CN-121033946-B - Digital psychological behavior twitch monitoring method and system

CN121033946BCN 121033946 BCN121033946 BCN 121033946BCN-121033946-B

Abstract

The invention provides a digital psychological behavior twitch monitoring method and system, which relate to the technical field of digital psychological behavior twitch monitoring, and are characterized in that a twitch monitoring coordinate system is established, an initial static node is determined, node dynamic behavior track information of the initial static node is acquired, origin displacement information is deleted, displacement behavior track information is acquired, differential anomaly analysis and soft clustering anomaly analysis are carried out on the initial static node through an AI chip, anomaly confidence analysis is further carried out, corresponding anomaly nodes are acquired according to the acquired differential confidence analysis result and soft clustering confidence analysis result, twitch confidence judgment is carried out, twitch confidence judgment information is acquired, and digital anomaly analysis on human psychological behavior twitch can be realized through the AI chip, so that analysis efficiency and accuracy are greatly improved.

Inventors

NI XIN
ZHANG GUOJUN
CUI YONGHUA
LI YING
ZHAO QING
SHI LINYU

Assignees

首都医科大学附属北京儿童医院

Dates

Publication Date: 20260508
Application Date: 20251028

Claims (7)

1. A method for monitoring digital psychology and behavioral twitches, the method comprising: s1, acquiring multi-frame twitch monitoring images according to remote video information through an AI chip, establishing a twitch monitoring coordinate system, determining an initial static node, and acquiring behavior acquisition information; S2, acquiring node dynamic behavior track information of an initial static node according to behavior acquisition information, deleting origin displacement information, and acquiring displacement behavior track information; S3, performing differential anomaly analysis and soft clustering anomaly analysis on the initial static node through an AI chip, further performing anomaly confidence analysis, acquiring a corresponding anomaly node according to the acquired differential confidence analysis result and the soft clustering confidence analysis result, performing twitch confidence judgment, and acquiring twitch confidence judgment information; wherein, the S2 includes: Acquiring displacement change information of each initial static node according to the behavior acquisition information, and further acquiring node dynamic behavior track information of the initial static node; acquiring synchronous displacement information of node dynamic behavior track information of a plurality of initial static nodes by using an AI chip synchronicity analysis method; Acquiring origin displacement information of a geometric center point; Judging whether the synchronous displacement information corresponds to the original point displacement information or not, and obtaining position corresponding judgment information; Deleting the synchronous displacement information in the node dynamic behavior track information according to the position correspondence judgment information to obtain displacement behavior track information of the initial static node; Wherein, the S3 includes: Acquiring preset behavior track information of an initial static node; Comparing the preset behavior track information with the displacement behavior track information to obtain difference information of the preset behavior track information and the displacement behavior track information; Comparing the difference information with a preset difference threshold value to obtain a behavior difference comparison result; Determining a difference static node according to the behavior difference comparison result; Calculating the difference confidence coefficient of the difference static node according to the comparison information through the AI chip; Performing difference confidence analysis according to the difference confidence to obtain a difference confidence analysis result; wherein, the S3 further comprises: setting various soft clustering condition information; Performing iterative soft cluster analysis on displacement behavior track information of a plurality of initial static nodes for a plurality of times according to the plurality of soft cluster condition information until the change of the soft cluster information is smaller than a preset change threshold value, and obtaining physiological nodes corresponding to physiological displacement information and pathological nodes corresponding to pathological displacement information; calculating the physiological confidence coefficient of the physiological node according to the physiological displacement information through the AI chip; calculating the pathology confidence coefficient of the pathology node according to the pathology displacement information through the AI chip; and carrying out soft clustering confidence analysis according to the physiological confidence coefficient and the pathological confidence coefficient through an AI chip to obtain a soft clustering confidence coefficient analysis result.
2. The method for monitoring digital psychomotor-behavioral tic pumping according to claim 1, wherein S1 comprises: loading remote video information through a video stream analysis module of an AI chip, and acquiring multi-frame pumping monitoring images according to the remote video information; Acquiring an initial twitch monitoring image of a plurality of frames of twitch monitoring images; Human body model information extraction is carried out on the initial twitch monitoring image based on a human body key point detection model of the AI chip, and initial human body model information is obtained; establishing a rectangular coordinate system by taking a geometric center point of initial human body model information as an origin through a coordinate operation unit of the AI chip to obtain a pumping monitoring coordinate system; According to the initial human body model information, enabling the AI chip to perform initial static point position marking on key points in the pumping monitoring coordinate system through a node marking method to obtain initial static nodes; and acquiring displacement change information of the initial static node in the multi-frame pumping monitoring image through a real-time data acquisition module of the AI chip to obtain behavior acquisition information.
3. The method for monitoring digital psychology and behavioral tic pumping according to claim 1, wherein performing a difference confidence analysis based on the difference confidence comprises: comparing the difference confidence coefficient with a preset difference confidence threshold; When the difference confidence coefficient is larger than a preset difference confidence threshold value, judging that the difference confidence coefficient is qualified; and when the difference confidence coefficient is smaller than or equal to a preset difference confidence threshold value, judging that the difference confidence coefficient is unqualified.
4. The method for monitoring digital psychological behavior twitch according to claim 1, wherein the soft clustering confidence analysis is performed by an AI chip according to the physiological confidence and the pathological confidence, and the soft clustering confidence analysis result is obtained, comprising: comparing the physiological confidence coefficient with a preset physiological confidence coefficient threshold value, and judging that the physiological confidence coefficient is qualified when the physiological confidence coefficient is larger than the preset physiological confidence coefficient threshold value; Comparing the pathology confidence coefficient with a preset pathology confidence coefficient threshold, and judging that the pathology confidence coefficient is qualified when the pathology confidence coefficient is larger than the preset pathology confidence coefficient threshold; And when the physiological confidence is qualified and the pathological confidence is qualified, judging that the soft clustering confidence is qualified.
5. The method for monitoring digital psychomotor-behavioral pumping according to claim 1, wherein S3 further comprises: When the difference confidence is qualified and the soft cluster confidence is qualified, obtaining a difference static node and a pathological node; corresponding the difference static node and the pathological node to obtain a corresponding abnormal node; determining abnormal twitch position information according to the corresponding abnormal node; and carrying out twitch confidence judgment according to the corresponding abnormal node data to obtain twitch confidence judgment information.
6. The method for monitoring the digital psychology and behavior twitch according to claim 5, wherein the step of determining the twitch confidence according to the corresponding abnormal node data to obtain twitch confidence determination information comprises the steps of: Comparing the number of the abnormal node data with a preset abnormal node number threshold value; When the number of the abnormal node data is larger than a preset abnormal node number threshold value, judging that the pumping confidence is qualified; and when the number of the abnormal node data is smaller than or equal to a preset abnormal node number threshold value, judging that the pumping confidence coefficient is unqualified.
7. A digital psychology behavioral tic pumping monitoring system, the system comprising: The node determining module is used for acquiring multi-frame twitch monitoring images according to remote video information through the AI chip, establishing a twitch monitoring coordinate system, determining an initial static node and acquiring behavior acquisition information; The interference deleting module is used for acquiring node dynamic behavior track information of the initial static node according to the behavior acquisition information, deleting origin displacement information and acquiring displacement behavior track information; The anomaly analysis module is used for carrying out differential anomaly analysis and soft clustering anomaly analysis on the initial static node through the AI chip, further carrying out anomaly confidence analysis, acquiring a corresponding anomaly node according to the acquired differential confidence analysis result and the soft clustering confidence analysis result, carrying out twitch confidence judgment, and acquiring twitch confidence judgment information; wherein, the interference deletion module includes: Acquiring displacement change information of each initial static node according to the behavior acquisition information, and further acquiring node dynamic behavior track information of the initial static node; acquiring synchronous displacement information of node dynamic behavior track information of a plurality of initial static nodes by using an AI chip synchronicity analysis method; Acquiring origin displacement information of a geometric center point; Judging whether the synchronous displacement information corresponds to the original point displacement information or not, and obtaining position corresponding judgment information; Deleting the synchronous displacement information in the node dynamic behavior track information according to the position correspondence judgment information to obtain displacement behavior track information of the initial static node; Wherein, the anomaly analysis module includes: Acquiring preset behavior track information of an initial static node; Comparing the preset behavior track information with the displacement behavior track information to obtain difference information of the preset behavior track information and the displacement behavior track information; Comparing the difference information with a preset difference threshold value to obtain a behavior difference comparison result; Determining a difference static node according to the behavior difference comparison result; Calculating the difference confidence coefficient of the difference static node according to the comparison information through the AI chip; Performing difference confidence analysis according to the difference confidence to obtain a difference confidence analysis result; Wherein, the anomaly analysis module further comprises: setting various soft clustering condition information; Performing iterative soft cluster analysis on displacement behavior track information of a plurality of initial static nodes for a plurality of times according to the plurality of soft cluster condition information until the change of the soft cluster information is smaller than a preset change threshold value, and obtaining physiological nodes corresponding to physiological displacement information and pathological nodes corresponding to pathological displacement information; calculating the physiological confidence coefficient of the physiological node according to the physiological displacement information through the AI chip; calculating the pathology confidence coefficient of the pathology node according to the pathology displacement information through the AI chip; and carrying out soft clustering confidence analysis according to the physiological confidence coefficient and the pathological confidence coefficient through an AI chip to obtain a soft clustering confidence coefficient analysis result.

Description

Digital psychological behavior twitch monitoring method and system Technical Field The invention provides a digital psychological behavior twitch monitoring method and system, relates to the technical field of twitch monitoring, and particularly relates to the technical field of digital psychological behavior twitch monitoring. Background Tic disorder is a common neurological disorder disease, traditional tic monitoring relies on direct offline observation of clinicians or self-description of patients, and has the problems of strong subjectivity, limited monitoring time and place, difficulty in capturing sporadic tic behaviors and the like, and remote and long-term dynamic monitoring cannot be realized through an AI chip. The prior art focuses on single dimension feature extraction, is easily influenced by interference factors, and has low twitch recognition accuracy and high misdiagnosis rate. Meanwhile, a multidimensional abnormality detection and confidence quantification mechanism is lacked, and mild twitch and accidental actions are difficult to distinguish. Disclosure of Invention The invention provides a digital psychological behavior twitch monitoring method and a digital psychological behavior twitch monitoring system, which are used for solving the problems: the invention provides a digital psychological behavior twitch monitoring method and a digital psychological behavior twitch monitoring system, wherein the method comprises the following steps: s1, acquiring multi-frame twitch monitoring images according to remote video information through an AI chip, establishing a twitch monitoring coordinate system, determining an initial static node, and acquiring behavior acquisition information; S2, acquiring node dynamic behavior track information of an initial static node according to behavior acquisition information, deleting origin displacement information, and acquiring displacement behavior track information; s3, performing differential anomaly analysis and soft clustering anomaly analysis on the initial static node through the AI chip, further performing anomaly confidence analysis, acquiring a corresponding anomaly node according to the acquired differential confidence analysis result and the soft clustering confidence analysis result, performing twitch confidence judgment, and acquiring twitch confidence judgment information. Further, the step S1 includes: loading remote video information through a video stream analysis module of an AI chip, and acquiring multi-frame pumping monitoring images according to the remote video information; Acquiring an initial twitch monitoring image of a plurality of frames of twitch monitoring images; Human body model information extraction is carried out on the initial twitch monitoring image based on a human body key point detection model of the AI chip, and initial human body model information is obtained; establishing a rectangular coordinate system by taking a geometric center point of initial human body model information as an origin through a coordinate operation unit of the AI chip to obtain a pumping monitoring coordinate system; According to the initial human body model information, enabling the AI chip to perform initial static point position marking on key points in the pumping monitoring coordinate system through a node marking method to obtain initial static nodes; and acquiring displacement change information of the initial static node in the multi-frame pumping monitoring image through a real-time data acquisition module of the AI chip to obtain behavior acquisition information. Further, the step S2 includes: Acquiring displacement change information of each initial static node according to the behavior acquisition information, and further acquiring node dynamic behavior track information of the initial static node; acquiring synchronous displacement information of node dynamic behavior track information of a plurality of initial static nodes by using an AI chip synchronicity analysis method; Acquiring origin displacement information of a geometric center point; Judging whether the synchronous displacement information corresponds to the original point displacement information or not, and obtaining position corresponding judgment information; and deleting the synchronous displacement information in the node dynamic behavior track information according to the displacement corresponding judgment information to obtain the displacement behavior track information of the initial static node. Further, the step S3 includes: Acquiring preset behavior track information of an initial static node; Comparing the preset behavior track information with the displacement behavior track information to obtain difference information of the preset behavior track information and the displacement behavior track information; Comparing the difference information with a preset difference threshold value to obtain a behavior difference comparison result; Determining a differen