Abstract
In an effort to ease staffing burdens and potentially improve patient outcome in an intensive care unit (ICU) environment, we are developing a real-time system to accurately and efficiently diagnose cardiopulmonary emergencies. The system is being designed to utilize all relevant routinely-monitored physiological data in order to automatically diagnose potentially fatal events. The initial stage of this project involved formulating the overall system design and appropriate methods for real-time data acquisition, data storage, data trending, waveform analysis, and implementing diagnostic rules. Initially, we defined a conceptual analysis of the minimum physiologic data set, and the monitoring time-frames (trends) which would be required to diagnose cardiopulmonary emergencies. Following that analysis, we used a fuzzy logic diagnostic engine to analyze physiological data during a simulated arrhythmic cardiac arrest (ACA) in order to assess the validity of our diagnostic methodology. We used rate, trend, and morphologic data extracted from the following signals: expired CO2 time-concentration curve (capnogram), electrocardiogram, and arterial blood pressure. The system performed well: The fuzzy logic engine effectively diagnosed the likelihood of ACA from the subtle hemodynamic trends which preceded the complete arrest. As the clinical picture worsened, the fuzzy logic-based system accurately indicated the change in patient condition. Termination of the simulated arrest was rapidly detected by the diagnostic engine. In view of the effectiveness of this fuzzy logic implementation, we plan to develop additional fuzzy logic modules to diagnose other cardiopulmonary emergencies.
Full text
PDF
