Cardiac output estimation using ballistocardiography: a feasibility study in healthy subjects

Area of Science:

• Biomedical Engineering
• Cardiovascular Physiology
• Medical Instrumentation

Background:

• Reliable, non-invasive circulation monitoring is lacking for prehospital emergency medicine.
• Current cardiac output (CO) monitors are impractical for prehospital use.
• Ballistocardiography (BCG) offers a non-invasive method to assess heart contractility and track CO changes.

Purpose of the Study:

• To assess the feasibility of estimating CO using morphological features from non-invasive BCG signals.
• To develop and validate a machine learning model for CO estimation based on BCG data.
• To explore the potential of BCG as a reliable, automated CO monitoring solution for prehospital settings.

Main Methods:

• Recorded ECG, carotid/abdominal BCG, and invasive arterial blood pressure in 20 healthy subjects.
• Processed BCG signals to isolate circulatory components (CCc, CCa) and extracted 66 beat-to-beat morphological features.
• Developed a machine learning model using a 75% development set and validated performance on a 25% validation set.

Main Results:

• The machine learning model achieved a mean absolute error of 0.83 L/min for CO estimation.
• The model demonstrated a percentage error of 30.2% and 95% limits of agreement of -2.18-1.89 L/min in the validation set.
• BCG analysis showed potential for reliable CO estimation and tracking.

Conclusions:

• BCG-based CO estimation is a promising non-invasive approach.
• This method represents a significant step towards automated, reliable CO monitoring for prehospital emergencies.
• Further testing is warranted to validate BCG's utility in real-world prehospital scenarios.