A three class treatment of the FHR classification problem using latent class analysis labeling

Area of Science:

• Obstetrics and Gynecology
• Medical Informatics
• Biostatistics

Background:

• Cardiotocography (CTG) is standard for fetal monitoring but suffers from subjective interpretation and high inter/intra-observer variability.
• The need for more objective fetal assessment methods is a significant challenge in perinatal care.
• Current interpretation of electronic fetal monitoring requires specialized expertise, leading to potential inconsistencies.

Purpose of the Study:

• To develop a more objective classification approach for electronic fetal monitoring data.
• To improve the reliability of fetal assessment by reducing subjective interpretation in cardiotocography.
• To explore latent class analysis for objective labeling of training data in fetal monitoring.

Main Methods:

• A classification approach utilizing latent class analysis (LCA) for objective data labeling.
• Integration of LCA with logistic regression in both standard multi-class and ordinal classification frameworks.
• Application of the proposed methods to electronic fetal monitoring (cardiotocography) data.

Main Results:

• The proposed latent class analysis combined with logistic regression demonstrated promising results for objective fetal monitoring.
• The methods showed potential in producing more reliable and objective classifications of fetal monitoring patterns.
• Both standard multi-class and ordinal logistic regression schemes yielded encouraging outcomes.

Conclusions:

• The developed classification approach shows significant potential for enhancing the objectivity of electronic fetal monitoring interpretation.
• Further research and development in this area are warranted to refine and validate the proposed methods.
• This approach could lead to more consistent and accurate fetal assessments, improving perinatal outcomes.