中国组织工程研究

中国组织工程研究 ›› 2022, Vol. 26 ›› Issue (32): 5085-5090.doi: 10.12307/2022.891

• 骨组织构建 bone tissue construction •    下一篇

中年男性骨量减少列线图预测模型的构建

沈炼伟1，王  维2   

  1. 1锦州医科大学附属第一医院， 辽宁省锦州市   121000；2锦州医科大学附属第一医院康复科， 辽宁省锦州市   121000
  • 收稿日期:2021-11-25 接受日期:2021-12-04 出版日期:2022-11-18 发布日期:2022-05-12
  • 通讯作者: 王维，硕士，副主任医师，锦州医科大学附属第一医院康复科， 辽宁省锦州市 121000
  • 作者简介:沈炼伟，男，1995年生，广东省潮安县人，汉族，锦州医科大学康复医学与理疗学在读硕士，主要从事肌骨康复的研究。

Construction of a nomogram model for predicting bone mass loss in middle-aged men

Shen Lianwei1, Wang Wei2   

  1. 1The First Affiliated Hospital of Jinzhou Medical University, Jinzhou 121000, Liaoning Province, China; 2Department of Rehabilitation, the First Affiliated Hospital of Jinzhou Medical University, Jinzhou 121000, Liaoning Province, China
  • Received:2021-11-25 Accepted:2021-12-04 Online:2022-11-18 Published:2022-05-12
  • Contact: Wang Wei, Master, Associate chief physician, Department of Rehabilitation, the First Affiliated Hospital of Jinzhou Medical University, Jinzhou 121000, Liaoning Province, China
  • About author:Shen Lianwei, Master candidate, The First Affiliated Hospital of Jinzhou Medical University, Jinzhou 121000, Liaoning Province, China

PDF

40

摘要：

文题释义：
K折交叉验证：数据集分割成K个子样本，一个单独的子样本被保留作为验证模型的数据，其他K-1个样本用来训练。交叉验证重复K次，每个子样本验证一次，平均K次的结果或者使用其它结合方式，最终得到一个单一估测。这个方法的优势在于，同时重复运用随机产生的子样本进行训练和验证，每次的结果验证一次，充分利用每一个数据，适用于样本量较小的数据集。
Hosmer-Lemeshow检验(HL检验)：为模型拟合指标，其原理在于判断预测值与真实值之间的gap情况，如果P > 0.05，则说明通过HL检验，即说明预测值与真实值之间并无非常明显的差异；反之如果P < 0.05，则说明没有通过HL检验，预测值与真实值之间有着明显的差异，即说明模型拟合度较差。

背景：列线图预测模型是基于多元回归分析得出的独立影响因素绘制的，由各种影响因素综合得出的总分可直观计算出患者患有骨量减少的概率。
目的：基于回归分析出的中年男性骨量减少的独立影响因素，建立列线图预测模型。
方法：收集2021-06-15/07-15及2021-10-15/11-15锦州医科大学附属第一医院体检中心279名男性体检者的临床资料。使用自制《中年男性骨量调查问卷》对受试者进行调查及使用DiscoveryW型双能X射线骨密度仪测量左侧髋部的骨密度值。按照收集资料的前后时段分为训练集(214名)及验证集(65名)。基于训练集单因素分析和多因素Logistics回归筛选骨量减少的独立影响因素，制作骨量减少列线图预测模型，采用C指数验证及校准曲线初步评价模型区分度及校准度。模型的内部验证采用K折交叉验证，由验证集进行模型外部验证，采用C指数、Hosmer-Lemeshow检验及Calibration校准曲线进一步评价列线图模型表现。
结果与结论：①单因素分析出训练集在吸烟、奶制品、运动、早餐等因素上存在统计学差异，多因素Logistics回归分析筛查出吸烟为骨量减少的独立危险因素，奶制品及运动为保护因素；②根据独立影响因素及现有理论，构建骨量减少列线图预测模型，模型的C指数为0.671(95%CI：0.600-0.750)，校准曲线拟合较好；③K折交叉验证得出C指数均值为0.708；验证集外部验证的C指数为0.689(95%CI：0.600-0.742)，Hosmer-Lemeshow检验中P=0.09 > 0.05，Calibration校准曲线拟合较好；④结果说明，该骨量减少列线图预测模型，具有较好的检验效能，有助于临床上筛查患有骨量减少的中年男性患者。

https://orcid.org/0000-0002-7658-9152 (沈炼伟) 

中国组织工程研究杂志出版内容重点：组织构建；骨细胞；软骨细胞；细胞培养；成纤维细胞；血管内皮细胞；骨质疏松；组织工程

关键词: 骨量减少, 列线图, 预测模型, 中年男性

Abstract: BACKGROUND: A nomogram prediction model can be plotted based on independent influencing factors derived from multiple regression analyses. The probability of bone mass loss in patients can be intuitively calculated based on the total score obtained from various influencing factors.
OBJECTIVE: To establish a nomogram prediction model based on the independent factors of bone mass loss in middle-aged men. 
METHODS: The clinical data of 279 middle-aged men who underwent physical examinations in the Physical Examination Center of the First Affiliated Hospital of Jinzhou Medical University from June 15 to July 15, 2021 and from October 15 to November 15, 2021 were collected and divided into training set (214 cases) and verification set (65 cases) according to different time periods of data collection. The self-made “Middle-aged Male Bone Mass Survey Questionnaire” was used in all subjects and the DiscoveryW dual-energy X-ray bone densitometer was used to measure the bone mineral density of the left hip. Based on the single-factor analysis and multivariate Logistics regression analysis of training set, the independent influencing factors of bone mass loss were screened, and the nomogram model for predicting bone mass loss was made. C-index verification and calibration curve were used to preliminarily evaluate the differentiation and calibration degree of the model. The internal validation of the model was performed by k-fold cross validation, and the external validation of the model was performed by the validation set. The performance of the nomogram model was further evaluated by C index, Hosmer-Lemeshow test, and calibration curve.
RESULTS AND CONCLUSION: Results from the single-factor analysis showed that there were statistical differences in smoking, dairy products, exercise, and breakfast in the training set. Multivariate Logistics regression analysis results showed that smoking was an independent risk factor for bone mass loss, while dairy products and exercise were protective factors. According to the independent factors and existing theories, the nomogram model for predicting bone mass loss was established with a C-index of 0.671 (95% confidence interval: 0.600-0.750), indicating that the calibration curve fitted well. K-fold cross verification showed that the mean value of C-index was 0.708. The C-index of external validation was 0.689 (95% confidence interval: 0.600-0.742), P=0.09 > 0.05 in the Hosmer-Lemeshow test, and the calibration curve was fitted well. To conclude, this nomogram model for predicting bone mass loss has well inspection efficiency and can help screen middle-aged male patients with boss mass loss.

Key words: bone mass loss, nomogram, prediction model, middle-aged man

中图分类号: