SBIR-STTR Award

The proposed research has two aims: 1) to develop a real-time algorithm that uses non-invasive physiological signals to quickly and accurately detect severity of acute blood loss; and 2) to develop a complimentary algorithm that uses all or a subset of these signals (depending on their availability and situation specific appropriateness) to forewarn providers of an injured soldier’s predicted risk for hemodynamic decompensation. The algorithms will be developed for use during transport and higher echelon medical care. These research efforts will be based on human subject Lower Body Negative Pressure (LBNP) experiments done in collaboration with the US Army Institute for Surgical Research, with proposed real-time testing of the developed algorithms. The research effort utilizes standard and novel machine learning, statistical and signal processing algorithms, in an aim to identify the most robust algorithms for prediction of hypovolemia, given noisy physiological data.

Keywords:
Hemorrhagic Shock, Estimated Blood Loss, Injury Severity, Machine Learning, Trauma, Hypovolemia, Vital Signs, Non-Invasive

On the battlefield, medics must quickly determine injury severity, treat the greatest threats to life, diagnose hemorrhage and establish a triage order. The objective of this research project is to apply our active, long-term learning technology to the task of modeling and prediction of central blood volume parameters from extremely large, and variable physiological datasets. In Phase I, we applied our feature extraction and modeling techniques to data gathered from human lower body negative pressure studies simulating acute blood loss at the U.S. Army Institute of Surgical Research. We discovered several previously hidden hemodynamic relationships that are predictive of acute blood loss volume and an individual"s risk for cardiovascular collapse. This work has led to the development of an entirely new type of intelligent, non-invasive medical device, which we call CipherSensor. In Phase II we will integrate the CipherSensor technology into the FDA approved Nexfin monitor and Flashback"s own portable device based on pulse oximetry (CipherOx). These prototype devices will be used to collect acute blood loss data in clinical and experimental settings. The collected data will be used to build more robust and accurate models of early hemorrhage detection and resuscitation effectiveness, as well as preparation for pre-clinical tests.