Assessment of the Need to Perform Life-Saving Interventions Using Comprehensive Analysis of the Electrocardiogram and Artificial Neural Networks.

摘要

This work addresses the need for advanced medical technologies with a high fielding potential, specifically, development of vital sign monitoring technology. The US Army Combat Critical Care Engineering Task Area aims to improve care in the Battlefield Critical Care Environment (BCCE) by developing new decision support systems that take better advantage of the large data stream available from casualties. In this short review we present new descriptors of physiologic status suitable for decision support in the BCCE. These variables can be extracted from biosignals such as the electrocardiogram (EKG) (featured in this manuscript). They can also be extracted from other waveforms, processed using similar analysis tools. This manuscript reviews our work to date on the development of such decision support systems. We focus on the use of new vital signs derived from heart rate complexity (HRC) and traditional heart rate variability (HRV) analysis in machine learning technology such as artificial neural networks (ANN). We present data from 262 prehospital and emergency department trauma patients, in whom noninvasive vital signs derived from EKG analysis were fed to a commercially available feed- forward back-propagation ANN (NeuralWare, Carnegie, PA). The endpoint was prediction of the performance of life-saving interventions (LSIs) such as intubation, cardiopulmonary resuscitation (CPR), chest-tube placement, needle pneumothorax decompression and blood transfusion. Sensitivity and specificity were explored via construction of receiver operating characteristic (ROC) curves.