Study maps early lung cancer symptoms to improve diagnosis odds
Researchers analyzed how patient-reported symptoms and background factors differ between early-stage and advanced lung cancer, using machine learning to predict which signs warrant urgent screening. The findings could help clinicians and health systems identify cancers earlier, when treatment is more effective and less costly.
Originaltitel: PEX-LC-stage-symptoms
This study characterised and compared background factors and symptoms at diagnosis across patients with non-advanced lung cancer, advanced lung cancer, and cancer-free controls. Univariate logistic regression and multivariate machine learning models (regularized logistic regression, random forest, and extreme gradient boosting) were used to identify variables that contribute to the detection of early- and late-stage lung cancer, and to assess the potential predictive value of detailed patient-reported symptom data and background factors. The current release contains all code used to produce the results and visualizations in the submitted manuscript after revision round 2.