Agriculture in India is undergoing a paradigm shift driven by the integration of Artificial Intelligence (AI) and the Internet of Things (IoT), collectively termed the Artificial Intelligence of Things (AIoT). Positioned within the broader transformation toward Agriculture 5.0, AIoT offers a pathway for developing resilient, sustainable, and highly efficient farming systems capable of addressing India’s pressing socio-economic and environmental challenges. This research paper examines the technological foundations, layered architectures, and enabling innovations - such as edge computing, deep learning, computer vision, blockchain, ambient IoT, and emerging 6G connectivity - that underpin the AIoT ecosystem. It further evaluates major national initiatives including Agri-Stack, ADeX, ADMF, and state-led pilots such as Telangana’s Saagu Baagu program, demonstrating measurable improvements in yield, resource-use efficiency, and farmer income. The study provides a comprehensive comparative analysis of AIoT applications in soil monitoring, variable rate fertilization, pest surveillance, livestock management, and post-harvest logistics, revealing strong evidence for large-scale economic benefits. However, challenges persist across digital literacy, infrastructure deficits, data heterogeneity, security vulnerabilities, and the affordability barriers faced by smallholder farmers. To address these gaps, the paper highlights future research directions focused on developing scalable, interpretable AI models, standardizing datasets, enhancing federated learning for privacy preservation, and deploying ambient IoT and 6G networks for massive connectivity. By presenting a unified technical, policy, and systems-level perspective, this research outlines a future-ready roadmap for realizing an equitable, data-driven, and globally competitive Agriculture 5.0 ecosystem in India. Received: December 14, 2025Revised: February 9, 2026Accepted: February 11, 2026
The integration of Internet of Things (IoT) and Artificial Intelligence (AI) technologies within social work applications presents significant opportunities to enhance service delivery while maintaining client confidentiality. This study proposes a novel, scalable, and privacy-preserving IoT-AI system architecture specifically designed for social work environments, incorporating federated learning methods alongside low-power communication protocols. The framework effectively addresses critical challenges concerning data privacy, system scalability, and energy efficiency in distributed social work contexts. It employs a hierarchical federated learning approach combined with adaptive Low-Power Wide-Area Network (LPWAN) protocols to enable collaborative model training across geographically dispersed client nodes, thereby safeguarding sensitive social work data. Empirical results demonstrate that the proposed architecture achieves a classification accuracy of 94.7% in social work case analyses and reduces communication energy consumption by 68% compared to centralized approaches. Additionally, the system exhibits strong scalability, supporting over 1,000 concurrent client nodes with minimal performance degradation, making it highly suitable for large-scale social work implementations. This article contributes to the advancement of privacy-preserving distributed AI systems and provides a robust foundation for the development of future technological platforms in social work. Received: November 8, 2025Accepted: December 6, 2025
The swift propagation of misinformation across digital platforms underscores the imperative for automated detection systems capable of real-time operation. Conventional methodologies predominantly focus on textual analysis, often neglecting essential visual indicators and dissemination dynamics. This paper presents an innovative multimodal framework based on convolutional neural networks (CNNs) that effectively integrates textual, visual, and network-structural features to improve the identification of false information. The proposed architecture adopts a multi-stream design: textual data are encoded using BERT embeddings augmented with attention mechanisms; visual content is examined through VGG-16 transfer learning complemented by forensic feature extraction; and information diffusion patterns are captured via graph neural networks (GNNs). A late-fusion technique amalgamates these diverse feature representations for the final classification task. Empirical evaluation on the PHEME and MediaEval datasets reveals that the framework attains an accuracy of 93% and an F1-score of 0.93, markedly surpassing unimodal baseline models. Additionally, the system is optimized for practical deployment, achieving a processing latency under 200 milliseconds and a throughput exceeding 5,000 posts per minute. These findings substantiate the model’s effectiveness for real-time surveillance and prompt mitigation of coordinated disinformation efforts. Received: January 11, 2026Accepted: February 9, 2026
The global demographic transition toward an aging population presents unprecedented challenges for long-term care systems, with critical workforce shortages affecting 92% of nursing homes and 70% of assisted living facilities. This mixed-methods study investigates the effectiveness of AI-enhanced human-machine collaboration in improving long-term care service efficiency and quality. Following PRISMA and STROBE guidelines, we conducted a systematic review of 105 studies and controlled trials across 218 facilities (94 intervention, 124 control) over 18 months. The AI-enhanced system analyzed 150 daily clinical data points per patient, providing real-time alerts for condition changes, fall risk assessment, and medication monitoring. Results demonstrated significant improvements in 89% of quality measures, including a 9% reduction in major falls $(p = 0.034)$, 22% decrease in ADL dependency $(p < 0.001)$, and 97.7% diagnostic accuracy. Operational efficiency improved substantially with 62.5% reduction in documentation time, enabling 5 additional hours of direct care daily. Annual cost savings averaged \$800,000 per facility with 14.2-month break-even point. Contrary to displacement concerns, staff job satisfaction increased by 278% $(p < 0.001)$ and burnout decreased by 34% $(p < 0.001)$. Workforce modeling suggests AI collaboration could reduce projected additional workforce needs by 50% by 2040. These findings provide compelling evidence that AI-enhanced human-machine collaboration, when properly implemented with human autonomy preservation and comprehensive training, significantly enhances long-term care efficiency and quality while maintaining compassionate, person-centered service delivery. Received: October 18, 2025Accepted: November 3, 2025
The importance of telecommunication systems in the medical field is immense especially as it relates to monitoring of cardiovascular conditions as well as taking the heart beat rate of expectant mothers and that of the fetus. Fetal monitoring during pregnancy enables the physician to diagnose and monitor pathological conditions especially asphyxia. The Electrocardiogram (ECG) is the simplest non-inversive diagnostic method used to solve various heart diseases. In this study, the use of Poisson probabilistic algorithm is employed to predict R-R intervals (a valid and reliable assessment of the time between two successive heartbeats, measured in milliseconds, and it is the most crucial for a scientific and practical use of HRV) in both maternal and fetal ECG signals for a set of 72 ECG heart rates for both the mother and her fetus. The application of Poisson technique has demonstrated promising results in error rates and better monitoring accuracy. 72 ECG signals for a certain R-R timing ranging from 0.66 to 0.99 in seconds were done, and ECG monitoring for important performance metrics, such as throughput, packet loss, error rate, and energy consumption was recorded. Using the Poisson forecast, a ‘P’ error amplitude ranging from 0.7735 eV to 1.305 eV for an R-R timing of 0.66 to 0.99 sec was obtained. From the results, the eV error amplitude proves to be more error prone from the ECG graphs obtained when compared with that of the actual data for Fetal Electrocardiogram (FECG) and Mother Electrocardiogram (MECG). The results from the research work compete favourable when compared with the wireless network error rate, throughput, energy consumption and energy efficiency, with and without the Poisson forecast. The proposed model in the work was also compared with an energy efficient wireless network system that applied Poisson algorithm to substantiate the effectiveness and accuracy of our system. Received: October 3, 2025Accepted: October 16, 2025
