Wireless body area networks are essentially constrained by the energy required to send data from one node to another. While most of times the whole data is sent, data compression can be used to decrease the amount of data and therefore the energy consumption. However, the compression is energy consuming and its efficiency depends on data size and data type. To tackle this challenge, the proposed Adaptive Near Sensor Compressing (ANSC) algorithm performs pertinent feature extraction on the sensed data, and selects the best compressor by estimating the energy required by the hardware platform to perform both data compression and transmission. The transmission power is also dynamically tuned to reach a specific quality of service according to node location and propagation channel. ANSC has been evaluated on a real platform dedicated to wireless body area network and has shown an energy gain of up to 49% compared to full transmission.