Objective

To assess expert inter-observer variability in antepartum cardiotocography (CTG) interpretation and compare it with the Dawes-Redman (DR) computerised system.

Methods

Dataset: 105 selected 15-minute segments of antepartum CTG traces.

Participants: 5 expert clinicians (3 midwives, 2 obstetricians, >10 years’ experience).

Annotations: Baseline, variability, accelerations and decelerations collected via a structured digital tool.

Majority Voting: Median participant baselines, mode of selected variability, accelerations and decelerations identified by majority have start and end points averaged.

DR Comparison: Same 105 CTG segments analysed using the Dawes-Redman system and compared to majority voted results.

Interpretation & Implications

Clinician Judgement: Baseline interpretation was consistent, but variability and deceleration assessments showed substantial disagreement.

DR System: Accurate for baseline and accelerations; poor sensitivity and PPV for decelerations, and limited agreement on variability.

Clinical Relevance: Discrepancies interpreting decelerations and variability may lead to missed fetal compromise or unnecessary interventions.

Recommendation: Improved training, standardised displays and hybrid AI-clinician approaches may enhance consistency and clinical reliability.

Conclusion

While expert and algorithmic interpretation show high concordance for baseline assessment, variability and decelerations remain inconsistently classified, particularly for patterns associated with fetal compromise. These findings underscore the need for structured training, harmonised display formats, and integrated clinical AI workflows to strengthen antepartum CTG interpretation and decision making.