目的 基于LASSO回归模型分析影响维持性血液透析(maintenance hemodialysis，MHD)患者肌肉衰减综合征(muscle wasting syndrome，MWS)发生的高危因素。 方法 选取江苏省苏北人民医院收治的MHD患者，参照亚洲肌肉衰减综合征诊断标准分为肌肉正常组及肌肉衰减组，比较2组患者基线资料。通过LASSO回归分析影响MHD患者MWS发生的危险因素，ROC曲线验证预测模型的准确度。 结果 共纳入280例患者，其中肌肉正常组152例，肌肉衰减组128例。LASSO回归分析显示年龄(OR=1.164，95% CI：1.043～1.299，P=0.007)、透析后体质量(OR=0.749，95% CI：0.610～0.920，P=0.006)、上臂围(OR=1.802，95% CI：1.299～2.500，P＜0.001)、腿围(OR=2.787，95% CI：1.463～5.311，P=0.002)、肌肉量(OR=0.680，95% CI：0.532～0.868，P =0.002)、肌酐(OR=0.456，95% CI：0.232～0.897，P=0.023)、尿素氮(OR=0.162，95% CI：0.070～0.373，P＜0.001)是MHD患者MWS的危险因素。ROC曲线验证LASSO回归结果的AUC为0.713(95% CI：0.654～0.773)，敏感度为76.60%，特异度为64.60%，Youden指数J=0.412。 结论 LASSO-Logistic回归模型可预测MHD患者MWS发生的高危因素且具有良好的预测效能，为临床干预和治疗提供了重要参考。

Objective  To analyze the risk factors of Muscle Wasting Syndrome (MWS) in Maintenance Hemodialysis (MHD) patients based on LASSO regression model.  Methods   Subei People's Hospital, Jiangsu Province of 280 MHD patients were selected and divided into normal muscle group and sarcopenia group according to the Asian sarcopenia comprehensive diagnostic criteria. The baseline data of the two groups were compared, and the risk factors of MWS in MHD patients were analyzed by LASSO regression. The Receiver Operating Characteristic (ROC) curve was used to verify the accuracy of the prediction model.  Results  There were 152 patients in the normal muscle group and 128 patients in the sarcopenia group. LASSO regression analysis showed that age (OR 1.164, 95% CI:1.043～1.299, P=0.007), weight after dialysis (OR 0.749, 95% CI:0.610～0.920, P=0.006), upper arm circumference (OR 1.802, 95% CI:1.299～2.500, P＜0.001), leg circumference (OR 2.787, 95% CI:1.463～5.311, P=0.002), muscle mass (OR 0.680, 95% CI:0.532～0.868, P＜0.001), and muscle mass (OR 0.680, 95% CI:0.532～0.868, P=0.002), creatinine (OR 0.456, 95% CI:0.232～0.897, P=0.023) and urea nitrogen (OR 0.162, 95% CI:0.070～0.373, P＜0.001) were risk factors for MWS in MHD patients. ROC Curve showed that the Area Under Curve (AUC) of LASSO regression was 0.713 (95% CI: 0.654～0.773), the sensitivity was 76.60%, the specificity was 64.60%, and the Youden index J=0.412. Conclusions  The LASAS-Logistic regression model can predict the high-risk factors of MWS in MHD