Grant Number:	1R01CA304878-01A1 Interpret this number
Primary Investigator:	Salahudeen, Ameen
Organization:	University Of Illinois At Chicago
Project Title:	A Multimodal Approach to Increase Accuracy and Generalizability in Lung Cancer Early Detection
Fiscal Year:	2026

Annual use of low-dose CT (LDCT) and Lung-RADs for reporting and managing of LDCT findings reduces lung cancer mortality, but is imperfect with frequent false positives and negatives and eligibility criteria that do not benefit all populations. Indeed, we and others have observed a greater rate of lung cancers within parts of the general population. In that regard, current LDCT eligibility criteria intrinsically promotes cancer-related differences because risk is variable among certain populations. There is, therefore, a critical need to develop a new paradigm for individualized early detection of lung cancer to minimize cancer-related differences. In the absence of individualized testing, lung cancer mortality in screening and survival are likely to remain intransigent. Our overall objective for this application is to validate a multimodal risk assessment strategy for the early detection of lung cancer. Advances in artificial intelligence, specifically computer vision and machine learning, have enabled accurate detection of lung cancer prior to visible radiographic disease. Our central hypothesis is that clinical features, computer vision inference from CT images, and circulating DNA-based fragmentomics integrated into multimodal risk assessment provides greater accuracy and fewer false positives compared to standard of care. The rationale for the proposed research is that validation of multimodal screening with greater accuracy than annual LDCT would provide a strong basis for its continued development and implementation as individualized screening with broader reach than current LDCT eligibility criteria. To this end, our proposed approach will be evaluated in patients undergoing lung cancer screening under current eligibility criteria in a prospective manner. In addition, we will carry out pilot studies in our patient population – the majority of whom exhibit cancer-related differences and at high risk of lung cancer, but would otherwise not be eligible to undergo lung cancer screening under current guidelines. Given that there are no data standards for lung cancer risk assessments from computer vision, circulating DNA-based screening, or integrated risk results, we will develop informatics tools and resources to facilitate health care management and interoperability. We will also identify obstacles to implementing this digital health approach in the community and across all healthcare settings. Our research will be supported by a multi-pronged community engagement strategy, with a community partner on our study team, a community advisory board, and community townhalls to disseminate findings and solicit insight for interpretation and implications for implementation. These results are intended to have an important positive impact as they would provide the framework for the design and execution of multicenter interventional studies to validate the benefit of improved screening, early detection, and survival for all.





None