Abstract P191: Use of Low Level Physiologic Signals and Machine Learning to Derive Important Hemodyamic Variables During Acute Volume Loss
Introduction: Monitoring traditional hemodynamic signals continuously is difficult. This has important implications in emergency care in the setting of the combat, EMS, mass casualties, ED overcrowding, remote monitoring, and in-patient detection of decompensation. Small wearable wireless devices capable of capturing hemodynamics could be useful. We report on a battery-operated armband to monitor low level physiologic signals followed by transformation of these signals into traditional hemodynamic parameters using machine learning (ML).
Methods: A small armband was attached to the left upper arm of 21 volunteers undergoing lower body negative pressure (LBNP) as a hemorrhage mimetic. The LBNP protocol consisted of a 5-min rest period (0 mm Hg) followed by 5 min of decompression of the lower body to −15, −30, −45, and −60 mm Hg and additional increments of −10 mm Hg every 5 min until cardiovascular collapse. Subjects were monitored continuously using a beat-to-beat noninvasive blood pressure device (FINOMETER®) allowing for continuous measurement of heart rate, blood pressure, stroke volume, and shock index. The armband measures one-lead ECG, galvanic skin response, skin heat flux, and motion. The models developed are simple regression trained with all parameters verified using by-subject cross-validation to eliminate over fitting. Agreement between methods was assessed using Bland-Altman analysis.
Results: All derived hemodynamic data using the model was significantly correlated to those measured using the FINOMETER® and there was no significant difference between actual and predicted means. Bland-Altman analysis suggests interchangability between armband and FINOMETER® derived hemodynamics.
Conclusion: Using ML, important hemodynamic parameters can be derived and tracked using easily obtained low level signals and without the need for traditional sensors. This may have important logistical implications for emergency and combat casualty care.