Combining latent class analysis labeling with multiclass approach for fetal heart rate categorization

Area of Science:

• Obstetrics and Gynecology
• Medical Informatics
• Biostatistics

Background:

• Cardiotocogram (CTG) is the standard for assessing fetal well-being during delivery.
• Current CTG interpretation suffers from significant inter- and intra-observer variability.
• Objective methods are needed to improve the reliability of fetal monitoring.

Purpose of the Study:

• To develop and evaluate an objective labeling system for cardiotocogram (CTG) interpretation.
• To address the limitations of subjective expert judgment and biochemical markers in fetal well-being assessment.
• To investigate the utility of latent class analysis (LCA) and ordinal classification for CTG analysis.

Main Methods:

• Utilized a well-documented, open-access database with CTG annotations from nine expert obstetricians.
• Applied latent class analysis (LCA) to derive objective class labels from CTG data.
• Employed an ordinal classification scheme to capture the severity spectrum of fetal conditions.

Main Results:

• Latent class analysis (LCA) demonstrated potential for generating more objective CTG class labels.
• Ordinal classification effectively explored the natural ordering and severity representation in CTG findings.
• The proposed methodology showed promising results in enhancing CTG interpretation objectivity.

Conclusions:

• The integration of LCA and ordinal classification offers a promising avenue for objective CTG analysis.
• Further research into this approach is warranted to improve fetal well-being assessment during delivery.
• This method has the potential to reduce diagnostic variability and enhance clinical decision-making in obstetrics.