M. Jabri, S. Pickard, P. Leong, Z. Chi, B. Flower, Y. Xie
Current Intra-Cardia defibrillators make use of simple classification algo(cid:173) rithms to determine patient conditions and subsequently to enable proper therapy. The simplicity is primarily due to the constraints on power dissipa(cid:173) tion and area available for implementation. Sub-threshold implementation of artificial neural networks offer potential classifiers with higher perfor(cid:173) mance than commercially available defibrillators. In this paper we explore several classifier architectures and discuss micro-electronic implementation issues.