Development and validation of a novel AI-derived index for predicting COPD medical costs in clinical practice

Background: Chronic Obstructive Pulmonary Disease (COPD) is a major contributor to global morbidity and healthcare costs. Accurately predicting these costs is crucial for resource allocation and patient care. This study developed and validated an AI-driven COPD Medical Cost Prediction Index (MCPI) to forecast healthcare expenses in COPD patients. Methods: A retrospective analysis of 396 COPD patients was conducted, utilizing clinical, demographic, and comorbidity data. Missing data were addressed through advanced imputation techniques to minimize bias. The final predictors included interactions such as Age x BMI, alongside Tumor Presence, Number of Comorbidities, Acute Exacerbation frequency, and the DOSE Index. A Gradient Boosting model was constructed, optimized with Recursive Feature Elimination (RFE), and evaluated using 5-fold cross-validation on an 80/20 train-test split. Model performance was assessed with Mean Squared Error (MSE), Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), and R-squared (R2). Results: On the training set, the model achieved an MSE of 0.049, MAE of 0.159, MAPE of 3.41 %, and R2 of 0.703. On the test set, performance metrics included an MSE of 0.122, MAE of 0.258, MAPE of 5.49 %, and R2 of 0.365. Tumor Presence, Age, and BMI were identified as key predictors of cost variability. Conclusions: The MCPI demonstrates strong potential for predicting healthcare costs in COPD patients and enables targeted interventions for high-risk individuals. Future research should focus on validation with multicenter datasets and the inclusion of additional socioeconomic variables to enhance model generalizability and precision.