Heemun Kwok1, Jason Coult2, Chenguang Liu3, Jennifer Blackwood4, Peter J Kudenchuk5, Thomas D Rea6, Lawrence Sherman7. 1. Center for Progress in Resuscitation, University of Washington, Seattle, WA, United States; Department of Medicine, University of Washington School of Medicine, Seattle, WA, United States. Electronic address: heemun@uw.edu. 2. Center for Progress in Resuscitation, University of Washington, Seattle, WA, United States; Department of Bioengineering, University of Washington, Seattle, WA, United States. 3. Philips Healthcare, Bothell, WA, United States. 4. Center for Progress in Resuscitation, University of Washington, Seattle, WA, United States; King County Emergency Medical Services, Seattle King County Department of Public Health, Seattle, WA, United States. 5. Center for Progress in Resuscitation, University of Washington, Seattle, WA, United States; Department of Medicine, University of Washington School of Medicine, Seattle, WA, United States; King County Emergency Medical Services, Seattle King County Department of Public Health, Seattle, WA, United States; Division of Cardiology, University of Washington School of Medicine, Seattle, WA, United States. 6. Center for Progress in Resuscitation, University of Washington, Seattle, WA, United States; Department of Medicine, University of Washington School of Medicine, Seattle, WA, United States; King County Emergency Medical Services, Seattle King County Department of Public Health, Seattle, WA, United States. 7. Center for Progress in Resuscitation, University of Washington, Seattle, WA, United States; Department of Medicine, University of Washington School of Medicine, Seattle, WA, United States; Department of Bioengineering, University of Washington, Seattle, WA, United States.
Abstract
OBJECTIVE: Real-time feedback improves CPR performance. Chest compression data may be obtained from an accelerometer/force sensor, but the impedance signal would serve as a less costly, universally available alternative. The objective is to assess the performance of a method which detects the presence/absence of chest compressions and derives CPR quality metrics from the impedance signal in real-time at 1s intervals without any latency period. METHODS: Defibrillator recordings from cardiac arrest cases were divided into derivation (N=119) and validation (N=105) datasets. With the force signal as reference, the presence/absence of chest compressions in the impedance signal was manually annotated (reference standard). The method classified the impedance signal at 1s intervals as Chest Compressions Present, Chest Compressions Absent or Indeterminate. Accuracy, sensitivity and specificity for chest compression detection were calculated for each case. Differences between method and reference standard chest compression fractions and rates were calculated on a minute-to-minute basis. RESULTS: In the validation set, median accuracy was 0.99 (IQR 0.98, 0.99) with 2% of 1s intervals classified as Indeterminate. Median sensitivity and specificity were 0.99 (IQR 0.98, 1.0) and 0.98 (IQR 0.95, 1.0), respectively. Median chest compression fraction error was 0.00 (IQR -0.01, 0.00), and median chest compression rate error was 1.8 (IQR 0.6, 3.3) compressions per minute. CONCLUSION: A real-time method detected chest compressions from the impedance signal with high sensitivity and specificity and accurately estimated chest compression fraction and rate. Future investigation should evaluate whether an impedance-based guidance system can provide an acceptable alternative to an accelerometer-based system.
OBJECTIVE: Real-time feedback improves CPR performance. Chest compression data may be obtained from an accelerometer/force sensor, but the impedance signal would serve as a less costly, universally available alternative. The objective is to assess the performance of a method which detects the presence/absence of chest compressions and derives CPR quality metrics from the impedance signal in real-time at 1s intervals without any latency period. METHODS: Defibrillator recordings from cardiac arrest cases were divided into derivation (N=119) and validation (N=105) datasets. With the force signal as reference, the presence/absence of chest compressions in the impedance signal was manually annotated (reference standard). The method classified the impedance signal at 1s intervals as Chest Compressions Present, Chest Compressions Absent or Indeterminate. Accuracy, sensitivity and specificity for chest compression detection were calculated for each case. Differences between method and reference standard chest compression fractions and rates were calculated on a minute-to-minute basis. RESULTS: In the validation set, median accuracy was 0.99 (IQR 0.98, 0.99) with 2% of 1s intervals classified as Indeterminate. Median sensitivity and specificity were 0.99 (IQR 0.98, 1.0) and 0.98 (IQR 0.95, 1.0), respectively. Median chest compression fraction error was 0.00 (IQR -0.01, 0.00), and median chest compression rate error was 1.8 (IQR 0.6, 3.3) compressions per minute. CONCLUSION: A real-time method detected chest compressions from the impedance signal with high sensitivity and specificity and accurately estimated chest compression fraction and rate. Future investigation should evaluate whether an impedance-based guidance system can provide an acceptable alternative to an accelerometer-based system.
Authors: Jason Coult; Jennifer Blackwood; Thomas D Rea; Peter J Kudenchuk; Heemun Kwok Journal: IEEE J Biomed Health Inform Date: 2019-05-24 Impact factor: 5.772
Authors: Sofía Ruiz de Gauna; Jesus María Ruiz; Jose Julio Gutiérrez; Digna María González-Otero; Daniel Alonso; Carlos Corcuera; Juan Francisco Urtusagasti Journal: PLoS One Date: 2020-09-30 Impact factor: 3.240