ObjectiveThe objective of this study was to create a mapping algorithm by utilizing traditional regression analyses and a machine learning approach to estimate EQ-5D-5 L values based on EORTC QLQ-LC43 data in the absence of direct EQ-5D-5 L measurements.MethodsData for EQ-5D-5 L and EORTC QLQ-LC43 were collected from patients with lung cancer at the Departments of Thoracic Surgery, Medical Oncology, and Radiation Oncology at Sichuan Cancer Hospital. Mapping algorithms were applied using the ordinary least squares model (OLS), Tobit model, Beta mixture regression (BM), the adjusted limited dependent variable mixture model (ALDVMM), and ridge regression (RR) as a machine learning model to map QLQ-LC43 results based on EQ-5D-5 L scores. To develop these models, dimension scores, squared items, and interaction items were incorporated. Performance metrics, including R-2, root mean square error (RMSE), and mean absolute error (MAE), were used to identify the optimal model. The stability of the models was assessed using five-fold cross-validation (CV).ResultsThe Beta mixture regression model (BETAMIX M1A), incorporating all dimensions of QLQ-C30 and QLQ-LC13 as covariates, exhibited the best mapping performance. The final prediction metrics were R-2=0.816, RMSE = 0.125, MAE = 0.083, AIC=-717.810, and BIC=-482.609. The BM model has good explanatory ability and low prediction error. Five-fold cross-validation (CV) results also demonstrated that the BM model had the best mapping power.ConclusionsThis study developed an optimized mapping algorithm to predict the utility index from the QLQ-LC43 to the EQ-5D-5 L, offering an effective alternative for estimating EQ-5D-5 L values when preference-based health utility data are unavailable.