Double-sided asymmetric method for automated fetal heart rate baseline calculation

Area of Science:

• Medical signal processing
• Obstetrics and Gynecology
• Computational Biology

Background:

• Fetal heart rate (FHR) monitoring is vital for assessing fetal well-being during labor.
• Manual FHR interpretation exhibits significant inter- and intra-observer variability, impacting clinical decisions.
• Accurate FHR baseline determination is critical for reliable interpretation.

Purpose of the Study:

• To develop and validate an automated method for FHR baseline determination.
• To reduce the variability associated with manual FHR interpretation.
• To create an algorithm that mimics expert obstetrician baseline determination.

Main Methods:

• The Auto-Regressed Double-Sided Improved Asymmetric Least Squares (ARDSIAsLS) method was developed.
• Physiological gap interpolation was incorporated to handle missing FHR data.
• A weighted algorithm combining asymmetric and symmetric least squares smoothing models was employed.

Main Results:

• The ARDSIAsLS method achieved a mean absolute error of 2.54 bpm and a root mean square error of 2.89 bpm.
• The algorithm demonstrated superior performance compared to 11 previously published methods.
• The automated method was found to be non-inferior to expert obstetrician annotations.

Conclusions:

• Automating FHR baseline calculation can significantly decrease practitioner discordance.
• The ARDSIAsLS method provides a reliable and accurate tool for FHR baseline determination.
• This automation aids in consistent and effective clinical decision-making in the delivery room.