类风湿性关节炎中医证候的计算机模式识别法应用研究

【作者】 李建婷；

【导师】 邓兆智；

【作者基本信息】 广州中医药大学 ， 中医内科学， 2000， 硕士

【摘要】 研究目的：类风湿性关节炎（RA）的发生发展是多种致病因素共同作用的结果，其病机复杂，病情反复缠绵，病程冗长，证候繁多复杂且多变，临床中医辨证较难准确掌握，妨碍了辨证技能和临床疗效的提高。我国卫生部药政局颁布的《中药新药临床研究指导原则》对RA制订了七种中医证候的分类标准，无疑对RA的临床研究有一定的指导意义，为了使RA中医证候建立在更为严格的、科学的基础上，更好地指导临床诊疗水平，我们拟应用计算机模式识别法，对RA的证候诊断学及其实验客观指标进行相关研究，通过观测RA患者的临床症状、体征及各项实验室检测指标的变化，分析这些变化与中医证候的关系，阐明RA患者中医证候间的内在联系，最终建立辨证分型“量化指标”的计算机软件，并通过这一软件量化标准来判断RA患者与及其所属证候，为治疗RA提供可靠依据，使中医证候具体化、数据化，为计算机在中医领域的应用，为中医现代化开辟新途径。 研究地点：广州中医药大学第二附属医院。 研究对象：1997年9月至1999年9月在广州中医药大学第二附属医院风湿病专科连续收治的类风湿性关节炎病人。西医诊断标准：按1987年美国风湿病学会制定的RA诊断标准。中医诊断标准：中医证候分类以《中药新药临床指导原则》、《实用中医风湿病学》、《中医证候诊断治疗学》为基础进行辨证分类。 主要检测方法：由风湿病专科的医师制定统一的标准和调查表，对病人的临床症状、体征、实验室检测数据以及医生对RA证候的判别结果进行收集整理，把上述资料输入电脑，采用计算机模式识别法把计算机判别的结果与临床医师对证候的判别结果进行研究、比较，因篇幅所限，本论文仅对202例类风湿性关节炎患者的临床中医证候进行比较分析，建立RA广州地区常见的五种证候的预测模型。 结果分析：计算机模式识别法结果显示不能用单一指标对RA的证候进行分型诊断，只能用多个指标所提供的信息对I（进行证候判别。计算机模式识别法提示怕冷、自觉发热、关节肿对五个证候样本的判别作用最大，其次是触热、晨僵、局部紫黯、倦怠。用马氏距离判别法对未知样本分类，对其结果进行检测，结果显示寒热错杂证判别准确率 96．4％；湿热阻络证￥临u准确率96．3％：气阴两虚证判别准确率89．36％：寒湿阻络证判别准确率97．73％；痰瘀阻络证判别准确率100％。总训练样本202例，判别准确率 96．53 ％，预报样本 40例，判别准确率 92．5％，总242例，判别准确率95．87％。以痰瘀阻络证预报准确率最高，气阴两虚证最低。与中医临床常规诊断方法和结果是基本一致。证明计算机模式识别法能对RA中医证候作出较为客观的判定与分类，通过该法可建立起一套能有效地判别健康人是否患RA、属RA何种中医证候的客观的诊断程序，为中医理论走向世界提供可靠科学的理论依据。 结论：本研究证明计算机模式识别法对RA证候的判别分类所得出的结果和临床医生常规判断所得出的结果基本一致，从而证实了中医证候理论的客观性和科学性，证明该法能使中医的辨证分型诊断在保持中医特色的前题下得以标准化和规范化，并在某种程度上减少了临床医师的主观性和臆测性，我们预测依据传统的中医辨证、计算机处理结果，制订统一的判别公式、方程，变宏观辨证为微观辨证，变定性诊断为定量诊断，在临床上广泛椎广应用，有利于中医辨证分型诊断的标准化、规范化和逐步向现代化迈进。更多还原

【Abstract】 Objective: Rheumatic arthritis (RA) is a repeatedly paroxysmal disease which caused by many etiological factors. The superiority and soul of Traditional Chinese Medicine (TCM) are diagnosis and treatment on the basis of an overall analysis of the illness and the patient’s condition, which is in terms of syndrome of the patient. Syndrome is a dynamic process. It is complicated in clinic and can easily be affected by the outside environment or by the doctor’s subjectivity, which restrict the improvement of the differentiation technique and the clinical curative effect. In order to establish an effective system based on the modern science, directing the clinical diagnosis and treatment, we apply the computer pattern recognition to develop predictive models that can differentiate the syndrome of mixed cold and heat, obstructed channel by cold and dampness, deficiency of qi and yin etc, illuminating the relations between the syndromes through observing the clinical symptom, laboratory detection index of the RA patient. This may promote the measurement standard of the syndrome of RA. We hope we could, through this study, find an appropriate way of measuring the five syndromes of RA.Setting: the second hospital affiliated to Guangzhou University of Traditional Chinese Medicine.Patients/participants: All the patients involved in the study were the successive outpatients and inpatients of the department of rheumatic disease of the second hospital affiliated to Guangzhou University of Traditional Chinese Medicine from July, 1997 to Octomber, 1999. The selected cases were those meeting the revised ARA criteria for rheumatoid arthritis, 1987,30:17 and the revised criteria based on the clinical golden rule of Traditional Chinese Medicine.Main Outcome Measures: We apply computer pattern recognition to analyze the clinical symptom, physical sign and laboratory detection index of 202 RA patients,establishing a predictive model thai, can differentiate* I he syndrome of RA, illuminating the relations between the syndromes.Result: The computer pattern recognition shows that tve can’t use one parameter to differentiate the syndrome of RA hut the ififonniilioM supplied by many parameters. Allergic to cold(冷), self-sensitivity of heat in joint (自觉发热), swelling of’ joint (肿) were important variables to develop the model. Mahalanobis distance method shows the distinguishing accuracy of mixed cold and heat; is 96. the obstruction of dampness and heat in vessels and channels is 9(i. .’J%; the deficiency of <|i and yin is , which is the lowest: the obstruction of dampness and cold in vessels and channels is 97. 73%; the obstruction of phlegm and blond stasis in vessels and channels is 100%, which is the highest. The general accuracy of the 202 cases is 9. 5?, the predicting accuracy is 92.5%, and the general accuracy of 242 cases is 95.87%,, So we considered thai, the predictive models could be used to predict the syndrome of RA.Conclusions: The present study shows that the computer pattern recognition can be used to predict the differentiation of RA avoiding the subjective and hypotheses of the doctors, reducing the disturbance of subjective factors, turning the macroscopic differentiation into microcosmic differentiation, normal mng the RA diagnostic. So it can be broadly used in the clinic.更多还原