Work

AI-Powered CURA Application for Identifying At-Risk Pregnancies in Obstetrics Management (CURAte): protocol for a randomised controlled trial

Year: 2026

Type: article

Abstract: Background Early identification of high-risk pregnancies is a cornerstone of obstetric care, yet antenatal risk assessment in routine practice remains imprecise and highly variable. Evidence from contemporary cohort studies suggests that only a minority of pregnancies that subsequently experience adverse maternal or fetal/neonatal outcomes are identified as high risk in the first trimester, with reported detection rates in the range of 20%–30%. Despite rapid advances in machine-learning (ML) methods for clinical risk prediction, no ML-based approach for general early antenatal risk categorisation has yet been evaluated against standard care within a randomised controlled trial (RCT) to assess impact on maternity outcomes. We hypothesise that adoption of an artificial intelligence (AI)-assisted antenatal risk stratification approach will improve the accuracy of categorising pregnancies into high-risk and low-risk groups and will reduce the composite incidence of adverse maternal and fetal/neonatal outcomes. Methods and analysis We will conduct an RCT comparing outcomes between participants allocated to an intervention arm employing AI-assisted risk stratification and those in the control arm receiving standard antenatal care without AI support. 1700 participants will be randomised in a 1:1 ratio using block randomisation, stratified by ethnicity and parity. Participants will be blinded to allocation. AI-assisted risk assessment using maternal sociodemographic, clinical and obstetrical data, will be applied at two time points: (1) a first trimester antenatal visit and (2) between 31 and 33 weeks’ gestation. AI categorisation outputs (intervention: ‘high-risk’ or ‘low-risk’; control: ‘undisclosed’) will be displayed to attending clinicians at both time points, and subsequent management will follow a predefined clinical protocol, one written for pregnancies categorised as high-risk and one for low-risk. For any pregnancy categorised as high-risk, the clinician works through predefined referral criteria for the maternal-fetal medicine-led high-risk clinic and completes the specified actionable investigations and interventions before referral. Participants are followed until 6 weeks after the end of pregnancy. The primary composite outcome is extracted from medical records 6 weeks after the end of pregnancy. The two arms are compared on an intention-to-treat basis once the database has been locked. There is no interim analysis of the primary outcome. Ethics and dissemination Ethical approval was obtained from the National Healthcare Group Domain Specific Review Board (ECOS 2024-3832). Written informed consent will be obtained from all participants, with data handled in accordance with institutional data-protection requirements and the Personal Data Protection Act. The AI tool functions as clinical decision support and does not replace clinician judgement; subgroup analyses will assess potential differential effects across key demographic and obstetric groups. Trial registration number NCT06974188.

Source: BMJ Open

DOI: 10.1136/bmjopen-2026-124115