Enhancing agreement in cardiotocography interpretation between midwives and obstetricians through a rule-based AI...

Area of Science:

• Medical Informatics
• Obstetrics and Gynecology
• Artificial Intelligence in Healthcare

Background:

• Cardiotocography (CTG) interpretation is crucial for fetal well-being during pregnancy.
• Interrater variability in CTG interpretation can impact clinical decision-making.
• Standardized interpretation guidelines, like the NICHD 2008 guidelines, aim to reduce variability.

Purpose of the Study:

• To assess if a rule-based artificial intelligence (AI) program can enhance interrater agreement in cardiotocography (CTG) interpretation.
• To evaluate the impact of AI support on the agreement between nurse-midwives and obstetricians.
• To measure changes in interpretation time and user satisfaction with AI assistance.

Main Methods:

• A rule-based AI program was developed using National Institute of Child Health and Human Development (NICHD) 2008 guidelines.
• Twenty nurse-midwives interpreted CTG tracings twice, with and without AI support.
• Interrater agreement was measured using quadratic weighted kappa and intraclass correlation coefficients (ICC) against an obstetrician consensus standard.

Main Results:

• AI significantly improved agreement for NICHD category interpretation (κ from 0.548 to 0.906).
• Substantial improvements were noted in baseline variability, fetal heart rate category, prolonged decelerations, and acceleration count.
• Interpretation time decreased by an average of 6.7 minutes, and 70% of midwives reported high satisfaction with the AI's clinical utility.

Conclusions:

• A rule-based AI program shows promise in enhancing interrater agreement for CTG interpretation between midwives and obstetricians.
• The AI tool led to reduced interpretation times and high user satisfaction in this exploratory study.
• Further research in larger cohorts is needed to confirm generalizability and assess the impact on perinatal outcomes.