Animal disease prediction algorithm

Abstract

The invention discloses an animal disease prediction algorithm, and relates to the technical field of animal disease identification. The algorithm is composed of a two-layer neural network structure,a sub-layer neural network is composed of a convolutional neural network or a recurrent neural network and is responsible for sound recognition, and a main layer is composed of a recurrent neural network and is responsible for disease prediction according to physical sign parameters such as heart rate, blood oxygen, body temperature and the sound recognition result of the sub-layer. Sound and conventional sign data are separately processed, the data volume of data processing is reduced while data processing details are kept, the difficulty of data training of the two network structures is reduced, and a technical means is provided for predicting animal diseases through heart rate, blood oxygen, body temperature and sound.

Description

technical field [0001] The invention relates to the technical field of animal disease prediction, in particular to an animal disease prediction algorithm. Background technique [0002] Signs of animals, such as body temperature, blood oxygen, heart rate, movement, cry, etc., are closely related to animal diseases, reproduction and production efficiency, and are of great significance to the study of improving animal growth efficiency and preventing diseases. For example, through changes in body temperature and heart rate, it is possible to analyze whether there are problems with the feed and growth environment of pigs during the breeding process; through changes in body temperature, blood oxygen, and heart rate, it is possible to predict diseases of pigs. At present, conventional physiological sign detection mainly adopts contact methods, such as measuring body temperature with a thermometer and measuring blood oxygen with an oximeter. The efficiency is low, and long-term un...

AI Technical Summary

Problems solved by technology

[0006] (1) Compared with the data of body temperature, heart rate, and blood oxygen, sound has a larger data volume group. If the four types of data are put into a neural network structure, the calculation amount of the algorithm is very large, and due to the huge amount of sound data, it is easy to The overfitting phenomenon of the neural network is caused, and the training is very difficult;

[0007] (2) Unified processing of body temperature, heart rate, blood oxygen, and sound as a data vector is extremely difficult, and there is currently a lack of relevant experimental results to support;

[0008] (3) Because body temperature, heart rate, blood oxygen, and sound data are mainly collected by low-quality sensors, the data error fluctuates greatly, and the data is not perfect, etc., so the algorithm needs to have strong robustness

Images