Statistical shape modeling-based algorithm for replacing missing beats in blood pressure signals.

Area of Science:

• Biomedical Engineering
• Signal Processing
• Medical Informatics

Background:

• Continuous monitoring of vital signs in Intensive Care Units (ICUs) is essential for detecting patient deterioration.
• Clinical data often suffers from noise, artifacts, and recording gaps, hindering accurate signal assessment.
• Accurate estimation of missing data is critical for reliable patient monitoring and timely intervention.

Purpose of the Study:

• To develop an automatic algorithm for reconstructing arterial blood pressure (ABP) signal waveforms.
• To address the challenge of missing or corrupted data in continuous physiological monitoring.
• To improve the accuracy of vital sign assessment in critical care settings.

Main Methods:

• Proposed an algorithm based on statistical shape modeling for waveform reconstruction.
• Combined average beat templates with estimated waveform shape variations to fill data gaps.
• Validated the algorithm using arterial blood pressure recordings from the MIMIC-III Waveform Database.

Main Results:

• The algorithm demonstrated excellent performance in reconstructing arterial blood pressure waveforms.
• Achieved overall averages for Q1 and Q2 metrics greater than 0.85 in validation tests.
• Successfully reconstructed signal segments compared to original data using established metrics.

Conclusions:

• The developed algorithm effectively reconstructs arterial blood pressure waveforms, even with missing data.
• This method holds significant potential for enhancing the reliability of patient monitoring in ICUs.
• Accurate signal reconstruction can lead to earlier detection of patient deterioration and improved clinical decision-making.