Volume 7 Issue 3 2026

Machine Learning has massive influence in the automotive industry whichleads to evolution of intelligent routing for traffic management in wireless network via autonomous vehicles(AVs). The evolution of numerous automotive platforms for efficient traffic management has been the current trend.A method called, Bayesian Neural Visual Odometry and Polynomial Regression (BNVO-PR) with vehicle to vehicle communication for traffic management ensuring intelligent routing for autonomous driving is proposed. This method has three sections, namely, perception, localization and prediction.Spatio-Temporal Motion Control-based Perception is applied to the raw Udacity Self Driving Car dataset provided as input to obtain robust point of interest object detection with which intelligent routing can be ensured. Second, the obtained point of interest results are subjected to Bayesian Neural Visual Odometry-based Localization, ensuring computationally efficient object recognition for significant traffic management. Finally, with the recognized objects obtained, Polynomial Regression-based Autonomous Vehicle Prediction is designed to interpret actuate kinematic maneuvers in AVs, ensuringintelligent routing for efficient traffic management in wireless network concurrently. There is a reduction inrouting overhead and performance evaluation like precision, routing accuracy and routing time during autonomous driving using our network when compared with the existingmethods, leads to higher routing accuracy during inference, achieving accurate autonomous driving.

Background: Social media has become a routine channel through which patients and the general public seek, share, and interpret health information, with potential influence on treatment choices, provider selection, and health behaviours, alongside persistent concerns about information reliability. Methods: A cross-sectional descriptive study was conducted at Prem Niketan Hospital, Jaipur, and among the general public residing within a 3 km radius of the hospital. Quantitative data were collected from 150 respondents (75 hospital patients and 75 from the general public) using a pre-tested semi-structured questionnaire containing 25 items (multiple-choice, Likert-scale, and open-ended questions). Data collection occurred between April 15 and May 15, 2024, using both online and offline modes, and analysis relied on descriptive statistics (frequencies, percentages, means, standard deviations). Results: The findings indicate that social media exerts a substantial influence on healthcare decision-making, including treatment decisions, choice of healthcare providers, and health behaviours. At the same time, respondents expressed scepticism about the reliability of health information available on social media, with perceived credibility shaped by source credibility, user reviews/comments, shares/likes, and visual content quality. Key challenges reported include misinformation, lack of credible information, information overload, and privacy concerns, while opportunities include peer support, exposure to diverse opinions, and timely access to health-related information. Conclusion: Social media functions as a high-reach but uneven-quality health information ecosystem. Strengthening health literacy, improving professional presence on social platforms, developing guidance for trustworthy information sharing, and enabling fact-checking and public awareness initiatives are central to improving patient engagement and supporting more informed healthcare decisions.

Digital transformation has significantly reshaped business operations, particularly in Business-to-Business (B2B) sales processes. Traditional B2B selling relied heavily on personal relationships, face-to-face meetings, and relationship-based negotiations. However, the rapid adoption of digital technologies such as Customer Relationship Management (CRM) systems, artificial intelligence (AI), big data analytics, cloud computing, and digital communication platforms has transformed the nature of buyer–seller interactions. These technologies enhance operational efficiency, improve communication, enable data-driven decision-making, and allow firms to deliver personalized customer experiences. This study examines the impact of digital transformation on B2B sales processes and its influence on buyer– seller relationships. The research explores how digital tools affect relationship quality, trust, communication, and long-term collaboration between organizations. A quantitative research design is adopted using a structured questionnaire administered to B2B sales professionals and organizational buyers. The collected data are analyzed using statistical techniques such as descriptive analysis, correlation, and regression to examine the relationship between digital transformation and relational outcomes. The findings are expected to demonstrate that digital transformation significantly improves sales efficiency, transparency, and communication while simultaneously altering traditional relationship dynamics in B2B markets. While digital platforms facilitate faster information exchange and customer engagement, the absence of personal interaction may reduce relational depth if not managed effectively. The study highlights the importance of adopting hybrid sales models that integrate digital technologies with human interaction to maintain trust and long-term partnerships. The research contributes to the growing body of literature on digital sales transformation and provides strategic insights for organizations seeking to enhance customer relationships in increasingly digital B2B environments.

Tomato leaf diseases are critical factors affecting tomato yield and quality, yet their accurate identification remains challenging in complex field environments. This study aims to develop an improved YOLOv12 model for robust tomato leaf disease recognition. First, we constructed and annotated a comprehensive multi-environment dataset covering 10 common leaf diseases. To address YOLOv12's limitations, we introduced three key innovations: (1) Incorporating a Self-Attention (SE) module to enhance disease feature representation in cluttered backgrounds; (2) Replacing standard convolutions with GhostConv to reduce computational load while preserving feature discriminative power; (3) Adopting a scale-adaptive WIoU_v2 loss function to optimize gradient distribution across varying data quality. Ablation studies confirm these improvements synergistically enhance the model's adaptability to varying disease scales and environmental conditions. The refined model achieves a 0.9% mAP@0.5 improvement and a 1.9% mAP@0.5:0.95 increase compared to the original model, while reducing computational demands. The proposed system achieves an optimal balance among detection accuracy, inference speed, and lightweight design, offering a novel solution for automated tomato disease recognition.

Background: The Personal Evolution Model (PEM) utilizes kingdom themes and homeopathy to explore their combined potential in improving treatment strategies. PEM establishes a connection between a patient's inner age and their actual biological age, with specific issues manifesting as physical symptoms of disease. Aim: To access the patient’s inner age and their biological age by using homeopathic treatment. Methods: PEM framework offers a unique perspective for investigating the origins of various clinical conditions, including endocrine disorders. Homeopathy, characterized by its holistic and individualized approach, takes into account both mental and physical symptoms, as well as the overall patient experience. Results: The advantages of integrating the nuanced concept of inner age, as expressed through the analysis of kingdom themes in PEM, into the broader, symptom-focused methodology of homeopathy. The goal of this integration is to enhance the thoroughness of patient evaluations, particularly regarding issues related to inner age. The findings suggest that while PEM, through kingdom charts, can yield significant insights, it should serve to complement rather than replace the comprehensive strategies employed in homeopathic treatment. Conclusion: Present study recommended to assess the effects of combining PEM with homeopathic practices on patient outcomes, especially in the context of endocrine disorders, thereby supporting and potentially advancing the tradition of holistic and personalized patient care.

Fetal movement monitoring is important to measure fetal and maternal health but the traditional ones fail to produce consistent readings, slow processing speed, and disturbance with the environment. This work gives a proposal of an intelligent alerting system and real-time tracking of fetal movements to reduce the risks of pregnancy. The system uses an ESP32 microcontroller that is coupled with several sensors, which include an ECG sensor of fetal heart rate, DS18B20 of maternal temperature, MAX30100 of fetal oxygen saturation, the force sensors with avoiding pressure detection, and the vibration sensor with detecting fetal movement. This information is somehow processed and transferred to an IoT platform through an ESP8266 NodeMCU WiFi board allowing remote access by medical professionals. Adaptivebased intelligent alerting mechanisms also improve the detection of risks at an early stage. Experimental analysis proves that there is a good acquisition of data, smooth wireless device communication, and prompt alerts. The system is a non-invasive, continuous, and proactive prenatal care solution, whose benefits include higher maternal-fetal health outcome as well as providing the basis to future smart pregnancy monitoring solutions.

Dependable and reliable electronic voting is one of the most significant challenges as it is fraught with centralization risk, privacy is at risk, coercion, and manipulation of the results. To solve these problems, the paper describes a decentralized, privacy-preserving e-voting system based on blockchains, smart contracts, and differential privacy, and using self-sovereign identity (SSI). The framework eliminates the use of centralized authorities, and it provides voter eligibility, anonymity, vote immutability, and verifiable tallying. It presents four new algorithms: Decentralized Identity Consistency Verification, which is used to verify SSI credentials without disclosing personal information; Adaptive Differential Noise Calibration, which finds a balance between preventing an inference attack and maintaining statistical accuracy; Smart Contract-driven Temporal Vote Locking, which eliminates the problem of double voting; and Consensus-Aware Verifiable Tallying, which is used to provide auditable aggregation under blockchain consensus. Experimental analysis indicates that the proposed system is much more transparent, discloses greater privacy protection, and does not readily yield to coercion, which is the case in digital election settings of large scale with high trust.

Interactive resource implementation, multi-tenancy, and massive data transfer present an ever more complicated risk to cloud computing environments. Conventional intrusion detection systems are not effective in detecting new and sophisticated attacks within this environment. This paper attempts to solve this issue by suggesting an artificial intelligence-powered intrusion detection and mitigation system that is specially implemented in cloud environments. The methodology incorporates 4 complicated and rarely applied algorithms: Hierarchical Temporal Memory, Bayesian Attack Graph Inference, Artificial Immune Systems with Negative Selection and Clonal Expansion, and Quantum-inspired Evolutionary Algorithms, to ensure improved anomaly detection, attack prediction and adaptive response. The framework has a layered way of working, which allows the creation of ongoing learning and real-time mitigation. As per experimental evidence, it has also been revealed that the framework will be characterized by much better detects, fewer false alarms and faster mitigation than traditional AI-based methods, which opens the possibility of enhancing cloud web security resilience.

The personalized online interaction experience could not be achieved without the recommender systems, which can be eroded by the fake users and the manipulated rating. The proposed research will solve these issues by coming up with a powerful model of rating predictions that combines fake user identification with the fusion of multi-layer features. There are four new algorithms namely Adaptive Anomaly Profiling Algorithm (AAPA) to detect suspicious users, Trust-Aware Noise Filtering Algorithm (TNFA) to eliminate unreliable interactions, Hierarchical Feature Blending Algorithm (HFBA) to merge user, item, and contextual features, and the Resilient Rating Estimation Algorithm (RREA) to predict accurate ratings. The model has been useful in isolating the malicious acts, using the real user data, and in modeling nonlinear relationships among fused features. Benchmark evaluations involving experiments indicate a higher level of prediction accuracy and strength in a number of attacks as compared to the traditional methods. The suggested framework makes the recommendations stable and reliable, which improves their user trust and the performance of the systems.

The Cloud computing address is providing scalable, flexible and cost efficient data storage and data processing solutions, but is posing major problems of data privacy and regulations compliance. The sensitive data stored in third-party infrastructures are vulnerable to unauthorized access use, data breach, and control loss. In this paper, a fully developed framework using four new algorithms is proposed involving Adaptive Privacy Encryption (APE) to dynamically encrypt data, Compliance Risk Predictor (CRP) to predict compliance in real time, Cross-Border Access Control (CBAC) to perform location-sensitive authorization, and Audit Trail Optimizer (ATO) to generate tamper resistant log. The framework systematically oversees the privacy of data, anticipates compliance risks, implements secure access as well as automates the audit procedures. The experiments prove the usefulness of the framework, realizing the 99.34%-accuracy in the detection of privacy and compliance violation. The suggested solution gives companies an effective guideline towards safe cloud implementation and seals the loophole between the performance effectiveness and regulatory compliance and the increased confidence in the cloud setup.

The emergence of social networking sites has further promoted the issue of spambots and counterfeiting followers, and these artificially inflate popularity rates, propagating false information, and losing the trust of the site user. The conventional methods of detection find it difficult to adapt to the changing spambot patterns and are not transparent in decision making. The paper proposes an explainable AI machine learning model at detecting spambots and fake followers with the help of gradient boosting tools, i.e., CatBoost, LightGBM, and XGBoost. The suggested approach combines profile attributes, behavioral patterns, and network-based characteristics to maximize the performance of detentions. Through the importance of the feature analysis, the interpretation of the model is clearly understood by establishing the factors that have influence in the predictions. It is experimentally proven that ensemble boosting models are better than baseline models in terms of precision and resiliency because both LightGBM and XGBoost show a higher classification rate. The results prove the hypothesis that high-performance machine learning combined with interpretability is an effective and reliable tool to fight malicious social network accounts.

ASPIRO AI is a smart mentoring and career guidance program aimed at helping students to determine their strengths, weaknesses and most suitable careers, based on a deep analysis held by AI. In the system, students are assessed on aptitude, communication and coding tests and automated analysis of their resume to create precise career suggestions, skills needed, and related courses. Role based architecture facilitates easy interaction between the students, mentors and the administration. Individualized guidance and feedback and career insights, as well as mentors, offer skill verification, career mentoring, and ongoing encouragement, are provided to students. Administrators monitor the whole system and regulate users, access and present role specific notifications to users. ASPIRO AI is an effective, scalable, and real-world career planning solution that integrates both artificial intelligence and mentorship as well as centralized administration. The platform will increase the accuracy of decision-making, make students more prepared, and act as an excellent academic and professional project, which can be used during college assessment and job interviews.

REST API testing is challenging since there is a tremendous number of combinations of operations, parameters, and dependencies, usually resulting in low coverage of the code and unnoticed faults. In this paper, a Semantic Property Dependency Graph (SPDG), an automated tool called AutoRestTest-HSL is presented, based on a novel algorithm: Hybrid SymbolicLearning Guided Multi-agent Testing (HSL-MAT). The HSL-MAT algorithm is a multi-agent reinforcement algorithm based on symbolic reasoning on API constraints, where the agents are enabled to give importance to operation sequences and parameter values that fulfill the dependencies as they explore a wide range of test environments. A combination of heuristic input and symbolic constraint solving is used to produce parameter values, which have valid and boundary-case inputs. It has a command-line interface to allow configuration, successful operations monitoring, server error monitoring and test duration. Experimental testing illustrates that AutoRestTest-HSL has higher fault detection effectiveness and better code coverage, which gives a comprehensive report in identifying the operations that were exercised and the faults that were identified, providing an effective and feasible used structure of automated testing on the REST API.

Organizations and educational institutions frequently use email for communication. One-by-one email correspondence takes time. Bulk email communication handled by the tools is ineffective. The Advance Bulk Mail Sender's design is explained in this paper. A web-based tool that simplifies mass email is the Advance mass Mail Sender. You can create messages using templates using the Advance Bulk Mail Sender. You can write a message by speaking with the Advance Bulk Mail Sender. You can schedule the email's delivery time with the Advance Bulk Mail Sender. The messages are recorded by the Advance Bulk Mail Sender. By enabling configurable password security for saved messages, the system increases privacy, decreases human labor, and improves usability. The suggested strategy aims to enhance user interaction while offering a scalable framework for backend integration.

Attendance management of schools in educational facilities is difficult because of the huge number, proxy attendance and absence of classes when a student reports to be present in school. The old manual orID-based systems are not effective, they can be easilycorrupted, and are not monitored in real-time, whichinhibits the free-will of the administrators to maintain accurate tracking of attendance and detect recurringabsenteeism. The current paper will suggest an AI-basedsurveillance system to cover the entire campus withYOLOv8 face detection and recognition. The cameras areplaced in classes and key areas on campuses and providelive video streams which are analyzed by the system toidentify the students automatically and capture theirattendance. The system makes a distinction between thosestudents who are on campuses and those who are inclasses, which create centralized records of attendance andauto-notifications to authorities in case of discrepancies.The system employs the deep learning technique,automated reporting, and real-time monitoring tominimize proxy attendance, increase accountability, anddeliver actionable analytics about student behavior. Thisexperience shows how AI can be used to revolutionizeclassroom management and enhance a disciplined andactive learning experience.

The fast increase in urban settlements has only aggravated the problems related to the traditional means of fuel delivery, such as the presence of long queues at the gas stations, poor logistic performance, and reduced supply in the event of an emergency. The solutions to these problems provided by REFUELX lie in the mobile based on-demand fuel delivery service which will allow its customers to obtain the fuel at their doorstep. The system proposed consists of the mobile application technology combined with GPS-based vehicle tracking, safe online payments, order monitoring in real time, and automated dispatching. This is supported by a strong backend architecture that handles users authentication, fuel truck assignment, optimal routes, scheduling and verification of delivery with high standards of safety and compliance with strict safety and compliance standards by using IoT-based monitoring. The contribution of the system is that they have modernized the fuel distribution system through enhancing convenience, operational efficiency, and safety, minimizing the wastage of fuel, and the congestion at the fuel stations. The results reveal that the experience of customers, fuel logistics, and the supply of fuel in emergency situations could be positively stimulated using REFUELX, which proves its prospects of being a scalable system to aviate fuel distribution in cities and semi-urban areas.