OP15.04: Validation of the first trimester machine learning model for predicting pre‐eclampsia in Asian populations

Le résumé fourni par la source

To evaluate the performance of a machine learning (ML) model for first trimester screening for pre-eclampsia (PE) in a large Asian population. This was a multicentre prospective cohort study in 10,935 women with singleton pregnancies undergoing routine assessment at 11-14 weeks of gestation. We applied the ML model for the first trimester prediction of preterm PE (< 37 weeks), term PE (≥37 weeks) and any PE, which was derived and tested in a cohort of pregnant participants in the United Kingdom (Model-1). This model comprises maternal factors with measurements of mean arterial pressure, uterine artery pulsatility index, and serum placental growth factor. The model was further retrained with adjustment for analysers used for biochemical testing (Model-2). The Delong test was used to compare the receiver operating characteristic curves of Model-1, Model-2, and the Fetal Medicine Foundation competing risk model (FMF model). The predictive performance of Model-1 was significantly lower than that of FMF model in predicting preterm PE (0.82 (95% confidence interval [CI] 0.77-0.87) vs 0.86 (95%CI 0.811-0.91), p = 0.019), term PE (0.75 (95%CI 0.71-0.80) vs 0.79 (95%CI 0.75-0.83), p = 0.006), and any PE (0.78 (95%CI 0.74-0.81) vs 0.82 (95%CI 0.79-0.84), p < 0.001). Following the retraining of data with adjustment for the PlGF analysers, the performance of Model-2 for predicting preterm PE, term PE, and any PE was significantly improved. The AUC values increased to 0.84 (95%CI 0.80-0.89), 0.77 (95%CI 0.73-0.81), and 0.80 (95%CI 0.76-0.83), respectively. There were no differences in AUCs between Model-2 and FMF model in predicting preterm PE (p = 0.135) and term PE (p = 0.084). However, Model-2 was inferior to FMF model in predicting any PE (p = 0.024). This study has demonstrated that after adjusting for biochemical marker analysers, the predictive performance of the first trimester ML prediction model for PE is comparable to that of the FMF model in Asian populations.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.