Abstract: Objective  To explore the important risk factors of medical decision-making delay in patients with acute ischemic stroke (AIS) and construct a nomogram prediction model, aiming to provide an individualized assessment tool for early identification of high-risk patients with medical decision-making delay.
Methods  AIS patients admitted to three hospitals in Shenzhen between September 2024 and April 2025 were prospectively enrolled. Patients were divided into a delay group (>1 h) and a non-delay group (≤1 h) based on the time interval from symptom onset or recognition to the first decision to seek medical help. Multivariate logistic regression analysis was used to identify independent influencing factors of medical decision-making delay, and a nomogram prediction model was constructed. The model performance was evaluated using ROC curves, Hosmer-Lemeshow goodness of fitting test, calibration curves, and decision curve analysis. 
Results  A total of 344 AIS patients were enrolled in this study, among whom 201 had medical decision-making delay. Multivariate logistic regression analysis showed that compared with the non-delay group, patients in the delay group had a younger mean age (OR 2.303, 95%CI 1.201-4.416, P=0.012), lower educational level (OR 3.908, 95%CI 1.460-10.463, P=0.007), higher proportion of diabetes mellitus history (OR 1.923, 95%CI 1.054-3.509, P=0.033), lower NIHSS scores at onset (OR 3.245, 95%CI 1.700-6.191, P<0.001), higher proportion of poorer understanding of stroke (OR 3.262, 95%CI 1.247-8.532, P=0.016), and higher proportion of tendency to adopt negative coping style (OR 11.436, 95%CI 6.069-21.550, P<0.001). The AUC of the nomogram prediction model constructed based on these six factors was 0.861 (95%CI 0.820-0.902). The Hosmer-Lemeshow goodness of fitting test indicated that the model had a good fit (χ²=8.064, P=0.427). The calibration curves showed a high degree of consistency between the predicted values and observed values of the model. Decision curve analysis showed that the model yielded a high clinical net benefit within a wide range of risk thresholds.
Conclusions  The nomogram prediction model constructed with age, educational level, history of diabetes mellitus, NIHSS score at onset, understanding of stroke, and disease coping style as core variables can serve as a valid predictive tool for clinically assessing the risk of medical decision-making delay in patients with AIS.

Key words: Acute ischemic stroke; Medical decision-making delay; Nomogram; Prediction model

摘要： 目的 探讨急性缺血性卒中（acute ischemic stroke，AIS）患者就医决策延迟的重要危险因素，构建列线图预测模型，为早期识别就医决策延迟高风险患者提供个体化评估工具。
方法 前瞻性纳入2024年9月—2025年4月深圳市3家医院的AIS患者，根据其症状出现或被识别至首次决定寻求医疗帮助的时间间隔分为延迟组（＞1 h）与未延迟组（≤1 h）。采用多因素logistic回归分析识别就医决策延迟的独立影响因素，并构建列线图预测模型。通过ROC曲线、Hosmer-Lemeshow拟合优度检验、校准曲线及决策曲线分析评估模型性能。 
结果 本研究共纳入344例AIS患者，其中发生就医决策延迟的患者201例。多因素logistic回归分析结果显示，延迟组患者的平均年龄更小（OR 2.303，95%CI 1.201～4.416，P=0.012）、受教育程度更低（OR 3.908，95%CI 1.460～10.463，P=0.007）、有糖尿病病史的比例更高（OR 1.923，95%CI 1.054～3.509，P=0.033）、发病时NIHSS评分更低（OR 3.245，95%CI 1.700～6.191，P<0.001）、卒中了解程度为不清楚的比例更高（OR 3.262，95%CI 1.247～8.532，P=0.016），以及倾向采取消极应对方式的比例更高（OR 11.436，95%CI 6.069～21.550，P<0.001）。基于上述6项影响因素构建的列线图预测模型的AUC为0.861（95%CI 0.820～0.902）；Hosmer-Lemeshow拟合优度检验显示模型拟合良好（χ2＝8.064，P＝0.427）；校准曲线显示模型预测值与观察值一致性较高；决策曲线分析表明模型在较宽风险阈值范围内具有较高的临床净收益。
结论 以年龄、受教育程度、糖尿病病史、发病时NIHSS评分、卒中了解程度和疾病应对方式为核心变量构建的列线图预测模型，可作为临床评估AIS患者就医决策延迟风险的有效预测工具。

关键词: 急性缺血性卒中; 就医决策延迟; 列线图; 预测模型