| 摘要: |
| ［摘要］ 目的 通过对老年易跌倒患者的肌肉功能指标进行主成分分析与聚类分析，提取关键肌群功能区域，对患者进行分类，为临床个性化康复干预提供依据。方法 收集2018年1月至2023年12月于南京医科大学第一附属医院住院的老年患者资料，共纳入265例，其中跌倒组（有跌倒史）36例，非跌倒组（无跌倒史）229例。采用超声诊断仪测量桡侧腕屈肌、尺侧腕屈肌、腹肌、腰部肌群、胫骨后肌、腓骨长肌、腓肠肌外侧头、腓肠肌内侧头的肌肉厚度，并收集患者的一般资料。应用SPSS 22.0统计软件进行主成分分析提取关键成分，并采用聚类分析对患者进行分类。通过受试者工作特征（ROC）曲线分析各主成分及其综合指标对跌倒的预测效能。采用多因素logistic回归分析跌倒的影响因素。结果 主成分分析提取出4个特征值>1的主成分，分别是主成分1（桡侧腕屈肌、尺侧腕屈肌、腹肌）、主成分2（胫骨后肌、腓骨长肌、BMI）、主成分3（腓肠肌内侧头、腓肠肌外侧头）、主成分4（腰部肌群、年龄），累积贡献率为69.181%。聚类分析将患者分为2个大类、4个亚类，各类别在肌群功能分布上具有明显差异。ROC曲线分析显示，单一主成分预测跌倒的曲线下面积（AUC）较低，而包括4个主成分的综合指标预测效能更佳。多因素logistic回归分析显示，年龄较大是促进跌倒发生的独立危险因素（P<0.05），腹肌增厚是抑制跌倒发生的独立保护因素（P<0.05）。结论 通过主成分分析和聚类分析可以有效识别老年易跌倒患者的肌群功能特征与个体差异，对各部位肌群功能状态作出客观评价，预测跌倒的风险，为制订个体化肌力训练和跌倒预防方案提供数据支持。 |
| 关键词: 老年人 跌倒 肌群功能 主成分分析 聚类分析 |
| DOI：10.3969/j.issn.1674-3806.2026.04.04 |
| 分类号:R 337 |
| 基金项目:国家自然科学基金项目（编号：82570129，82070093） |
|
| Exploration on functional characteristics of diverse muscle groups in the fall-prone elderly population using principal component analysis and cluster analysis |
|
CHEN Haili1, SHENG Yunlu2, DUAN Yu2, WANG Yichun1, WANG Kaiming1, LU Junhao3, DING Ning1, LIN Wei2*
|
|
1.Department of Pulmonary and Critical Care Medicine, the First Affiliated Hospital of Nanjing Medical University, Nanjing 210029, China; 2.Department of Geriatric Endocrinology, the First Affiliated Hospital of Nanjing Medical University, Nanjing 210029, China; 3.School of Food and Pharmaceutical Engineering, Nanjing Normal University, Nanjing 210029, China
|
| Abstract: |
| ［Abstract］ Objective To extract key muscle group function regions and classify the fall-prone elderly patients using principal component analysis(PCA) and cluster analysis on muscle function indicators, and to provide a basis for individualized rehabilitation interventions in clinical practice. Methods The data of elderly patients who were hospitalized in the First Affiliated Hospital of Nanjing Medical University from January 2018 to December 2023 were collected, and a total of 265 cases were included, with 36 cases in the fall group(with a history of falls) and 229 cases in the non-fall group(without a history of falls). The thicknesses of the radial wrist flexor, ulnar wrist flexor, abdominal muscles, lumbar muscle groups, posterior tibial muscle, long peroneal muscle, lateral head of gastrocnemius and medial head of gastrocnemius were measured by ultrasonography, and the patients′ general data were collected. Principal component analysis was performed using SPSS 22.0 statistical software to extract the key components, and cluster analysis was adopted to classify the patients. The efficacy of each principal component and their comprehensive indicators for predicting falls was analyed using receiver operating characteristic(ROC) curve. The influencing factors of falls were analyzed using multivariate logistic regression. Results A total of four principal components with eigenvalues greater than 1 were extracted by using principal component analysis, which were the principal component 1(radial wrist flexor, ulnar wrist flexor, and abdominal muscles), principal component 2(posterior tibial muscle, long peroneal muscle, and BMI), principal component 3(medial head of gastrocnemius and lateral head of gastrocnemius), principal component 4(lumbar muscle groups and age), with a cumulative contribution rate of 69.181%. The patients were divided into 2 major categories and 4 subcategories through cluster analysis, and there were significant differences in the functional distributions of muscle groups between the categories and among the subcategories. ROC curve analysis revealed that a single principal component had lower area under the curve(AUC) values for predicting falls, and the efficacy of combining the 4 principal components for predicting falls was better than a single principal component. Multivariate logistic regression analysis indicated that older age was an independent risk factor for promoting the occurrence of falls(P<0.05), and thickening of abdominal muscles was an independent protective factor for inhibiting the occurrence of falls in the patients(P<0.05). Conclusion Principal component analysis and cluster analysis can effectively identify the functional characteristics and individual differences of muscle groups in the fall-prone elderly patients, provide an objective evaluation on the functional status of muscle groups in various parts of the human body, predict the risk of falls, and offer data support for developing individualized muscle strength training and plans for preventing falls. |
Key words: The elderly Falls Muscle group function Principal component analysis Cluster analysis
*共同通信作者 |
