Are You Bleeding? Validation of a Machine-learning Algorithm for Determination of Blood Volume Status: Application to Remote Triage

Abstract

Due to limited remote triage monitoring capabilities, combat medics cannot currently distinguish bleeding soldiers from those engaged in combat unless they have physical access to them. The purpose of this study was to test the hypothesis that low-level physiological signals can be used to develop a machine-learning algorithm for tracking changes in central blood volume that will subsequently distinguish central hypovolemia from physical activity. Twenty-four subjects underwent central hypovolemia via lower body negative pressure (LBNP), and a supine-cycle exercise protocol. Exercise workloads were determined by matching heart rate responses from each LBNP level. Heart rate and stroke volume (SV) were measured via Finometer. ECG, heat flux, skin temperature, galvanic skin response, and two-axis acceleration were obtained from an armband (SenseWear Pro2) and used to develop a machine-learning algorithm to predict changes in SV as an index of central blood volume under both conditions. The algorithm SV was retrospectively compared against Finometer SV. A model was developed to determine whether unknown data points could be correctly classified into these two conditions using leave-one-out cross-validation. Algorithm vs. Finometer SV values were strongly correlated for LBNP in individual subjects (mean r 0.92; range 0.75 0.98), but only moderately correlated for exercise (mean r 0.50; range 0.23 0.87). From the first level of LBNP/exercise, the machine-learning algorithm was able to distinguish between LBNP and exercise with high accuracy, sensitivity, and specificity (all 90%). In conclusion, a machine-learning algorithm developed from low-level physiological signals could reliably distinguish central hypovolemia from exercise, indicating that this device could provide battlefield remote triage capabilities.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jan 09, 2014
Accession Number
ADA614872

Entities

People

  • Caroline A.
  • Chelsea R. Barrera
  • David Andre
  • Gennifer M. Hurst
  • Kathy L. Ryan
  • Kevin R. Ward
  • Nisarg Vyas
  • Victor A Convertino

Organizations

  • United States Army Institute of Surgical Research

Tags

Communities of Interest

  • Autonomy
  • Biomedical

DTIC Thesaurus Topics

  • Accuracy
  • Algorithms
  • Blood Volume
  • Cardiovascular Physiological Phenomena
  • Data Analysis
  • Health Services
  • Heart Rate
  • Learning
  • Machine Learning
  • Measurement
  • Medical Personnel
  • Physical Activity
  • Physiological Monitoring
  • Validation
  • Vital Signs
  • Volume
  • Warning Systems

Readers

  • Cardiovascular Physiology
  • Exercise and Sports Science.
  • Sensor Fusion and Tracking Systems.

Technology Areas

  • AI & ML
  • AI & ML - Neural Networks