This study presents a hybrid federated learning and explainable artificial intelligence framework, termed RobustFL, designed to mitigate adversarial attacks in medical imaging while incorporating privacy-preserving mechanisms. Focusing on chest X-ray analysis, the research systematically examined adversarial vulnerabilities and data privacy challenges in distributed healthcare systems. A simulation-based methodology was implemented using a 500-sample dataset partitioned across multiple clients in a non-independent and identically distributed setting. The framework extends federated averaging by integrating projected gradient descent-based adversarial training, differential privacy stochastic gradient descent, and SHAP-based interpretability. Experimental results demonstrate that RobustFL maintains stable performance under adversarial conditions, achieving 70% accuracy against FGSM and PGD attacks, with an adversarial success rate of approximately 30–31%. A balanced privacy budget of ε = 1.0 produced an F1-score of 0.412, indicating a trade-off between privacy and model utility. Interpretability consistency, measured via SHAP outputs, served as a proxy for trust assessment, while privacy risk indicators were used to infer potential user concern. The framework provides a practical pathway toward secure, transparent, and privacy-aware healthcare AI systems.

The Residue Number System (RNS) is a non-positional number system that represent number in the form of residues modulo a set of coprime moduli. This nature of RNS makes it advantageous in achieving carry-free arithmetic. However, due to the non-positional nature of RNS, direct number comparisons to determine the magnitude of RNS numbers is nearly impossible. In this paper, a magnitude comparison scheme is proposed and is theoretically proven and evaluated. The proposed scheme compares the relative magnitude of any two RNS numbers via sign detection. It first computes the modular differences of the residues of the two RNS numbers, and , and then computes the sum of the results, ,which is then proven to be equal to zero. The final implementation of the technique requires the reverse conversion of . Then sign detection is performed on and the results determine whether or . Further analysis proves that the proposed scheme accurately compares the magnitude of any two RNS numbers, whether a signed or unsigned number. The proposed scheme was evaluated and compared with another state-of-the-art scheme based on three key matrices, Area, Delay and Area-Delay product. The proposed scheme outperformed the state-of-the-art scheme in all three-matrix evaluations.

Customer lifetime value is an important measure for understanding long term profitability in e-commerce, where firms depend on effective customer retention and development in highly competitive digital markets. Traditional statistical approaches have supported early customer lifetime value estimation, but they often rely on assumptions that limit their ability to represent complex and changing customer behavior. This systematic review examined existing studies that applied machine learning methods to customer lifetime value prediction in e-commerce. Guided by the PRISMA framework, the review searched Scopus and IEEE Xplore for studies published between 2020 and 2026, and 12 studies met the inclusion criteria. The findings show that ensemble and deep learning approaches dominate recent research, with stacking, hybrid, and recurrent neural models frequently reporting strong predictive performance. However, the evidence also reveals important limitations, including restricted data accessibility, limited interpretability, high computational demands, and inconsistent evaluation methods, which reduce comparability across studies. The strong predictive performance of these models suggests their potential to support more effective customer segmentation and improved allocation of marketing resources. Future progress in customer lifetime value prediction will depend not only on improved model performance, but also on more transparent methods, publicly available benchmark datasets, and stronger attention to interpretability and practical applicability.

Accurate predictive models are essential for analysing COVID-19 mortality trends. This study evaluates the effect of a novel customised data pre-processing pipeline on ten machine learning models predicting COVID-19 mortality in India, using data from Our World in Data (OWID). Our pipeline diverges significantly from a standard pre-processing pipeline in four main steps. First, it corrects administrative reporting delays by converting weekly reported totals into daily figures, reducing reporting biases and delivering precise estimates. Second, it applies momentum-preserving local outlier processing to retain data variability and improve accuracy. Third, it enforces mathematical relationships among features to ensure strict data consistency. Finally, it integrates an iterative feature selection process to optimise the feature set and boost model performance. Results demonstrate considerable improvement with the custom pipeline: the MLP Regressor achieved a test RMSE of 83.663 and a test R² of 0.986, outperforming the Gradient Boosting Regressor from the standard pipeline, which had a test RMSE of 171.525 and a test R² of 0.859. A primary contribution of this research is the formal validation of model stability via a newly introduced diagnostic: RMSE Variance. This metric quantifies the consistency of predictive performance across multiple iterations, distinguishing authentic generalisability from stochastic success. These results highlight the necessity of tailored pre-processing, providing a transferable framework for global health datasets facing non-stationary noise and reporting inconsistencies.

Telemedicine requires the provision of medical images over unstable networks in a secure manner. Traditional hybrid cryptographic algorithms e.g. ECC-LSB watermarking provide confidentiality, but are often limited by a large computational cost hence limiting its use to real time. This paper, in turn, offers a lightweight security architecture that integrates Least Significant Bit (LSB) watermarking, Elliptic Curve Integrated Encryption Scheme (ECIES) and the Advanced Encryption Standard in Counter Mode (AES-CTR) to provide fast and secure encryption of medical images. The red channel of every image is watermarked through LSB watermarking to store Patient Electronic Health Records (EHR), and the RGB channels are encrypted with AES-CTR independently, making it possible to perform parallel processing. ECIES is used to transfer symmetric encryption keys. The security of the proposed framework is evaluated in terms of confidentiality, integrity preservation, and resistance to common attacks. Experimental results, validated against conventional ECC-based LSB watermarking schemes, demonstrate superior image quality with MSE of 0.0022, PSNR of 74.64 dB, and UIQI of 1.0, while achieving encryption and decryption times below 0.3 seconds, throughput at 1.51MB/s and 1.91MB/s respectively which makes it suitable to the real-time telemedicine and Picture Archiving and Communication System (PACS) settings.

This study critically evaluates WhatsApp’s security ecosystem through a shared responsibility framework, examining the interaction between platform safeguards, usability, and user accountability. A mixed-methods approach was employed, integrating four open-access cybersecurity datasets with primary data from a structured questionnaire completed by 835 respondents and semi-structured interviews with five members of an alumni WhatsApp group affected by an account breach. Analytical techniques included sentiment–feature mapping using logistic regression, risk distribution profiling with Relative Risk Ratios, exploratory factor analysis with Cronbach’s Alpha, multivariate time-series correlation with Granger causality, and thematic analysis of interview transcripts. Results indicate that two-step verification recorded the strongest positive sentiment and highest odds ratio (OR = 2.34), yet adoption remained low, with fewer than 40 percent of users enabling the feature. Investment fraud accounted for 61 percent of high-loss scam cases, while questionnaire responses revealed persistent engagement in risky behaviours such as joining unverified groups. Qualitative findings further demonstrate that trust dynamics, misunderstanding of security prompts, and limited practical awareness significantly contributed to the observed account breach. Behavioural accountability constructs explained 80 percent of variance in security outcomes; however, only 24 percent of users consistently exhibited high accountability. Platform-level interventions such as key transparency were associated with reductions in banned accounts (r = −0.62, p = 0.04), while spikes in abuse incidents triggered reactive increases in two-step verification adoption. These findings underscore the limitations of encryption-centric security models and highlight the need for improved usability, targeted awareness strategies, and strengthened user accountability. The study provides actionable insights for platform designers, policymakers, and users seeking to enhance digital resilience in encrypted communication environments.

Flood events are becoming more frequent and severe with rapid urbanization and climate change. With rising risks, accurate real-time assessments are critical for disaster management and operational decision-making, such as emergency response and resource allocation. Implementing artificial intelligence (AI), particularly deep learning (DL), has advanced the automation of estimating the damage from floods using satellite and sensor data. However, despite strong results in controlled environments, these models often suffer significant performance degradation under real-world conditions. This article assesses the state of the art in using AI to measure the severity of floods, paying specific attention to the robustness of the models and their general applicability to various environmental conditions. More than fifty primary studies published from 2019 through 2024 were analysed within the systematic literature review framework. The review proposes a diagnostic framework centered on three interconnected domains of difficulty: the quality and diversity of data, the model structure and ‘interpretability, and the authenticity of the environment, which includes turbidity, meteorological interference, and intricate topographical challenges. Results show that predictive reliability decreases significantly under sensor noise and unseen environmental conditions, although benchmark performance is usually high. There is a lack of analytic tools to measure the degree of these shortcomings. This review focuses on the growing “robustness gap” in artificial intelligence literature on flooding. It presents a vision for developing precise, adaptable, and functionally resilient AI capable of reliable deployment in operational flood disaster management.

This study explores the application of gradient descent optimization and loss functions for depression prediction through multidimensional analysis of behavioral, psychological, and social factors. A logistic regression model was trained on a set of 1,631 records. Experimental results indicate an overall accuracy of 91.17 %, a weighted precision of 0.92, a recall of 0.91, and an F1 score of 0.91. These performances demonstrate the robustness of the model and its potential for early detection of depression. This research highlights the role of artificial intelligence in decision support and prevention in mental health. This article explores in depth the gradient descent algorithm, a fundamental pillar of machine learning, focusing on its essential interaction with the loss function (or cost function). Far from a simple theoretical analysis, this talk applies these fundamental concepts to a crucial public health problem: the prediction of depression. Depression is a complex disorder, and its diagnosis and early prediction remain major challenges. To develop reliable predictive models of depression, it is imperative to identify and integrate a diverse set of relevant factors. We will rely on a set of 24 input variables, grouped into 8 symptomatic and contextual categories (such as concentration difficulties, feelings of guilt, sleep disturbances, intense fatigue, etc.), to build our predictive model. The challenge is to navigate the space of these multidimensional parameters in order to find the optimal combination minimizing the prediction error of depression. This paper details how gradient descent, guided by a carefully chosen loss function, addresses this prediction challenge: Modeling of input variables, Application and optimization of gradient descent and Convergence and performance analysis. By integrating these specific variables into a rigorous gradient descent optimization approach, this paper will seek to demonstrate the feasibility and effectiveness of a machine learning model for more accurate and meaningful prediction of depression.

Criminal activities in the Kaduna-Abuja Highway have been an enormous security risk to the commuters, residents, and economic growth in Northern Nigeria. Conventional law enforcement approaches are still reactive and lack data, which restricts the anticipatory and preemptive actions. This paper describes Data Mining Analytics Approach-Based System of forecasting the occurrence of crime along the Kaduna-Abuja highway, using developed models of machine learning. The analysis used three supervised learning algorithms which include Support Vector Machine (SVM), K-Nearest Neighbor (KNN) and Random Forest (RF) to represent crime patterns and predict potential types of crime, time and location. These algorithms were selected because they are widely used for classification problems, handle and have shown strong performance in similar prediction studies. Artificial data of 1,500 entries that cited five years of criminal incidences was created and prepared. Class imbalance was dealt with with the help of SMOTE (Synthetic Minority Oversampling Technique), hyperparameters optimization with the help of the GridSearchCV. The experimental outcomes showed that the KNN classifier performed best having an accuracy of 71% then SVM and Random Forest with 66.5% respectively. Analysis of the importance of features showed that the most significant predictors were month, day of the week, number of victims, year, and time of day. Study evidences that predictive analytics based on data can complement the proactive policing practices and help the security agencies to predict the high-risk times and locations along the transport corridors in Nigeria that are critical.

Globally, lot quantities of data exist in medical, industry and other related fields. An interested area of research in bioinformatics a case study in medicinal plants is the application and improvement of data mining techniques to solve biological problems involved. The quantity and variety of data in natural sciences enhanced quickly. Therefore, data abstraction, data manipulation and pattern discovery techniques are require in dealing with such large varieties of medicinal plants. Analyzing huge biological medicinal plants’ data sets requires making sense of the data by inferring structure or generalizations from the data. Integration among different sources of data is likewise of most important interest, as complex relations may happen. In this research paper, some basic concepts of bioinformatics, data mining and WEKA explorer tool are discussed. This research supports the development and validation of herbal treatments by comparing their efficacy with orthodox medicines that enhances decision making for herbal practitioners through structured databases and software tools that provide accurate treatment recommendations. Also it strengthens the integration of bioinformatics with herbal medicine by facilitating gene expression analysis, sequence interpretation, and knowledge discovery, ultimately promoting more efficient herbal drug development and application which is the main goal. Therefore, the area of application of data mining in the field of bioinformatics was clearly stated in this research and the current opportunities and challenges of data mining in medicinal plants bioinformatics are discussed as well.