Cardiovascular diseases are the most common type of diseases with the highest mortality rate, and the most common of these diseases is myocardial infarction. Regarding the prediction of myocardial infarction, the traditional method is judged by the expertise and experience of medical personnel, but this method may lead to inaccurate results due to differences in judgment criteria, while using deep learning methods to study myocardial infarction facilitates the processing of features and can improve the accuracy of diagnosing the disease. In this paper, we propose a CNN_SVM model of convolutional neural network combined with support vector machine to predict myocardial infarction. A total of 87 features are obtained based on the vital signs and biochemical examinations of patients in the MIMIC data set. For these features, the CNN_SVM model firstly performs PCA dimensionality reduction by mapping the data from high dimension to low dimension to reduce the dimensionality of the features and retain as much information as possible to complete the dimensionality reduction, and finally obtains 45 data features; secondly, the last layer of the CNN model, Softmax classification layer, is replaced with SVM classification, and the final results are two kinds, namely myocardial infarction and non-myocardial infarction, respectively. The model construction was completed. The experiments showed that the accuracy of CNN_SVM model was 96.67%: the accuracy was improved by 7.60%, 7.15% and 17% when compared with CNN, XGBoost and MLP, respectively.
The key problem of event extraction in the medical field is that the cost of medical data labeling is too high, and the labeled samples are scarce, making medical event extraction difficult. In response to this problem, this paper proposes to perform a partial synonymous replacement of training samples to expand the data, and the data and the original data together constitute new electronic medical record data (i.e. EDA data enhancement); In addition, the unlabeled data is predicted by the medical event extraction model to generate labeled data, and then the accurate labeled data is filtered out and added to the original data to form new electronic medical record data, thereby realizing data enhancement (i.e. UDF data enhancement) , to a certain extent, to solve the problem of the scarcity of medical data samples. Based on augmented data, a medical event extraction model (i.e. TEC_MEE model) based on Transformer Encoder and CRF are constructed to extract attributes of specified events from unstructured Chinese electronic medical record text. The experimental results show that, compared with the baseline model, the TEC_MEE model proposed in this paper obtains better medical event extraction results after data enhancement.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.