ICAHN SCHOOL OF MEDICINE AT MOUNT SINAI

Agency: United States Department of Health and Human Services
NAICS: 541620 - Environmental Consulting Services
Period start: 2026-06-01
Period end: 2031-03-31
Budget: 819506.0
Place of performance: NEW YORK, NEW YORK 10029, UNITED STATES

Description

ASSESSING PATHWAY-SPECIFIC STROKE RISKS: A MULTIFACETED APPROACH INCORPORATING GENERATIVE AI, ENVIRONMENTAL EXPOSURES, AND POPULATION CHARACTERISTICS. - STROKE IS A MAJOR CAUSE OF DEATH GLOBALLY. SEVERAL STUDIES FOUND A HIGHER STROKE RISK ASSOCIATED WITH FINE PARTICULATE MATTER (PM2.5), BUT CRITICAL KNOWLEDGE GAPS PERSIST, NECESSITATING URGENT INVESTIGATION. WE ARE UNIQUELY POSITIONED TO INVESTIGATE THE ASSOCIATIONS OF PATHWAY-SPECIFIC HEMORRHAGIC AND ISCHEMIC (I.E., LARGE ARTERY ATHEROSCLEROSIS, CARDIO AORTIC EMBOLISM, SMALL ARTERY OCCLUSION, OTHER, UNDETERMINED) STROKES, WITH SHORT- AND LONG-TERM PM2.5 EXPOSURE AND ITS COMPOSITION. WE WILL LEVERAGE NOVEL ARTIFICIAL INTELLIGENCE (AI) MODELS TO CLASSIFY PATHWAY-SPECIFIC STROKES USING 13 YEARS OF ELECTRONIC HEALTH RECORDS (EHR) FROM THE MOUNT SINAI HEALTH SYSTEM (MSHS). THIS COMPREHENSIVE HEALTH AND COVARIATES DATA INCLUDES CLINICAL NOTES, IMAGING, LABS, MEDICATIONS, AND DIAGNOSES. MSHS, THE LARGEST HEALTH CARE PROVIDER IN THE NEW YORK CITY METROPOLITAN AREA, SERVES AN URBAN AND SUBURBAN POPULATION, INCLUDING ALL DEMOGRAPHIC GROUPS. AIM 1 WILL BUILD GENERATIVE AI-BASED MODELS TO CATEGORIZE PATHWAY-SPECIFIC STROKE CASES IN THE EHR. THIS MODEL WILL BE GENERALIZABLE TO MULTIPLE HEALTHCARE SYSTEMS AND IMPROVE STROKE RESEARCH, WHICH OFTEN COMBINES DISTINCT SUBTYPES INTO A SINGLE ENDPOINT DESPITE THEIR DIFFERING PATHOPHYSIOLOGY. BY CONNECTING AIR POLLUTION EXPOSURE TO SPECIFIC STROKE PATHWAYS, WE WILL PROVIDE EVIDENCE ON THE UNDERLYING BIOLOGICAL MECHANISMS DRIVING THESE ASSOCIATIONS. AIM 2 WILL ASSESS THE ASSOCIATION OF PM2.5 MASS AND COMPOSITION WITH PATHWAY-SPECIFIC STROKE. UNLIKE MOST STUDIES THAT OBTAIN PM2.5 COMPONENT MEASUREMENTS FROM GROUND MONITORING SITES, WE HAVE DEVELOPED INNOVATIVE, HIGHLY SPATIOTEMPORALLY RESOLVED MODELS THAT SIGNIFICANTLY MINIMIZE THE EXPOSURE MEASUREMENT ERROR AND IMPROVE OUR ABILITY TO ESTIMATE EXPOSURE- RESPONSE ASSOCIATIONS. WE WILL USE AN EXPOSURE-MIXTURE APPROACH TO ESTIMATE THE COMBINED ASSOCIATIONS OF THE CHEMICAL COMPONENTS AND IDENTIFY THE COMPONENTS ATTRIBUTING MOST TO EACH PATHWAY-SPECIFIC STROKE RISK. FINALLY, ALTHOUGH THE EPA HAS LOWERED THE ANNUAL PM2.5 STANDARD FROM 12 TO 9 ΜG/M3, THIS THRESHOLD IS STILL MARKEDLY HIGHER THAN THE THRESHOLD RECOMMENDED TO PROTECT HUMAN HEALTH (5 ΜG/M3). OUR PREVIOUS ANALYSES SHOW THAT PM2.5 EXPOSURE <9 ΜG/M3 IS STILL ASSOCIATED WITH INCREASED CARDIOVASCULAR RISK. WE WILL ASSESS WHETHER THE ASSOCIATIONS REMAIN CONSIDERING DIFFERENT ANNUAL PM2.5 STANDARDS. AIM 3 WILL ASSESS WHETHER THE ASSOCIATION BETWEEN PM2.5 AND PATHWAY-SPECIFIC STROKE VARIES BY DEMOGRAPHIC AND CLINICAL CHARACTERISTICS. WE WILL QUANTIFY EXPOSURE VARIATIONS AND EVALUATE THE EXTENT TO WHICH THE ASSOCIATIONS BETWEEN PM2.5 AND STROKE DIFFER ACROSS INCOME AND FINANCIAL HARDSHIP LEVELS CONSIDERING LOWER PM2.5 STANDARDS. IN SUMMARY, THIS STUDY WILL ADVANCE THE USE OF ADMINISTRATIVE DATA FOR LARGE-SCALE ENVIRONMENTAL HEALTH RESEARCH. WE WILL DEVELOP AN AI MODEL THAT CAN PROVIDE ACCURATE STROKE IDENTIFICATION AND CLASSIFICATION FOR PATIENTS NATIONWIDE. OUR FINDINGS WILL PROVIDE SCIENTIFIC EVIDENCE ON THE HEALTH EFFECTS OF INDIVIDUAL PM2.5 COMPONENTS TO INFORM TARGETED REGULATIONS AND AIR QUALITY STANDARDS. ADDITIONALLY, THIS STUDY WILL HELP SHAPE POLICIES TO REDUCE AIR POLLUTION EXPOSURES AND SUPPORT PREVENTION AND TREATMENT STRATEGIES TO PROTECT CEREBROVASCULAR HEALTH, WITH EMPHASIS ON SUSCEPTIBLE GROUPS.

Source notice