基于神经网络模式识别的人体大便便意识别模型研究

Study on the Identification Model of Human Defecation Intention Based on Neural Network Pattern Recognition

针对医院患病老人长期卧床、缺乏自理能力,尤其是大便护理难以解决的问题,提出了基于神经网络的大便便前监测模型。根据人体便前90 s生理参数会发生明显变化的特点,通过采集人体有便意和无便意状态下的生理参数制定数据集,构建大便便意识别模型并使用数据集对其进行训练,最终通过训练后的模型实现对卧床失能老人的大便便意识别和便前监测。临床实验结果表明:该预测模型进行便意识别和便前监测的准确率在80%~87%之间,具有较高精度和实用价值。

The article proposes a neural network-based fecal pre-monitoring model for elderly patients who are bedridden and lack self-care ability,especially for those with difficult fecal care in hospitals.According to the characteristic that physiological parameters of the human body change significantly before defecation within 90 seconds,a dataset was developed by collecting physiological parameters in states of defecation urge and non-defecation urge.Then,a fecal urge recognition model was constructed and trained using the dataset,and ultimately,the trained model was used to achieve fecal urge recognition and pre-monitoring for bedridden elderly patients.Clinical experimental results show that the accuracy of the predictive model for fecal urge recognitionnd pre-monitoring is between 80%~87%,which has high accuracy and practical value.