Detecting cardiomyopathies in pregnancy and the postpartum period with an electrocardiogram-based deep learning model

Abstract

Aims Cardiovascular disease is a major threat to maternal health, with cardiomyopathy being among the most common acquired cardiovascular diseases during pregnancy and the postpartum period. The aim of our study was to evaluate the effectiveness of an electrocardiogram (ECG)-based deep learning model in identifying cardiomyopathy during pregnancy and the postpartum period. Methods and Results We used an ECG-based deep learning model to detect cardiomyopathy in a cohort of women who were pregnant or in the postpartum period seen at Mayo Clinic. Model performance was evaluated using area under the receiver operating characteristic curve (AUC), accuracy, sensitivity, and specificity. We compared the diagnostic probabilities of the deep learning model with natriuretic peptides and a multivariable model consisting of demographic and clinical parameters. The study cohort included 1,807 women; 7%, 10%, and 13% had left ventricular ejection fraction (LVEF) of 35% or less, less than 45%, and less than 50%, respectively. The ECG-based deep learning model identified cardiomyopathy with AUCs of 0.92 (LVEF ≤35%), 0.89 (LVEF <45%), and 0.87 (LVEF <50%). For LVEF of 35% or less, AUC was higher in Black (0.95) and Hispanic (0.98) women compared to White (0.91). Natriuretic peptides and the multivariable model had AUCs of 0.85 to 0.86 and 0.72, respectively. Conclusions An ECG-based deep learning model effectively identifies cardiomyopathy during pregnancy and the postpartum period and outperforms natriuretic peptides and traditional clinical parameters with the potential to become a powerful initial screening tool for cardiomyopathy in the obstetric care setting.

An artificial intelligence electrocardiogram analysis for detecting cardiomyopathy in the peripartum period

“…Given the distinct features of pregnancy, a study was necessary to show the suitability of an ECG-based DLM for pregnant patients compared with its effectiveness for ordinary patients. In August 2021, Adedinsewo et al conducted a study at the Mayo Clinic and reported that ECG-DLM effectively detects cardiomyopathy related to pregnancy [19]. Our model achieved an AUROC of 0.877 in the external validation test (LVEF 45% or less) compared with Mayo Clinic's AUROC of 0.89 (LVEF <45%); both studies manifest excellent performance of the DLM in the pregnancy group.…”

“…However, the few, previous studies present conflicting opinions of ECGs on PPCM diagnosis. Honigberg et al [18] show that ECG has a low sensitivity for PPCM diagnosis, but a recent study from the Mayo Clinic by Adedinsewo et al [19] demonstrates the high performance of ECG-based deep learning models in left ventricular dysfunction in perinatal women.…”

“…To date, few studies have investigated the utility of ECG-based deep learning models for PPCM. The only previous DLM study for PPCM [19] was not externally validated and included a small subset of individuals of Asian ethnicity. Therefore, our study aims to derivate an ECG-based DLM detecting cardiomyopathies occurring in the peripartum period in a predominantly Asian population with external validation.…”

Cardiovascular Disease Screening in Women: Leveraging Artificial Intelligence and Digital Tools

“…We recently evaluated the ability of an AI-enhanced ECG to detect cardiomyopathies during pregnancy and the postpartum period and found it to be effective (Figure 5A). 87 Other ML models specifically for use in detecting or predicting the risk for pregnancy-related disorders, such as gestational diabetes and preeclampsia, have also shown promise, albeit with limited external validation. 88 Potential AI/ digital tools that can be leveraged for screening during the peripartum period include not only the incorporation of ECG-based AI tools for cardiomyopathy detection 87 but also remote monitoring technologies for blood pressure assessments 89 and digital scales for weight assessments with automated data capture and transfer such that measurement reporting is less reliant on the patient who may also be navigating the challenges of being a new mother.…”

“…87 Other ML models specifically for use in detecting or predicting the risk for pregnancy-related disorders, such as gestational diabetes and preeclampsia, have also shown promise, albeit with limited external validation. 88 Potential AI/ digital tools that can be leveraged for screening during the peripartum period include not only the incorporation of ECG-based AI tools for cardiomyopathy detection 87 but also remote monitoring technologies for blood pressure assessments 89 and digital scales for weight assessments with automated data capture and transfer such that measurement reporting is less reliant on the patient who may also be navigating the challenges of being a new mother. Data captured from these monitoring tools can be incorporated into an ML algorithm to identify uncontrolled hypertension and elevated BMI and incorporated with smartphone-based apps to provide automated educational interventions on lifestyle modification or prompts to ensure medication compliance.…”