Federated Model Optimization for Real-Time Big Data Analytics in Smart Drug Delivery Edge Devices
Abstract
Abstract Smart drug delivery systems generate continuous streams of physiological and device telemetry, creating opportunities for real-time anomaly detection and intelligent clinical decision support. However, conventional cloud-centric machine learning approaches face significant challenges related to latency, bandwidth consumption, and patient privacy. This study proposes a Federated Model Optimization (FMO) framework for privacy-preserving real-time analytics in smart drug delivery edge devices. The framework integrates a two-stage gradient compression mechanism, reliability-weighted federated aggregation, and a hardware-aware convolutional recurrent neural architecture optimized for resource-constrained medical devices. By enabling local model training and transmitting only compressed model updates, FMO minimizes communication overhead while preserving sensitive patient data. Experimental evaluation in a simulated Internet of Medical Things environment demonstrates superior performance in drug dosage anomaly detection, achieving high classification accuracy, reduced network utilization, and low inference latency suitable for closed-loop clinical applications. The proposed framework offers a scalable, secure, and efficient solution for deploying federated intelligence across next-generation smart healthcare ecosystems.
Read the paper
Related studies
- Active Learning during Federated Learning for Object Detection
- Fairness-aware federated learning
- Decentralized Federated Learning
- Building a Personally Identifiable Information Recognizer in a Privacy Preserved Manner Using Automated Annotation and Federated Learning
- Preservation of Patient Level Privacy: Federated Classification and Calibration Models
- Efficient explainable federated learning
Go deeper in Evidence Hub
Evidence Hub adds a study data sheet (design, sample size, evidence level, key numbers), an AI summary and a chat grounded in the paper's text.