The hydrological prediction in transboundary river basins is difficult because of their heterogeneous data distribution, restrictions of data sovereignty, and the irregular nature of the discharge pattern. In this paper, Federated Learning (FL) with an LSTM neural network is proposed for next-day predictions of discharge in the Chu–Talas transboundary basin. The basin area belongs to both Kazakhstan and Kyrgyzstan. In an FL scenario, two hydrological stations from the basin were selected as client nodes, representing two different discharge regimes. Station 15125—the Chu main channel is characterized by the highest discharge regime among stations located in the basin, while Station 15233—Merke tributary represents a small catchment with an irregular regime. The federated LSTM model is compared against a centralized LSTM and a local-only LSTM baseline model. The training process is based on nearly three decades of daily discharge measurements. The preprocessing step includes synchronization, lag calculation, and windowing operation. The models are trained using three metrics: root mean square error, mean absolute error, and Nash–Sutcliffe Efficiency, as well as using Monte Carlo Dropout for estimation of the probabilistic uncertainty. The results demonstrate that the federated model demonstrates comparable performance with the centralized one for the Chu main channel. It also improves prediction accuracy for the smaller Merke tributary compared with both centralized and local-only models. These findings show that FL can work effectively with non-IID and heterogeneous hydrological data. The study makes three main contributions: (i) it implements the FedAvg algorithm on transboundary, heterogeneous hydrological data, proving that decentralized optimization can effectively capture autoregressive temporal hydrology without data centralization; (ii) it systematically compares federated, centralized, and local-only models, demonstrating that the federated approach eliminates the scale bias that traditionally neglects smaller, high-variance catchments; and (iii) it utilizes Monte Carlo Dropout to translate deterministic AI outputs into risk-aware probabilistic bounds. Ultimately, the results of this study demonstrate the practical and scientific usefulness of FL in operational water management, as the method presents a privacy-saving means of increasing predictive capacity and enabling risk-based decision-making in transboundary river basins.

The field of Artificial Intelligence (AI) is at an inflection point, shifting from the “One Model Fits All” philosophy represented by centralized models like ChatGPT to “Hyper-Personalized AI” that honors user data sovereignty while retaining the collaborative learning advantages. This review examines the convergence of Personalized Federated Learning (PFL), Large Language Models (LLMs), and privacy-preserving technologies to create truly sovereign personal AI agents. We review over 39 peer-reviewed articles (2020-2025) on these topics, with particular emphasis on three major areas: architectural innovations in personalization using Low-Rank Adaptation (LoRA) adapters and knowledge graphs, privacy-preserving technologies like Differential Privacy (DP) and secure aggregation, and frameworks that allow users' personal data to be used to train personal agents locally without requiring cloud access. Our review demonstrates that PFL is at the core of creating the “Sovereign Data Ecosystem” where users can retain personal data like their calendar, email, and health records locally while contributing to global model improvements through aggregated gradient sharing. We also outline future research directions towards Federated Agent Swarms and how PFL redefines the relationship between user data ownership, model personalization, and algorithmic transparency.

The Internet of Medical Things (IoMT) has the potential to change the healthcare industry with the help of sensing technologies. However, the increasing number of heterogeneous devices creates critical data silos due to the use of proprietary data formats, such as Apple Health and Fitbit. These data formats are not compatible with standardized electronic health record (EHR) systems. Fast Healthcare Interoperability Resources (FHIR) and Federated Learning (FL) have the potential to bridge the gap between the incompatibilities of data formats while maintaining the privacy of the patient. This systematic review has included the recent literature from the year 2020 to 2025. New findings on the integration of FL and FHIR have also been included in this systematic review. A novel FedFHIR architecture has been introduced with the help of on-device converters that standardize raw data before FL processing. This helps in overcoming the problems of interoperability, governance, and edge resource constraints. The regional issues faced by people living in rural and Central Asia have also been addressed in this systematic review. Semantic interoperability is a prerequisite for the scaling of privacy-preserving artificial intelligence (AI) in a distributed healthcare system.

Python web development is experiencing a paradigm shift from the traditional Synchronous Web Server Gateway Interface (WSGI) to the modern Asynchronous Server Gateway Interface (ASGI). While WSGI frameworks like Flask have existed for a long time as the industry standard for microservices, their blocking architecture inherently restricts performance under high-concurrency workloads. This research fills this specific gap completely by introducing a rigorous quantitative benchmark comparing Flask WSGI with FastAPI (ASGI). Through the construction of two functionally identical applications and rigorous controlled I/O-bound load testing using Apache JMeter, throughput, latency, and error rates are measured. Results show a stark contrast in stability; ASGI kept 100% reliability, WSGI failed by 76% under identical load. These findings have actionable implications for software developers that moving to asynchronous frameworks is not just an adaptation but a requirement in our current and high-throughput API environment.