A Research Of Detection Algorithm Of Myocardial Infarction Based On Machine Learning

Electrocardiogram is a convenient and quick way to detect heart disease clinically.Traditional ECG diagnosis is basically based on signal processing.First,the ECG signal is subjected to correlation transformation to extract features from it,and then the disease is detected according to the feature.The traditional methods have some drawbacks,such as high quality requirements for original ECG,difficulty in signal processing,difficulty in feature extraction,etc.,especially for diseases such as myocardial infarction with many features and some features are vague.Some methods of machine learning can solve these problems to a large extent.Deep convolutional neural networks can extract features from atypical multi-dimensional myocardial infarction with multi-dimensional features.For typical myocardial infarction with well-characterized features,decision trees can be used to make rapid diagnosis.Therefore,this paper proposes a machine learning-based myocardial infarction detection algorithm,which can not only effectively solve the problem of myocardial infarction disease detection,but also provide a new idea for other disease detection,which has certain research value.This paper first introduces the background and significance of the subject,and analyzes the shortcomings of traditional ECG diagnosis and the research status of myocardial infarction,and proposes an algorithm based on machine learning for myocardial infarction detection.Through the in-depth study of the related technologies of the subject,the whole process from ECG pre-processing,waveform detection to feature extraction and model construction and evaluation is established.Among them,ECG pre-processing mainly includes ECG precision conversion,re-sampling,filtering,smoothing,de-baseline and other related algorithms.In this part,the improved adaptive filtering algorithm is implemented and compared with other filtering algorithms.The waveform detection part mainly realizes the detection and starting point localization of each wave of ECG,and the focus is on the positioning of the starting point of the QRS complex.Then,the ECG was extracted with medical knowledge,and then the specific model was constructed according to the different characteristics of typical myocardial infarction and atypical myocardial infarction,and the model was trained.Finally,the model was evaluated and the experimental results were analyzed.According to the characteristics of myocardial infarction disease,this paper proposes a combination of decision tree and convolutional neural network,which not only greatly increases the accuracy of myocardial infarction detection,but also reduces the overall time complexity of the algorithm.In this paper,12-lead data is used for analysis,and the location of myocardial infarction can be initially located.At the end of the paper,the PTB database and the medical university database were tested separately,and compared with other algorithms.The experimental results show that the accuracy of the algorithm is improved in the detection of myocardial infarction,and the overall performance is better than other algorithms.