Steganography, BioPatterns and Independent Components

In this paper we discuss the development of steganography in the context of the electronic Patient Health Record. We exploit the signal processing aspects of a frame-based approach of expanding signals using a nonorthogonal basis of independent components derived from the data, and investigate the consequences of applying the framework to different dimensionality biopatterns. The integrity of the hidden message under different benign ’attacks’ is investigated along with the capacity of the embedded message. We will particularly concentrate on hiding information in biopatterns such as single channel EEG time series, since this is especially challenging in low dimensional problems such as single channel time series because information capacity and hence exploitable information redundancy is lower. We will conclude by investigating the additional security of the embedded message by the sensitivity of recovered messages to slight variations in the structure of the independent components, or knowledge of which components have been modified a topic not previously considered, and which augments any cryptographic approach for security.