| 346 | 0 | 109 |
| 下载次数 | 被引频次 | 阅读次数 |
目的：基于机器学习构建针灸联合常规治疗干预多囊卵巢综合征合并胰岛素抵抗（PCOS-IR）疗效的预测模型。方法：收集两项真实世界研究的数据，分别用于训练集（284例）和验证集（132例），训练集为2023年1月至2024年9月就诊的PCOS-IR患者，验证集为2024年9月至2025年2月就诊的PCOS-IR患者。结合逻辑回归（LR）和随机森林（RF）算法进行预测特征选择，并基于筛选结果建立5种具有不同原理的代表性机器学习模型。通过受试者工作特征曲线（ROC）、校准曲线及决策曲线分析（DCA）评估最佳模型的预测效果，最后采用沙普利加性解释（SHAP）框架对最佳模型的预测结果进行解释。结果：用于模型构建的预测特征包括空腹胰岛素（FINS）、总胆固醇（TC）、月经周期最长时限（UML）、体质量指数（BMI）和丙氨酸氨基转移酶（ALT）。RF模型在准确度、精密度、F1分数和曲线下面积（AUC）方面表现最为均衡，表明该模型在预测PCOS-IR疗效方面具有优异的综合性能。SHAP分析进一步揭示了5个预测特征在疗效预测中的重要性，特别是FINS作为胰岛素水平的指标，在疗效预测中贡献最为突出。结论：构建的针灸联合常规治疗干预PCOS-IR疗效的预测模型，为识别获益人群、优化治疗策略提供了重要的实证依据。
Abstract：Objective To construct an efficacy prediction model for polycystic ovary syndrome with insulin resistance(PCOS-IR) treated with acupuncture and moxibustion combined with conventional treatment based on machine learning. Methods Data from two real-world studies were collected for the training set(284 cases) and validation set(132 cases). The training set included the data of PCOS-IR patients visited from January 2023 to September 2024, and the validation set was composed of the patients visited from September 2024 to February 2025. Logistic regression(LR) and random forest(RF) algorithms were combined for predictive feature selection, and 5 representative machine learning models with different principles were built based on the screening results. The predictive effectiveness of the best model was assessed by receiver operating characteristic(ROC) curve, calibration curve, and decision curve analysis(DCA). Finally, the predictive results of the best model were interpreted using the Shapley additive interpretation(SHAP) framework. Results The predictive features used for model construction included fasting insulin(FINS), total cholesterol(TC), the upper limit of the menstrual cycle length cycle(UML), body mass index(BMI), and alanine aminotransferase(ALT). The RF model showed the most balanced performance in terms of accuracy, precision, F1 score, and area under curve(AUC), suggesting that it achieved the overall favorable performance in predicting the efficacy on PCOS-IR. The SHAP analysis further revealed the importance of these 5 predictive features in efficacy prediction; and in particular, FINS, as an indicator of insulin level, was conductive most significantly to the efficacy prediction. Conclusion The efficacy prediction model constructed for PCOS-IR treated with acupuncture and moxibustion, combined with conventional regimens, provides an important empirical evidence for identifying beneficiary population and optimizing treatment strategy.
[1]Joham AE, Norman RJ, Stener-Victorin E, et al. Polycystic ovary syndrome[J]. Lancet Diabetes Endocrinol, 2022, 10(9):668-680.
[2]Mastrototaro L, Roden M. Insulin resistance and insulin sensitizing agents[J]. Metabolism, 2021, 125:154892.
[3]Zhao H, Zhang JQ, Cheng XY, et al. Insulin resistance in polycystic ovary syndrome across various tissues:an updated review of pathogenesis, evaluation, and treatment[J]. J Ovarian Res, 2023, 16(1):9.
[4]Nabrdalik K, Skonieczna-Żydecka K, Irlik K, et al.Gastrointestinal adverse events of metformin treatment in patients with type 2 diabetes mellitus:a systematic review, meta-analysis and meta-regression of randomized controlled trials[J]. Front Endocrinol, 2022, 13:975912.
[5]Wen JY, Chen X, Yang Y, et al. Acupuncture medical therapy and its underlying mechanisms:a systematic review[J]. Am J Chin Med, 2021, 49(1):1-23.
[6]陈晨,陆丽明.针灸真实世界研究方法与实践[J].中山大学学报(医学科学版), 2023, 44(1):4-9.
[7]王驰,刘成勇,王晓秋,等.针刺通督养心组方治疗失眠症的人工智能辅助决策平台开发研究[J].中国针灸, 2025, 45(7):881-888, 922.
[8]高珍,崔梦洁,王海军,等.针刺治疗颈痛的适宜人群筛选:一项机器学习研究[J].中国针灸, 2025, 45(4):405-412.
[9]Chen WX, Miao JL, Chen JF, et al. Development of machine learning models for diagnostic biomarker identification and immune cell infiltration analysis in PCOS[J]. J Ovarian Res, 2025,18(1):1.
[10]Zhang WJ, Wu YL, Yuan YL, et al. Identification of key biomarkers for predicting atherosclerosis progression in polycystic ovary syndrome via bioinformatics analysis and machine learning[J]. Comput Biol Med, 2024, 183:109239.
[11]Teede HJ, Tay CT, Laven J, et al. Recommendations from the2023 international evidence-based guideline for the assessment and management of polycystic ovary syndrome[J]. Obstet Gynecol Surv, 2023, 78(12):742-743.
[12]Li R, Qiao J, Yang DZ, et al. Epidemiology of hirsutism among women of reproductive age in the community:a simplified scoring system[J]. Eur J Obstet Gynecol Reprod Biol, 2012, 163(2):165-169.
[13]Ascaso JF, Pardo S, Real JT, et al. Diagnosing insulin resistance by simple quantitative methods in subjects with normal glucose metabolism[J]. Diabetes Care, 2003, 26(12):3320-3325.
[14]Jiang XH, Liu RJ, Liao T, et al. A predictive model of live birth based on obesity and metabolic parameters in patients with PCOS undergoing frozen-thawed embryo transfer[J]. Front Endocrinol,2022, 12:799871.
[15]Li X, Liao MY, Shao JQ, et al. Plasma diaphanous related formin1 levels are associated with altered glucose metabolism and insulin resistance in patients with polycystic ovary syndrome:a case control study[J]. Mediators Inflamm, 2022, 2022:9620423.
[16]中国中西医结合学会妇产科专业委员会.多囊卵巢综合征中西医结合诊治指南[J].中国中西医结合杂志, 2024, 44(1):5-18.
[17]金春兰,庞然,徐立鹏,等.针灸治疗多囊卵巢综合征临床选穴组方规律[J].中国针灸, 2015, 35(6):625-630.
[18]Guo BJ, Shen YT, Dai ZY, et al. A nomogram to predict the risk of insulin resistance in Chinese women with polycystic ovary syndrome[J]. Front Endocrinol, 2024, 15:1446827.
[19]O'Connell NS, Jaeger BC, Bullock GS, et al. A comparison of random forest variable selection methods for regression modeling of continuous outcomes[J]. Brief Bioinform, 2025, 26(2):bbaf096.
[20]Baumann D, Baumann K. Reliable estimation of prediction errors for QSAR models under model uncertainty using double cross-validation[J]. J Cheminform, 2014, 6(1):47.
[21]Salditt M, Humberg S, Nestler S. Gradient tree boosting for hierarchical data[J]. Multivariate Behav Res, 2023, 58(5):911-937.
[22]Hu JC, Szymczak S. A review on longitudinal data analysis with random forest[J]. Brief Bioinform, 2023, 24(2):bbad002.
[23]Zhang J, Mucs D, Norinder U, et al. LightGBM:an effective and scalable algorithm for prediction of chemical toxicity-application to the Tox21 and mutagenicity data sets[J]. J Chem Inf Model,2019, 59(10):4150-4158.
[24]Wolak W, Plichta A, Orlicki H. Interpretable ensemble learning for tumor-type prediction with a SHAP-based evaluation of CatBoost and voting classifiers[J]. Sci Rep, 2025, 16(1):1401.
[25]Naimi AI, Balzer LB. Stacked generalization:an introduction to super learning[J]. Eur J Epidemiol, 2018, 33(5):459-464.
[26]Carrington AM, Fieguth PW, Qazi H, et al. A new concordant partial AUC and partial c statistic for imbalanced data in the evaluation of machine learning algorithms[J]. BMC Med Inform Decis Mak, 2020, 20(1):4.
[27]Ponce-Bobadilla AV, Schmitt V, Maier CS, et al. Practical guide to SHAP analysis:explaining supervised machine learning model predictions in drug development[J]. Clin Transl Sci, 2024, 17(11):e70056.
[28]Fife DA, D'Onofrio J. Common, uncommon, and novel applications of random forest in psychological research[J]. Behav Res Methods, 2023, 55(5):2447-2466.
[29]Li MW, Chi XW, Wang Y, et al. Trends in insulin resistance:insights into mechanisms and therapeutic strategy[J]. Signal Transduct Target Ther, 2022, 7(1):216.
[30]Dong HX, Wang Q, Wang Z, et al. Impact of low frequency electro-acupuncture on glucose and lipid metabolism in unmarried PCOS women:a randomized controlled trial[J]. Chin J Integr Med, 2021, 27(10):737-743.
[31]Zhang Q, Kong XJ, Yuan H, et al. Mangiferin improved palmitate-induced-insulin resistance by promoting free fatty acid metabolism in HepG2 and C2C12 cells via PPARα:mangiferin improved insulin resistance[J]. J Diabetes Res, 2019, 2019:2052675.
[32]Zhong YM, Luo XC, Chen Y, et al. Acupuncture versus sham acupuncture for simple obesity:a systematic review and meta-analysis[J]. Postgrad Med J, 2020, 96(1134):221-227.
[33]昌恩惠,李兰兰,刘红霞,等.补肾调经汤联合温针灸在多囊卵巢综合征中的应用[J].辽宁中医杂志, 2025, 52(2):122-126.
[34]Bai JJ, Ma Y, Zhao YD, et al. Mixed exposure to phenol, parabens,pesticides, and phthalates and insulin resistance in NHANES:a mixture approach[J]. Sci Total Environ, 2022, 851(Pt 2):158218.
[35]Wang MJ, Zhao D, Xu LZ, et al. Role of PCSK9 in lipid metabolic disorders and ovarian dysfunction in polycystic ovary syndrome[J]. Metabolism, 2019, 94:47-58.
[36]Liu CB, Shao M, Lu L, et al. Obesity, insulin resistance and their interaction on liver enzymes[J]. PLoS One, 2021, 16(4):e0249299.
[37]李知行,张海华,蓝丹纯,等.电针对高脂诱导胰岛素抵抗大鼠肝脏固醇调节元件结合蛋白-1c、脂肪酸合成酶的影响[J].针刺研究, 2018, 43(1):8-13.
基本信息:
DOI：10.13703/j.0255-2930.20250530-k0003
中图分类号:R711.75
引用信息:
[1]郭本婕,孙家豪,杨妍,等.基于真实世界数据的针灸联合常规治疗对PCOS-IR疗效的预测：机器学习与可解释性分析[J].中国针灸,2026,46(05):669-677.DOI:10.13703/j.0255-2930.20250530-k0003.
基金信息:
国家自然科学基金面上项目：82174489; 第七批全国老中医药专家传承工作项目：2023YL024-43
2026-02-02
2026-02-02
2026-02-02