Randomised study of human machine collaboration for cardiotocography interpretation during labour.

Area of Science:

• Obstetrics and Gynecology
• Perinatal Medicine
• Clinical Decision Support Systems

Background:

• Interpretation of cardiotocography (CTG) during labour is known for high interobserver variability.
• This variability limits the effectiveness of CTG in accurately predicting perinatal acidaemia.
• Objective assessment tools are needed to improve diagnostic accuracy and patient outcomes.

Purpose of the Study:

• To evaluate the impact of computerised CTG (cCTG) assistance on clinicians' predictive performance for perinatal acidaemia.
• To determine if cCTG improves diagnostic accuracy compared to unaided interpretation.
• To assess the effect of cCTG on inter-observer agreement and reliability.

Main Methods:

• Prospective, randomised multi-reader study design.
• 211 clinicians from 23 countries assessed 100 CTG recordings (50 with pH <7.15).
• Assessments were performed with and without cCTG assistance to predict perinatal acidaemia.

Main Results:

• cCTG assistance significantly improved overall prediction success rate (54.0% to 61.4%, p < 0.01).
• Sensitivity for detecting perinatal acidaemia increased significantly (49.3% to 61.7%, p < 0.01).
• Specificity showed no significant difference (58.7% vs 61.2%, p = 0.14), but cCTG was correct in 67.5% of discordant cases. Agreement and reliability improved across diverse clinician groups.

Conclusions:

• Computerised CTG (cCTG) assistance demonstrably enhances clinicians' ability to predict perinatal acidaemia.
• The use of cCTG improves diagnostic accuracy and consistency in labour monitoring.
• cCTG represents a valuable tool for improving the detection of fetal distress and adverse perinatal outcomes.