Study
NLST
(Learn more about this study)
Project ID
NLST-1125
Initial CDAS Request Approval
Oct 3, 2023
Title
Addressing equity in lung cancer early detection through identification of high risk clinical and radiographic features in minority and disadvantaged patients.
Summary
Lung cancer is the leading cause of cancer death globally, and advances in standard of care low dose CT (LDCT) screening has contributed to a reduction of lung cancer mortality through detection of early-stage and curable disease. In the United States disparities in lung cancer mortality exist for minority and disadvantaged populations, including African Americans and Hispanic Whites. These disparities are associated with barriers to accessing and continuing LDCT screening, as well as eligibility criteria for screening based on epidemiologic data from Non-Hispanic Whites. Current data suggests that lung cancer is detected in approximately 1% of the screened general population, whereas in African Americans and Hispanic Whites, the incidence is 1.5 to 2 times greater. This means that current screening guidelines disproportionately leave out larger numbers of at risk African American and Hispanic Whites. At UIC where nearly 60% of our lung cancer screening population is African American, the lung cancer diagnosis rate is 2.6 patients for every 100 patients – nearly twice the general population rate. To achieve health equity in our population, we need the ability to capture more lung cancer diagnoses through LDCT screening. This, in turn, requires the ability to identify and screen a broader group of at risk lung cancer patients in minority and disadvantaged populations.
Aims
We will utilize AI and machine learning techniques to predict cancer incidence in patients by analyzing clinical and demographic features. We will apply machine learning methods, such as Support Vector Machines, Logistic Regression, and Neural Network encoding, to identify high risk clinical features that predict lung cancer in our screening population. The population will be divided into a training, test, and validation cohorts comprising 2083 patients, allowing us to identify risk features beyond the primary lung cancer screening criteria of age and smoking status.
These training machine learning models will contribute to the development of a risk assessment tool for 1) further outreach and engagement in LDCT eligible UIC patients identified via EPIC EMR and 2) provide a rationale for further validation in minority and disadvantaged cohorts outside of UIC and 3) provide rationale and justification for a pilot program to be funded to enroll minority and disadvantaged patients in LDCT screening who may not be eligible according to USPSTF guidelines.
Collaborators
Abdul Zakkar, M.D., University of Illinois at Chicago
Ameen Salahudeen, M.D., Ph. D., University of Illinois at Chicago
Ryan Nguyen, D.O., University of Illinois at Chicago
Aly Azeem Khan, Ph.D., University of Chicago
