Fassihi Safa C, Mathur Abhay, Best Matthew J, Chen Aaron Z, Gu Alex, Quan Theodore, Wang Kevin Y, Wei Chapman, Campbell Joshua C, Thakkar Savyasachi C
Department of Orthopedic Surgery, George Washington Hospital, 2300 M St NW, Washington, DC, 20037, USA.
Johns Hopkins Department of Orthopaedic Surgery, Adult Reconstruction Division, 10700 Charter Drive, Suite 205, Columbia, MD, 21044, USA.
J Orthop. 2021 Nov 25;28:91-95. doi: 10.1016/j.jor.2021.11.013. eCollection 2021 Nov-Dec.
The purpose is to utilize an artificial neural network (ANN) model to determine the most important variables in predicting mortality following total hip arthroplasty (THA).
Patients that underwent primary THA were included from a national database. Demographic, preoperative, and intraoperative variables were analyzed based on their contribution to 30-day mortality with the use of an ANN model.
The five most important factors in predicting mortality following THA were preoperative international normalized ratio, age, body mass index, operative time, and preoperative hematocrit.
ANN modeling represents a novel approach to determining perioperative factors that predict mortality following THA.
利用人工神经网络（ANN）模型确定全髋关节置换术（THA）后预测死亡率的最重要变量。
从一个国家数据库中纳入接受初次THA的患者。使用ANN模型，根据人口统计学、术前和术中变量对30天死亡率的贡献进行分析。
THA后预测死亡率的五个最重要因素是术前国际标准化比值、年龄、体重指数、手术时间和术前血细胞比容。
ANN建模是一种确定预测THA后死亡率的围手术期因素的新方法。
