Mobile banking has been experiencing unsustainable growth in the last ten years, as the volume of digital payments worldwide is currently more than USD 8.49 trillion and is estimated to be more than USD 20 trillion by 2026. At the same time, fraudulent attacks have been directed toward mobile banking platforms to a significant extent, causing losses amounting to USD 485.6 billion worldwide alone in 2023. The current fraud detection architectures have been heavily based on a centralized cloud-based model which is inherently associated with a latency delay of between 200 to 800 milliseconds that introduces a time delay that advanced attackers can use to transact fraudulent transactions before the defensive countermeasures are activated. The paper will suggest a new Edge-to-Cloud Collaborative Fraud Detection (EC-CFD) model that allocates inference workloads to three hierarchical levels. The structure is a combination of differential privacy, mutual authentication, and federated learning to allow joints in improving the model without having to centralize sensitive financial information. The feature engineering has 18 dimensions which include transaction statistics, geo-location anomalies, device fingerprinting and behavioral biometrics. Synthetic dataset of 2.4 million transactions, experimented on with samples of PaySim and IEEE-CIS benchmarks on fraud detection methods to provide realistic behavioural diversification. The findings indicate that the suggested EC-CFD model is characterized by a 98.7 % detection rate of fraud, where the F1-score is 0.974 and the AUC-ROC value is 0.992. It is important to note that the hierarchical structure allows accomplishing 87.2% raw transactions at the edge level and achieves a mean detection latency of 8.3 ms. The proposed framework lowers false positives by 31.4 per cent, the average end-to-end latency by 73, and could be extended to support 50,000 active users without lowered performance, which proves the effectiveness of hierarchical edge intelligence as the primary paradigm of next-generation financial cybersecurity.

This bibliometric survey examines 200 highly cited publications spanning from 2018 to 2025, retrieved from the Scopus database using search terms targeting federated learning applications in healthcare, medical imaging, and clinical diagnostics. The analysis reveals a rapidly maturing interdisciplinary field at the convergence of distributed machine learning, privacy-preserving technologies, and biomedical informatics. Key findings indicate that federated learning has emerged as a transformative paradigm for collaborative AI model training across healthcare institutions without compromising patient data privacy. The United States, China, the United Kingdom, Germany, and India emerge as the most productive nations, with extensive international collaboration networks reflecting the global need for privacy-compliant healthcare AI. Thematic clustering identifies five major research domains: (1) federated learning architectures and algorithms for healthcare, (2) privacy-preserving techniques including differential privacy, homomorphic encryption, and blockchain integration, (3) medical imaging applications (radiology, pathology, dermatology, ophthalmology), (4) electronic health records and predictive analytics, and (5) wearable devices and remote patient monitoring. Emerging trends include the integration of federated learning with explainable AI for clinical interpretability, the convergence of federated learning with digital twins and the healthcare metaverse, and the development of personalized federated learning frameworks for handling non-IID medical data. This survey provides a comprehensive mapping of the intellectual landscape, identifies persistent challenges including data heterogeneity, communication efficiency, and security vulnerabilities, and proposes future directions for privacy-preserving healthcare AI.

At the edge of the network, the exponential increase of data produced along with the growing concerns over data privacy coming from regulations and society have all together triggered the rise of Federated Learning (FL) as the main approach in distributed machine learning (DML). Fed learning allows the model training to be performed on decentralized devices or data silos even without the raw data being transferred. Hence, FL is completely in line with the objectives of the privacy-preserving techniques. In this paper, we carry out a scientometric analysis on the 200 most cited papers, which are the first 200 papers at the intersection of "Distributed Machine Learning," "Federated Learning," and "Privacy-Preserving" published between 2020 and 2024, and the Scopus database is where they are indexed. The literature of publication trends, prominent authors and works, the thematic clusters, and research fronts that are changing are all systematically examined in this study; hence, the intellectual landscape of this fast developing field is mapped out. Our findings point to the existence of certain streams of research such as the algorithms with differential privacy being the mainstay, secure aggregation methods through the use of homomorphic encryption and multi-party computation, blockchain-based FL systems which ensure security and trust, and resource-efficient FL that supports IoT and edge computing. The results also show an area that is nearly enjoying a complete transformation as a result of the overpowering need to address the triad of model quality, data protection, and system efficiency. The review not only encourages researchers, and practitioners but also helps the policymakers by providing the current trend to which the key challenges can be identified and the future directions in privacy-preserving distributed intelligence anticipated.