Personalized care for prenatal stress reduction and preterm birth prevention

Univ Of North Carolina Chapel Hill, Chapel Hill NC

Investigators

Linked publications, trials & patents

Abstract

Modified Abstract Section There is an urgent need to understand the complex, interactive relationships between hundreds of chemical (e.g. water and air quality) and nonchemical (e.g social vulnerability) stressors in the environment and their impact on preterm birth (PTB) risk and disparities. Currently, generating insights is hindered by the substantial computing power, large analytics demand, and innovative visualization tools required to ingest, process, analyze, and interpret the plethora of datasets across different domains and data types. To address these barriers, in this competing revision, we will create a cloud-based template for PTB research, addressing the underuse of cloud computing in perinatal health disparities research. In Aim 1, we will generate the Preterm Birth Stressors Data Lake (PS-DL): a cloud-based data resource for PTB research that will adhere to the Findable, Accessible, Interoperable, and Reusable (FAIR) principles, providing an invaluable resource for other research. PS-DL will integrate hundreds of spatially and temporally variant data across different domains including chemical (air, soil, and water quality), climate (heat, precipitation, extreme weather events), and socioeconomic (social vulnerability, education) data. These datasets will each be linked in time and space to 1.3 million birth certificate records from North Carolina (NC). In Aim 2, we will design PTB-PREDICT: a reproducible ML pipeline that leverages PS-DL and a cloud-based, efficient machine learning (ML) platform to identify key mixtures of stressors impacting PTB, investigate disparities in NC, and predict PTB in real-time. In Aim 3, we will disseminate insights via a cloud- based informatics platform and web visualization tool, ENVIROSCAN 2.0 in which end users (physicians, patients, advocates, public health professionals, and researchers) will be able to interact, visualize and download PS-DL data and results from PTB-PREDICT. Through these aims, we will explore and test the use of the cloud to accelerate insights into environmental perinatal health challenges. We will assess the feasibility of leveraging cloud solutions for complex geospatial problems in perinatal health and disparities research, explore the benefits of cloud-based analytical pipelines for ML modeling in PTB research, and establish efficiencies gained through running analyses and visualizations in the cloud. Together, these aims will inform the context of the patient population in the parent grant and contribute to an enhanced understanding of the complex web of stressors that impact prenatal populations and the most important driving factors of PTB, which will, in turn, accelerate solutions-oriented research. This research directly addresses numerous key goals of the NIH Office of Data Science, including developing a modernized data ecosystem and developing accessible workflows and tools, and fits into the vision of the National Institute of Minority Health and Health Disparities by leveraging high-end computing and big data to identify complex interactions between chemical and nonchemical determinants of the national public health issue of PTB.