2024 World Conference on Lung Cancer (WCLC) - Posters-P4.04C.10 Validation of LungFlag™ Prediction Model Using Electronic Medical Records (EMR) On Taiwan Data

P4.04C.10 Validation of LungFlag™ Prediction Model Using Electronic Medical Records (EMR) On Taiwan Data

Pdf Summary

The document discusses a study evaluating the efficacy of the LungFlag model versus the PLCOm2012 model in identifying high-risk lung cancer (LC) populations in Taiwan using electronic medical records (EMR). Lung cancer is a leading cause of cancer deaths in Taiwan, with more than half affecting non-smokers. To address this, the Ministry of Health launched a screening program using low-dose computed tomography (LDCT) targeting high-risk groups.<br /><br />The study extracted data from 65,882 individuals (10,557 with LC and 50,194 without) from National Taiwan University Hospital's records, focusing on individuals aged 40 to 80 between 2010-2022. The models were compared based on metrics like Area Under the Curve (AUC), sensitivity, and odds ratio (OR) at various false positive rates (FPR) of 3% and 10%.<br /><br />Results indicated both LungFlag and PLCOm2012 surpassed the USPSTF criteria in detecting at-risk individuals within the ever-smoker population. However, LungFlag showed statistically significant superiority over PLCOm2012. Specifically, at a 3% FPR, LungFlag displayed an OR of 11.7 among ever-smokers and 9.0 among USPSTF eligible groups, compared to PLCOm2012's OR of 5.9 and 4.3, respectively. LungFlag also presented an OR of 9.9 for never-smokers.<br /><br />Overall, LungFlag demonstrated it could effectively utilize Taiwanese EMR data, outperforming the PLCOm2012 model. The study concludes that LungFlag could significantly support identifying high-risk populations for LC, especially among ever and never smokers. It suggests further local data retraining to enhance its utility for never-smokers in Taiwan.

Asset Subtitle

Eran Netanel Choman

Meta Tag

Speaker Eran Netanel Choman

Topic Screening & Early Detection

Keywords

LungFlag model

PLCOm2012 model

high-risk lung cancer

Taiwan

electronic medical records

low-dose computed tomography

Area Under the Curve

odds ratio

ever-smokers

never-smokers