Background The diagnosis of cardiac abnormalities based on heart sound signal is a research hotspot in recent years. The early diagnosis of cardiac abnormalities has a crucial significance for the treatment of heart diseases. Methods For the sake of achieving more practical clinical applications of automatic recognition of cardiac abnormalities, here we proposed a novel fuzzy matching feature extraction method. First of all, a group of Gaussian wavelets are selected and then optimized based on a template signal. Convolutional features of test signal and the template signal are then computed. Matching degree and matching energy features between template signal and test signal in time domain and frequency domain are then extracted. To test performance of proposed feature extraction method, machine learning algorithms such as K-nearest neighbor, support vector machine, random forest and multilayer perceptron with grid search parameter optimization are constructed to recognize heart disease using the extracted features based on phonocardiogram signals. Results As a result, we found that the best classification accuracy of random forest reaches 96.5% under tenfold cross validation using the features extracted by the proposed method. Further, Mel-Frequency Cepstral Coefficients of phonocardiogram signals combing with features extracted by our algorithm are evaluated. Accuracy, sensitivity and specificity of integrated features reaches 99.0%, 99.4% and 99.7% respectively when using support vector machine, which achieves the best performance among all reported algorithms based on the same dataset. On several common features, we used independent sample t-tests. The results revealed that there are significant differences (p < 0.05) between 5 categories. Conclusion It can be concluded that our proposed fuzzy matching feature extraction method is a practical approach to extract powerful and interpretable features from one-dimensional signals for heart sound diagnostics and other pattern recognition task.

Wearing masks has been generally recommended to reduce the spreading of COVID-19. However, little is known about its effects on metabolic VOC changes in human body. To explore how the duration of wearing masks influences VOC metabolism in the human body, the essay used a self-developed electronic nose to analyse exhaled breath samples from 10 healthy individuals in this study. Firstly, polytetrafluoroethylene sampling bags are used to collect breath samples after volunteers wearing masks for 1h, 2h, 3h, 4h, and 5h. Secondly, data pre-processing, including baseline calibration and normalization are carried out. Thirdly, the study used LDA for dimensionality reduction on the original data to extract 4 features. Fourthly, differences in the length of time of wearing masks are analysed. Then, 4 algorithms were applied for cluster analysis based on extracted features. Moreover, 3 supervised classification algorithms were used to recognize the duration of wearing masks. Finally, multi-dimensional linear regression is used to study the possibility of predicting the duration of wearing masks based on breath signals acquired through electronic noses. As a result, the first feature extracted by LDA significantly differs from each other in the duration of wearing masks (p<0.05). Cluster analysis results show that the optimal internal parameters Adjusted Rand Index, Adjusted Mutual Information, Homogeneity and V-measure reach 80.2%, 81.5%, 83.5% and 83.7% respectively. Using 5-fold cross-validation on the K nearest neighbour classification model, the best accuracy of recognizing durations of wearing a mask reaches 88%. R-square of multi-dimensional linear regression reaches 92.5%, which shows excellent fitting performance. It can be concluded that the VOC metabolism of the human may change with the duration of wearing masks. Further, “breath prints” obtained by electronic nose may have the potential to predict the effective time and even the quality of masks.