| Patient monitoring in anesthesia attempts to achieve three main purposes: fault detection, diagnosis and treatment. This research developed a graphic information display for the cardiovascular system. The object-oriented graphic display was used to represent the clinical data. Numerical information was represented by preattentively processed shapes, colors, and textures. Three-dimensional graphic metaphors were used to integrate variables. A Just Noticeable Differences (JND) test was used to quantify the data mapping based on clinical relevance. Data communication software was developed to drive the display by commercial anesthesia monitors or human simulators.; In the operating room, the anesthesiologists need to derive diagnostic information from many pieces of sensor data. The new display aimed at providing higher-level information to support decision making. Three hemodynamic indices for myocardial ischemia were evaluated in this study: subendocardial viability ratio (SEVR), rate pressure product (RPP), and pressure rate quotient (PRQ). In simulation study, the hemodynamic indices were able to indicate hypotension induced ischemia, but failed to identify ischemia caused by volume overload. Using real patient records in the Massachusetts General Hospital (MGH) database, we compared SEVR, RPP, and PRQ to ECG in detection of ischemia. All three indices had poor positive predictive value and sensitivity. Those hemodynamic indices are unreliable for predicting myocardial ischemia.; To estimate cardiovascular function of patients, we used a lumped ventricle-arterial model to estimate myocardial contractility, preload, and arterial compliance. Given the radial artery pressure and cardiac output, the model parameters were identified by minimizing the squared error between simulated and measured pressure waveforms. Transesophageal echocardiography (TEE) measurements of ventricular function were collected in the operating room to validate the model. Pressure waveforms from eight open heart surgical patients were used in the identification process. The model reliably reproduced the radial arterial pressure and aortic flow. The model based estimates of left ventricle end-diastolic volumes (LVEDV) were compared to 47 TEE measurements. The bias and standard deviation were 1.01ml and 9.69ml, respectively. The correlation coefficient between estimation results and the measurements was 0.92. This model identification scheme provides a functional characterization of the hemodynamic status of the left ventricle and systemic arterial load. |
