Unique insights into risk factors for antepartum stillbirth using explainable AI
AMERICAN JOURNAL OF OBSTETRICS AND GYNECOLOGY(2023)
Abstract
To use explainable AI to identify the most important risk factors for antepartum stillbirth. This retrospective study used data from 153,432 singleton births (20-43 weeks’ gestation) at 20 U.S. hospitals (01/2016 - 12/2021). An Explainable Boosting Machine (EBM) was trained to predict antepartum stillbirth using 30 features available early in pregnancy including patient demographics and medical history. Births at 13 hospitals (n=105,174) were used to train the EBM and births at 7 different hospitals (n=48,258) were used for external validation. A logistic regression (LR) model was also trained on the data for comparison. Stillbirth occurred in 610 (0.4%) pregnancies. The strongest predictor of stillbirth was BMI – risk was highest for patients with Class I obesity (Figure 1). Area of residence was also important. Risk was lowest for patients living in the most prosperous areas (DistressQuintile 1), and gradually rose to the highest risk for patients living in the most economically deprived areas (DistressQuintile 5). Maternal age, IVF, self-reported race, and nulliparity were also among the strongest predictors. Surprisingly, shorter maternal height was associated with increased stillbirth risk. EBMs yielded an AUC of 0.70 (95% CI 0.67 – 0.73), comparable to the LR AUC of 0.69 (95% CI 0.66 – 0.72). Sensitivity was 59% at a 20% screen positive rate. Experiments with other AI models on the same data suggested it is difficult to obtain higher sensitivity using only features available in early pregnancy. Our study demonstrates how the intelligibility of EBMs can reveal surprising risk factors, in this case highlighting maternal BMI and particularly Class I obesity as a crucial risk factor. Why risk was highest in Class 1 obesity is uncertain but it could be speculated that patients with Class 2 or 3 obesity receive different clinical management and increased surveillance thus mitigating stillbirth risk. Like other adverse pregnancy outcomes, socioeconomic factors are important for stillbirth.View Large Image Figure ViewerDownload Hi-res image Download (PPT)
MoreTranslated text
Key words
antepartum stillbirth,risk factors
AI Read Science
Must-Reading Tree
Example
![](https://originalfileserver.aminer.cn/sys/aminer/pubs/mrt_preview.jpeg)
Generate MRT to find the research sequence of this paper
Chat Paper
Summary is being generated by the instructions you defined