ECG INTERPRETATION TASK SHARING RULES IN A DISTRIBUTED WEARABLE MONITORING SYSTEM

摘要

This paper addresses the problem of interpretation task sharing between the central server and the remote wearable ECG recorder. Adaptive task sharing is presented as an alternative to the remote embedded and to the centralized interpretation commonly used today. The patient-dependent adaptation of the task sharing needs an extreme care, since it is advantageous only if based on medically justified rules. Three procedures are involved in optimization and control of the task sharing: human expert-derived diagnosis priority, statistic-based automatic assessment of result reliability and machine estimation of remote resources availability. The paper discusses these procedures and their impact to the distant interpretation performance and diagnosis reliability. Presented approach was designed and programmed in C++ language and consists an integral part of distributed architecture of the ECG interpretive software. The prototype implemented in the PDA-based ECG monitor and the stationary server enabled the assessment of the task sharing correctness for standard test signals. With the promising results achieved with the prototype we plan to transfer the task sharing rules to a multithreaded man-supervised server.