Artificial intelligence-based ultrasounds: accuracy in predicting mode of birth and monitoring labour

Background/Aims Artificial intelligence-based ultrasound technologies can be used in labour monitoring and predicting mode of birth. This review and meta-analysis aimed to assess the accuracy of these technologies globally. Methods The PubMed, CINAHL, Web of Science, Cochrane Library, Scopus, Medline (OVID) and Google Scholar databases were searched for studies that explored the use of artificial intelligence-based ultrasound technologies used to monitor labour and predict mode of birth. The technologies’ sensitivity, specificity and predictive values were calculated, as well as the pooled prevalence of prediction for mode of birth. Overall accuracy was estimated using random and fixed effects models. Results A total of 13 studies were included, encompassing 145 085 labour and birth records. The globally pooled prevalence of accurately predicted mode of birth was 90.54%. The overall accuracy for monitoring labour progress was 90.46%. The pooled positive predictive value was 89.95% and the true positive rate was 87.23%. Conclusions Artificial intelligence-based ultrasound technologies demonstrated high accuracy in labour monitoring and mode of birth prediction. Implications for practice Integrating these technologies into clinical practice is essential to enhance decision making and support safer, more effective and personalised obstetric management.