Use of artificial intelligence to identify cardiovascular compromise in a model of hemorrhagic shock.

摘要: Objective To determine whether a prototype artificial intelligence system can identify volume of hemorrhage in porcine model controlled hemorrhagic shock. Design Prospective vivo animal Setting Research foundation surgical suite; computer laboratories collaborating industry partner. Subjects Nineteen, juvenile, 25- to 35-kg, male and female swine. Interventions Anesthetized animals were instrumented for arterial systemic venous pressure monitoring blood sampling, splenectomy was performed. Following 1-hr stabilization period, hemorrhaged aliquots 10, 20, 30, 35, 40, 45, 50% total with 10-min recovery between each aliquot. Data downloaded directly from commercial into proprietary PC-based software package analysis. Measurements main results Arterial gas values, glucose, cardiac output collected at specified intervals. Electrocardiogram, electroencephalogram, mixed oxygen saturation, temperature (core blood), mean pressure, pulmonary artery central pulse oximetry, end-tidal CO(2) continuously monitored downloaded. Seventeen 19 (89%) died as direct result hemorrhage. Stored data streams analyzed by the system. For this project, identified compared three electrocardiographic features (R-R interval, QRS amplitude, R-S interval) nine unknown samples complex. We found that system, trained on only features, an average accuracy 91% (95% confidence 84-96%). Conclusions These experiments demonstrate based solely analysis R-R interval electrocardiogram, is able accurately lethal suggest technology may represent noninvasive means assessing physiologic state during immediately following Point care application improve outcomes earlier diagnosis better titration therapy