Wildfires are increasingly destructive natural disasters, annually consuming millions of acres of forests and vegetation globally. The complex interactions among fuels, topography, and meteorological factors, including temperature, precipitation, humidity, and wind, govern wildfire ignition and spread. This research presents a framework that integrates satellite remote sensing and numerical weather prediction model data to refine estimations of final wildfire sizes. A key strength of our approach is the use of comprehensive geospatial datasets from the IBM PAIRS platform, which provides a robust foundation for our predictions. We implement machine learning techniques through the AutoGluon automated machine learning toolkit to determine the optimal model for burned area prediction. AutoGluon automates the process of feature engineering, model selection, and hyperparameter tuning, evaluating a diverse range of algorithms, including neural networks, gradient boosting, and ensemble methods, to identify the most effective predictor for wildfire area estimation. The system features an intuitive interface developed in Gradio, which allows the incorporation of key input parameters, such as vegetation indices and weather variables, to customize wildfire projections. Interactive Plotly visualizations categorize the predicted fire severity levels across regions. This study demonstrates the value of synergizing Earth observations from spaceborne instruments and forecast data from numerical models to strengthen real-time wildfire monitoring and postfire impact assessment capabilities for improved disaster management. We optimize an ensemble model by comparing various algorithms to minimize the root mean squared error between the predicted and actual burned areas, achieving improved predictive performance over any individual model. The final metric reveals that our optimized WeightedEnsemble model achieved a root mean squared error (RMSE) of 1.564 km2 on the test data, indicating an average deviation of approximately 1.2 km2 in the predictions.

Open Source Software (OSS) has gained significant traction in the government sector due to its potential to reduce costs, enhance security, and offer diverse benefits. This study focuses on the adoption of OSS within the Madinah Development Authority (MDA), a Saudi Arabian governmental agency. It aims to explore the OSS adoption process, identify challenges, and propose solutions to maximize its benefits. Employing a hybrid approach, data were collected through preliminary interviews with managers and a structured questionnaire survey among MDA employees. A SWOT analysis was conducted to evaluate the organization's IT environment and staff capabilities. The study’s key contribution is the development of a phased strategy tailored for MDA to successfully adopt OSS, addressing identified challenges and optimizing the benefits of open-source solutions for government operations.

Literature confirms that the low success rate of eHealth systems is closely linked to inadequate computer knowledge. Therefore, this study aimed to assess knowledge and utilization of computers among healthcare workers in Benue South. An institution-based quantitative cross-sectional study design was conducted in 120 health facilities, including primary healthcare centres, and private clinics. This study involved 430 healthcare workers. Using Python programming, descriptive and multivariable logistic regression analyses were conducted to assess the level of computer knowledge and utilization among participants, as well as to identify factors influencing their computer knowledge and utilization. Of the 430 participants, 233 (54.19%) of healthcare workers passed the computer knowledge test, 216 (50.23%) of health workers have access to computers, and 221 (51.40%) of them had undergone formal computer training. The findings revealed that 263 (61.16%) of healthcare workers exhibit good computer utilization, while 167 (38.84%) demonstrated poor utilization. Access to a computer (adjusted odds ratio [AOR]=2.83, 95% CI 0.48-1.60), and prior computer training (AOR=3.34, 95% CI 0.65-1.76) were found to be significantly associated with computer knowledge, while Access to a computer (AOR=2.98, 95% CI 0.48-1.70), Sex (AOR=2.23, 95% CI 0.28-1.32), Department (AOR=1.06, 95% CI 0.00-0.11), and prior computer training (AOR=5.53, 95% CI 1.10-2.32) were found to be significantly associated with computer utilization. These findings imply that improved access to computers and comprehensive computer training for healthcare professionals is vital for improved service delivery.

This medical image segmentation plays a fundamental role in the diagnosis of diseases related to the correct identification of internal structures and pathological regions in different imaging modalities. The conventional fuzzy-based segmentation approaches, though quite useful, still have some drawbacks regarding handling uncertainty, parameter optimization, and high accuracy of segmentation with diverse datasets. Because of these facts, it generally leads to poor segmentations, which can give less reliability to the clinical decisions. In addition, the paper is going to propose a model, FTra-UNet, with advanced segmentation of medical images by incorporating fuzzy logic and transformer-based deep learning. The model would take complete leverage of the strengths of FIS concerning the handling of uncertainties in segmentation. Besides, it integrates SSHOp optimization technique to fine-tune the weights learned by the model to ensure improvement in adaptability and precision. These integrated techniques ensure faster convergence rates and higher accuracy of segmentation compared to state-of-the-art traditional methods. The proposed FTra-UNet is tested on BRATS, CT lung, and dermoscopy image datasets and ensures exceptional results in segmentation accuracy, precision, and robustness. Experimental results confirm that FTra-UNet yields consistent, reliable segmentation outcomes from a practical clinical application perspective. The architecture and implementation of the model, with the uncertainty handled by FIS and the learning parameters optimization handled by the SSHOp method, increase the power of this model in segmenting medical images.

Code-switching, which is the mixing of words or phrases from multiple, grammatically distinct languages, introduces semantic and syntactic complexities to sentences which complicate automated text classification. Despite code-switching being a common occurrence in informal text-based communication among most bilingual or multilingual users of digital spaces, its use to spread misinformation is relatively less explored. In Kenya, for instance, the use of code-switched Swahili-English is prevalent on social media. Our main objective in this paper was to systematically re- view code-switching, particularly the use of Swahili-English code-switching to spread misinformation on social media in the Kenyan context. Additionally, we aimed at pre-processing a Swahili-English code-switched dataset and developing a misinformation classification model trained on this dataset. We discuss the process we took to develop the code- switched Swahili-English misinformation classification model. The model was trained and tested using the PolitiKweli dataset which is the first Swahili-English code-switched dataset curated for misinformation classification. The dataset was collected from Twitter (now X) social media platform, focusing on text posted during the electioneering period of the 2022 general elections in Kenya. The study experimented with two types of word embeddings - GloVe and FastText. FastText uses character n-gram representations that help generate meaningful vectors for rare and unseen words in the code-switched dataset. We experimented with both the classical machine learning algorithms and deep learning algo- rithms. Bidirectional Long Short-Term Memory Networks (BiLSTM) algorithm showed the best performance with an f-score of 0.89. The model was able to classify code-switched Swahili-English political misinformation text as fake, fact or neutral. This study contributes to recent research efforts in developing language models for low-resource languages.

Polycystic Ovary Syndrome (PCOS) is a common endocrine disorder that affects women of reproductive age, leading to hormonal imbalances and ovarian dysfunction. Early detection and intervention are vital for effective management and prevention of complications. This study compares PCOS prediction using the XGBoost machine learning model against four traditional models: Logistic Regression (LR), Support Vector Machine (SVM), Decision Trees (DT), and Random Forests (RF). LR and SVM achieve accuracies of 95% and 96%, respectively, demonstrating strong predictive capabilities. In contrast, DT had a lower accuracy (82%), indicating limitations in PCOS data complexity. RF showed competitive performance with 96% accuracy, underscoring its effectiveness in ensemble learning. XGBoost achieves 98% accuracy with its parameter configuration. The scale pos weight parameter adjusts the positive class weight in imbalanced datasets, addressing under representation by assigning more weight to the minority class, and thereby improving the training focus. The gradient boosting framework incrementally builds models to address complex feature interactions and dependencies, enhancing the accuracy and stability in predicting intricate PCOS dataset. This analysis highlights the importance of advanced machine learning models such as XGBoost for accurate and reliable PCOS predictions. This research advances PCOS prediction, demonstrates the potential of machine learning in healthcare, and clarifies the strengths and limitations of different algorithms with complex medical datasets.

CAPTCHA is an acronym for Completely Automated Public Turing test to tell Human and Computer Apart. The main purpose of CAPTCHA is to differentiate between human and automated machine during online transaction. Text, image, audio and video are types of CAPTCHAs. However, text-based CAPTCHAs are available in the market in different languages i.e., English, Arabic, Urdu and Chinese but accented character-based text CAPTCHA system namely NAIJACAPTCHA is a newly introduce text-based CAPTCHA developed using Latin characters and accented characters from two Nigerian language: Yorùbá and Igbo. The usability of sixteen accented character-based CAPTCHAs was tested to see if they were suitable for human usage. The usability performance was measured using response time, solving time, accuracy, and success rate. A total of two hundred and twenty-two participants were selected for the study, and 1108 CAPTCHA codes were generated. The response time for Text Distortion with Coloured Background was the fastest, with 1.18×103 ms, while Coloured Texts with No Background (CTNB) had the least response time of 1.09 ms. With a solving time of 2.52×104 ms, Character Fragmentation with No Background was the fastest. The result showed that CTBN's response and problem-solving time is highly promising; as a result, its website application for authentication during online transactions to distinguish between humans and machines will be simple for human beings to solve and user requests will also be swiftly attended to. Lastly, the security aspect of the developed NAIJACAPTCHA will be looked into determine its vulnerability.