Development and validation of a risk prediction model for benign and malignant pulmonary nodules combined with artificial intelligence

Background: Lung cancer is the malignant tumor with the highest mortality rate in the world, and high-risk pulmonary nodules are of great significance for early diagnosis of lung cancer. We aimed to develop and validate an effective pulmonary nodule risk prediction model to improve the early diagnosis rate of lung cancer. Methods: A retrospective analysis was conducted on 610 patients with pulmonary nodules with histopathological results from May 2021 to August 2022, and variables assessing the benign and malignant nature of pulmonary nodules were screened through logistic regression to develop a nomogram. 120 patients with pulmonary nodules with histopathological results were again collected for external validation. Both internal verification and external verification adopt bootstrap sampling method. Results: The clinical prediction model achieved an accuracy of 84.13%, a sensitivity of 86.01%, a specificity of 79.28%, and an AUC of 0.896 in the training cohort. In the validation cohort, the accuracy was 83.10%, the sensitivity was 89.73%, the specificity was 68.66%, and the AUC was 0.856. The calibration curve demonstrated good agreement between the predicted and observed results. Decision curve analysis(DCA) further confirmed the clinical benefits of the early diagnosis model. In the independent validation cohort (n = 213), the AUC of this model was 0.856, outperforming the Mayo model (AUC=0.689) and the VA model (AUC=0.606). It was also found that the model performed well in predicting gender, nodule location (upper lobe or non-upper lobe), and age （≤45 years or >45 years). Conclusions：We developed and validated an effective multivariate model to predict the malignant risk of pulmonary nodules, which has good diagnostic performance and clinical practicability, and can provide a theoretical basis for judging the nature of pulmonary nodules in clinical practice.