Li Xia, Wei Ben-Kai, Li Fan, Yan Huan-Huan, Shen Jun
Department of Breast Surgery, the First People's Hospital of Lianyungang, The Affiliated Hospital of Xuzhou Medical University, Lianyungang, 222000, Jiangsu Province, People's Republic of China.
Department of General Surgery, the First People's Hospital of Lianyungang, The Affiliated Hospital of Xuzhou Medical University, Lianyungang, 222000, Jiangsu Province, People's Republic of China.
Psychol Res Behav Manag. 2025 Feb 15;18:315-329. doi: 10.2147/PRBM.S501127. eCollection 2025.
This study aims to develop and validate a predictive model for short-term post-treatment anxiety trajectories in patients with breast cancer, utilizing baseline patient characteristics and initial anxiety scores to inform precise clinical interventions.
Baseline characteristics were collected from 424 patients diagnosed with breast cancer who underwent surgical treatment at our hospital between January 1, 2021, and December 30, 2022. Anxiety levels were assessed using the Self-Rating Anxiety Scale (SAS) scores at admission and at 3-, 6-, 9-, and 12-months post-treatment. Distinct trajectories of SAS score changes were identified and categorized. Variables were screened, and multiple models were developed. The optimal model was identified through comparative analysis, and a nomogram was generated following model simplification.
We found three distinct trends in the trajectory of anxiety, but we grouped them into two broad categories: gradual reduction of anxiety and persistent anxiety. LM Model was established by logistic regression, and Model 1 and Model 2 were established by Random Forest (RF) and eXtreme Gradient Boosting (Xgboost) screening variables. The ROC curve areas in the validation set were 0.822 (0.757-0.887), 0.757 (0.680-0.834) and 0.781 (0.710-0.851), respectively. Model comparison, using Net Reclassification Improvement (NRI) and Integrated Discrimination Improvement (IDI), identified the Lm model as optimal, which underwent further simplification and value assignment. Decision Curve Analysis (DCA) and Clinical Impact Curve (CIC) analyses confirmed the superiority of model-based interventions over general interventions.
Distinct anxiety trajectories are observed in patients diagnosed with breast cancer during the first 12 months post-treatment. Predictive modeling based on baseline characteristics is feasible although though further research is warranted.
本研究旨在开发并验证一种针对乳腺癌患者治疗后短期焦虑轨迹的预测模型，利用患者的基线特征和初始焦虑评分来指导精准的临床干预。
收集了2021年1月1日至2022年12月30日期间在我院接受手术治疗的424例乳腺癌患者的基线特征。采用自评焦虑量表（SAS）在入院时以及治疗后3、6、9和12个月评估焦虑水平。识别并分类SAS评分变化的不同轨迹。筛选变量并开发多个模型。通过比较分析确定最佳模型，并在模型简化后生成列线图。
我们在焦虑轨迹中发现了三种不同趋势，但将它们分为两大类：焦虑逐渐减轻和持续性焦虑。通过逻辑回归建立了LM模型，通过随机森林（RF）和极端梯度提升（Xgboost）筛选变量建立了模型1和模型2。验证集中的ROC曲线面积分别为0.822（0.757 - 0.887）、0.757（0.680 - 0.834）和0.781（0.710 - 0.851）。使用净重新分类改善（NRI）和综合辨别改善（IDI）进行模型比较，确定LM模型为最佳模型，该模型进一步简化并赋值。决策曲线分析（DCA）和临床影响曲线（CIC）分析证实了基于模型的干预优于一般干预。
在确诊为乳腺癌的患者治疗后的前12个月观察到不同的焦虑轨迹。基于基线特征的预测建模是可行的，尽管仍需进一步研究。
