TASK-DEPENDENT ADAPTABILITY OF REMOTE RECORDER IN A CARDIAC SURVEILLANCE NETWORK

The paper discusses a reconfigurable software for automated ECG interpretation and reporting. The target application is a star-shaped wireless surveillance network for out-hospital patients and elderly people. The interpretation process is distributed between the remote recorder and the central server in order to provide maximum adaptability to the patient status at optimal diagnostic reliability and use of resources. Such approach implied significant communication cost and suffers from cumulative interpretation errors. Studying various existing applications we usually encountered a very similar architecture based on the upgradeable modules concept. The presented alternative approach considers medical data at various processing stage in context of their volume, priority, reliability and interdependence with a goal of optimal performance and adequacy of medical report to the expectations implied by the patient status. The aim of the newly proposed architecture is twofold: cut the error propagation chains and reduce the data volume at early stages of processing. Verification of the proposed prototype was based on the comparison of diagnoses made by distributed and static software, frequency and correctness of software reconfiguration attempts and on the analysis of dynamic response to a sudden event. As demonstrated in the tests, 768 (89,6%) of adaptation attempts were technically appropriate, among of them 643 (99,4%) software upgrades and 97 (80,2%) software replacements yielded diagnostic results similar to the reference observed from the experts' survey. The software adaptation during the seamless ECG monitoring made on a background of automated diagnosis verification and resources assessment yields the unprecedented advantages of the surveillance system.