2018-910 Prediction of Cardiac Arrest Using Continuous Electrocardiogram

SUMMARY

UCLA researchers in the Geffen School of Medicine have developed a method to accurately and timely predict cardiac arrest in hospital patients based on a patient’s own data. This method allows for fewer and more relevant alarms that improve the chance of survival from a cardiac arrest episode.

BACKGROUND

In the United States, over 200,000 patients suffer cardiac arrests in hospitals annually with fewer than 30% surviving to discharge. 70-80% of these cardiac arrests are attributed to pulseless electrical activity (PEA)/asystole arrest, with fewer than 12% of patients suffering these arrests surviving to discharge. Currently, intermittently obtained vital sign and telemetry measurements have been used to identify and evaluate the risk of patients for cardiac arrest. These methods have been largely unsuccessful at predicting in-hospital deterioration and cardiac arrest due to their inability to measure parameters that are not indicative of physiologic change. For successful monitoring and minimizing false alarms, new technologies are needed to identification and evaluate the risk of patients with cardiac arrest.

INNOVATION

UCLA researchers have developed a method to predict the occurrence of cardiac arrests in patients using a continuous electrocardiogram. Patients are monitored and the trends in electrocardiographic parameters are compared to create a patient-specific predictive model. Electrocardiographic waveform data is automatically processed and filtered to produce a time series of electrocardiographic parameters, to remove non-physiologic information obtained from artefactual measurements. Further data processing results in the determination of arrhythmias within time windows and the creation of a predictive model that evaluates a risk score for cardiac arrest. This method has been successfully developed and first trial model predicted cardiac arrest with 89.6% sensitivity and 90.1% specificity.

POTENTIAL APPLICATIONS

• Personalized cardiac arrest risk scores
• Cardiac arrest prevention
• Hospital instrumentation

ADVANTAGES

• Accurate
• Predictive model
• Personalized

RELATED MATERIALS

• Do DH, Kuo A, Lee ES, Mortara D, Elashoff D, Hu X, Boyle NG. Usefulness of Trends in Continuous Electrocardiographic Telemetry Monitoring to Predict In-Hospital Cardiac Arrest. Am J Cardiol. 2019 Oct 1;124(7):1149-1158. doi: 10.1016/j.amjcard.2019.06.032. Epub 2019 Jul 17. PMID: 31405547; PMCID: PMC6744991.
• Ding Q, Bai Y, Tinoco A, Mortara D, Do D, Boyle NG, Pelter MM, Hu X. Developing new predictive alarms based on ECG metrics for bradyasystolic cardiac arrest. Physiol Meas. 2015 Dec;36(12):2405-22. doi: 10.1088/0967-3334/36/12/2405. Epub 2015 Oct 26. PMID: 26502065; PMCID: PMC4838570.

STATUS OF DEVELOPMENT

This method was tested in a three-hour (only one time window) continuous trial with patients and resulted in high accuracy predictions. Further expansion of the model to multiple time windows is currently underway.