Could Computable Phenotypes Determine Eligibility for Lung Cancer Screening?

By: Chris Schimpf, MSW
Posted: Tuesday, February 18, 2025

Electronic health records–based computable phenotypes may be effective in identifying patients who are eligible for lung cancer screening, according to the results of a study published in JCO Clinical Cancer Informatics. Jiang Bian, PhD, of the University of Florida, Gainesville, and colleagues developed and validated two sets of computable phenotype algorithms for assessing individuals’ lung cancer screening eligibility in real-world settings. In their findings, the researchers stressed the potential of the practice to aid clinical decision-making and optimize patient care.

“[Our] findings demonstrate the robustness of computable phenotypes in effectively assessing lung cancer screening eligibility and highlight the benefits of integrating diverse data sources,” the investigators stated. “[They] underscore the critical need for improved documentation of smoking information in electronic health records, demonstrate the value of artificial intelligence techniques in enhancing computable phenotype performance, and confirm the effectiveness of electronic health records–based computable phenotypes in identifying lung cancer screening–eligible individuals.”

A total of 5,778 patients who underwent low-dose CT for lung cancer screening between 2012 and 2022 were included in the study. Using data from the University of Florida’s Health Integrated Data Repository, the researchers based the two sets of computable phenotype rules on the 2013 and 2020 lung cancer screening guidelines of the U.S. Preventive Services Task Force. Integrating both structured and unstructured data, the algorithms focused on age (55–80 years for 2013 and 50–80 years for 2020), smoking status (current, former, and other), and pack-years (≥ 30 for 2013 and ≥ 20 for 2020).

The investigators reported that the rules achieved high F1 scores of 0.75 (2013) and 0.84 (2020), using both structured and unstructured data. When these data were applied to the study’s cohort, 65.5% and 69.7% of participants were identified as eligible for lung cancer screening under the 2013 and 2020 guidelines, respectively.

Disclosure: Dr. Bian reported no conflicts of interest. For full disclosures of the other study authors, visit ascopubs.org.

JCO Clinical Cancer Informatics

---