Zhou Yu-Hong, Tu Guang, Wu Yan, Wu Juan, Shen Lei, Lei Yu-Ling
Nursing Department, Shanghai Public Health Clinical Center, Fudan University, Shanghai, China.
The Second Department of Interior, Lichuan County People's Hospital of Jiangxi Province, Fuzhou, China.
Medicine (Baltimore). 2025 Sep 5;104(36):e44353. doi: 10.1097/MD.0000000000044353.
Stroke is a severe neurological disorder that significantly impacts patients' recovery and quality of life. Stroke patients frequently experience sleep disorders, including difficulty initiating sleep, insomnia, vivid dreams, and sleep apnea. These disorders not only disrupt nighttime rest but also significantly affect stroke recovery and prognosis, increasing the risks of recurrence and mortality. Currently, there are few studies on this topic, and most rely on Logistic regression models, which can identify risk factors but cannot quantify risks. Therefore, it is essential to develop a tool that can comprehensively assess multiple risk factors and provide individualized predictions. Nomogram models can quantify risk factors and intuitively present them, thereby providing clinicians with comprehensive assessments. This study aims to develop and validate a new nomogram model to predict the risk of sleep disorders in stroke patients, enabling early identification and personalized interventions to support patient recovery and improve quality of life. A cohort of 156 stroke patients (January-August 2023) was utilized for model development, comprising 70 with sleep disorders and 86 without. An external validation set included 72 patients (September-December), with 34 experiencing sleep disorders. Patient data was analyzed using Lasso regression; the "rms" package in R facilitated model construction. Model performance was assessed through Hosmer-Lemeshow goodness-of-fit tests, calibration curves, and receiver operating characteristic analyses. Gender bias, co-morbidities (hypertension and coronary heart disease), depression, and anxiety scales differentiated the groups significantly. Key predictors included female gender, hypertension, coronary heart disease, and psychological distress. The model yielded impressive predictive capabilities, with area under the curves of 0.950 (modeling group) and 0.966 (validation group). Calibration curves matched closely with ideals, confirming robustness across both sets. Net benefit rates indicated strong utility over a wide probability spectrum. Female gender, specific co-morbidities, heightened depressive and anxiety states signify elevated sleep disorder risks poststroke. Our nomogram effectively predicts these conditions, offering valuable insights for timely detection and intervention in susceptible stroke survivors.
中风是一种严重的神经疾病，对患者的康复和生活质量有重大影响。中风患者经常出现睡眠障碍，包括入睡困难、失眠、多梦和睡眠呼吸暂停。这些障碍不仅会干扰夜间休息，还会显著影响中风的恢复和预后，增加复发和死亡风险。目前，关于这个主题的研究很少，大多数依赖逻辑回归模型，该模型可以识别风险因素，但无法量化风险。因此，开发一种能够全面评估多种风险因素并提供个性化预测的工具至关重要。列线图模型可以量化风险因素并直观地呈现它们，从而为临床医生提供全面评估。本研究旨在开发并验证一种新的列线图模型，以预测中风患者睡眠障碍的风险，实现早期识别和个性化干预，以支持患者康复并提高生活质量。一组156名中风患者（2023年1月至8月）用于模型开发，其中70名有睡眠障碍，86名没有。一个外部验证集包括72名患者（9月至12月），其中34名有睡眠障碍。使用套索回归分析患者数据；R中的“rms”包有助于模型构建。通过Hosmer-Lemeshow拟合优度检验、校准曲线和受试者工作特征分析评估模型性能。性别偏见、合并症（高血压和冠心病）、抑郁和焦虑量表在两组之间有显著差异。关键预测因素包括女性、高血压、冠心病和心理困扰。该模型具有令人印象深刻的预测能力，建模组曲线下面积为0.950，验证组为0.966。校准曲线与理想情况紧密匹配，证实了两组的稳健性。净受益率表明在广泛的概率范围内具有强大的效用。女性、特定合并症、抑郁和焦虑状态加剧表明中风后睡眠障碍风险升高。我们的列线图有效地预测了这些情况，为及时检测和干预易患中风的幸存者提供了有价值的见解。
