Rheumatic and autoimmune diseases represent a significant diagnostic challenge due to overlapping clinical presentations and the complexity inherent in their pathogenesis. In this work, we present a framework for explainable machine learning to improve the precision and interpretability of diagnoses for these illnesses. The dataset used in this study, which was collected from Iraq, consists of$\mathbf{1 2, 0 8 5}$samples for six primary diseases and normal conditions. The five machine learning algorithms are: Decision Tree, Random Forest, XGBoost, AdaBoost, and Light Gradient Boosting Machine, with parameters tuned for evaluation. Explainable global AI approaches were employed at the global level to enhance the visibility and interpretability of the model, allowing a better understanding of how significant each feature was used and contributed to the diagnostic results. For example, the Random Forest model achieved an accuracy of 83.867% and an AUC of 89.641%, which outperformed many of these approaches.

Fake news has become a major issue owing to its quick growth on the internet, difficulty in identifying it from true news, and people's reliance on social media platforms as primary sources. It has negative effects at several levels, including individual, communal, political, and economical. Detecting Arabic fake news requires significant effort owing to limited datasets and studies in the sector. In this study, Employ Perturbation Adversarial attacks was applied as a regularization technique for fake news classification. Adversarial examples are generated by perturbing the model's word embedding matrix. The AraBERTv2 model is utilized for preprocessing operations. Five different machine learning models were trained on clean data and tested using both clean and adversarial examples to evaluate the generalization capabilities of the classification models. To address the scarcity of Arabic datasets, A translated English fake news was utilized dataset Experimental results indicate that the LightGBM and AdaBoost algorithms exhibited the best performance (90%) compared to other classifiers.

the rapid spread of fake news online, coupled with the difficulty of distinguishing it from real news, has become a serious issue, especially with social media being a primary news source for many people. The fake news can have damaging effects on individuals, communities, and political and economic systems. Detecting Arabic fake news presents additional challenges due to the limited availability of relevant datasets and research. In this study, Perturbation Adversarial attacks was applied as a regularization technique for fake news detection, generating adversarial examples by modifying the model’s word embedding matrix. The AraBERTv2 model is used for preprocessing, and the lack of Arabic data was overcome by utilizing a translated English fake news dataset. A Recurrent Neural Network (RNN) model was trained on clean data, testing it with both clean and adversarial examples to evaluate its generalization capability. The results demonstrate that the RNN model performs effectively, achieving strong accuracy in Arabic fake news detection.

Predicting patient mortality risk in intensive care units (ICUs) is one of the tasks that has strategic significance in improving clinical decisions and health care outcomes. Disease mortality monitoring methods based on machine learning models have shown efficacy; however, their susceptibility to adversarial attacks in the input data presents reliability and robustness challenges. The present work addresses these challenges by introducing an effective ensemble model enriched with adversarial training to increase the performance of mortality prediction models in the ICU context. The developed methodology combines a variety of ensemble methods, such as random forest, extreme gradient boosting, bagging, AdaBoost, extra trees, and the light gradient boosting machine. These approaches work by combining several algorithms and employing adversarial training strategies that put the stakeholder’s data in their correct order and bar data tampering at all points of the model development ecosystem. The set of experiments performed with the help of real ICU datasets proved that this approach provides better accuracy, robustness, and reliability of predictions than standard models do. The extra trees algorithm achieved the best accuracy among the tested models.

The explosion of medical literature over the past decade has resulted in efficient and accurate techniques for text categorization to handle huge amount of data. This work combines ensemble learning methods with coupled multi-word embedding techniques to improve cancer text classification. The intricate semantic links present in medical tests are frequently outside the scope of traditional word embedding models, resulting in not ideal categorization results. To address this problem, we employ e continuous bag-of-words and Skip-gram approaches that yield more complete word representations capturing multiple linguistic nuances. Subsequently, such embeddings are passed through LGBM, CatBoost, and NGBoost ensemble learning classifiers to enhance classification accuracy. With 99.868% accuracy rate, LGBM and CatBoost were the most successful ensemble approaches examined. These approaches provide solid foundation for future work on the use of ensemble methods with complex word representations and for advancing the field of medical text classification.

The complexity of overlapping symptoms and interactions among multiple diseases makes it very difficult to accurately diagnose comorbidities. This article presents an innovative comorbidity diagnosis improvement approach that integrates conflicting group learning and numerous diverse machine learning algorithms such as Random Forest, Gradient Boosting, AdaBoost, Bagging and Extra Trees. The ensemble model is advantageous over individual models because it improves diagnostic accuracy while improving adversarial robustness. Effective adversarial training methods can also be used to strengthen the model against interference which may interfere with the diagnosis. Benchmark comorbidity datasets explore the effectiveness of the proposed method that is not solely more efficient in terms of accuracy compared to other methods, but similarly more efficient in its aptitude to endure adversarial instances. The current study is supposed to be merged into the investigative pipeline to allow for the development of vigorous diagnosis schemes. Of the algorithms tested in the study, the Extra Trees algorithm achieved 91.845% which was the highest performance obtained among the other algorithms.

Similar to most spoken languages worldwide, Arabic language is facing a significant problem with the rapid spread of fake information, which requires effective methods to detect and combat it. This research introduces a novel approach to detecting fake news in Arabic using Ensemble Techniques. The process involves gathering a comprehensive collection of Arabic news articles and carefully labeling them as either fake or authentic. To ensure the quality of the data, various text preprocessing techniques such as cleaning, tokenization, and stemming are applied. The detection accuracy is then enhanced by leveraging Ensemble Techniques. Multiple base classifiers, each with its own strengths, are employed to thoroughly analyze the text data for deceptive patterns. By utilizing this ensemble approach, the diversity of classifiers is harnessed to improve the reliability of detecting fake news. In this case, a robust ensemble learning framework that combines three states of the art algorithms are utilized for the detection of fake news. Two different methods for extracting features, Term Frequency and Term Frequency-Inverse Document Frequency are used to improve the accuracy of identifying fake news. The results show significant improvements in fake news detection accuracy, which provide valuable insights and methodologies for combating misinformation in the Arabic digital sphere, which is crucial for maintaining information integrity and promoting informed decision-making.

Detecting connections between genes and diseases is a vital endeavor in bioinformatics and genomics, carrying significant implications for the entire comprehension of the molecular underpinnings of various diseases. The rapid increase in the number of documents in the field of biomedicine has resulted in a significant burden and time requirement for manually curating relationships within this literature. In order to tackle this particular difficulty, the present study introduced a resilient ensemble machine-learning methodology that aimed at automating the classification of gene-disease relationships through the analysis of biomedical text. The proposed model was meant to leverage ensemble learning capabilities by integrating different base classifiers which are Decision Trees, Random Forest, AdaBoost, Bagging, CatBoost, Extra Trees and XGBoost with two feature extraction TF and TF-IDF. This ensemble architecture aimed to enhance the accuracy and dependability of gene-disease association predictions by utilizing a wide range of variables obtained from biomedical literature, including abstracts and various ensemble configurations and evaluating performance using standard metrics which are precision, recall, and F1-score, AUC, and accuracy. The study findings provided evidence supporting the efficacy of the ensemble methodology in enhancing both accuracy and resilience when compared to the performance of individual classifiers. The highest accuracy was achieved with XGBoost and TF 0.979%.

The sentiment mining field (also known as opinion mining, opinion extraction, sentiment analysis (SA), sentiment extraction, and so on) has seen significant growth in academia. Researchers have experimented with a variety of approaches to automate SA and other fields in the fields of machine learning (ML), data mining, and natural language processing. Our research aims to develop a model that extracts and categorizes words from a specific text. In this study, we used TF-IDF to select 500 to 20,000 words with vector. After pre-processing and frequency-dependent word extraction, constructs are generated. Four Naive Bayes models (complement, multinomial, Bernoulli, and Gaussian) were used. The kappa scale, precision and accuracy scores, and F1 score were used to evaluate the proposed model. The Naive Bayes multinomial system produced the most accurate results, with an accuracy rate of 86.46 percent, according to the findings.

Animal-Vehicle Accidents have shown deep increase in the middle east regions over the last decades. These collisions resulting from camels fleeing the wildlife and crossing the roads and hence endangering drivers and camel's lives and leading to habitat degradation. Additionality, the size, strength, and the unpredictable behavior of camels play a key role in high mortality rates in the camel-vehicle collisions. Various solutions and countermeasures such as warning signs and fences have been adopted in the past. However, several drawbacks are associated to them, and their effectiveness are reducing with time. Therefore, this study proposes a framework for the use of machine learning approaches and computer vision for the detection and recognition of camels. This can help to provide warning to drivers about potential animal crossings in an effort to mitigate camel-vehicle accidents.

The diagnosis of common diseases, in which people suffer from several bad health conditions, is a complex medical challenge. This study investigated the use of machine learning methods combined with the Chi-square feature selection technique to improve the accuracy and efficiency of comorbidity diseases diagnosis. Using various decision trees of machine learning algorithms, random forest, Gradient Boost, AdaBoost, Bagging, and extra trees, this research aimed to improve the identification and prediction of comorbid conditions and ultimately advance early detection and treatment strategies. Using the selection of Chi-Square features helped give priority to the most relevant attributes, reduce noise, and refine the models. The results showed that both the AdaBoost and Gradient Boost algorithms obtained an accuracy rate of 91.33%, confirming their efficacy in comorbidity diseases diagnosis. However, it is essential to ensure their reliability and efficacy in the real clinical environment, including the practical implementation and validation of these methods, to improve patient care and medical decisions.

In contemporary times, the proliferation of data dimensionality has introduced many challenges within the realm of machine learning. Consequently, identifying and selecting pertinent features have assumed paramount importance. Consequently, a diverse array of techniques for feature selection has been proffered. Among them, Metaheuristic techniques are particularly significant within this milieu, with a focus on the dragonfly algorithm garnering notable attention. Metaheuristic algorithms are rooted in the emulation of swarm intelligence, drawing inspiration from the collective behaviors exhibited by insects. This scholarly inquiry aims to enhance the efficacy of classification outcomes by leveraging the dragonfly algorithm to select the most salient features. In this paper, the proposed approach is rigorously evaluated across three distinct datasets, namely, “Ansur,” “Predict 5-Year Career Longevity for NBA Rookies,” and “Chronic Kidney Disease,” all of which were procured from the Kaggle repository. The study used four machine learning classifiers, namely, XGBoosting, K-Nearest Neighbors (KNN), Decision Tree, and Gaussian Naive Bayes (Gaussian-NB), for classification and evaluation. The empirical findings unveiled a notable enhancement in classification performance. Notably, the accuracy metrics exhibit substantial improvements across all classifiers. For instance, within chronic kidney disease classification, both XGBoost and Decision Tree classifiers yielded a remarkable accuracy rate of 100%, thereby underscoring the efficacy of the proposed dragonfly algorithm-based feature selection technique in augmenting the predictive capabilities of machine learning models.

Feature selection plays a crucial role in the domain of machine learning, serving as an essential mission. Eliminating redundant and irrelevant attributes can enhance the learning process by increasing the learning rate, improving accuracy, and enhancing classifier performance. The primary objectives of feature selection encompass the creation of models that are more understandable and user-friendly, the enhancement of data mining efficiency, and the facilitation of data cleaning and organization for subsequent analysis. Metaheuristic algorithms have garnered considerable attention due to their prowess in solving many optimization problems such as feature section, in this paper using Dragonfly algorithm. The effectiveness of the proposed methodologies is evaluated on three benchmark datasets obtained from the Kaggle repository: the heart disease dataset, the Vehicle-Coupon-Recommendation dataset, and the Predict Diabetes dataset. Three machine learning classifiers—namely, Light Gradient Boosting Machine (LightGBM), AdaBoost, and Bagging—are utilized in this assessment. This approach yields an impressive accuracy rate of 85.71%. In contrast, when all features are included in the training process, the accuracy drops to a lower value of 83.51%.

This paper presents the utilisation of dynamical recurrent neural network architectures in the purpose of classifying the Sickle Cell disorder data. It is indicted that recurrent neural networks such as the Jordan network produce a great improvement with clinical data sets and have helped in acquiring high accuracy. The main aim of this study is to provide a sophisticated model to differentiate applications of dynamical neural networks for medically related problems. We attempt to classify the amount of medications for each patient with Sickle Cell disorder. We use different recurrent neural network architectures in terms of examining performance for each model within this study. The motivation for the classification approach used in this study is to support medical sectors to offer proper therapy advice depending on the former data set. The outcomes yield from different classifiers during our experiments indicated that Elman and hybrid recurrent neural networks produced inferior results when compared to Jordan neural networks. Results have indicated that for the recurrent network models tested, the Jordan architecture was found to yield considerably better results over the range of performance measures that been selected for this research.