The continuous decline of the Dead Sea water level represents one of the most critical environmental challenges in the Middle East, primarily caused by reduced inflow from the Jordan River basin and increased regional water consumption. Predicting the long-term evolution of the Dead Sea is therefore essential for evaluating the potential effectiveness of large-scale restoration initiatives such as the Red Sea-Dead Sea (RSDS) conveyance project. In this study, a stochastic simulation framework is proposed to model and predict the long-term dynamics of the Dead Sea water level under multiple inflow scenarios. The framework integrates probabilistic increment generation, recursive hydrological updating, and stability-aware monitoring within a unified computational procedure. Annual water-level increments are generated using statistically fitted probability distributions derived from historical observations, enabling the model to capture the inherent variability and uncertainty of natural hydrological processes. The system state is iteratively updated across the simulation horizon, while a stability index is used to evaluate convergence behavior and detect potential equilibrium conditions. Five representative RSDS scenarios are investigated, including baseline recovery, moderate inflow, accelerated recharge, threshold-stabilized inflow, and high-variability environmental conditions. Simulation results indicate that under the baseline scenario the Dead Sea level increases gradually from approximately −434 m to about −60 m within 500 simulation steps, whereas accelerated recharge conditions can raise the level beyond 200 m due to higher early-stage inflow rates. The threshold-stabilized scenario reveals regime transitions in increment dynamics that lead to a near-equilibrium state, while the high-variability scenario exhibits strong stochastic fluctuations with annual increments ranging between 0.4 and 2.2 m. Overall, the results demonstrate that the proposed framework effectively captures diverse hydrological behaviors and provides a flexible and computationally efficient tool for long-term forecasting and policy evaluation of Dead Sea restoration strategies.

Preeclampsia is the most common medical disorder in pregnancy that is steadily increasing over time. Evidence indicates that early identification of women at high risk, followed by timely preventive interventions such as Aspirin and close surveillance, can significantly reduce disease incidence and severity, highlighting the importance of accurate early prediction. At the molecular level, preeclampsia is primarily a placental disorder associated with dysregulated microRNA (miRNA) expression. DICER, a key enzyme in miRNA biogenesis, and its single nucleotide polymorphisms (SNPs) have been shown to influence placental miRNA profiles and disease susceptibility. In this study, artificial intelligence-driven data mining and machine learning classifiers integrating genetic (DICER SNPs) and clinical features were developed and evaluated in a sample of Jordanian population. Six different classifiers were applied to predict the occurrence of preeclampsia, stratify disease severity, and distinguish early- from late-onset disease. This study included 224 pregnant women, with 137 controls and 87 women who had developed preeclampsia. Genotypes of rs3742330 and rs14035 SNPs in DICER and RAN genes respectively, were determined using polymerase chain reaction techniques, and clinical data was recruited. Six machine learning classifiers were used to build the prediction models. Model performance was estimated using repeated stratified k-fold cross-validation to provide robust, unbiased estimates given the limited sample size. The results showed that ensemble methods Random Forest and XGBoost outperformed other models with the highest accuracy, sensitivity, precision, F1-scores, and G-means. For occurrence prediction, Random Forest achieved a cross-validated accuracy of 0.973 ± 0.040, sensitivity of 0.953 ± 0.058 and F1-score of 0.965 ± 0.050, while XGBoost attained an accuracy of 0.978 ± 0.022, indicating that combining genetic with clinical factors provides strong discriminative ability compared to models relying solely on traditional risk factors. Thus, integrating genetic information, particularly the DICER rs3742330 polymorphism, with routine early pregnancy clinical features using data mining and machine learning techniques yields strong predictive performance for preeclampsia in the Jordanian population. These findings suggest that such AI-driven models could be translated into practical clinical applications to support early risk stratification.

Speaking disorders such as stuttering or stammering cause sounds, syllables, or phrases to be repeated or delayed, impairing natural speaking patterns. A person's quality of life may suffer as a result, and communication with others may become challenging. Over 70 million individuals globally experience stuttering. Many people with this illness struggle to fit in and lose confidence while speaking in front of others. Because of this, this research aims to create a thoroughly annotated dataset and, in the end, provide an extensive empirical comparison of deep convolutional neural network (CNN) architectures on the classification of stuttering or Stammering. Short-time fast Fourier transformation (STFT) is used in this study to convert audio data into a visual representation (spectrogram), and Mel Frequency Cepstral Coefficients (MFCC) are then used to extract many audio features to use CNNs to create a workable classification system. Matthew's correlation coefficient (MCC), time to build, and accuracy score are used to evaluate the classifiers' performance. The final findings showed that AlexNet, Le-Net5, ResNet50, LSTM-CNN, and VGG are the top five CNN classifiers that have been investigated. The four designs that are being used have an exceptionally high degree of accuracy, ranging from 90% to 97%.

The fashion industry increasingly relies on artificial intelligence technologies to enhance creative workflows and accelerate design innovation. This research presents a comprehensive framework that employs Generative Adversarial Net- works and advanced diffusion models to generate high-quality fashion imagery from textual descriptions. The proposed system integrates Stable Diffusion architecture with specialized text preprocessing pipelines to create diverse, photo realistic fashion designs that align with textual specifications while maintaining aesthetic coherence and commercial viability. The framework was evaluated using a dataset of 10,000 high-resolution fashion images, with systematic assessment conducted across multiple performance dimensions including creativity, aesthetic appeal, design diversity, and semantic consistency. Experimental results demonstrate exceptional performance in creative design generation, achieving average scores of 4.7 for originality and 4.5 for aesthetic quality based on comprehensive evaluation by thirty participants. The system successfully produces varied design alternatives from similar prompts, indicating robust exploration of design possibilities rather than repetitive pattern generation. While text prompt accuracy achieved a moderate score of 3.8, highlighting opportunities for enhanced semantic interpretation, the overall results validate the framework’s capability to support professional fashion design workflows. The research contributes to the growing body of knowledge in AI-assisted creative applications and demonstrates significant potential for transforming traditional fashion design processes through intelligent automation and creative augmentation technologies.

The effectiveness and efficiency of a machine learning model can be improved by feature selection, especially for high-dimensional datasets such as in cybersecurity. The proposed approach utilizes an enhanced version of the Rainbow agent with a memory storage structure. The suggested approach is assessed using two benchmark datasets namely RT-IoT2022 which is targeted towards IoT network security and the Android Malware Detection dataset which is meant for mobile security. The specification of the reinforcement learning model has been trained for 20 epochs and it is progressively enhanced through feature subsets to enhance classification accuracy. The results show that the AUC scores continuously increase were the one for RT-IoT2022 achieves 0.91 and Android at 0.93. Three well-known classifiers XGBoost, Random Forest and multi-layer perceptron (MLP) are used to test the power of the selected features. The outcome evaluation on RT-IoT2022 dataset shows that Random Forest achieved maximum accuracy (99.48%), followed by XGBoost (99.16%), while MLP secured 94.04% accuracy. In the Android malware dataset, XGBoost model gave the best accuracy of 89.50%, followed closely by Random Forest with 87.00% and MLP with 86.50%. This clearly shows that reinforcement learning based feature selection enhances accuracy and reduces computation. The research emphasizes utilizing dynamic feature selection in any cyber security application. The future will experiment with incorporating deep reinforcement learning as well as hybrid selection.