The Design And Realization Of Inteligent Diagnosis Of Abnormal ECG Waveforms

Cardiovascular diseases (CVDs) are on the top of death causes, and most countries face increasing and high rates of CVDs. Therefore early detection of cardiovascular diseases is of great importance. At the basics of carefully studied the drawbacks of the current problems in the ECG auto diagnosis system, this paper proposed a new auto diagnosis system which based on the semantic features rather than the traditional geometry features, and make a further studied concentrated on several topics including judging rules, inference mechanism and the design and implement of this system. The main studies and contributions summarized below:(1) The paper proposed a new knowledge representation concept:semantic features, and make a detail description about how to withdraw semantic features from ECG wave data. Semantic features are a higher level features than geometric features, it not only contain all the information the geometric features have, but also more interpretative and structured.(2) The paper proposed a new data training method:a process oriented data training method. It can deal with the situation that all the data arrived on intermittent, and it can make sure that the final result has nothing to do the coming order of the date. And make a further a prove on math method.(3) This paper designed and build a knowledge database for the expert system make a strategic decision. Tables including patient table, repository table, disease table, constant tables, ecgpoints tables. All this tables make up the reserve unit of expert knowledge base system.(4) This paper designed a new inference mechanism which used forward and backward inference according to whether this record has this semantic features. Finally we can get the diagnosis result through several computing process.(5) This paper make a initial implementation of auto diagnosis system on the windows which laid the foundation for further function extension and beautify the interface.