Quality Control Department, The Third Hospital of Quzhou, China.
Psychiatry Department, The Third Hospital of Quzhou, China.
Biomed Res Int. 2021 Jul 17;2021:9201235. doi: 10.1155/2021/9201235. eCollection 2021.
Depression is highly prevalent in non-Hodgkin's lymphoma (NHL) patients undergoing chemotherapy. The social stress associated with malignancy induces neurovascular pathology promoting clinical levels of depressive symptomatology. The purpose of this study was to establish an effective depressive symptomatology risk prediction model to those patients.
This study included 238 NHL patients receiving chemotherapy, 80 of whom developed depressive symptomatology. Different types of variables (sociodemographic, medical, and psychosocial) were entered in the models. Three prediction models (support vector machine-recursive feature elimination model, random forest model, and nomogram prediction model based on logistic regression analysis) were compared in order to select the one with the best predictive power. The selected model was then evaluated using calibration plots, ROC curves, and -index. The clinical utility of the nomogram was assessed by the decision curve analysis (DCA).
The nomogram prediction has the most efficient predictive ability when 10 predictors are included (AUC = 0.938). A nomogram prediction model was constructed based on the logistic regression analysis with the best predictive accuracy. Sex, age, medical insurance, marital status, education level, per capita monthly household income, pathological stage, SSRS, PSQI, and QLQ-C30 were included in the nomogram. The -index was 0.944, the AUC value was 0.972, and the calibration curve also showed the good predictive ability of the nomogram. The DCA curve suggested that the nomogram had a strong clinical utility.
We constructed a depressive symptomatology risk prediction model for NHL chemotherapy patients with good predictive power and clinical utility.
化疗的非霍奇金淋巴瘤（NHL）患者中普遍存在抑郁。与恶性肿瘤相关的社会压力会导致神经血管病理学，从而促进临床抑郁症状水平。本研究的目的是为这些患者建立有效的抑郁症状预测模型。
本研究纳入了 238 名接受化疗的 NHL 患者，其中 80 名患者出现了抑郁症状。不同类型的变量（社会人口统计学、医学和心理社会）被纳入模型。比较了三种预测模型（支持向量机递归特征消除模型、随机森林模型和基于逻辑回归分析的列线图预测模型），以选择预测能力最佳的模型。然后使用校准图、ROC 曲线和 - 指数评估选定的模型。通过决策曲线分析（DCA）评估列线图的临床实用性。
当纳入 10 个预测因子时，列线图预测具有最高的预测效率（AUC = 0.938）。基于具有最佳预测准确性的逻辑回归分析构建了列线图预测模型。性别、年龄、医疗保险、婚姻状况、教育程度、人均月家庭收入、病理分期、SSRS、PSQI 和 QLQ-C30 被纳入列线图。 - 指数为 0.944，AUC 值为 0.972，校准曲线也显示了列线图的良好预测能力。DCA 曲线表明列线图具有很强的临床实用性。
我们为 NHL 化疗患者构建了具有良好预测能力和临床实用性的抑郁症状风险预测模型。
