Abstract：【Objective】 To explore the use of machine learning eXtreme Gradient Boosting(XGBoost) algorithm combined with the SHapley Additive exPlanations(SHAP) interpretability method to analyze the impact of post-general anesthesia(GA) sore throat nursing indicators on postoperative patient comfort.【Methods】 A retrospective analysis was conducted on the clinical data of 191 patients who underwent general anesthesia. The patients were divided into a comfortable group and an uncomfortable group according to their comfort level. An XGBoost classification prediction model was constructed, and model parameters were optimized via the grid search method. SHAP values were used to quantify the contribution of each nursing indicator to patient comfort.【Results】 The accuracy of the model was 49%, and the Area Under the Curve(AUC) of the Receiver Operating Characteristic(ROC) curve was 0.41. SHAP analysis showed that cuff pressure monitoring(mean SHAP value: 0.22), prophylactic medication administration(mean SHAP value: 0.18), and application of laryngeal pain assessment tools(mean SHAP value: 0.15) contributed the most to predicting postoperative discomfort.【Conclusion】 The XGBoost-based algorithm combined with SHAP can identify key nursing factors affecting postoperative patient comfort. In clinical nursing practice, priority should be given to cuff pressure management and prophylactic medication use to improve the postoperative comfort of patients with sore throat after general anesthesia.
[1] 卓眉秀,王宜庭,周美云,等.外科医护人员对全身麻醉术后病人咽喉痛认知及态度的质性研究[J].全科护理,2022,20(5):668-671.
[2] 王宜庭,包磊,周英凤,等.全麻气管插管患者术后咽喉疼痛预防最佳证据总结[J].护理学杂志,2021,36(18):82-86.
[3] 洪博洋,陈红,余遥,等.全身麻醉气管插管病人术后咽喉疼痛危险因素的研究进展[J].护理研究,2024,38(5):832-836.
[4] FRITZ B A, CUI Z C, ZHANG M H, et al. Deep-learning model for predicting 30-day postoperative mortality[J].Br J Anaesth,2019, 123(5):688-695.
[5] LUNDBERG S M, NAIR B, VAVILALA M S, et al. Explainable machine-learning predictions for the prevention of hypoxaemia during surgery[J].Nat Biomed Eng,2018, 2(10):749-760.
[6] MEHTA D, GONZALEZ X T, HUANG G, et al. Machine learning-enhanced perioperative nursing interventions: systematic review and meta-analysis[J].Br J Anaesth,2024, 133(6):1159-1172.
[7] 张亚菲,李墅明,杨建军,等.气管插管全麻术后咽痛的研究进展[J].临床麻醉学杂志,2020,36(5):510-513.
[8] 杨扬,陈德凤,李蓓,董旭辉,等.人工智能用于慢性伤口护理的研究进展[J].护理学杂志,2024,39(5):18-21.
[9] 刘超,吉林,刘存明.腹腔镜结直肠手术中气管导管套囊压力控制对术中血流动力学及术后咽喉痛的影响[J].南京医科大学学报(自然科学版),2024,44(8):1100-1105.
[10] 楼佳烨,王艳梅,潘欣欣, 等.基于机器学习的糖尿病足发病风险预测模型构建[J].护理学杂志,2025,40(9):26-30.
[11] 戚馨如,宋玉磊,吕玉婵, 等.基于XGBoost算法的面部特征对轻度认知障碍风险的预测作用[J].护理学杂志,2025,40(7):95-99.
[12] 黄子菁,郭晓贝,何梅,等.机器学习在跌倒风险评估中的应用进展[J].护理学杂志,2022,37(21):110-113.
[13] 李彩福,赵伟,叶秀春,等.基于机器学习算法的社区老年衰弱前期风险预测模型构建[J].护理学杂志,2022,37(15):84-88.
2026, Vol. 43 
