A Human Defecation Prediction Method Based on Multi-Domain Features and Improved Support Vector Machine

The difficulty of defecation seriously affects the quality life bedridden elderly. To solve problem that it is difficult to know time elderly, this paper proposed a human pre-defecation prediction method based on multi-domain features and improved support vector machine (SVM) using bowel sound as original signal. includes three stages: extraction, feature optimization, prediction. In stage statistical analysis, fast Fourier transform (FFT), wavelet packet are used extract information in domain, frequency time-frequency domain. symmetry signal domain will change when has urge defecate. optimization stage, Fisher Score (FS) algorithm introduced select meaningful sensitive according importance each feature, aiming remove redundant improve computational efficiency. prediction, SVM optimized by gray wolf (GWO) realize Finally, experimental analysis data collected during study carried out. result shows could achieve an accuracy 92.86% which proves effectiveness method.