretable classifier embedded in a conversational front-end can nonetheless make standardized CDR-style staging more widely accessible while preserving clinical inspectability; the resulting system is positioned as a screening-stage adjunct to, and not a replacement for, clinician-administered CDR assessment.

Background/Objectives : Brain tumor analysis using magnetic resonance imaging (MRI) remains a challenging task due to tumor heterogeneity, complex anatomical structures, and reliance on expert interpretation. Although deep learning approaches have shown promising results in medical image analysis, many existing studies focus on either tumor classification or segmentation independently, limiting their applicability in comprehensive automated brain tumor analysis workflows. This study proposes an integrated dual-task deep learning framework for automated brain tumor classification and segmentation using MRI scans. The framework aims to provide complementary diagnostic support by combining tumor-type prediction and tumor boundary delineation within an integrated workflow. Methods: The proposed framework utilizes EfficientNet-based convolutional neural networks for multi-class brain tumor classification and U-Net++ architectures with EfficientNet encoders for tumor segmentation. Experiments were conducted using the BRISC2025 dataset, consisting primarily of 6000 T1-weighted 2D MRI slices collected from axial, coronal, and sagittal planes. Standard preprocessing, augmentation, transfer learning, and selective fine-tuning strategies were applied. Multiple architectures were systematically evaluated using evaluation metrics. Results: EfficientNet-B1 achieved a classification accuracy of 99.70% with near-perfect precision, recall, and F1-scores across glioma, meningioma, pituitary tumor, and no-tumor classes. For segmentation, U-Net++ with an EfficientNet-B1 encoder achieved a Dice score of 0.9055, an IoU score of 0.8442, and an HD95 value of 12.21 pixels on the held-out test set. The proposed framework demonstrated robust performance in detecting small and low-contrast tumor regions while maintaining strong generalization performance across diverse MRI samples. Conclusions: The proposed integrated framework demonstrated strong performance in both brain tumor classification and segmentation tasks, effectively detecting small and low-contrast tumor regions while maintaining good generalization across diverse MRI samples. These findings suggest that the framework may serve as a reliable decision-support tool for automated brain tumor analysis in clinical practice.

Microarray gene expression data are characterized by high dimensionality and small sample sizes, which complicates cancer classification tasks. To address these challenges, this study proposes a hybrid gene selection approach that integrates a filter-based dimensionality reduction method with a metaheuristic optimizer. Specifically, the method applies the F-score statistical filter to rank and reduce gene features, followed by Nuclear Reaction Optimization (NRO) to refine the selection. This combination is referred to as the F-score-based Nuclear Reaction Optimization method or F-NRO. The performance of F-NRO was evaluated on six publicly available microarray cancer datasets (Colon, Leukemia1, Leukemia2, Lung, Lymphoma, and SRBCT) using Support Vector Machines (SVMs) and Leave-One-Out Cross-Validation (LOOCV). Comparative analysis against several existing hybrid gene selection algorithms demonstrates that F-NRO achieves high classification accuracy, including perfect accuracy on five datasets, using compact gene subsets. These results suggest that F-NRO is an effective and interpretable solution for gene selection in cancer classification tasks.

The classification of cancer based on gene expression profiles is a central challenge in precision oncology due to the high dimensionality and low sample size inherent in microarray datasets. Effective gene selection is crucial for improving classification accuracy while minimizing computational overhead and model complexity. This study introduces Genetic-Embedded Nuclear Reaction Optimization (GNR), a novel hybrid metaheuristic that enhances the conventional Nuclear Reaction Optimization (NRO) algorithm by embedding a genetic uniform crossover mechanism into its fusion phase. The proposed algorithm leverages a two-stage process: an initial F-score filtering step to reduce dimensionality, followed by GNR-driven optimization to identify compact, informative gene subsets. Evaluations were conducted on six widely used microarray cancer datasets, with Support Vector Machines (SVM) employed as classifiers and performance assessed via Leave-One-Out Cross-Validation (LOOCV). Results show that GNR consistently outperforms the original NRO and several benchmark hybrid algorithms, achieving 100% classification accuracy with significantly smaller gene subsets across all datasets. These findings confirm the efficacy of the genetic-embedded fusion strategy in enhancing local exploitation while preserving the global search capabilities of NRO, thereby offering a robust and interpretable approach for gene selection in cancer classification.

Background/Objectives: Cancer classification using microarray datasets presents a significant challenge due to their extremely high dimensionality. This complexity necessitates advanced optimization methods for effective gene selection. Methods: This study introduces and evaluates the Nuclear Reaction Optimization (NRO)—drawing inspiration from nuclear fission and fusion—for identifying informative gene subsets in six benchmark cancer microarray datasets. Employed as a standalone approach without prior dimensionality reduction, NRO was assessed using both Support Vector Machine (SVM) and k-Nearest Neighbors (k-NN). Leave-One-Out Cross-Validation (LOOCV) was used to rigorously evaluate classification accuracy and the relevance of the selected genes. Results: Experimental results show that NRO achieved high classification accuracy, particularly when used with SVM. In select datasets, it outperformed several state-of-the-art optimization algorithms. However, due to the absence of additional dimensionality reduction techniques, the number of selected genes remains relatively high. Comparative analysis with Harris Hawks Optimization (HHO), Artificial Bee Colony (ABC), Particle Swarm Optimization (PSO), and Firefly Algorithm (FFA) shows that while NRO delivers competitive performance, it does not consistently outperform all methods across datasets. Conclusions: The study concludes that NRO is a promising gene selection approach, particularly effective in certain datasets, and suggests that future work should explore hybrid models and feature reduction techniques to further enhance its accuracy and efficiency.

Cancer classification based on gene expression data is a critical challenge in modern bioinformatics, requiring efficient and accurate feature selection methods. This study explores the performance of hybrid bio-inspired algorithms and deep learning techniques for gene selection and cancer classification. Hybrid bio-inspired methods, inspired by natural optimization processes, have demonstrated significant advantages in navigating high-dimensional genomic data. Meanwhile, deep learning models excel in pattern recognition and automated feature extraction, offering a complementary approach to traditional gene selection techniques. This paper systematically evaluates both approaches, highlighting their strengths and limitations in terms of classification accuracy, computational efficiency, and feature selection effectiveness. Our findings reveal that hybrid bio-inspired methods, such as Grey Wolf Optimizer and Harris Hawks Optimization, achieve high classification accuracy with minimal selected genes, making them computationally efficient for clinical applications. Conversely, deep learning models, including convolutional neural networks and autoencoders, demonstrate superior feature extraction but often require larger datasets and higher computational resources. By providing a comparative analysis, this study aims to guide researchers and clinicians in selecting the most suitable approach for cancer classification tasks. The results underscore the potential of hybrid methodologies in advancing precision oncology while identifying opportunities for future improvements in deep learning-based feature selection.

Machine learning (ML) has increasingly been utilized in healthcare to facilitate disease diagnosis and prediction. This study focuses on predicting Alzheimer’s disease (AD) through the development and comparison of ML models using Support Vector Machine (SVM), Random Forest (RF), and Logistic Regression (LR) algorithms. Additionally, feature selection techniques including Minimum Redundancy Maximum Relevance (mRMR) and Mutual Information (MI) were employed to enhance the model performance. The research methodology involved training and testing these models on the OASIS-2 dataset, evaluating their predictive accuracies. Notably, LR combined with mRMR achieved the highest accuracy of 99.08% in predicting AD. These findings underscore the efficacy of ML algorithms in AD prediction and highlight the utility of the feature selection methods in improving the model performance. This study contributes to the ongoing efforts to leverage ML for more accurate disease prognosis and underscores the potential of these techniques in advancing clinical decision-making.

In this study, our aim is to address the problem of gene selection by proposing a hybrid bio-inspired evolutionary algorithm that combines Grey Wolf Optimization (GWO) with Harris Hawks Optimization (HHO) for feature selection. The motivation for utilizing GWO and HHO stems from their bio-inspired nature and their demonstrated success in optimization problems. We aim to leverage the strengths of these algorithms to enhance the effectiveness of feature selection in microarray-based cancer classification. We selected leave-one-out cross-validation (LOOCV) to evaluate the performance of both two widely used classifiers, k-nearest neighbors (KNN) and support vector machine (SVM), on high-dimensional cancer microarray data. The proposed method is extensively tested on six publicly available cancer microarray datasets, and a comprehensive comparison with recently published methods is conducted. Our hybrid algorithm demonstrates its effectiveness in improving classification performance, Surpassing alternative approaches in terms of precision. The outcomes confirm the capability of our method to substantially improve both the precision and efficiency of cancer classification, thereby advancing the development of more efficient treatment strategies. The proposed hybrid method offers a promising solution to the gene selection problem in microarray-based cancer classification. It improves the accuracy and efficiency of cancer diagnosis and treatment, and its superior performance compared to other methods highlights its potential applicability in real-world cancer classification tasks. By harnessing the complementary search mechanisms of GWO and HHO, we leverage their bio-inspired behavior to identify informative genes relevant to cancer diagnosis and treatment.

With the growing significance of the tourism industry and the increasing desire among travelers to discover new destinations, there is a need for effective recommender systems that cater to individual interests. Existing tourism mobile applications incorporate recommendation systems to alleviate information overload. However, these systems often overlook the varying importance of different items, resulting in suboptimal recommendations. In this research paper, a novel approach is proposed: a weighted parallel hybrid recommendation system. By considering item weights and leveraging parallel processing techniques, this method significantly enhances the accuracy of the similarity between items, leading to improved recommendation quality and precision. With this approach, users can efficiently and effectively explore new destinations that align with their unique preferences and interests, thereby enhancing their overall tourism experience and satisfaction. To evaluate the effectiveness of the proposed weighted parallel hybrid recommendation system, we conducted experiments using a dataset consisting of 20 users. The results demonstrated that the proposed approach achieved an impressive classification accuracy of 80%. A comparative analysis revealed that the proposed approach outperformed that of existing systems and achieved the best results in terms of classification accuracy. This finding highlights the effectiveness and efficiency of the proposed method in generating and promoting sustainable travel experiences by developing a personalized recommendations system for the unique preferences and interests of individual users.