Daily Archives: March 31, 2025

Abstract-Assistive technology plays a crucial role in enhancing the learning experiences and outcomes for physically challenged and disabled pupils, thereby leveraging educational opportunities tailored to their specific levels of disability. This paper emphasizes the importance of assistive technological solutions that facilitate participation in inclusive education, highlighting how innovative technologies can contribute to mainstreaming education and fostering societal integration for special-category learners. By providing tools such as software, adaptive devices, and personalized learning applications, assistive technology supports academic achievement and promotes independence and agency among students with disabilities. Integrating these technologies into educational settings is increasingly essential for creating an equitable learning environment where all students can thrive optimally, ensuring that barriers to education are minimized and that every learner can succeed.

Abstract-Waste from construction and demolition—especially in post-conflict reconstruction—had severe implications for resources and the environment. This research used Waste Demolished Aggregate (WDA) as a replacement for natural aggregates to examine its sustainability in concrete mixtures, assessing workability and compressive strength with varying replacement percentages. The study correlated experimental results from three concrete mixing batches to provide a comprehensive overview of the impact of WDA on concrete functionality. Overall, it demonstrated how such a strategy can maximize environmental protection by minimizing landfill waste and the need for virgin aggregates, which brings economic benefits through such reconstruction. Minimal differences in workability were observed across the three batches based on concrete mixing slump test results. The first batch slump ranged from 75mm to 80mm, the second batch from 79mm to 80mm, and the third batch from 79mm to 82mm. However, at higher WDA percentages, there was a slight difference in workability and slump. The trend observed in compressive strength tests for the three batches indicated a declining pattern of strength with higher WDA content, consistent with the literature, which attributed this observation to the porous structure and irregular shape of WDA particles. The statistical testing demonstrated a significant reduction in strength at both the 25% and 50% WDA replacement levels. The findings indicated that 25% WDA and 50% WDA resulted in a reduction of compressive strength of approximately 23% and 31%, respectively, compared to natural aggregates. These results had significant implications for the use of WDA in structural applications. Although non-structural applications for WDA such as drainage systems, pathways or non-load-bearing walls suggest potential, WDA may well have limited application at best in structural concrete without further adjustment of the mix to overcome strength losses.

Abstract-Enterprise Resource Planning systems are critical for streamlining business processes, but the increasing complexity of modern enterprises demands greater flexibility and customization. Cloud-based custom objects have emerged as a transformative solution in ERP implementation, offering enhanced scalability, adaptability, and integration capabilities. This paper explores the role of cloud-based custom objects in ERP systems, examining how they enable organizations to tailor ERP solutions to specific business needs without compromising system integrity or performance. By leveraging cloud infrastructure, companies can efficiently create, modify, and manage custom objects, ensuring better alignment with dynamic business environments. Through a case study analysis and a review of current best practices, this paper highlights the benefits, challenges, and potential impacts of adopting cloud-based custom objects within ERP frameworks. The findings demonstrate that, when implemented strategically, cloud-based custom objects can significantly improve the agility and effectiveness of ERP systems, ultimately contributing to enhanced operational efficiency and business success.

Abstract-Rotator cuff muscle tears are among the most prevalent musculoskeletal injuries, and ultrasound imaging serves as an effective diagnostic tool. However, the interpretation of these scans requires specialized expertise, often leading to significant delays in diagnosis. This study introduces an AI-driven approach to accelerate the detection of full-thickness rotator cuff tears, reducing assessment time from months to mere minutes. The proposed method consists of two key steps: first, segmentation of the humeral cortex and subacromial bursa, followed by classification of tears based on these identified regions. Automated segmentation in ultrasound imaging poses challenges due to speckle noise, low contrast, and image artifacts. To overcome these, we employ a CNN-based autoencoder that directly predicts the boundary contour points of relevant anatomical structures instead of traditional pixel-wise semantic segmentation. This approach enhances interpretability by focusing on clinically significant landmarks rather than relying on a black-box classifier. The study utilized a dataset of 206 patients, comprising 10,080 training images and 2,520 evaluation images. The proposed segmentation model outperformed the conventional UNet, achieving a Dice coefficient of 94.2% and a Hausdorff Distance of 2.8 mm, compared to UNet’s 90.5% DC and 6.8 mm HD. Following segmentation, a VGG-16-based classification model achieved an accuracy of 81.0%, with a sensitivity of 78.5% and specificity of 76.2%. The implementation of AI-powered ultrasound for rotator cuff tear detection has the potential to facilitate early and precise diagnosis, significantly improving patient outcomes. This automated system can be deployed in primary care settings such as general practitioner clinics and emergency departments, empowering lightly trained personnel to perform initial assessments efficiently.

Abstract-To enhance the accuracy of predictions and enable real-time monitoring of mechanical parts’ operational status, a deep learning model was initially developed using a convolutional neural network (CNN) structure to extract features from the mechanical components. Subsequently, another deep learning model was designed to process these extracted features through a fully connected layer for data fusion and classification, facilitating the prediction of lifespan and monitoring of health status. This trained model was then integrated with a monitoring system, creating a comprehensive solution for predicting the lifespan and tracking the health of mechanical parts. Finally, the system underwent continuous optimization and updates to improve both its prediction accuracy and real-time responsiveness, while also adapting to various operating conditions and environmental factors. The results demonstrated that the deep learning model achieved a mean absolute error (MAE) of 2.1, a root mean square error (RMSE) of 2.5, and a mean absolute percentage error (MAPE) of 10%, reflecting strong performance. This approach holds significant potential for practical application in the engineering field.

Abstract-This study offers a comprehensive investigation into the field of gait recognition in biometric analysis, focusing on the specific challenges associated with using gait as a biometric feature. The research evaluates various machine learning (ML) algorithms, including Individual Node Evaluation, Statistical Inference, Regression Modelling, Support Vector Machines, Nearest Neighbour Classification, Decision Tree Structures, Random Forest Ensembles, and Multilayered Neural Networks. Thorough testing is conducted to assess the performance of each model in accurately identifying individuals based on their unique gait characteristics. The approach emphasizes extensive preprocessing to maintain data quality and relevance. Additionally, Sequential Backward Selection (SBS) is employed for feature selection, along with dimensionality reduction techniques such as Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA), which improve the model’s focus on key features. The research also investigates deep learning models, analysing different architectures to assess their effect on gait recognition accuracy. A detailed comparative analysis evaluates the advantages and limitations of each method, providing valuable insights for the field. By exploring a variety of ML and DL approaches, this study sets a benchmark for future developments in biometric security. It highlights the potential of gait recognition as a reliable, non-invasive identification method, paving the way for the creation of more advanced and precise biometric systems that are crucial for the evolving needs of security and personal identification.

Abstract-Artificial intelligence holds significant potential to revolutionize healthcare. Machine learning (ML) and deep learning (DL) techniques have been increasingly utilized for predicting and diagnosing a wide range of diseases. In addition, social media platforms such as Twitter, Facebook, and Reddit have become popular outlets for individuals to share their emotions and experiences. Following the COVID-19 pandemic, mental health concerns have escalated, prompting numerous studies that apply ML and DL models to analyses social media data for predicting mental health issues. This research aims to offer an in-depth review of the ML and DL algorithms applied to the prediction of various mental health disorders. It presents an extensive overview of 37 research papers, analysing and compiling a table of the accuracy of these algorithms across four key mental health conditions: Depression, Anxiety, Bipolar Disorder, and ADHD. The study is intended to serve as a foundational resource for future researchers and practitioners, offering insights into the performance of different ML and DL approaches. Additionally, it includes a compilation of publicly available datasets, providing valuable resources for ongoing research in this area.

Abstract-Effective energy management is essential for promoting sustainability, reducing carbon emissions, conserving resources, and cutting costs. However, traditional energy forecasting methods often fall short in terms of accuracy, indicating a need for more advanced solutions. Artificial intelligence (AI) has emerged as a valuable tool for energy forecasting, but its lack of transparency and interpretability makes it difficult to understand its predictions. To address this, Explainable AI (XAI) frameworks have been developed to enhance the transparency and interpretability of AI models, particularly those considered “black-box” models. This paper examines household energy consumption predictions by comparing various forecasting models using evaluation metrics such as the Coefficient of Determination (R2), Root Mean Squared Error (RMSE), Mean Squared Error (MSE), and Mean Absolute Error (MAE). After testing with unseen data, the best-performing model is selected, and its predictions are explained through two XAI techniques: Local Interpretable Model-Agnostic Explanations (LIME) and Shapley Additive Explanations (SHAP). These methods help identify key factors influencing energy consumption forecasts, such as current consumption patterns and previous energy usage. The study also highlights the importance of XAI in developing predictive models that are both reliable and consistent.

Abstract-Banks are essential to the global financial system, and one of their primary sources of income comes from loan interest. However, if borrowers fail to repay these loans, it can turn profits into substantial losses, highlighting the importance of assessing the risk of default before approving a loan. Machine learning techniques can be an effective method for quickly and accurately evaluating whether a credit risk should be approved. This study explored six machine learning models—Decision Tree, Random Forest, Support Vector Machine (SVM), Multi-layer Perceptron (MLP) Artificial Neural Network, Naive Bayes, and a stacking ensemble model—to predict the credit risk associated with a loan. Using a dataset of twenty factors typically found in loan applications, the stacking ensemble model achieved the highest accuracy at 78.75%. The Random Forest model, though slightly less accurate at 78.15%, was more efficient while yielding comparable results. Key factors such as credit amount, account status, age, loan duration, and loan purpose were identified as the most influential indicators of credit risk. The findings of this research further support the efficacy of machine learning models for predicting loan default risk.

Abstract-The spread of fake news has become a significant concern in today’s society, as misleading information can easily damage reputations and lives. To address this issue, researchers have developed fake news detection systems using machine learning techniques. The identification of fake news is rapidly gaining traction and is increasingly being adopted by various industries, either for their own use or to offer as a service to others. Machine learning (ML) and deep learning (DL) are two prominent approaches employed to determine the authenticity of news. There are various methods available for detecting false news through both ML and DL techniques. This paper presents a comprehensive analysis of fake news detection using machine learning approaches. Upon thorough examination, it was found that several ML and DL algorithms have been applied in this domain, with the Support Vector Machine (SVM) being the most commonly used ML method, and Long Short-Term Memory (LSTM) being the most widely applied DL technique.