Abstract: Background　Falls are common in patients with Alzheimer's disease (AD) and are associated with increased risks of severe trauma, infection, disability, and death, often leading to unplanned hospitalization and placing a heavy burden on families and society. Objective　To develop and validate a risk prediction model for fall-related hospitalization in patients with AD, and to identify the risk factors for fall-related hospitalization, thereby assisting physicians in accurately identifying high-risk individuals and implementing early interventions. Methods　A total of 987 patients with AD from the dementia database of Hubei Provincial Hospital of Traditional Chinese Medicine between January 2020 and October 2025 were included. They were randomly divided into a training set (n=690) and a validation set (n=297) at a ratio of 7:3. Data on general characteristics, assessment scales, laboratory indicators, and medication use were extracted. Fall-related hospitalization was the outcome variable. Potential predictors were selected using LASSO regression. A nomogram prediction model was established using multivariable Logistic regression. Model discrimination, calibration, and clinical utility were evaluated using the area under the receiver operating characteristic curve (AUC), calibration curves, and decision curve analysis (DCA), respectively. Results　Among the 987 patients with AD, 276 (27.96%) experienced fall-related hospitalization. Multivariable Logistic regression based on variables selected by LASSO regression showed that disease duration (OR=2.843, 95%CI=1.862-4.342), Clinical Dementia Rating (CDR) score (OR=1.275, 95%CI=1.010-1.610), history of falls (OR=7.779, 95%CI=3.515-17.213), osteoarthritis (OR=1.757,95%CI=1.068-2.892), osteoporosis (OR=2.481, 95%CI=1.692-3.638), high-risk behavioral and psychological symptoms of dementia (BPSD) (OR=2.193, 95%CI=1.229-3.914), albumin (ALB) level (OR=1.781, 95%CI=1.128-2.814), and high-risk medication use (OR=1.466, 95%CI=1.191-1.805) were independent risk factors for fall-related hospitalization in patients with AD (P<0.05). Receiver operating characteristic analysis showed that the AUCs of the training set and validation set were 0.753 (95%CI=0.711-0.795) and 0.794 (95%CI=0.734-0.853), respectively. Calibration curves showed good agreement between the predicted and ideal curves in both sets. Decision curve analysis showed that the nomogram provided a net benefit greater than 0 when the predicted probability of fall-related hospitalization ranged from 0.1 to 0.8. Conclusion　Disease duration, CDR score, history of falls, osteoarthritis, osteoporosis, high-risk BPSD, ALB level, and high-risk medication use were identified as risk factors for fall-related hospitalization in patients with AD. The nomogram model constructed in this study can be used to predict the risk of fall-related hospitalization in this population.

Key words: Alzheimer's disease, Falls, Hospitalization, Prediction model, Nomogram, Logistic regression

摘要： 背景　跌倒是阿尔茨海默病（AD）患者的常见临床事件，不仅增加严重创伤、感染、失能与死亡风险，也常导致非计划性住院，给家庭与社会带来沉重负担。目的　构建并验证AD患者跌倒相关住院的风险预测模型，并分析AD患者跌倒相关住院的危险因素，以辅助医师精准识别高危个体并进行干预。方法　共纳入987名2020年1月—2025年10月收录于湖北省中医院痴呆数据库的AD患者，将其按照7：3随机拆分为训练集（n=690）和验证集（n=297）。提取研究对象的一般资料、评估量表、实验室检查指标及用药相关数据。以AD患者是否发生跌倒相关住院事件作为结局变量。以LASSO回归筛选潜在预测变量，并采用多因素Logistic回归建立列线图预测模型。采用受试者工作特征曲线（ROC）下面积（AUC）、校准曲线和决策曲线（DCA）检验模型的区分度、校准度和临床实用性。结果　987例AD患者中，跌倒相关住院者276例（27.96%）。在LASSO回归基础上行多因素Logistic回归分析结果显示，病程（OR=2.843，95%CI=1.862~4.342）、临床痴呆评定量表（CDR）评分（OR=1.275，95%CI=1.010~1.610）、跌倒史（OR=7.779，95%CI=3.515~17.213）、骨关节炎（OR=1.757，95%CI=1.068~2.892）、骨质疏松（OR=2.481，95%CI=1.692~3.638）、高危痴呆的精神行为症状（BPSD）（OR=2.193，95%CI=1.229~3.914）、白蛋白（ALB）水平（OR=1.781，95%CI=1.128~2.814）及高风险用药（OR=1.466，95%CI=1.191~1.805）是AD患者跌倒相关住院的独立影响因素（P<0.05）。ROC结果显示，训练集与验证集AUC分别为0.753（95%CI=0.711~0.795）和0.794（95%CI=0.734~0.853）；校准曲线结果显示，训练集与验证集的预测曲线及理想曲线拟合度较好；DCA分析结果显示：当列线图预测AD患者跌倒相关住院风险概率在0.1~0.8阈值范围内，患者的净获益率>0。结论　病程、CDR评分、跌倒史、骨关节炎、骨质疏松、高危BPSD、ALB、高风险用药是AD患者跌倒相关住院的影响因素，本研究构建的列线图模型可用于预测AD患者跌倒相关住院风险。

关键词: 阿尔茨海默病, 跌倒, 住院, 预测模型, 列线图, Logistic 回归