Analysis of factors associated with diabetes mellitus in Shanxi Province based on Bayesian network model
-
摘要:
目的 针对2015年山西省糖尿病调查数据,利用最大最小爬山(max-min hill-climbing, MMHC)算法构建糖尿病相关因素的贝叶斯网络模型,探索糖尿病及其相关因素间的网络关系,通过网络模型推理反映各影响因素对糖尿病的影响程度。 方法 采用单因素及多因素Logistic回归分析模型对2015年山西省≥18岁居民的糖尿病调查数据进行变量初筛,再以MMHC算法构建贝叶斯网络模型,参数估计采用极大似然估计法。 结果 2015年山西省糖尿病的检出率是9.5%。经Logistic回归分析模型对变量进行筛选后,年龄、职业、日均摄油量、高血压、高脂血症、BMI和心率被纳入贝叶斯网络模型;贝叶斯网络模型结果显示:年龄、高脂血症、高血压与糖尿病直接相关,BMI通过影响高脂血症与糖尿病间接相关,日均摄油量通过影响BMI和高脂血症与糖尿病间接相关。 结论 贝叶斯网络模型能很好地揭示糖尿病及其相关因素间复杂的网络关系,在分析疾病相关因素上具有较好的适用性和应用前景。 Abstract:Objective For the survey data on diabetes in Shanxi Province in 2015, a Bayesian network model of diabetes-related factors was constructed using the max-min hill-climbing (MMHC) algorithm to explore the network relationships between diabetes and its related factors, and the strength of each influencing factor on diabetes was reflected through network model inference. Methods Single-factor analysis and multi-factor logistic regressions were used to initially screen the variables for survey data on diabetes mellitus among residents aged 18 years and above in Shanxi Province. Afterwards, a Bayesian network was constructed with the MMHC algorithm, and the parameters were estimated by great likelihood estimation. Results The detection rate of diabetes mellitus in Shanxi Province in 2015 stood at 9.5%. After logistic regression feature screening, eight variables, namely age, occupation, average daily oil intake, hypertension, hyperlipidaemia, BMI and heart rate, were finally entered into the model. The Bayesian network model demonstrated that age, hyperlipidaemia and hypertension were directly related to diabetes; BMI was indirectly related to diabetes by hyperlipidaemia, and the average daily oil intake indirectly affected diabetes by BMI and hyperlipidaemia. Conclusion Bayesian network models can well reveal the complex network relationships between diabetes and its associated factors and have a good applicability and prospects in the analysis of disease-related factors. -
表 1 2015年山西省DM患病率的单因素分析[n (%)]
Table 1. Univariate analysis of the prevalence of diabetes mellitns in Shanxi Province in 2015 [n (%)]
变量 N(%) 糖尿病 χ2值 P值 变量 N(%) 糖尿病 χ2值 P值 是 否 是 否 民族 1.998 0.157 日均摄油量(g) 5.941 0.015 汉族 4 548(99.6) 433(9.5) 4 115(90.5) ≤25 238(5.2) 50(7.0) 662(93.0) 少数民族 19(0.4) 0(0.0) 19(100.0) >25 4 329(94.8) 383(9.9) 3 472(90.1) 性别 2.296 0.130 日均摄盐量(g) 0.098 0.754 男 2 236(49.0) 197(8.8) 2 039(91.2) ≤6 236(5.2) 21(8.9) 215(91.1) 女 2 331(51.0) 236(10.1) 2 095(89.9) >6 4 331(94.8) 412(9.5) 3 919(90.5) 年龄(岁) 86.867 < 0.001 高血压 40.103 < 0.001 18~ < 40 549(12.0) 24(4.4) 525(95.6) 否 2 500(54.7) 158(6.3) 2 342(93.7) 40~ < 60 2 202(48.2) 148(6.7) 2 054(93.3) 是 2 067(45.3) 275(13.3) 1 792(86.7) ≥60 1 816(39.8) 261(14.4) 1 555(85.6) 高脂血症 34.012 < 0.001 婚姻状况 9.329 0.009 否 4 134(90.5) 175(7.1) 2 278(92.9) 未婚 160(3.5) 9(5.6) 151(94.4) 是 433(9.5) 258(12.2) 1 856(87.8) 已婚 4 178(91.5) 391(9.4) 3 787(90.6) 心率 12.872 0.002 离异/丧偶 229(5.0) 33(14.4) 196(85.6) 过缓 156(3.5) 14(9.0) 142(91.0) 职业 64.142 < 0.001 正常 4 156(92.2) 381(9.2) 3 775(90.8) 农牧渔业 2 324(50.9) 175(7.5) 2 149(92.5) 过速 196(4.3) 33(16.8) 163(83.2) 其他劳动 1 025(22.4) 89(8.7) 936(91.3) 慢性消化系统疾病 0.135 0.713 未就业 127(2.8) 14(11.0) 113(89.0) 无 752(16.5) 74(9.8) 678(90.2) 自由职业 766(16.8) 87(11.4) 679(88.6) 有 3 815(83.5) 359(9.4) 3 456(90.6) 离退休 325(7.1) 68(20.9) 257(79.1) 慢性泌尿系统疾病 3.258 0.071 文化水平 0.419 0.518 无 325(7.1) 40(12.3) 285(87.7) 高中以下 3 692(80.8) 345(9.3) 3 347(90.7) 有 4 242(92.9) 393(9.3) 3 849(90.7) 高中及以上 875(19.2) 88(10.1) 787(89.9) 中心性肥胖 35.490 < 0.001 吸烟 8.890 0.003 否 1 966(43.0) 128(6.5) 1 838(93.5) 否 3 184(69.7) 329(10.3) 2 855(89.7) 是 2 601(57.0) 305(11.7) 2 296(88.3) 是 1 383(30.3) 104(7.5) 1 279(92.5) BMI (kg/m2) 46.716 < 0.001 饮酒 12.428 0.006 < 18.5 127(2.8) 6(4.7) 121(95.3) 从不 4 247(93.0) 420(9.9) 3 827(90.1) 18.5~ < 24.0 1 926(42.2) 136(7.1) 1 790(92.9) 偶尔 40(0.9) 3(7.5) 37(92.5) 24.0~ < 28.0 1 715(37.6) 170(9.9) 1 545(90.1) 经常 151(3.3) 5(3.3) 146(96.7) ≥28.0 799(17.5) 121(15.1) 678(84.9) 每天 129(2.8) 5(3.9) 124(96.1) 身体活动度 10.394 0.004 不足 1 032(22.6) 110(10.7) 922(89.3) 达标 2 381(52.1) 241(10.1) 2 140(89.9) 充足 1 154(25.3) 82(7.1) 1 072(92.9) 表 2 2015年山西省DM患病的多因素Logistic回归分析
Table 2. Multi-factor Logistic regression analysis of diabetes mellitns prevalence in Shanxi Province in 2015
变量 β值 sx值 Wald χ2值 OR(95% CI)值 P值 变量 β值 sx值 Wald χ2值 OR(95% CI)值 P值 年龄(岁) 高脂血症 18~ < 40 1.000 否 1.000 40~ < 60 0.426 0.237 3.232 1.532(0.962~2.438) 0.072 是 0.487 0.109 19.949 1.627(1.314~2.015) < 0.001 ≥60 1.200 0.241 24.866 3.321(2.072~5.324) < 0.001 心率 职业 过缓 1.000 农牧渔业 1.000 正常 0.109 0.294 0.138 1.115(0.627~1.983) 0.710 其他劳动 0.361 0.143 6.417 1.435(1.085~1.898) 0.011 过速 0.790 0.353 5.013 2.204(1.104~4.403) 0.025 未就业 0.645 0.315 4.193 1.907(1.028~3.537) 0.041 BMI (kg/m2) 自由职业 0.405 0.144 7.882 1.499(1.130~1.988) 0.005 < 18.5 1.000 离退休 0.934 0.166 31.680 2.545(1.838~3.523) < 0.001 18.5~ < 24.0 0.464 0.434 1.143 1.590(0.679~3.722) 0.285 日均摄油量(g) 24.0~ < 28.0 0.710 0.435 2.669 2.035(0.868~4.772) 0.102 ≤25 1.000 ≥28.0 1.085 0.442 6.043 2.961(1.246~7.035) 0.014 >25 0.370 0.164 5.116 1.448(1.051~1.996) 0.023 常数 -5.279 0.633 69.566 0.005 < 0.001 高血压 否 1.000 是 0.402 0.115 12.207 1.494(1.193~1.872) < 0.001 表 3 2015年山西省DM节点的条件概率表
Table 3. Conditional probability table of diabetes mellitus nodes in Shanxi Province in 2015
高脂血症 年龄(岁) DM患病风险 否 18~ < 40 0.018 否 40~ < 60 0.051 否 ≥60 0.112 是 18~ < 40 0.072 是 40~ < 60 0.086 是 ≥60 0.182 -
[1] 赵振华, 童亚慧, 杨青敏. 国内外糖尿病自我管理的研究与进展[J]. 上海护理, 2015, 15(4): 68-71. https://www.cnki.com.cn/Article/CJFDTOTAL-SHHL201504028.htmZhao ZH, Tong YH, Yang QM. Research and progress in diabetes self-management at home and abroad [J]. Shanghai Nursing, 2015, 15(4): 68-71. https://www.cnki.com.cn/Article/CJFDTOTAL-SHHL201504028.htm [2] Cho NH, Shaw JE, Karuranga S, et al. IDF diabetes atlas: global estimates of diabetes prevalence for 2017 and projections for 2045 [J]. Diabetes Res Clin Pract, 2018, 138: 271-281. DOI: 10.1016/j.diabres.2018.02.023. [3] 宁光. 中国糖尿病防治的现状及展望[J]. 中国科学: 生命科学, 2018, 48(8): 810-811. https://www.cnki.com.cn/Article/CJFDTOTAL-JCXK201808002.htmNing G. Status quo and prospect of prevention and control of diabetes in China [J]. Scientia Sinica Vitae, 2018, 48(8): 810-811. https://www.cnki.com.cn/Article/CJFDTOTAL-JCXK201808002.htm [4] Aurich H, H Häberle, Wirbelauer C, et al. Classification of postural profiles among mouth-breathing children by learning vector quantization [J]. Methods Inf Med, 2011, 50(4): 349-357. DOI: 10.3414/ME09-01-0039. [5] 魏珍, 张雪雷, 饶华祥, 等. 禁忌搜索算法的贝叶斯网络模型在冠心病影响因素分析中的应用[J]. 中华流行病学杂志, 2016, 37(6): 895-899. DOI: 10.3760/cma.j.issn.0254-6450.2016.06.031.Wei Z, Zhang XL, Rao HX, et al. Using the Tabu-search-algorithm-based Bayesian network to analyze the risk factors of coronary heart disease [J]. Chin J Epidemiol, 2016, 37(6): 895-899. DOI: 10.3760/cma.j.issn.0254-6450.2016.06.031. [6] LMD Campos. Independency relationships and learning algorithms for singly connected networks [J]. J Exp Theor Artif Intell, 1998, 10(4): 511-549. DOI: 10.1080/095281398146743. [7] Spirtes P, Glymour C, Scheines R. Causation, prediction and search [M]. 2th ed. Cambridge: MIT Press, 2000. [8] Heckerman D, Geiger D, Chickering DM. Learning Bayesian networks: the combination of knowledge and statistical data [J]. Mach Learn, 1995, 20(3): 197-243. DOI: 10.1007/BF00994016. [9] Tsamardinos I, Brown LE, Aliferis CF. The max-min hill-climbing Bayesian network structure learning algorithm [J]. Mach Learn, 2006, 65(1): 31-78. DOI: 10.1007/s10994-006-6889-7. [10] 庞邵杰. 成年人血脂及磷脂谱与胰岛素抵抗的关系研究[D]. 北京: 中国疾病预防控制中心, 2018.Pang SJ. Associations of lipid parameters and phospholipid profiles with insulin resistance among adults [D]. Beijing: Chinese Center for Disease Control and Prevention, 2018. [11] Rao H, Wu E, Fu S, et al. The higher prevalence of truncal obesity and diabetes in American than Chinese patients with chronic hepatitis C might contribute to more rapid progression to advanced liver disease [J]. Aliment Pharmacol Ther, 2017, 46(8): 731-740. DOI: 10.1111/apt.14273 [12] Hu M, Yi W, Yu L, et al. Prevalence, awareness, treatment, and control of hypertension and associated risk factors among adults in Xi'an, China: a cross-sectional study[J]. Medicine, 2016, 95(34): e4709. DOI: 10.1097/MD.0000000000004709. [13] Huang Y, Gao L, Xie X, et al. Epidemiology of dyslipidemia in Chinese adults: Meta-analysis of prevalence, awareness, treatment, and control[J]. Popul Health Metr, 2014, 12(1): 28. DOI: 10.1186/s12963-014-0028-7. [14] Liu X, Li Y, Li L, et al. Prevalence, awareness, treatment, control of type 2 diabetes mellitus and risk factors in Chinese rural population: the RuralDiab study[J]. Sci Rep, 2016, 6(1): 31426. DOI: 10.1038/srep31426. [15] Huang X, Zhou Z, Liu J, et al. Prevalence, awareness, treatment, and control of hypertension among China's Sichuan Tibetan population: a cross-sectional study[J]. Clin Exp Hypertens, 2016, 38(5): 457. DOI: 10.3109/10641963.2016.1163369. [16] Koch D, Eisinger RS, Gebharter A. A causal Bayesian network model of disease progression mechanisms in chronic myeloid leukemia [J]. J Theor Biol, 2017, 433: 94-105. DOI: 10.1016/j.jtbi.2017.08.023. [17] 姚洁. 基于启发式搜索的贝叶斯网络结构学习研究[D]. 浙江: 浙江师范大学, 2016.Yao J. Research on structural learning based on heuristic search in Bayesian networks [D]. Zhejiang: Zhejiang Normal University, 2016. [18] Stajduhar I, Dalbelo-Basic B, Bogunovic N. Impact of censoring on learning Bayesian networks in survival modelling [J]. Artif Intell Med, 2009, 47(3): 199-217. DOI: 10.1016/j.artmed.2009.08.001. [19] 陈璐, 陈适, 许可, 等. 糖尿病患者行为方式与血糖控制关系[J]. 中国公共卫生, 2017, 33(10): 1501-1503. DOI: 10.11847/zgggws2017-33-10-21.Chen L, Chen S, Xu K, et al. Relationship between behavior style and glycemic control in diabetic patients [J]. Chin J Publ Heal, 2017, 33(10): 1501-1503. DOI: 10.11847/zgggws2017-33-10-21. [20] 陈名道. 胰岛β细胞的"糖毒性"、"脂毒性"与"糖脂毒性" [J]. 中华内分泌代谢杂志, 2009, 25(1): 5-8. DOI: 10.3760/cma.j.issn.1000-6699.2009.01.002.Chen DM. Effects of "glucotoxicity", "lipotoxicity" and "glucolipotoxicity" on islet β-cells [J]. Chin J Endocrinol Metab, 2009, 25(1): 5-8. DOI: 10.3760/cma.j.issn.1000-6699.2009.01.002. [21] 李碧汐, 李耘, 刘力松. 高血压病合并高尿酸血症与2型糖尿病的相关性研究[J]. 心肺血管病杂志, 2019, 38(8): 830-832, 837. DOI: 10.3969/j.issn.1007-5062.2019.08.003.Li BX, Li Y, Liu LS, Relationship between hypertension complicated with hyperuricemia and type 2 diabetes mellitus [J]. Journal of Cardiovascular and Pulmonary Diseases, 2019, 38(8): 830-832, 837. DOI: 10.3969/j.issn.1007-5062.2019.08.003. [22] 王维波, 李振卿, 王继美, 等. 2型糖尿病合并高血压、血脂异常住院患者的药学科普干预研究[J]. 中国药师, 2019, 22(8): 1464-1467. DOI: 10.3969/j.issn.1008-049X.2019.08.021.Wang WB, Li ZQ, Wang JM, et al. Effects of pharmaceutical science intervention service in type 2 diabetes patients with hypertension and dyslipidemia [J]. China Pharmacist, 2019, 22(8): 1464-1467. DOI: 10.3969/j.issn.1008-049X.2019.08.021. [23] Zhou X, Ji L, Luo Y, et al. Risk factors associated with the presence of diabetes in Chinese communities in Beijing [J]. Diabetes Res Clin Pract, 2009, 86(3): 233-238. DOI: 10.1016/j.diabres.2009.09.014. [24] 潘磊磊, 卢春明, 吴明, 等. 辽宁省35~75岁居民糖尿病患病率及影响因素[J]. 中华疾病控制杂志, 2020, 24(6): 670-675. DOI: 10.16462/j.cnki.zhjbkz.2020.06.010.Pan LL, Lu CM, Wu M, et al. Analysis on the diabetes prevalence and its influencing factors in population aged 35-75 years in Liaoning Province [J]. Chin J Dis Control Prev, 2020, 24(6): 670-675. DOI: 10.16462/j.cnki.zhjbkz.2020.06.010. [25] 梁森, 韩冰, 范雷, 等. 河南省35~74岁人群糖尿病患病率及相关因素分析[J]. 中华疾病控制杂志, 2018, 22(6): 569-572, 589. DOI: 10.16462/j.cnki.zhjbkz.2018.06.007.Liang S, Han B, Fan L, et al. Prevalence of diabetes mellitus and associated risk factors in population aged 35-74 years in Henan Province [J]. Chin J Dis Control Prev, 2018, 22(6): 569-572, 589. DOI: 10.16462/j.cnki.zhjbkz.2018.06.007. [26] 刘先锋. 重庆市高脂血症患病率及影响因素研究[D]. 重庆: 重庆医科大学, 2007.Liu XF. Analysis on prevalence and factors related to hyperlipidemia in Chongqing [D]. Chongqing: Chongqing Medical University, 2007.