Classification and Characteristics of TCM Syndromes of Chronic Respiratory Failure Based on Self-adaptive Fuzzy Inference System

Abstract

To explore classification and characteristics of the common syndromes of chronic respiratory failure (CRF ) based on self-adaptive fuzzy inference system. The methods applied were in the following: collecting the data of patients with CRF from four 3 A-level hospitals and establishing a database with Epidata software; selecting the artificial neural network, fuzzy system to build up a self-adaptive fuzzy inference system and then programming with MATLAB6.5 software; testing the model’s reliability by means of Fisher-iris data and eventually obtaining characteristics of the common syndromes of CRF based on clinical data mining results. Finally the rationality was tested. Through the rule conversion for the main and secondary symptom screening, seven syndromes and their corresponding main and secondary symptoms were determined, including syndrome of accumulated phlegm-heat in the lung, stagnated phlegm obstructing the lung, edema due to yang deficiency, accumulated phlegm-dampness in the lung, deficiency of both qi and yin, stagnated phlegm obstructing the lung accompanied by yin deficiency pattern, mental confusion due to phlegm. The coincident diagnostic rate reached 74.% approved by the test results of syndrome diagnostic criteria. By the Fisher-iris data test of the model, acquisition of fuzzy classification rules reflected accurately regularity of centralized learning samples, which suggests reliability of the model and thus it can be applied to study of classification and characteristics of TCM syndromes.