模糊决策在医疗诊断中的应用
|
摘要
由于存储在医疗数据库中的知识和数据具有广泛性，这就对数据的存储、连接、分析及被存储的知识和数据能够有效地使用的专业工具提出更高的要求。在Edward H. Shortliffe文章“人工智能在医疗中正处在青春期”中提出影响人工智能成功应用于医疗的三个因素：训练的增强、国际标准和信息的底层结构。从1993年起，信息的底层结构就已经较其余两个因素有很大发展。实际上，医疗资料学（研究资料的搜集、分类、储存等的学科）已经成为医疗机构的重要组成部分。许多现代化医院和健康中心都配有监视装置和数据收集设施，使数据能被收集且在连接的网络上共享。现代化的医院正快速发展其信息系统，从前曾是一个单独的数据库或单独的实验室信息系统，现在都被连接成一个大规模的医疗信息系统。
数据量的增加引起决策支持引用有用信息的困难。传统的手动数据分析已经远远不能满足要求，基于计算机的有效分析方法成为必不可少的技术。例如，在智能数据分析领域中发展的技术，尤其是数据的抽象和数据的挖掘。智能数据分析（IDA）包括统计、模式识别、机器学习、数据抽象和可视化工具。他们的作用是数据收集和数据理解的桥梁，目的是更有效地执行医疗工作。信息革命的到来使之通过电子媒介收集和存储大量数据成为可能。
如今，由于数据比较容易获得，使合成函数在医疗领域得到应用。当合成数值型数据时，主要的算法有两个WM和OWA。这两个函数的不同之处在于权的涵义和如何根据输入数据合成权。确切地说，WM算法中的权衡量的是信息源的可靠性，而不管数据源所提供的数据。相反，OWA算法中的权衡量的是这些数据之间的相对关系而不管数据是由什
么来提供的。为了利用两套权的优势，我们提出了WOWA算法，它合成了两种函数的优势，既考虑了使用者加权信息的可靠性，同时又考虑了数据间的相对位置。
本文把医疗与数据分析结合起来，用模糊知识和智能合成技术解决医疗诊断的难题。即睡眠呼吸暂停症状的检测。
呼吸暂停是难以诊断的病例，目前还没有比较满意的标准统计模型技术来解决此病例。我们认为主要问题是由于“问卷回答”作为数据的内在模糊性。本文用模糊法来对“问卷回答”进行表示，把所获得的知识作为病人回答问题的隶属度，并且作为用于睡眠暂停呼吸症状诊断的最后合成值。采用可靠性和相关性权的形式表示知识，用两种不同的数据合成技术（WOWA，OWA）熔合每个病例诊断中不同的变量，OWA（ordered weighted average, OWA）算子考虑数据之间的相对关系，WOWA（weighted ordered weighted average, OWA）算子用两组权考虑两方面的因素，从两套权计算一套新的权。同时，我们提供了获得相关权的方法——遗传算法（genetic algorithms,简称GA），并与医疗专家所定义的相关权相比较。
Op是WOWA预测的诊断，Or是标准化RDI值，是一个实际的诊断。对于所有患者病例，目的是减少在总数Op和总数Or之间的差值。即：
Min(ΣOp-ΣOr)
我们能用遗传算法技术从医院所得数据值获得ρ（相关性）权系数。遗传算法模仿自然选择的过程，从当前一代到以后选择最适合的生存。遗传算法有一套输入和输入数据。一套可以更改的值，一套强制限制（在本文中为加权系数总和为1）和一个目的函数，在我们这个病例中是减少预测诊断和实际诊断的差值。我们通过遗传算法技术使输入和输出相接近找到相关性权，而减少目的函数。
Extensive amounts of knowledge and data stored in medical databases require the development of specialized tools for storing and accessing of data, data analysis, and effective use of stored knowledge and data. In this excellent article on “the adolescence of AI in Medicine”, Edward H. Shortliffe exposes three factors that may influence the successful integration of AI systems: enhancement of training, international standards, and information infrastructure. Since 1993, information infrastructure has certainly advanced more than the other two factors. In fact, medical informatics has become an integral part of successful medical institution. Many modern hospitals and health care institutions are now well equipped with monitoring and other data collection devices, and data is gathered and shared in inter- and intra-hospital information systems. Modern hospitals are rapidly advancing their information systems. What was before and isolated database or a laboratory information system is now integrated in a larger scale medical information system.
The increase in data volume causes difficulties in extracting useful information for decision support. The traditional manual data analysis has become insufficient, and methods for efficient computer-based analysis are indispensable, such as the technologies developed in the area of intelligent data analysis, in particular data abstraction and of data mining. Intelligent data analysis(IDA) encompasses statistical, pattern recognition, machine learning, data abstraction and visualization tools to support the analysis of data and discovery of principles that are encoded within the data. Their role is clearly that of an intelligent assistant that tries to bridge the gap between data
gathering and data comprehension, in order to enable the physician to perform his task more efficiently and effectively. The information revolution made it possible to collect and store large volumes of data from diverse sources on electronic media.
In these days, the need of combination functions in medicine fields. When the objects to synthesise are numeric values, There are two combination functions (WM,OWA). Both functions, the weighted mean and the OWA operator are to combine values according to a set of weights. However, the meaning of these weights is different in both functions. The weighted mean computes a value that synthesizes the ones of the information sources taking into account the reliability of these sources. The OWA operator, instead, combines the information allowing to weight the values in relation to their ordering position. In order to take advantage of both sets of weights, We propose the WOWA operator that combines the advantages of both combination functions. The WOWA allows the user to weight the reliability of the information source and the values in relation to their relative position.
In this article, joint medical and data analysis expertise is brought to bear using fuzzy knowledge representation and “intelligent” aggregation techniques to solve a difficult medical diagnosis problem, that of sleep apnea syndrome screening.
Screening of Apnea cases is a difficult diagnosis problem, at present not satisfactorily resolved by standard statistical modeling techniques. We propose that part of the problem is due to the inherent fuzzy nature of a significant part of the data: questionnaire.
In this article a fuzzy representation is proposed for the questionnaire.We can consider as derived knowledge the membership grades
of the patient responses, and the final aggregated value used for diagnosis of sleep apnea syndrome. The knowledge is represented in the form of reliability and relevance weight for each variable. Two contrasting data aggregation techniques (WOWA,OWA) “fuse” the variables into a diagnosis for each case. We present a way of learning information about the relevance of the data, that is genetic algorithms. comparing this with the definition of the information by the medical expert.
If Op is the diagnosis predicted by the aggregation
数据量的增加引起决策支持引用有用信息的困难。传统的手动数据分析已经远远不能满足要求，基于计算机的有效分析方法成为必不可少的技术。例如，在智能数据分析领域中发展的技术，尤其是数据的抽象和数据的挖掘。智能数据分析（IDA）包括统计、模式识别、机器学习、数据抽象和可视化工具。他们的作用是数据收集和数据理解的桥梁，目的是更有效地执行医疗工作。信息革命的到来使之通过电子媒介收集和存储大量数据成为可能。
如今，由于数据比较容易获得，使合成函数在医疗领域得到应用。当合成数值型数据时，主要的算法有两个WM和OWA。这两个函数的不同之处在于权的涵义和如何根据输入数据合成权。确切地说，WM算法中的权衡量的是信息源的可靠性，而不管数据源所提供的数据。相反，OWA算法中的权衡量的是这些数据之间的相对关系而不管数据是由什
么来提供的。为了利用两套权的优势，我们提出了WOWA算法，它合成了两种函数的优势，既考虑了使用者加权信息的可靠性，同时又考虑了数据间的相对位置。
本文把医疗与数据分析结合起来，用模糊知识和智能合成技术解决医疗诊断的难题。即睡眠呼吸暂停症状的检测。
呼吸暂停是难以诊断的病例，目前还没有比较满意的标准统计模型技术来解决此病例。我们认为主要问题是由于“问卷回答”作为数据的内在模糊性。本文用模糊法来对“问卷回答”进行表示，把所获得的知识作为病人回答问题的隶属度，并且作为用于睡眠暂停呼吸症状诊断的最后合成值。采用可靠性和相关性权的形式表示知识，用两种不同的数据合成技术（WOWA，OWA）熔合每个病例诊断中不同的变量，OWA（ordered weighted average, OWA）算子考虑数据之间的相对关系，WOWA（weighted ordered weighted average, OWA）算子用两组权考虑两方面的因素，从两套权计算一套新的权。同时，我们提供了获得相关权的方法——遗传算法（genetic algorithms,简称GA），并与医疗专家所定义的相关权相比较。
Op是WOWA预测的诊断，Or是标准化RDI值，是一个实际的诊断。对于所有患者病例，目的是减少在总数Op和总数Or之间的差值。即：
Min(ΣOp-ΣOr)
我们能用遗传算法技术从医院所得数据值获得ρ（相关性）权系数。遗传算法模仿自然选择的过程，从当前一代到以后选择最适合的生存。遗传算法有一套输入和输入数据。一套可以更改的值，一套强制限制（在本文中为加权系数总和为1）和一个目的函数，在我们这个病例中是减少预测诊断和实际诊断的差值。我们通过遗传算法技术使输入和输出相接近找到相关性权，而减少目的函数。
Extensive amounts of knowledge and data stored in medical databases require the development of specialized tools for storing and accessing of data, data analysis, and effective use of stored knowledge and data. In this excellent article on “the adolescence of AI in Medicine”, Edward H. Shortliffe exposes three factors that may influence the successful integration of AI systems: enhancement of training, international standards, and information infrastructure. Since 1993, information infrastructure has certainly advanced more than the other two factors. In fact, medical informatics has become an integral part of successful medical institution. Many modern hospitals and health care institutions are now well equipped with monitoring and other data collection devices, and data is gathered and shared in inter- and intra-hospital information systems. Modern hospitals are rapidly advancing their information systems. What was before and isolated database or a laboratory information system is now integrated in a larger scale medical information system.
The increase in data volume causes difficulties in extracting useful information for decision support. The traditional manual data analysis has become insufficient, and methods for efficient computer-based analysis are indispensable, such as the technologies developed in the area of intelligent data analysis, in particular data abstraction and of data mining. Intelligent data analysis(IDA) encompasses statistical, pattern recognition, machine learning, data abstraction and visualization tools to support the analysis of data and discovery of principles that are encoded within the data. Their role is clearly that of an intelligent assistant that tries to bridge the gap between data
gathering and data comprehension, in order to enable the physician to perform his task more efficiently and effectively. The information revolution made it possible to collect and store large volumes of data from diverse sources on electronic media.
In these days, the need of combination functions in medicine fields. When the objects to synthesise are numeric values, There are two combination functions (WM,OWA). Both functions, the weighted mean and the OWA operator are to combine values according to a set of weights. However, the meaning of these weights is different in both functions. The weighted mean computes a value that synthesizes the ones of the information sources taking into account the reliability of these sources. The OWA operator, instead, combines the information allowing to weight the values in relation to their ordering position. In order to take advantage of both sets of weights, We propose the WOWA operator that combines the advantages of both combination functions. The WOWA allows the user to weight the reliability of the information source and the values in relation to their relative position.
In this article, joint medical and data analysis expertise is brought to bear using fuzzy knowledge representation and “intelligent” aggregation techniques to solve a difficult medical diagnosis problem, that of sleep apnea syndrome screening.
Screening of Apnea cases is a difficult diagnosis problem, at present not satisfactorily resolved by standard statistical modeling techniques. We propose that part of the problem is due to the inherent fuzzy nature of a significant part of the data: questionnaire.
In this article a fuzzy representation is proposed for the questionnaire.We can consider as derived knowledge the membership grades
of the patient responses, and the final aggregated value used for diagnosis of sleep apnea syndrome. The knowledge is represented in the form of reliability and relevance weight for each variable. Two contrasting data aggregation techniques (WOWA,OWA) “fuse” the variables into a diagnosis for each case. We present a way of learning information about the relevance of the data, that is genetic algorithms. comparing this with the definition of the information by the medical expert.
If Op is the diagnosis predicted by the aggregation
引文
1、刘正纾，等. 医学哲学概论—医学的主体、客体和整体. 北京：中国医药科技出版社，1992
2、Jhon E. Simpson. XSL技术实践. 机械工业出版社. 2002,7
3、L.A.Zadeh, Knowledge representation in fuzzy logic, In: R.R.Yager and L.A.Zadeh eds., An introduction to fuzzy logic. Applications in intelligent systems (Kluwer Academic Publisher, Boston, 1992)2-25.
4、J.C.Fodor and M. Roubens, Aggregation and scoring procedures in multicriteria decision-making methods, Proceedings of the IEEE International Conference on Fuzzy Systems, San Diego, 1992 1261-1267
5、J.C.Fodor and M. Roubens, Fuzzy Preference Modelling and Multicriteria Decision Support, Theory and Decision Library, Series D, system Theory, Knowledge Engineering and Problem Solving, Vol. 14, Kluwer, Dordrecht, 1994.
6、Yager, R.R.,Families of OWA operators, Fuzzy Set and systems, 59(1993):125-148
7、徐泽水等. 一种不确定型OWA算子及其在群决策中的应用. 东南大学学报，
2002，32（1）：147-150
8、Torra,V., Cortes, U., Towards and automatic consensus generator tool: EGAC, IEEE Transactions on Systems, Man and Cybernetics, 25:5(1995)888-894.
9、Torra, V.,The Weighted OWA Operator.International Journal of Intelligent Systems,
Vol. 12,153-166. John Wiley& Sons(1997)
10、Chen, J.E., OTTO, K.N., Constructing membership functions using interpolation and measurement theory, Fuzzy Sets and Systems, 73:3 (1995)313-327.
11、玄光男，程润伟. 遗传算法与工程设计. 北京：科学出版社，2000
12、孔维佳主编. 耳鼻咽喉科学. 人民卫生出版社，2002
13、郑中立主编. 耳鼻咽喉科诊断学. 人民卫生出版社，1989
14、康健. 睡眠呼吸障碍的检查技术[J]. 中国实用内科杂志， 2002，22：1-2
15、Hoffstein, V., Szalai J.P. Predictive value of clinical features in diagnosing obstructive sleep apnea. Sleep 1993,16:118-122
16、Katz I., Stradling J., Slutsky A.S., et al. Do patients with sleep apnea have thick necks? American Review of Respiratory Diseases, 1990, 141:1228-1231.
2、Jhon E. Simpson. XSL技术实践. 机械工业出版社. 2002,7
3、L.A.Zadeh, Knowledge representation in fuzzy logic, In: R.R.Yager and L.A.Zadeh eds., An introduction to fuzzy logic. Applications in intelligent systems (Kluwer Academic Publisher, Boston, 1992)2-25.
4、J.C.Fodor and M. Roubens, Aggregation and scoring procedures in multicriteria decision-making methods, Proceedings of the IEEE International Conference on Fuzzy Systems, San Diego, 1992 1261-1267
5、J.C.Fodor and M. Roubens, Fuzzy Preference Modelling and Multicriteria Decision Support, Theory and Decision Library, Series D, system Theory, Knowledge Engineering and Problem Solving, Vol. 14, Kluwer, Dordrecht, 1994.
6、Yager, R.R.,Families of OWA operators, Fuzzy Set and systems, 59(1993):125-148
7、徐泽水等. 一种不确定型OWA算子及其在群决策中的应用. 东南大学学报，
2002，32（1）：147-150
8、Torra,V., Cortes, U., Towards and automatic consensus generator tool: EGAC, IEEE Transactions on Systems, Man and Cybernetics, 25:5(1995)888-894.
9、Torra, V.,The Weighted OWA Operator.International Journal of Intelligent Systems,
Vol. 12,153-166. John Wiley& Sons(1997)
10、Chen, J.E., OTTO, K.N., Constructing membership functions using interpolation and measurement theory, Fuzzy Sets and Systems, 73:3 (1995)313-327.
11、玄光男，程润伟. 遗传算法与工程设计. 北京：科学出版社，2000
12、孔维佳主编. 耳鼻咽喉科学. 人民卫生出版社，2002
13、郑中立主编. 耳鼻咽喉科诊断学. 人民卫生出版社，1989
14、康健. 睡眠呼吸障碍的检查技术[J]. 中国实用内科杂志， 2002，22：1-2
15、Hoffstein, V., Szalai J.P. Predictive value of clinical features in diagnosing obstructive sleep apnea. Sleep 1993,16:118-122
16、Katz I., Stradling J., Slutsky A.S., et al. Do patients with sleep apnea have thick necks? American Review of Respiratory Diseases, 1990, 141:1228-1231.