EMAI-COPD: An Ethical Multimodal AI Model for Preventive, Diagnostic, and Therapeutic Interventions in Chronic Obstructive Pulmonary Disease

Abstract
Chronic Obstructive Pulmonary Disease (COPD) is a major cause of morbidity and mortality in the United
States. We propose to integrate existing omics data (genetic, transcriptomic, epigenetic, proteomic,
metabolomics, and spirometrics) from multiple TOPMed cohorts, medical imaging data (CT), wearable
data (Fitbit and Apple), social determinants of health (SDOH), and clinical research data (text and tabular)
to develop EMAI-COPD, a novel and open-source ethics-aware multimodal AI (MAI) model to enable a
variety of downstream applications in preventive, diagnostic and therapeutic interventions in COPD. These
include medical applications, such as disease progression and prediction modeling, disease subtyping and
pathway analysis, and early exacerbation detection. We envision EMAI-COPD as both a blueprint and a
starting model to be adopted and adapted to generate a family EMAI-X models for other diseases (i.e., “X”).
Our team is well positioned to take on this task. The team constitutes of 1) AI experts with extensive
research experience in introducing novel AI models as well as agile model development lifecycle for large-
scale models, 2) domain experts with decades of research and clinical experience in COPD interventions,
and 3) medical ethics experts that could help us form the ethics framework required to develop EMAI-
COPD as an intrinsically ethics-aware tool, including awareness of bias, fairness, privacy, accountability
and transparency. Along with a wide set of stakeholders, which have worked with the team in the past, this
team will adopt an agile approach to co-design and enhance EMAI-COPD to meet all ethical and
performance requirements of the model.

Grant Number: 1OT2OD038074-01
NIH Institute/Center: NIH
Principal Investigator: Farnoush Banaei-Kashani