贝叶斯网络模型在老年人认知功能评价队列研究中的应用
发布时间：2018-07-16 08:05
【摘要】：目的:基于阿尔茨海默病(Alzheimer disease,AD)的疾病自然史,将贝叶斯网络模型应用于老年人认知评价研究,建立老年人认知功能评价的推理及预测模型,探讨影响老年人认知功能的因素,为制定AD不同发展阶段的防治措施提供理论依据。同时从实际应用角度出发,挖掘贝叶斯统计在队列研究数据分析中的优势,为其他慢性病的进程研究提供方法学借鉴,将精准医学的概念扩展至疾病风险预测。方法:采用项目组前期的调查随访数据,以2014年5月的数据为基线数据,对301名调查对象以正常老化→轻度认知损害(Mild Cognitive Impairment,MCI)、MCI→中重度认知损害、中重度认知损害→AD是否转移构建Logistic回归模型,筛选对认知退化有影响的因素,将进入模型的所有变量构建贝叶斯网络模型,分析各变量间的概率依赖关系,并采用10折交叉验证法对建模数据和2016年5月的随访数据进行模型预测效果评价,实现老年人认知功能评价的推理及预测。结果:1、Logistic回归模型结果显示,年龄(OR:1.794,95%CI:1.200-2.682)、性别(OR:4.125,95%CI:2.017-8.436)、受教育程度(OR:0.633,95%CI:0.448-0.894)、抑郁程度(OR:4.458,95%CI:1.915-10.377)、高血压(OR:2.346,95%CI:1.086-5.069)对正常老化→MCI的转移有影响;年龄(OR:2.450,95%CI:1.212-4.953)、性别(OR:0.118,95%CI:0.031-0.442)、受教育程度(OR:0.614,95%CI:0.375-1.004)、性格(OR:0.092,95%CI:0.013-0.662)、婚姻状况(OR:0.272,95%CI:0.086-0.862)、家庭人均收入(OR:0.456,95%CI:0.273-0.762)、身体活动(OR:0.631,95%CI:0.407-0.980)和读书看报(OR:0.432,95%CI:0.188-0.992)、抑郁程度(OR:97.144,95%CI:21.452-439.909)、高血压(OR:0.304,95%CI:0.077-1.199)、脑外伤史(OR:0.188,95%CI:0.037-0.959)均对MCI→中重度认知损害转移有影响;性别(OR:0.328,95%CI:0.087-1.234)、婚姻状况(OR:0.102,95%CI:0.043-0.243)、离休前职业(OR:7.799,95%CI:1.242-48.955)、饮酒(OR:0.126,95%CI:0.016-0.997)、抑郁程度(OR:3.560,95%CI:0.998-12.705)在中重度认知损害→AD转移中有统计学意义。2、用于构建贝叶斯网络结构的节点包括认知功能、抑郁程度、性别、年龄、性格、受教育程度、婚姻状况、家庭人均收入、离休前职业、读书看报、身体活动、饮酒、脑外伤史、高血压。结果显示模型的期望损失为10.28,且贝叶斯网络的预测效能优于其他分类器效能,对认知功能的预测准确率为77.14%,灵敏度为0.869,特异度为0.770,抑郁情况的预测准确率为80.07%,灵敏度为0.801,特异度为0.648。对老年人认知功能有直接作用的是高血压、受教育程度、离休前职业和抑郁程度,性别、家庭人均收入、婚姻状况、读书看报、身体活动、性格分别通过受教育程度、离休前职业、抑郁程度、高血压间接作用于认知功能。结论:1、高血压、受教育程度、离休前职业和抑郁程度与老年人认知功能可能有直接因果关系,性别、家庭人均收入、婚姻状况、读书看报、身体活动、性格可能分别通过受教育程度、离休前职业、抑郁程度、高血压间接作用于认知功能。老年人在平时的生活中多读书、进行适当的身体活动、培养外向的性格和适当饮酒可能延缓认知和记忆的退化。2、将贝叶斯网络应用于老年人认知队列研究,可以直观地了解认知功能影响因素之间的相互关系,并实现了因果推断和个体的疾病风险预测,表明其在医学研究中的优势和在其他疾病中实践的可行性。
[Abstract]:Objective: Based on the natural history of Alzheimer disease (AD), the Bias network model was applied to the cognitive evaluation of elderly people, the reasoning and prediction model of cognitive function evaluation for the elderly was established, and the factors affecting the cognitive function of the elderly were discussed, and the theoretical basis for the formulation of the prevention measures at different stages of AD was provided. At the same time, from the practical point of view, we excavate the advantages of Bayesian statistics in the cohort study data analysis, provide methodology for other chronic disease process research, and extend the concept of precision medicine to the risk prediction of disease. The subjects with normal aging, mild cognitive impairment (Mild Cognitive Impairment, MCI), MCI to moderate to severe cognitive impairment, moderate to severe cognitive impairment and whether or not AD transfer to construct Logistic regression model, select factors that affect cognitive degradation, and will enter all variables of the model to construct Bayesian network model and analyze the probability of each variable. Rate dependence, and using the 90% off cross validation method to evaluate the model data and the follow-up data of May 2016, to realize the reasoning and prediction of cognitive function evaluation for the elderly. Results: 1, the results of Logistic regression model show that age (OR:1.794,95%CI: 1.200-2.682), sex (OR:4.125,95%CI:2.017-8.436), education degree (OR:0.633,95%CI:0.448-0.894), the degree of depression (OR:4.458,95%CI:1.915-10.377), hypertension (OR:2.346,95%CI:1.086-5.069) had an effect on the transfer of normal aging to MCI; age (OR:2.450,95%CI:1.212-4.953), sex (OR:0.118,95%CI:0.031-0.442), educational level (OR:0.614,95%CI:0.375-1.004), character (OR:0.092,95%CI:0.013-0.662), Marital status (OR:0.272,95%CI:0.086-0.862), family per capita income (OR:0.456,95%CI:0.273-0.762), physical activity (OR:0.631,95%CI:0.407-0.980) and reading and reading (OR:0.432,95%CI:0.188-0.992), degree of depression (OR:97.144,95%CI:21.452-439.909), hypertension (OR:0.304,95%CI:0.077-1.199), and history of brain trauma (OR:0.188,95%CI:0.037-0.959) The influence of MCI to moderate severe cognitive impairment; sex (OR:0.328,95%CI:0.087-1.234), marital status (OR:0.102,95%CI:0.043-0.243), pre retirement occupation (OR:7.799,95%CI:1.242-48.955), drinking (OR:0.126,95%CI:0.016-0.997), and degree of depression (OR:3.560,95% CI:0.998-12.705) were statistically significant in moderate to severe cognitive impairment to AD metastasis. .2, the nodes used to build Bayesian network structure include cognitive function, degree of depression, sex, age, personality, education, marital status, family per capita income, pre retirement occupation, reading and reading, physical activity, drinking, brain trauma history, hypertension. The expected loss of the model is 10.28, and the predictive effectiveness of the Bayesian network is better than that of the Bayesian network. The accuracy of the cognitive function was 77.14%, the sensitivity was 77.14%, the sensitivity was 0.869, the specificity was 0.770, the accuracy of the depression was 80.07%, the sensitivity was 0.801. The specificity was the hypertension, the degree of education, the degree of occupation and depression before the rest, the sex, and the family per person. Income, marital status, reading and reading, physical activity, and character respectively through the degree of education, pre occupation, depression, and hypertension indirect effect on cognitive function. Conclusion: 1, hypertension, education, the degree of occupation and depression before retirement and the cognitive work of the elderly may have direct causal relationship, sex, family per capita income, marriage The status of reading, reading, reading, physical activity and character may be indirectly influenced by the degree of education, the occupation before the rest, the degree of depression, and the hypertension. The elderly in the normal life are reading more, carrying out proper physical activity, developing the extroverted character and drinking wine, which may delay the degenerative.2 of cognition and memory, and the Bias network Applied to the cognitive cohort study of the elderly, we can intuitively understand the relationship between the factors affecting the cognitive function, and realize the causal inference and the individual's disease risk prediction, indicating its advantages in medical research and the feasibility of practice in other diseases.
【学位授予单位】：山西医科大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R749.16;R181.3
本文编号：2125773
[Abstract]:Objective: Based on the natural history of Alzheimer disease (AD), the Bias network model was applied to the cognitive evaluation of elderly people, the reasoning and prediction model of cognitive function evaluation for the elderly was established, and the factors affecting the cognitive function of the elderly were discussed, and the theoretical basis for the formulation of the prevention measures at different stages of AD was provided. At the same time, from the practical point of view, we excavate the advantages of Bayesian statistics in the cohort study data analysis, provide methodology for other chronic disease process research, and extend the concept of precision medicine to the risk prediction of disease. The subjects with normal aging, mild cognitive impairment (Mild Cognitive Impairment, MCI), MCI to moderate to severe cognitive impairment, moderate to severe cognitive impairment and whether or not AD transfer to construct Logistic regression model, select factors that affect cognitive degradation, and will enter all variables of the model to construct Bayesian network model and analyze the probability of each variable. Rate dependence, and using the 90% off cross validation method to evaluate the model data and the follow-up data of May 2016, to realize the reasoning and prediction of cognitive function evaluation for the elderly. Results: 1, the results of Logistic regression model show that age (OR:1.794,95%CI: 1.200-2.682), sex (OR:4.125,95%CI:2.017-8.436), education degree (OR:0.633,95%CI:0.448-0.894), the degree of depression (OR:4.458,95%CI:1.915-10.377), hypertension (OR:2.346,95%CI:1.086-5.069) had an effect on the transfer of normal aging to MCI; age (OR:2.450,95%CI:1.212-4.953), sex (OR:0.118,95%CI:0.031-0.442), educational level (OR:0.614,95%CI:0.375-1.004), character (OR:0.092,95%CI:0.013-0.662), Marital status (OR:0.272,95%CI:0.086-0.862), family per capita income (OR:0.456,95%CI:0.273-0.762), physical activity (OR:0.631,95%CI:0.407-0.980) and reading and reading (OR:0.432,95%CI:0.188-0.992), degree of depression (OR:97.144,95%CI:21.452-439.909), hypertension (OR:0.304,95%CI:0.077-1.199), and history of brain trauma (OR:0.188,95%CI:0.037-0.959) The influence of MCI to moderate severe cognitive impairment; sex (OR:0.328,95%CI:0.087-1.234), marital status (OR:0.102,95%CI:0.043-0.243), pre retirement occupation (OR:7.799,95%CI:1.242-48.955), drinking (OR:0.126,95%CI:0.016-0.997), and degree of depression (OR:3.560,95% CI:0.998-12.705) were statistically significant in moderate to severe cognitive impairment to AD metastasis. .2, the nodes used to build Bayesian network structure include cognitive function, degree of depression, sex, age, personality, education, marital status, family per capita income, pre retirement occupation, reading and reading, physical activity, drinking, brain trauma history, hypertension. The expected loss of the model is 10.28, and the predictive effectiveness of the Bayesian network is better than that of the Bayesian network. The accuracy of the cognitive function was 77.14%, the sensitivity was 77.14%, the sensitivity was 0.869, the specificity was 0.770, the accuracy of the depression was 80.07%, the sensitivity was 0.801. The specificity was the hypertension, the degree of education, the degree of occupation and depression before the rest, the sex, and the family per person. Income, marital status, reading and reading, physical activity, and character respectively through the degree of education, pre occupation, depression, and hypertension indirect effect on cognitive function. Conclusion: 1, hypertension, education, the degree of occupation and depression before retirement and the cognitive work of the elderly may have direct causal relationship, sex, family per capita income, marriage The status of reading, reading, reading, physical activity and character may be indirectly influenced by the degree of education, the occupation before the rest, the degree of depression, and the hypertension. The elderly in the normal life are reading more, carrying out proper physical activity, developing the extroverted character and drinking wine, which may delay the degenerative.2 of cognition and memory, and the Bias network Applied to the cognitive cohort study of the elderly, we can intuitively understand the relationship between the factors affecting the cognitive function, and realize the causal inference and the individual's disease risk prediction, indicating its advantages in medical research and the feasibility of practice in other diseases.
【学位授予单位】：山西医科大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R749.16;R181.3
【参考文献】
相关期刊论文 前10条
1 焦怡琳;王吉春;张群;何广学;;中国在精准医学领域面临的机遇与挑战[J];中国公共卫生管理;2015年05期
2 冯荣芳;刘惠卿;贾艳丽;王建华;李美锡;孙素娟;孙占用;;MMSE和MoCA在不同类型轻度认知障碍筛查中的应用[J];脑与神经疾病杂志;2015年05期
3 杨振;张冬梅;陈任;胡志;王晶晶;陈若陵;;我国五省(市)老年痴呆照料现状调查分析[J];安徽医学;2013年06期
4 唐丹;;简版老年抑郁量表(GDS-15)在中国老年人中的使用[J];中国临床心理学杂志;2013年03期
5 王广州;戈艳霞;;中国老年人口丧偶状况及未来发展趋势研究[J];老龄科学研究;2013年01期
6 李建民;杜鹏;桂世勋;张翼;;新时期的老龄问题我们应该如何面对[J];人口研究;2011年04期
7 高中宝;王炜;尚延昌;柏秀娟;吴卫平;;蒙特利尔认知量表中文版诊断老年轻度认知功能损害的应用研究[J];中华保健医学杂志;2011年03期
8 于大林;肖军;;阿尔茨海默病的流行病学调查现况[J];实用医院临床杂志;2011年03期
9 唐娟娟;肖军;;蒙特利尔认知评估量表与简易精神状态量表在认知功能障碍筛查中的应用与比较[J];实用医院临床杂志;2011年02期
10 田金洲;时晶;苗迎春;王平;孔明望;;阿尔茨海默病的流行病学特点及其对公共卫生观念的影响[J];湖北中医学院学报;2009年01期
相关博士学位论文 前2条
1 胡春玲;贝叶斯网络结构学习及其应用研究[D];合肥工业大学;2011年
2 宫秀军;贝叶斯学习理论及其应用研究[D];中国科学院研究生院（计算技术研究所）;2002年
相关硕士学位论文 前1条
1 陈固胜;基于动态贝叶斯网络的战场信息预测与评估[D];南京理工大学;2013年
，本文编号：2125773
本文链接：https://www.wllwen.com/yixuelunwen/jsb/2125773.html
最近更新
教材专著
