The Third Affiliated Hospital of Kunming Medical University, Kunming, China.
Children's Hospital Affiliated of Kunming Medical University, Kunming Children's Hospital, Kunming, Yunnan, China.
Medicine (Baltimore). 2024 Nov 29;103(48):e40673. doi: 10.1097/MD.0000000000040673.
During the progression of lung cancer, cancer pain is a common complication. Currently, there are no accurate tools or methods to predict the occurrence of cancer pain in lung cancer. Our study aims to construct a predictive model for lung cancer pain to assist in the early diagnosis of cancer pain and improve prognosis. We retrospectively collected clinical data from 300 lung cancer patients between March 2013 and March 2023. First, we compared the clinical data of the groups with and without cancer pain. Significant factors were further screened using random forest analysis (IncMSE% > 2) to identify those with significant differences. Finally, these factors were incorporated into a multifactorial logistic regression model to develop a predictive model for lung cancer pain. The predictive accuracy and performance of the model were assessed using receiver operating characteristic (ROC) curves, calibration curves, and decision curve analysis (DCA) analysis. Our study collected data from 300 lung cancer patients, including 100 in the pain-free group and 200 in the pain group. Subsequently, we conducted univariate analysis on 22 factors and selected statistically significant factors using random forest methods. Ultimately, lymphocytes(LYM) percentage, bone metastasis, tumor necrosis factor alpha (TNFα), and interleukin-6 (IL6) were identified as key factors. These 4 factors were included in a multivariate logistic regression analysis to construct a predictive model for lung cancer pain. The model demonstrated good predictive ability, with an area under the curve (AUC) of 0.852 (95% CI: 0.806-0.899). The calibration curve indicated that the model has good accuracy in predicting the risk of lung cancer pain. DCA further emphasized the model's high accuracy. The model was finally validated using 5-fold cross-validation. We developed a reliable predictive model for cancer pain in lung cancer. This can provide a theoretical basis for future large-sample, multi-center studies and may also assist in the early prevention and intervention of cancer pain in lung cancer.
在肺癌的发展过程中，癌痛是常见的并发症。目前，没有准确的工具或方法来预测肺癌中癌痛的发生。我们的研究旨在构建一个肺癌疼痛预测模型，以协助早期诊断癌痛并改善预后。我们回顾性地收集了 2013 年 3 月至 2023 年 3 月间 300 例肺癌患者的临床资料。首先，我们比较了有癌痛和无癌痛两组的临床资料。使用随机森林分析（IncMSE%>2）进一步筛选有显著差异的显著因素，然后将这些因素纳入多因素逻辑回归模型，以建立肺癌疼痛预测模型。使用受试者工作特征（ROC）曲线、校准曲线和决策曲线分析（DCA）分析评估模型的预测准确性和性能。我们的研究共收集了 300 例肺癌患者的数据，其中无痛组 100 例，疼痛组 200 例。随后，我们对 22 个因素进行了单因素分析，并使用随机森林方法选择了有统计学意义的因素。最终，淋巴细胞（LYM）百分比、骨转移、肿瘤坏死因子-α（TNFα）和白细胞介素-6（IL6）被确定为关键因素。这 4 个因素被纳入多因素逻辑回归分析，以构建肺癌疼痛预测模型。该模型显示出良好的预测能力，曲线下面积（AUC）为 0.852（95%CI：0.806-0.899）。校准曲线表明该模型在预测肺癌疼痛风险方面具有良好的准确性。DCA 进一步强调了该模型的高准确性。该模型最终通过 5 折交叉验证进行了验证。我们开发了一个可靠的肺癌癌痛预测模型。这可为未来的大样本、多中心研究提供理论依据，也可能有助于早期预防和干预肺癌癌痛。
