In this work two radios with heterogeneous characteristics have been paired to enable mobile sensor nodes to achieve a significantly higher reliable connectivity at a lower total energy cost, compared to a single radio. A Patient Monitoring Network (PMN) is developed to elucidate the need for a multi-radio platform. Nordic's nRF51822 Bluetooth Low Energy (BLE) and Digi's XBee-Pro 868MHz modules have been integrated into the developing platform as the communication standards. Furthermore, PMN requires low energy consumption due to wearable nodes, and reliable data transfer. We have shown in this work that both requirements can be achieved by using a multi-radio system with a switching algorithm based on reinforcement-learning. Eventually it is shown that in the application scenario in this work a performance enhancement is achieved. The multi-radio system is twice more reliable in data transfer than BLE and consumes two times less energy compared to ZigBee.