Prediction of radiation pneumonia after radiotherapy for esophageal cancer using a unified fractional dosiomics combined model.

ABSTRACT:

Objective

This study aimed to construct an optimal model to predict radiation pneumonia (RP) after radiotherapy for esophageal cancer using unified fractional dosiomics and to investigate the improvements in the prediction efficiency of each model for RP.

Methods

The clinical data, DVH, pre-treatment CT, and dose distribution of 182 patients were retrospectively analyzed.The independent risk factors were screened using univariate and multivariate logistic regression. The mutual information (MI),least absolute shrinkage and selection operator (LASSO), and recursive feature elimination (RFE) methods were used to screen the omics features. The AUC values of ROC, calibration curves, and clinical decision curves were calculated to evaluate the efficacy and trends of each model.

SUBMITTER: Yang T 

PROVIDER: S-EPMC10646633 | biostudies-literature | 2023 Dec

REPOSITORIES: biostudies-literature