A SOFTWARE SENSOR USING NEURAL NETWORKS FOR DETECTION OF PATIENT WORKLOAD

The morphology of intracardiac electrograms (IEGMs) was used for pacem aker patient workload estimation. The body posture also was studied as another characteristic. The IEGMs were obtained and recorded via temp orary transcutaneous leads connected to the implanted pacemaker. IEGMs were recorded during exercise and at rest. Recordings at rest were pe rformed in different body positions. The morphology was analyzed visua lly in order to observe changes due to workload and posture. The recor dings were digitized and processed by a computer-simulated neural netw ork. The network was used as an automatic IEGM classifier based on the morphology. Our results show that the morphology of the IEGM may be u sed as an indicator of patient workload and body posture. The necessar y information is found mainly in the ST segment. We conclude that neur al networks seem to be useful in an active cardiac device.