Ransomware continues to pose a critical threat to modern information systems, particularly due to the increasing prevalence of polymorphic and behaviorally evasive variants. This paper proposes a reproducible, forensically auditable machine learning pipeline for multi-class ransomware family classification based on dynamic behavioral features. Using the RanSAP-2022 dataset comprising ten ransomware families, the pipeline automates data ingestion, exploratory analysis, preprocessing, and model evaluation under deterministic controls. Exploratory Data Analysis reveals significant class imbalance, with the largest family accounting for 32.4% of samples and an imbalance ratio of 4.6:1 (Fig. 1), as well as 27% aggregate missingness concentrated in network-related features (Fig. 2). Feature correlation analysis (Fig. 3) identifies moderate behavioral coupling among file system, registry, and API activity features without severe multicollinearity. Six machine learning models are comparatively evaluated using macro-averaged F1-score as the primary metric. Boosting-based ensemble models achieve the strongest performance, with XGBoost attaining 96.1% accuracy and a macro-F1 score of 0.948 (Table 7). Confusion matrix analysis (Fig. 4) demonstrates high precision and recall for majority families, while highlighting reduced recall for minority families due to behavioral overlap. These findings confirm that behavior-based ensemble learning, when embedded in a reproducible and leakage-resistant pipeline, provides both high predictive performance and forensic reliability for ransomware family classification.

Keywords: Ransomware classification; behavioral malware analysis; machine learning; reproducible pipelines; digital forensics

Generative AI (GAI) has now been widely integrated into higher education. While its impact on cognitive abilities (cognitive offloading) has been well-documented, its influence on creativity is less well-understood. This study analysed 12 video game design projects created by 36 second-year computer science students using generative AI tools. Despite the creative power of the tools, the results reveal a strong thematic convergence, with 50-58% of projects set in post-apocalyptic worlds, 75-83% featuring male protagonists and conventional narrative structures. This article discusses the pedagogical implications of this finding and proposes ways to integrate GAI judiciously.

Keywords: generative AI; creativity; cognitive offloading; creative homogenisation; game design; higher education

With the advancing time more and more vehicles usage are increasing. The demand for vehicle manufacturing companies for the production are increasing. However, these vehicles use fossil fuels (petrol or diesel) for operation. As it is known that these fossil fuels are limited in quantities. Hence alternative options have come to use electric vehicles which uses electric power to store energy in batteries. The electric power can come from either renewable or non renewable sources. But these electric vehicles have its own merits and demerits.

This research paper introduces two new constraint types and four subtypes of database constraints added to our (Elementary) Mathematical Data Model, which are the duals of the existence and non-existence ones. They are formally defined, characterized, and exemplified with real-life instances. The well-formedness, satisfiability, coherence, and minimality of sets of all 7 subtypes of existence constraints are studied. Corresponding SQL-embedded pseudocode algorithms for managing such sets are provided and proved to be of constant complexity, sound, complete, and optimal. Also provided are algorithms for enforcing these new types of constraints, called inexistence and anti-existence. Their characterization proves that they have linear complexity in the sum of the arities of the function (Cartesian products) involved, and are sound, complete, and optimal as well. All these algorithms were implemented in both versions of our intelligent data and knowledge base management system prototype MatBase, which automatically generates code for enforcing all the 7 subtypes of existence constraints.

Keywords: (Elementary) Mathematical Data Model; MatBase; anti-existence constraint; inexistence constraint; existence constraint; non-existence constraint; conceptual data modeling; database design; non-relational constraints enforcement; null values management

In digital contact tracing, a user’s device exchanges proximity data with nearby devices, and this data includes identifiers associated with the user. When a user tests positive, the proximity data stored on their device is analysed to identify others who may have been exposed. Once these individuals are identified, manual contact tracing is initiated for further follow-up. Multiple contact-tracing solutions exist today: some are designed for closed user groups, some for national populations, and others are open systems without clearly defined boundaries. However, these isolated solutions are insufficient for a world with high levels of global mobility. Ideally, a contact-tracing system should be global, interoperable, and capable of supporting seamless mobility across regions. Designing such a system presents several challenges, including scalability. In this study, we propose a proximity-data encoding approach that enables scalable centralized, distributed, and federated contact-tracing architectures. Our federated and distributed models support the coexistence of diverse identifiers and proprietary encryption methods. The federated approach enables user mobility while allowing each country to maintain control over its citizens’ data.

Keywords: Contact Tracing; Encoding; TraceTogether; Exposure Notification; Proximity Identifier; Federated; Distributed; Bluetooth Low Energy; Global mobility

In recent years, bioinformatics has emerged as a transformative force in drug development, streamlining processes that were once laborious and resource-intensive. By integrating computational tools with biological data, bioinformatics enables researchers to analyze complex biological systems, predict drug interactions, and accelerate the discovery of new therapeutic agents. As the demand for personalised medicine grows, bioinformatics has become an indispensable tool for ensuring more precise, effective, and safer drug development.

To prevent the disease and degradation of health and to cure the diseases are the two main objectives of Ayurveda. Ayurveda is one of the ancient holistic healing pathy arouse in India about five thousand years back. It harbours treatments for several diseases, ways to live an effective and longer life, ways to attain good health, to lead a prosperous and happy life, to overcome the seasonal variations, to protect nature, and to attain one's higher self. In a nutshell it encompasses a healthier and sustainable means of living a splendid life.

Water scarcity and inefficient irrigation practices remain pressing challenges in Indian agriculture, particularly in rural areas with limited internet connectivity. Existing smart irrigation solutions predominantly rely on cloud-based processing and continuous network availability, making them unsuitable for remote farming environments. This paper presents SmartIrrig, a cost-effective, edge-based intelligent irrigation system designed for small-scale agricultural use. The proposed architecture integrates soil moisture sensing, real-time weather condition monitoring, and a local decision-making engine to autonomously control water pump operation without dependence on cloud services. A predictive irrigation module leverages threshold-based logic enhanced by lightweight rule-based reasoning to anticipate water requirements before critical soil dryness occurs. The system provides farmer-friendly feedback through an LCD display and optional mobile application. Evaluated through controlled field trials under varied soil and weather conditions, SmartIrrig demonstrates 34% reduction in water consumption compared to conventional timer-based systems, with an average pump decision latency of under 200 milliseconds. The system's offline-first design addresses the primary limitation of existing solutions, establishing edge computing as a viable paradigm for agricultural IoT in resource-constrained environments.
Keywords: smart irrigation; IoT; edge computing; soil moisture sensing; precision agriculture; embedded systems; water management; Arduino

Chronic Kidney Disease (CKD) is a progressive condition characterized by a gradual decline in kidney function, leading to the accumulation of metabolic waste and fluid imbalance. The leading cause of CKD is persistent hyperglycemia due to diabetes mellitus. High blood glucose levels damage renal microvasculature, particularly the glomeruli, resulting in proteinuria. Hyperfiltration, intraglomerular hypertension, and the formation of advanced glycation end products (AGEs) further exacerbate renal injury through inflammation, fibrosis, and vascular damage. Hypertension, often coexisting with diabetes, accelerates CKD progression by increasing intrarenal pressure and perpetuating a cycle where elevated blood pressure and renal dysfunction worsen each other. This vicious cycle significantly heightens the risk of cardiovascular events, including myocardial infarction and stroke.
In the Greco-Arabic (Unani) medicine, CKD is described as a chronic imbalance in the body’s humors (Akhlat)—particularly involving a derangement of the kidneys' cold and moist temperament (Mizaj). Conditions such as Zauf-e-Kulya (renal weakness), Daffe-e-Baul ka ikhtilal (urinary dysfunction), and Istisqa (dropsy) accentuate the manifestations of CKD. Scholars like Hippocrates, Galen, and Avicenna emphasized the role of humoral imbalance and renal purification in systemic health. Avicenna notably linked Fasad-e-Mizaj (temperamental derangement) with renal dysfunction and advised dietary regulation and herbal interventions (Muqawwi-e-Kilya) for prevention and management of CKD. This integrated perspective highlights the importance of understanding CKD both through modern pathophysiological mechanisms and traditional holistic frameworks for comprehensive care.
Keywords: Chronic Kidney Disease (CKD); Advanced Glycation End Products (AGEs); Humoral Imbalance; Dietary intervention and Greco-Arabic management

Background: Hospital management remains a major challenge for ensuring quality care in a context marked by multiple health issues. This study aimed to identify the components of hospital management and assess the practices and approaches implemented at the District Referral Hospital Center (DRHC) of Moramanga.
Methods: An evaluative, descriptive, cross-sectional study with a mixed-methods approach was conducted using retrospective data from 2022 to 2024. All staff members with at least six months of seniority were included. Data were collected through standards-based assessment grids and staff interviews, then analyzed according to the Donabedian model focusing on structure, process, and outcomes of care.
Results: Results showed a very high budget execution rate of approximately 99.5% over the study period, reflecting effective resource management despite a slight decline. Equipment coverage was satisfactory overall, reaching 80.94%. Reception and patient intake services were adequately organized, with six of eight required standards available, including triage systems, dedicated staff, and structured reception areas. Hospital activity trends revealed marked increase in outpatient consultations in 2023 followed by a slight decline in 2024. Efficiency indicators showed a stable average length of stay of around 2.4 days, while the bed occupancy rate slightly decreased from 20.6 in 2022 to 19 in 2023. The hospital utilization rate also progressively declined from 0.8% to 0.52%.
Conclusion: The need to strengthen medical staffing and increase revenue through investments, particularly by creating improved inpatient accommodation conditions, is evident in order to improve the quality of care and hospital performance, while taking into account the specific needs of patients and staff.
Keywords: donabedian; hospital; management; madagascar

Spam detection remains a critical challenge in natural language processing (NLP) and cybersecurity, with over 50% of global email traffic consisting of unwanted messages. This comprehensive study presents an extensive comparative analysis of machine learning algorithms and text vectorization techniques for spam classification, evaluating seven distinct machine learning models across four feature engineering approaches using multiple large-scale datasets comprising over 15,000 messages. Our experimental results demonstrate that XGBoost achieves the highest overall performance with 94.4% accuracy and 95.4% precision, while ensemble methods consistently outperform traditional approaches by 5-7%. The research reveals that text vectorization techniques show minimal performance variance (less than 0.3% accuracy difference), with Bag of Words (BoW) achieving slightly superior results at 87.9% accuracy. These findings highlight the critical importance of algorithmic sophistication over feature complexity for spam detection systems, providing evidence-based guidance for practical deployment in cybersecurity applications. The study contributes novel insights into ensemble method superiority and establishes comprehensive benchmarks for spam detection research.
Keywords: Spam Detection; Machine Learning; Ensemble Methods; XGBoost; Random Forest; Text Vectorization; Cybersecurity; Natural Language Processing

The rambling-trembling decomposition of the center-of-pressure (CoP) trajectory during quiet standing separates slow supraspinal drift (rambling) from fast peripheral oscillation (trembling). The decomposition is empirically robust, has substantial clinical reproducibility, and is sensitive to cognitive load, aging, and neurological disease, yet a quarter century after its introduction [Zatsiorsky and Duarte, 2000] it remains mechanistically unexplained. Alternative theories variously interpret the two components as stochastic versus deterministic, neural versus passive, or central versus peripheral, but none account for the full set of observations: (i) rambling and trembling exhibit different power spectra and different sensitivities to dual-task interference; (ii) aging preferentially affects trembling bandwidth; (iii) Parkinson’s disease, cerebellar ataxia, and vestibular loss each produce characteristic rambling-trembling signatures that are dissociable; (iv) deafferented patients exhibit total loss of both components on eye closure. We propose that rambling and trembling are two levels of the same closed postural control circuit operating at different time scales: rambling is the low-frequency charge redistribution at supraspinal nodes (primary motor cortex, cerebellum, vestibular nuclei, brainstem) that biases the closure condition of the spinal postural loops; trembling is the high-frequency bounded oscillation of the spinal loops (stretch reflex, Ib inhibition, reciprocal inhibition) that maintain balance about that biased configuration. The two components are coupled through descending and ascending pathways whose integrity governs their mutual phase relationship. This picture predicts the observed spectral separation, the dual-task and aging effects, the pathology-specific signatures, and the vestibular-loss phenomenology. We develop the prediction quantitatively, specify a multi-sensor wearable protocol (smartphone/smartwatch inertial measurement unit [IMU], photoplethysmography [PPG], wrist-worn electromyography [EMG], ankle-mounted force-sensing insole) for non-invasive extraction of both components, derive consistency conditions across sensors, and report validation results from numerical simulations that reproduce the spectral structure of human postural sway, the dual-task effect (40-55% sway increase under cognitive load), the rambling-trembling spectral separation (> 90% correlation with ground-truth decomposition), and the predicted signatures of deafferentation, Parkinsonian rigidity, and cerebellar ataxia. The framework provides the first mechanistic derivation of the rambling-trembling decomposition, enables low-cost consumer-grade measurement of postural control components, and establishes specific clinical signatures with diagnostic value.
Keywords: postural control; rambling; trembling; quiet standing; wearable sensors; inertial measurement units; center of pressure; closed-loop motor control; dual-task; Parkinson’s disease; cerebellar ataxia; consumer health technology

We are becoming more and more dependent on digital communication, which makes protecting data from cyber threats a top priority. Threats from quantum computing, complex cyberattacks, and weaknesses in IoT devices are only a few examples of the modern security issues that traditional cryptographic models and network security protocols frequently fail to handle [1, 2]. Cybercriminals are using vulnerabilities in antiquated security systems more frequently as technology develops, which makes the need for creative defence tactics to grow. The Hybrid Adaptive Security Model (HASM) proposed in this study integrates cutting-edge cryptographic innovations such blockchain-based authentication methods, artificial intelligence (AI)-driven anomaly detection, and quantum-resistant encryption [3, 4]. HASM provides a multi-layered, dynamic defence strategy that adjusts to threats in real time while preserving system confidentiality and integrity. Post-quantum cryptography (PQR) algorithms like New Hope, machine learning methods like Isolation Forest for anomaly detection, and smart contracts for decentralised identity verification are some of the elements of this architecture that we develop and assess. When compared to traditional systems, the testing results demonstrate a significant improvement in threat detection accuracy (96%), increased encryption strength, and decreased susceptibility in authentication processes. According to the survey, there is an increasing demand for a unified cybersecurity framework that uses cutting-edge technologies to safeguard private information in increasingly intricate digital environments. For future applications in industries where security is crucial, HASM provides a solid framework.
Keywords: Post-Quantum Cryptography (PQC); AI Security; Blockchain Authentication; IoT Security; Hybrid Security Framework; HASM

This study demonstrates that applying artificial intelligence (AI) to chemical-mechanical planarization (CMP) process control significantly reduces rework rates and enhances manufacturing efficiency. We propose a hierarchical AI CMP controller featuring an auto-tuning capability. This integrated hierarchical reinforcement learning (iHRL) framework is designed to mitigate CMP variations arising from removal rate decay and pattern density fluctuations. The iHRL agent utilizes a structured wafer-to-wafer and lot-to-lot control architecture. Results demonstrate that the self-learning auto-tuning mechanism effectively minimizes the CMP rework rate. Furthermore, comprehensive simulations indicate that the proposed controller enhances average manufacturing efficiency by more than double digits compared to the conventional run-to-run controller.
Keywords: semiconductor manufacturing; chemical mechanical planarization; artificial intelligence

There is a concern whether social networks are the beast as envisioned by the eschatologist Jean, greedy of social interactions in practical terms rather than simply connecting better humans in the real world. This paper gives insights on the improbability that Internet social networks eased the emergence of a new bright connected planet empowering technology to brighten every horizon of the planet, on the shoulders of two decades of practicing social networks from Hi5 to TikTok through Facebook/Instagram/Threads via Orkut. Our results reveal that the interactions facilitated by these social networks are very real. However, these social networks sometimes generate an overload of information seriously affecting the analysis and attention capacity of users of various digital literacies. This is particularly the case of Youtube, and worse TikTok, on which we rely to illustrate this fact by gathering a set of data from the activity of @hanifrabbanihaqqani account private datasets, broadcasting since 2023g, 1445h as the contemporary imam of the rawdah, streaming from french maturity without precedent-ness recorded.