Abstract
Introduction: Early detection and timely treatment (Tx) initiation are critical to improving lung cancer (LC) outcomes. This study assessed the natural language processing (NLP)-assisted incidental pulmonary nodule (IPN) evaluation program, which employs chest computer tomography (CT) report analysis as an LC diagnostic screening (LCS) tool to identify suspicious lung findings (SLF) necessitating further investigation, and evaluated its impact on prognosis and diagnostic work-up and Tx timelines for patients with LC. Materials and Methods: Consecutive LC patients (n = 200) diagnosed at Assuta Medical Centers (AMC) between January 2019 and December 2022 were retrieved from the AMC electronic database using the MDClone big data platform, and divided into two groups: group A (NLP-assisted IPN evaluation, n = 100) and group B (traditional referral for evaluation of SLF by the community physician, n = 100). Stage at diagnosis, different diagnostic work-up and Tx timelines, and overall survival (OS) were assessed. Results: The NLP-assisted IPN evaluation program led to a significant stage shift (stage I disease: 48% vs. 27% in groups A and B, respectively, p = 0.013). Although the time from imaging to Tx initiation was similar (2.1 ± 5.3 months vs. 2.6 ± 5.9 months in groups A and B, respectively, p = 0.654), the time to systemic Tx (p = 0.035) and the time to radiotherapy (p = 0.044) were significantly shorter in group A. Conclusions: Implementing an NLP-assisted IPN evaluation program may enable earlier LC detection, driving a stage shift towards earlier diagnosis, improved diagnostic efficiency, and expedited time-critical interventions.