Key result
A novel pulse signal classification method using EPNCC and wavelet scattering feature fusion with a convolutional neural network achieved 98.3% accuracy in identifying three clinical conditions.
Why the study?
The study was conducted to rapidly obtain complete characterization information of pulse signals and evaluate their sensitivity and validity for clinical diagnosis of related diseases.
An improved EPNCC and wavelet scattering-based convolutional neural network achieved 98.3% accuracy in classifying pulse signals of CHF pulmonary edema, respiratory failure, and cardiogenic shock.
May aid AI differentiation of acute cardiac states from pulse signals; leaves open prospective validation before clinical use.
To rapidly obtain the complete characterization information of pulse signals and to verify the sensitivity and validity of pulse signals in the clinical diagnosis of related diseases. In this paper, an improved PNCC method is proposed as a supplementary feature to enable the complete characterization of pulse signals. In this paper, the wavelet scattering method is used to extract time-domain features from impulse signals, and EEMD-based improved PNCC (EPNCC) is used to extract frequency-domain features. The time-frequency features are mixed into a convolutional neural network for final classification and recognition. The data for this study were obtained from the MIT-BIH-mimic database, which was used to verify the effectiveness of the proposed method. The experimental analysis of three types of clinical symptom pulse signals showed an accuracy of 98.3% for pulse classification and recognition. The method is effective in complete pulse characterization and improves pulse classification accuracy under the processing of the three clinical pulse signals used in the paper.
No takes yet. Share an insight, caveat, or question.
Chen et al. (2022) studied CHF pulmonary edema, respiratory failure, and cardiogenic shock (n=39). EPNCC and wavelet scattering feature fusion with Convolutional Neural Network vs. Single-domain features and other feature extraction methods was evaluated on Classification accuracy. A novel pulse signal classification method using EPNCC and wavelet scattering feature fusion with a convolutional neural network achieved 98.3% accuracy in identifying three clinical conditions.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: