Objective: To analyze the affecting factors of hemoglobin changes in apheresis red blood cells (RBCs), and to establish a predictive model for the evaluation of apheresis.
Methods: The clinical data of 130 patients undergoing selective surgery for apheresis autologous RBCs from January 2017 to December 2018 were collected. The change of hemoglobin and its affecting factors before and after apheresis were analyzed. The predictive model of the hemoglobin change was established by machine learning algorithm and compared with the theoretical predictive model.
Results: The average Hb level in the 300 ml autologous RBC group decreased by 22.61±8.85 g/L, and the average Hb in 400 ml group decreased by 29.08±7.25 g/L. The change of Hb was mainly affected by Hb level before apheresis and peripheral circulation blood volume (P＜0.05). Sex, age, and the interval time between blood collection and operation not significantly influenced Hb change (P＞0.05). The initially established predictive model by the machine learning (MAE 6.27) is superior to the theoretical predictive model (MAE 8.11).
Conclusion: The predictive model established by the machine learning can provide a reference for more accurate evaluation of apheresis autologous red blood cells.
题目: 单采自体红细胞患者采集前后血红蛋白变化预测模型的建立.
目的: 分析单采自体红细胞后患者血红蛋白变化的影响因素, 初步建立单采自体红细胞采集后血红蛋白变化预测模型.
方法: 收集本院2017年1月到2018年12月期间130例行术前单采自体红细胞择期手术患者的临床资料, 分析采集前后血红蛋白变化规律及影响因素, 通过机器学习算法初步建立单采自体红细胞血红蛋白变化预测模型并与理论预测模型进行比较.
结果: 采集300、400 ml 自体红细胞组患者血红蛋白水平分别平均下降22.61±8.85和29.08±7.25 g/L。采集后血红蛋白变化值主要受采血前患者血红蛋白水平和外周血容量影响（P＜0.05）。患者的性别、年龄及采血-手术间隔时间对采集后血红蛋白变化值无明显影响（P＞0.05）。机器学习算法初步建立的预测模型（平均绝对误差6.27）优于理论预测模型（平均绝对误差8.11）.
结论: 利用机器学习算法初步建立的单采自体红细胞预测模型可为更准确地评估自体红细胞的采集提供参考.