Distributed Edge Intelligence for Rapid In-Vehicle Medical Emergency Response in Internet of Vehicles

Abstract

The unparalleled possibilities of Internet of Vehicles (IoV) development prompt the enhancement of in-vehicle medical emergency response. Nevertheless, the IoV environment is still affected by data privacy, latency, and network instability, which hamper effective and reliable emergency medical systems. In this regard, this article suggests the emergency-aware distributed edge intelligence (DEI) for medical response (EDEM) framework, a novel approach leveraging DEI to address these challenges. Specifically, EDEM introduces a hierarchical edge collaborative computing architecture that dynamically constructs learning domains based on a comprehensive medical data capability model. The framework incorporates an in-vehicle medical data reliability model and tailored latency and energy consumption models to optimize resource allocation and response times. Then, a deep-reinforcement-learning-based node selection algorithm ensures efficient task distribution across the network. Finally, EDEM’s dual-layer federated learning model features an emergency-aware adaptive aggregation mechanism and an adaptive medical model updating scheme for cross-domain scenarios, complemented by an emergency-weighted asynchronous model fusion approach. The superiority of EDEM over state-of-the-art methods is demonstrated through simulation results showing up to a 15% increase in model accuracy, a 30% reduction in response times, and a 20% better resource utilization efficiency. This implies that it can greatly enhance speed, accuracy, and reliability for in-vehicle emergency responses within IoV environments.