Pain information processing method and device, equipment and storage medium

Abstract

The invention discloses a pain information processing method and device, equipment and a storage medium. The pain information processing method includes the steps: obtaining pain expression video stream of a target user; obtaining an expression image frame corresponding to each sampling moment according to the video stream; extracting expression feature vectors corresponding to the expression image frames corresponding to the sampling moments; grouping the expression feature vectors corresponding to the sampling moments; obtaining a plurality of expression feature vector groups; and obtaininga pain level corresponding to the target user based on the plurality of expression feature vector groups, thereby fully fusing the influence of the time dimension on the pain expression into the expression features, and taking the time dimension as a calculation dimension for judging the pain degree, thereby effectively improving the accuracy of analyzing the pain degree of the user.

Description

technical field [0001] The present application relates to the field of pain information processing, in particular to a pain information processing method, device, equipment and storage medium. Background technique [0002] Pain assessment is a necessary diagnostic method for clinical acute pain, chronic pain and refractory pain. The clinical environment can understand the patient's pain situation through the patient's chief complaint, but the ratio of doctors to patients is seriously unbalanced. For example, the annual operation volume in China alone in a certain year reaches 60 million units, while there are less than 100,000 anesthesiologists in China, resulting in a large number of patients whose pain cannot be treated in time. [0003] There is a strong correlation between facial expression and pain. The existing technology is based on the pain level analysis of a single facial expression image. For example, a single facial expression image is used as input, and a convol...

AI Technical Summary

Problems solved by technology

[0002] Pain assessment is a necessary diagnostic method for clinical acute pain, chronic pain and intractable pain. The clinical environment can understand the patient's pain situation through the patient's chief complaint, but the ratio of doctors to patients is seriously unbalanced. For example, the annual operation volume in China alone in a certain year reaches 60 million units, while there are only less than 100,000 anesthesiologists in China, resulting in a large number of patients whose pain cannot be treated in time

[0003] There is a strong correlation between facial expression and pain. The existing technology is based on the pain level analysis of a single facial expression image. For example, a single facial expression image is used as input, and a convolutional neural network is used to extract the pain characteristics of a single image, and finally through a fully connected layer. Train to get the pain level corresponding to the image, such as Figure 1A As shown, and because the patient's facial expression changes rapidly, it is difficult to capture the real expression. Therefore, the accuracy of analysis based on a single expression image is low, which is not conducive to the treatment of patients.

Images