Physiological data is generated from signals received from one or more leads associated with the IMD. The physiological data is sampled to generate a plurality of data samples. A predictive encoding algorithm is applied to the plurality of data samples to generate a plurality of encoded data samples each corresponding to one of the plurality of data samples. An entropy encoding algorithm is then applied to the plurality of encoded data samples to generate a plurality of code words each corresponding to one of the plurality of encoded data samples. The code words are then stored in a memory in the IMD.