A dynamic online nomogram for predicting depression risk in cancer patients based on NHANES 2007-2018

Abstract

Background: Cancer, recognized as a significant global public health issue, exhibits a notably elevated prevalence of depression among its patient population. This study aimed to construct a nomogram to predict depression risk in cancer patients.

Methods: In this study, the training set comprises 70 % of the dataset, while the test set comprises 30 %. On the training set, we employed the least absolute shrinkage and selection operator (LASSO) regression in conjunction with multivariable logistic regression to identify key variables, subsequently constructing a prediction model. ROC curves, calibration tests, and decision curve analysis (DCA) were used to evaluate model performance.

Results: A total of 2604 participants were included in this study. The nomogram predictors encompassed age, poverty-income ratio (PIR), sleep disorder, and food security. We have developed a web-based dynamic nomogram incorporating these factors (available at https://xiaoshuweiya.shinyapps.io/DynNomapp/). The area under the model's ROC curve (AUC) was 0.803 and 0.766 when evaluated on the training and test sets, respectively. These AUC values highlight the model's robustness and reliability in making accurate predictions across different datasets. The calibration curves demonstrated consistency between the model's predicted and actual results. Additionally, the decision curve analysis further substantiated the potential clinical utility of the nomograms.

Conclusions: This study developed a nomogram to help clinicians identify high-risk populations for depression among cancer patients, providing a scientific method for early detection and assessment of depression risk.

Keywords: Cancer; Depression; Nomogram; Prediction model.