Abstract: The medical diagnosis expert system is a variant of the artificial intelligence which has been used for the medical diagnosis and the medical consultation. It supplements the shortage of human experts and can effectively solve a variety of clinical problems. However, the existing diagnostic reasoning models are too dependent on the weight set and doctors experience value, diagnosis results are entirely dependent on the weights given quality, and until now there is not a practical weight setting and automatic assignment method. In order to solve the problem of weight setting and automatic assignment, this paper presents a method of medical diagnosis based on correlation analysis of features. Firstly, the feature correlation matrix is used to prune the features to avoid unnecessary calculation of the invalid features. Then, a practical mathematical model of weight setting based on the analysis of medical data is present. Secondly, the group decision method is used to calculate the similarity between the medical records. Finally, it simulates the proposed method. The experimental results show that the methods can effectively solve the problem of automatic weight setting and meet the needs of practical application.

Key words: medical diagnosis expert system, correlation analysis, group decision

摘要： 医疗诊断专家系统作为人工智能重要应用领域之一，已被广泛应用于医疗诊断和网络医疗咨询等方面。其补充了人类专家的不足，并能有效地解决各种临床问题。然而，现有诊断推理模型在病例推理过程中过于依赖权值设定和医生经验值，诊断结果完全依赖于权值给定质量，并且目前尚未发现较实用的权值设定和自动赋值方法。为了解决权值设定和自动赋值的难题，提出了一种基于属性相容性分析的医疗诊断方法并进行了探讨。利用基于属性的相容性分析对医疗数据进行数字化处理，并根据属性之间的关联度构造属性相关矩阵和对属性进行剪枝处理，避免了无效属性的不必要运算；在相容性分析基础上提出了较为实用的权值设定数学模型；在进行病例相似度计算时，采用群体决策策略来完成诊断。针对所提方法进行了模拟。实验结果表明，该方法能有效解决权值自动设定问题和满足实际应用需求。

关键词: 医疗诊断专家系统, 相容性分析, 群体决策