Abstract:

Objective To develop and validate a predictive model for malnutrition risk in patients with hematologic neoplasms，providing a reference for the early identification of high-risk patients. Methods Using a convenience sampling method，292 patients with hematologic neoplasms admitted to the Department of Hematology of a tertiary hospital in Qingdao between January and March 2025 were enrolled as a modeling cohort. Concurrently，126 patients admitted to another tertiary hospital in Qingdao during the same period served as a validation cohort. The influencing factors were screened by LASSO regression and Logistic regression，and the nomogram was constructed. Internal validation was performed using the Bootstrap method. The model’s predictive performance was evaluated using the area under the receiver operating characteristic（ROC） curve and the Hosmer-Lemeshow test. Results The incidence of malnutrition was 27.40% in the modeling cohort and 25.40% in the validation cohort. Logistic regression analysis identified age，body mass index（BMI），presence of gastrointestinal reactions，serum albumin level，C-reactive protein level，depression score，and performance status score as independent risk factors for malnutrition. In the modeling cohort，the area under the ROC curve（AUC） was 0.924（95%CI：0.888~0.960），with a sensitivity of 81.3% and a specificity of 90.6%. The Hosmer-Lemeshow test indicated good model fit （χ²=6.349，P=0.608）. In the validation cohort，the AUC was 0.895（95%CI：0.838~0.952），with a sensitivity of 81.3% and a specificity of 85.1%，and the Hosmer-Lemeshow test also showed good calibration（χ²=5.456，P=0.708）. Conclusion The developed risk prediction model demonstrates favorable predictive performance，facilitating the early identification of patients with hematologic neoplasms at high risk for malnutrition and offering a practical basis for healthcare providers to implement tailored nutritional interventions.

Key words: Hematologic Neoplasms, Malnutrition, Root Cause Analysis, Nomograms, Nursing Care

摘要：

目的 构建并验证血液肿瘤患者营养不良风险预测模型，为早期识别高风险患者提供参考。 方法 采用便利抽样法，选取2025年1—3月青岛市某三级甲等医院血液内科收治的292例血液肿瘤患者作为建模组，并以同期青岛市另1所同级医院收治的126例患者作为验证组。通过LASSO回归、Logistic回归筛选影响因素，并绘制列线图。采用Bootstrap法进行内部验证，采用受试者操作特征曲线下面积和Hosmer-Lemeshow检验评估模型的预测性能。 结果 建模组和验证组的营养不良发生率分别为27.40%和25.40%。Logistic回归分析结果显示，年龄、BMI、有无胃肠道反应、血清白蛋白、C反应蛋白、抑郁得分、体能状态得分是血液肿瘤患者营养不良的影响因素。建模组的受试者操作特征曲线下面积为0.924（95%CI：0.888~0.960），灵敏度为81.3%，特异度为90.6%；Hosmer-Lemeshow检验结果显示，χ²=6.349，P=0.608。验证组的受试者操作特征曲线下面积为0.895（95%CI：0.838~0.952），灵敏度为81.3%，特异度为85.1%，Hosmer-Lemeshow检验结果显示，χ²=5.456，P=0.708。 结论 该研究构建的风险预测模型具有良好的预测性能，有助于早期识别血液肿瘤患者营养不良高风险人群，为医护人员实施个体化营养干预提供依据。

关键词: 血液肿瘤, 营养不良, 影响因素分析, 列线图, 护理