Raja Majid Mehmood

Depression in students is a serious public health issue that affects their learning and daily life activities. Early detection of students who are at risk is important however many studies focus on high accuracy without checking fairness and interpretability. This study developed a transparent and fairness aware framework to predict the risk of non-clinical depression using publicly available Kaggle Dataset from 27,900 anonymized student survey answers. Three models were evaluated which are Logistic Regression, Random Forest, and XGBoost. The models showed very similar performance with accuracy around 84.5% to 84.8%. XGBoost had a slight difference with a ROC AUC of 0.923 but Logistic Regression was almost similar at 0.922, suggesting there’s little difference between the two in terms of overall performance suggesting that the data patterns are simple not complex.A post hoc gender fairness evaluation showed minimal performance differences across male and female students. SHAP and LIME explanations revealed that psychosocial stress factors, particularly suicidal thoughts, academic pressure and financial stress, were the strongest contributors to predicted risk. In summary, this research shows that fairness evaluation and interpretability as well as prediction accuracy could be Incorporated into one responsible framework for student mental health risk screening embedding ethical consideration.

Soil nutrient analysis is essential for sustainable agriculture and precision farming. Traditional chemical tests for nitrogen, phosphorus, and potassium are reliable but costly, time-consuming, and often unavailable in remote areas. The objective of this study is to develop a novel approach that combines laboratory chemical extraction, controlled image acquisition, image processing, machine learning, and deep learning to classify soil nitrogen, phosphorus, and potassium levels. 50 independent soil samples were analyzed, chemically extracted, and classified as Low, Medium, or High for each nutrient. A specially designed black-box prototype was developed to capture high-quality images of soil extract samples under consistent lighting conditions. The images were used to extract color and texture information, which was evaluated using multiple classical machine-learning algorithms, such as K-nearest neighbours, support vector machines, logistic regression, random forest, extra trees, gradient boosting, and Gaussian naive Bayes. A convolutional neural network-based model was also used for comparison. To minimize data leakage, all validation procedures were performed at the physical-sample level, and augmented images were restricted to the training data. The best models obtained using nested cross-validation for nitrogen, phosphorus, and potassium were KNN, Random Forest, and Logistic Regression, respectively, with accuracies of 60%, 82%, and 48%. The nitrogen model correctly identified the High category, and the phosphorus model correctly identified the dominant Low category each time. For potassium, Logistic Regression made predictions of all three classes and outperformed the majority class baseline. The Classical machine-learning models slightly outperformed the compact CNN for nitrogen and potassium, while both achieved the same accuracy for phosphorus; however, overall performance remained limited by the small and imbalanced dataset. The findings suggest that the proposed framework could contribute to image-assisted decision-making in smart farming and support improved soil management practices, particularly in regions where laboratory testing facilities are limited.

Background/Objectives: The human brain is responsible for controlling various physiological functions, and hence, the presence of tumors in the brain is a major concern in the medical field. The correct identification and categorization of tumors in the brain using Magnetic Resonance Imaging (MRI) is a major requirement for the diagnosis and treatment of a tumor. The proposed research will focus on designing a CNN model that is optimized for tumor image classification. Methods: This research proposes an optimized CNN model featuring strategically placed dropout layers and hyperparameter optimization. This study uses a dataset of 640 MRI scans (320 tumor and 320 non-tumor) collected from a private hospital in Saudi Arabia. The proposed method utilizes a learning rate of 0.001 in combination with the Adam optimizer to ensure stable and efficient convergence. Its performance was benchmarked against established architectures, including VGG-19, Inception V3, ResNet-10, and ResNet-50, with evaluation based on classification accuracy and computational cost. Results: The experimental results show that the optimized CNN proposed in this work performs much better than the deeper architectures. The network reached a maximum training accuracy of 97.77% and a final test accuracy of 95.35% with a small test loss of 0.2223. The test accuracy of the optimized VGG-19 and Inception V3 networks was much lower, with a training time per epoch that was several orders of magnitude higher. The validation stability of the proposed network was high (92.25% to 95.35%) during the final stages of training. Conclusions: The conclusion drawn from this study is that hyperparameter optimization and strategic regularization are more advantageous for tumor classification using MRI images than the mere depth of the model. The accuracy of 95.35% with low computational complexity makes this lightweight CNN model a feasible solution for real-time applications.

Large-scale earthquakes can cause huge loss of life and material losses. After an earthquake, highways are the most commonly used type of transportation for the delivery of the necessary aid teams and materials to the scene of the event. If the highways are not well maintained, it may cause serious disruption of transportation after the earthquake or aftershocks. In this study, field studies were conducted in the provinces where the earthquake was felt severely after the earthquakes in Turkey on February 6, 2023. In these studies, images were collected according to the condition of asphalt cracks on the highways. These images were labeled as in need of urgent maintenance (Major) and not in need of urgent maintenance (Minor) and a new dataset was created. The classification performance of popular pre-trained CNN models is evaluated on this dataset. First, classification algorithms other than softmax were used to improve the classification performance. The Combined Metaheuristic Optimization-Relieff (CMO-R) algorithm was designed to improve the classification performance by one more level. Extensive experiments were conducted on the dataset, and the VGG16 model demonstrated superior performance, reaching an accuracy of 80.32% without encountering overfitting.

Affective Computing is one of the central studies for achieving advanced human-computer interaction and is a popular research direction in the field of artificial intelligence for smart healthcare frameworks. In recent years, the use of electroencephalograms (EEGs) to analyze human emotional states has become a hot spot in the field of emotion recognition. However, the EEG is a non-stationary, non-linear signal that is sensitive to interference from other physiological signals and external factors. Traditional emotion recognition methods have limitations in complex algorithm structures and low recognition precision. In this article, based on an in-depth analysis of EEG signals, we have studied emotion recognition methods in the following respects. First, in this study, the DEAP dataset and the excitement model were used, and the original signal was filtered with others. The frequency band was selected using a butter filter and then the data was processed in the same range using min–max normalization. Besides, in this study, we performed hybrid experiments on sash windows and overlays to obtain an optimal combination for the calculation of features. We also apply the Discrete Wave Transform (DWT) to extract those functions from the preprocessed EEG data. Finally, a pre-trained k-Nearest Neighbor (kNN) machine learning model was used in the recognition and classification process and different combinations of DWT and kNN parameters were tested and fitted. After 10-fold cross-validation, the precision reached 86.4%. Compared to state-of-the-art research, this method has higher recognition accuracy than conventional recognition methods, while maintaining a simple structure and high speed of operation.