Taking into account the current feature extraction speed and recognition effect of intelligent diagnosis of menopausal women's health care behavior, this paper proposes to use a cross-layer convolutional neural network to extract behavior features autonomously and use support vector machine multiclass behavior classifier to classify behavior. Compared with the feature images extracted by traditional methods, the behavioral features extracted in this paper are related to the individual menopausal women and have better semantic information, and the feature description ability in the time domain and the space domain has been enhanced. Through Matlab software, using the database established in this paper to compare its feature extraction time, test classification time, and final recognition accuracy with ordinary convolutional neural networks, it is concluded that the cross-layer CNN-SVM model can ensure the speed of feature extraction. It proves that the method in this paper can be applied to the behavioral intelligent diagnosis system for intelligently nursing menopausal women and has good practical value. This paper designs a home care bed intelligent monitoring system, which can automatically detect the posture of the care bed, and not only can change the posture of the bed under the control of personnel, but also can automatically complete the posture conversion according to the setting. At the same time, the system has the function of monitoring the physical condition of the person being cared for and can detect the heart rate, blood oxygen, and other physiological indicators of the bedridden person. In addition, the system can also provide a remote diagnosis function, allowing nursing staff to remotely view the current state of the nursing bed and the physical condition of the person. After testing, the system works stably, improves the automation and safety of the nursing bed control, and enriches the functions of the nursing bed.Copyright © 2021 Qing Chao et al.