Shock decision algorithm for use during load distributing band cardiopulmonary resuscitation

Area of Science:

• Cardiology
• Biomedical Engineering
• Medical Device Technology

Background:

• Load distributing band (LDB) chest compressions create electrocardiogram (ECG) artefacts.
• These artefacts interfere with automated external defibrillator (AED) shock decisions.
• Developing reliable algorithms for LDB compression artefact management is crucial for effective resuscitation.

Purpose of the Study:

• To develop and validate a reliable shock decision algorithm for use during LDB chest compressions.
• To compare the performance of the new algorithm against a commercial defibrillator algorithm.
• To characterize and address the unique challenges of LDB artefacts in ECG signals.

Main Methods:

• Utilized a dataset of 5813 ECG segments from 896 cardiac arrest patients undergoing LDB compressions.
• Annotated ECG segments as shockable or non-shockable.
• Employed adaptive filters to remove LDB artefacts and characterized them against manual artefacts.
• Developed a machine learning algorithm for shock decision-making post-filtering.

Main Results:

• LDB artefacts exhibited lower compression frequencies (80/min) and higher amplitudes (5.5 µV) compared to manual artefacts.
• The machine learning algorithm achieved superior shock decision accuracy: 92.1% sensitivity and 96.8% specificity.
• This significantly outperformed the commercial defibrillator algorithm (91.4% sensitivity, 87.1% specificity).

Conclusions:

• LDB artefacts present unique challenges due to their amplitude and frequency characteristics.
• The developed machine learning algorithm provides clinically reliable shock decisions during LDB compressions.
• This represents a significant advancement in AED technology for patients receiving LDB CPR.