This month, NEXUS is proud to spotlight the Electronic Health Records-Informed Lagrangian Method for Precision Public Health (EHRLICH) and Modeling Outcomes Using Surveillance Data and Scalable Artificial Intelligence for Cancer (MOSSAIC).
Author: Dr. Heidi Hanson
Dr. Heidi Hanson of Oak Ridge National Laboratory utilizes advanced artificial intelligence models to automatically code unstructured clinical documents. Through two key projects—Modeling Outcomes Using Surveillance Data and Scalable Artificial Intelligence for Cancer ( MOSSAIC) and the Electronic Health Records-Informed Lagrangian Method for Precision Public Health ( EHRLICH) — she combines clinical records with multimodal health and environmental data.
The MOSSAIC and EHRLICH teams leverage the Department of Energy’s world-class computing facilities, powered by exascale supercomputers capable of executing over 1 quintillion calculations per second. These systems enable the development of tools that facilitate deep-learning capabilities for large-scale exposomic research by synthesizing data and making possible population-level simulations while ensuring patient privacy and data security. Prospective digital twins for population health will allow for in-depth investigations into how environmental factors affect health outcomes.
Historically, the integration of geospatial exposure measures—such as assessments of land, water, and airborne toxins—with human-subject studies has been challenged by a lack of appropriate tools, reference datasets, privacy concerns, and expertise. Additionally, the absence of standardized exposure measures tailored for health research restricts researchers’ ability to compare and pool data across studies. These barriers hinder efforts to understand how the exposome influences the risk and progression of cancer and other diseases.
A pivotal component of the MOSSAIC and EHRLICH projects is the Centralized Health and Exposomic Resource (C-HER), which is designed to unify a wide range of area-based and environmental exposure datasets with population health data. This resource comprises state-of-the-art computational tools, algorithms, datasets, and documentation aimed at advancing the integration of exposomics into human-subjects-based cancer and health research.
C-HER serves as the AI-ready foundational platform for the computational tools developed by the MOSSAIC and EHRLICH teams, accelerating exposomic research and revealing valuable insights into the interplay between environmental factors and health. It is also being employed by the National Cancer Institute (NCI) and the National Institute for Environmental Health Sciences (NIEHS) to bolster additional exposomic research initiatives. For instance, residential histories for cancer cases recorded in the NCI Surveillance, Epidemiology, and End Results (SEER) registries from 1995 to the present have been geocoded and linked to multiple environmental exposure datasets. Researchers who gain approval through the SEER registries will have access to these longitudinal measures, enhancing our understanding of the exposome’s impact on cancer incidence, treatment response, and survival.
NEXUS is thrilled to be collaborating with Dr. Hanson in the larger mission of advancing exposomics for precision environmental health. The work of the MOSSAIC and EHRLICH teams is essential for advancing geospatial capabilities for the exposome, in particular for assessing the totality of external exposures at scale, to enable comprehensive understanding of causes and risk factors of disease over the life course.
Learn more about MOSSAIC
https://datascience.cancer.gov/collaborations/nci-department-energy-collaboration