# Development and validation of predictive model for emesis in cervical cancer patients receiving concurrent chemoradiotherapy based on multi-institutional retrospective study

## TL;DR

This study created a model to predict chemotherapy-induced nausea and vomiting in cervical cancer patients undergoing chemoradiotherapy, which could help tailor antiemetic treatments.

## Contribution

The study introduces a validated predictive model for CINV in cervical cancer patients receiving CCRT, which is a novel tool for personalized antiemetic strategies.

## Key findings

- The model achieved an ROC-AUC of 0.772 in the training dataset and 0.808 in the validation dataset.
- The model includes predictors such as age, smoking history, and cancer stage, showing good discrimination and calibration.
- The model may help individualize antiemetic strategies and improve patient care.

## Abstract

Chemotherapy-induced nausea and vomiting (CINV) remains a significant barrier to treatment adherence and quality of life in cervical cancer patients receiving chemoradiotherapy. No validated predictive models exist to assess CINV risk in this population. We aimed to develop and temporal validate a predictive model for CINV incidence in cervical cancer patients receiving concurrent chemoradiotherapy (CCRT). This multi-institutional, retrospective cohort study analyzed 921 patients who received CCRT with weekly cisplatin (40 mg/m2) between January 2016 and March 2024. Candidate predictors were selected through expert consultations and literature reviews. A multivariate logistic regression model was developed using training data, and the model with the highest receiver operating characteristic-area under the curve (ROC-AUC) was tested using validation data. The model (age, smoking history, total radiation dose, chemotherapy history, 5-hydroxytryptamine 3 receptor antagonist use, cancer stage) showed good discrimination. In the training dataset, the model achieved an ROC-AUC of 0.772 (95% confidence interval [CI], 0.717–0.827). In the validation dataset, the model showed high discriminative ability (ROC-AUC, 0.808; 95%CI, 0.763–0.853) and good calibration (intraclass correlation coefficient, 0.826; p < 0.001). We developed and validated a clinically useful CINV prediction model for cervical cancer patients receiving CCRT. This tool may individualize antiemetic strategies and improve care.

The online version contains supplementary material available at 10.1038/s41598-025-21494-5.