Arterial Pressure Estimation based on Chest Compression Waveform and Electrocardiogram for Cardiopulmonary Resuscitation

Le résumé fourni par la source

Arterial blood pressure (ABP) is a vital hallmark reflecting the quality of cardiopulmonary resuscitation (CPR). Clinical examination of ABP relies on invasive measurement using catheters inserted into the thoracic aorta, which is too complex to be used in CPR operation. This study aimed to propose a novel non-invasive method for ABP estimation using machine-learning algorithms based on chest compression (CC) and electrocardiographic (ECG) features during CPR. Eight pigs were used in this study to construct cardiac arrest model. Two-lead ECG, ABP and CC signal were simultaneously recorded during CPR. Systolic blood pressure (SBP) and diastolic blood pressures (DBP) were extracted from the ABP waveform. Three machine-learning algorithms, namely Least Absolute Shrinkage and Selection Operator (LASSO), Support Vector Machine (SVM) and Random Forest (RF) were used for estimation, with seven features from CC waveform and twelve features from ECG signal as input, SBP and DBP as output. The result showed that SVM model with a combined ECG and CC features achieved better estimation of SBP and DBP than the other algorithms. Specifically, the mean absolute error was 3.466 mmHg for SBP and 1.424 mmHg for DBP, the root mean square error was 5.769 mmHg for SBP and 2.487 mmHg for DBP and the adjusted R2was 0.968 for SBP and 0.939 for DBP. A strong correlation was found between the non-invasive estimation and invasive measurement of ABP, with the correlation coefficients of 0.985 (95% confidence interval (CI): 0.975-0.989, p < 0.001) for SBP and 0.971 (95% CI: 0.955 - 0.980, p < 0.001) for DBP. The results suggested that non-invasive ABP estimation can be achieved during CPR based on CC and ECG signal with high correlation and low bias. The method developed in this study may facilitate the monitoring and improvement of CPR quality.Clinical Relevance— This study established a noninvasive ABP estimation model based on CC and ECG features. The model achieved a high consistency between the estimated and measured ABP values, which may play a role in guaranteeing high quality of CPR.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.