Machine learning: a new era for cardiovascular pregnancy physiology and cardio-obstetrics research

Abstract

The maternal cardiovascular system undergoes functional and structural adaptations during pregnancy and postpartum to support increased metabolic demands of offspring and placental growth, labor, and delivery, as well as recovery from childbirth. Thus, pregnancy imposes physiological stress upon the maternal cardiovascular system, and in the absence of an appropriate response it imparts potential risks for cardiovascular complications and adverse outcomes. The proportion of pregnancy-related maternal deaths from cardiovascular events has been steadily increasing, contributing to high rates of maternal mortality. Despite advances in cardiovascular physiology research, there is still no comprehensive understanding of maternal cardiovascular adaptations in healthy pregnancies. Furthermore, current approaches for the prognosis of cardiovascular complications during pregnancy are limited. Machine learning (ML) offers new and effective tools for investigating mechanisms involved in pregnancy-related cardiovascular complications as well as the development of potential therapies. The main goal of this review is to summarize existing research that uses ML to understand mechanisms of cardiovascular physiology during pregnancy and develop prediction models for clinical application in pregnant patients. We also provide an overview of ML platforms that can be used to comprehensively understand cardiovascular adaptations to pregnancy and discuss the interpretability of ML outcomes, the consequences of model bias, and the importance of ethical consideration in ML use.

Keywords: artificial intelligence; cardiovascular; machine learning; maternal health; pregnancy.

Figure 1.

Conceptual diagram of the canonical machine learning (ML) model development process. ML first “learns” underlying conditions within a given data set (training) that lead to an outcome of interest, validates the model developed on unseen data (validation), and then conducts a final evaluation of model performance when applied to novel observations. Once the model reaches an acceptable performance threshold, the “learned” model can be fit to data of interest and (ideally) achieve the optimal solution for a problem of interest. BP, blood pressure; EMR, electronic medical record; HR, heart rate; UADF, umbilical artery Doppler-flow.

Figure 2.

Doppler-echo ultrasonography provides a wealth of information about the physiological condition of the circulatory system. Vascular remodeling induces changes in vessel stiffness and wall thickness, which result in changes in the shape of the flow velocity envelope. As shown, we process video recordings of the Doppler ultrasound into strip-chart images, segment these into the flow data related to each cardiac epoch, and then extract physiologically related variables such as peak and minimum flow velocities and rates of acceleration, as well as the overall shape of the flow envelope as an image. These features are then used to train a machine-learning classifier so that it can classify new ultrasound recordings with respect to disease state.

Figure 3.

Schematic illustrating how machine learning (ML) may be implemented to integrate multiple data types and sources (e.g., clinical, genetic, biomarker) for predicting adverse cardiovascular-related pregnancy outcomes. ML pipelines may include various methods for feature selection and learning to develop predictive models capable of enabling more precise patient care for improved outcomes. CV, cardiovascular.