Human activity recognition from on-body sensor networks is a challenging task for many healthcare applications because it offers continuous monitoring capability. However, these systems still face a major energy issue that prevent their wide adoption. Indeed, in healthcare applications, continuous sampling and communication tasks quickly deplete sensors' battery reserves, and frequent battery replacement are not convenient. Therefore, there is a need to develop energy-efficient solutions for long-term monitoring applications in order to foster the acceptance of these technologies by the patients. In this paper, we present a new energy-efficient architecture for health monitoring where sensors send the sensed data mainly via the user mobile phone, but they have the possibility to send data to access points when they are available in the user environment. The objective of this strategy is to relieve a unique base station from receiving all the data, and it enables sensor nodes to select the best communication technology in terms of energy consumption. Contrary to existing approaches, our solution allows to optimize the lifetime of both the sensor and the base station. The simulation results show that the proposed approach effectively improves the energy efficiency of the system compared to traditional architectures where wearable sensors can only communicate with the user mobile phone.