Objective: This study aimed to establish and verify a model for predicting death within 2 days after spontaneous cerebral hemorrhage based on the patient's characteristics at the time of admission.Methods: During 2015–2017, the records of a cohort of 397 patients with clinically diagnosed cerebral hemorrhage were collected for model development. Minimum absolute contraction and the selection operator (lasso) regression model were used to determine factors that most consistently and correctly predicted death after cerebral hemorrhage. Discrimination and calibration were used to evaluate the performance of the resulting nomogram. After internal validation, the nomogram was further assessed during 2017–2018 using a different cohort of 200 consecutive subjects. Results: The nomogram included four predictors from the lasso regression analysis: Glasgow Coma Scale, hematoma location, hematoma volume, and primary intraventricular hemorrhage. The nomogram showed good discrimination and good calibration for both training and verification cohorts. Decision curve analysis showed that the prediction nomogram was clinically useful. Conclusion: This prediction model can be used for early, simple, and accurate prediction of early death following cerebral hemorrhage.

目的：本研究旨在根据入院时患者的特征建立并验证自发性脑出血后两天内死亡的预测模型。方法：在2015-2017年期间，收集了397例临床诊断为脑出血的患者队列，以进行模型开发。最小绝对收缩和选择算子（套索）回归模型用于确定最一致，最正确地预测脑出血后死亡的因素。鉴别和校准用于评估所得列线图的性能。经过内部验证后，在2017-2018年期间，使用200个连续受试者的不同队列进一步评估了列线图。结果：列线图包括来自套索回归分析的四个预测因子：格拉斯哥昏迷量表，血肿位置，血肿量和原发性脑室内出血。列线图显示出对训练和验证队列的良好区分和良好校准。决策曲线分析表明，预测列线图在临床上是有用的。结论：该预测模型可用于脑出血后早期死亡的早期，简单，准确的预测。