A Predictive Model for Lung Cancer Screening Non-Adherence in a Community Setting Healthcare Network
Menée à l'aide d'un algorithme d'apprentissage automatique et de données portant sur 1 875 personnes ayant un subi un examen de dépistage du cancer du poumon par tomographie numérique, cette étude évalue la performance d'un modèle, basé sur des données cliniques et démographiques, pour prédire le risque de non adhésion au dépistage
Background : Lung Cancer Screening (LCS) decreases lung cancer mortality. However, its benefit may be limited by non-adherence to screening. While factors associated with LCS non-adherence have been identified, to the best of our knowledge, no predictive models have been developed to predict LCS non-adherence. The purpose of this study is to develop a predictive model leveraging a machine learning (ML) model to predict LCS non-adherence risk.
Methods : A retrospective cohort of patients who enrolled in our LCS program between 2015 to 2018 was used to develop a model to predict the risk of non-adherence to annual LCS after the baseline examination. Clinical and demographic data were used to fit logistic regression, random forest, and gradient boosting models that were internally validated based on accuracy and Area Under the Receiver Operating Curve (AUC).
Results : A total of 1,875 individuals with baseline LCS were included in the analysis with 1,264 (67.4%) as non-adherent. Non-adherence was defined based on baseline chest CT findings. Clinical and demographic predictors were used based on availability and significance. The gradient boosting model had the highest AUC (0.89, 95% CI: 0.87-0.90) with a mean accuracy of 0.82. Referral specialty, insurance type, and baseline LungRADS score were the best predictors of non-adherence to LCS.
Conclusions : We developed an ML model using readily available clinical and demographic data to predict LCS non-adherence with high accuracy and discrimination. After further prospective validation, this model can be used to identify patients for interventions to improve LCS adherence and decrease lung cancer burden.
JNCI Cancer Spectrum , article en libre accès, 2022