A low-power network for on-line diagnosis of heart patients

Implantable cardioverter defibrillators detect and treat dangerous cardiac arrhythmias. Current ICDs, however, cannot distinguish between some potentially fatal arrhythmias and benign conditions. Our system classifies intracardiac electrograms to detect such arrhythmias and uses analog techniques to meet the strict power and area requirements of implantable systems. A robust neural network architecture reduces the impact of noise, drift, and offsets inherent in analog approaches