Animal state rapid discrimination method and system based on hierarchical behavior information

Abstract

The invention discloses an animal state rapid discrimination method and system based on hierarchical behavior information, and is applied to the technical field of animal behavior analysis, and the method comprises the steps: obtaining an animal spontaneous behavior; performing behavior analysis on the animal spontaneous behavior, and decomposing the animal spontaneous behavior into an action module; and calculating the action conversion frequency of the animal in the preset time and the behavior ratio of the sniffing action and the carding action, and judging the animal behavior state based on the data of the original database. According to the method, the animal does not need to be trained in advance, and the behavior state of the animal can be identified only by recording the spontaneous behavior of the animal and carrying out fine action analysis and calculation.

Description

Technical Field

[0001] The present invention relates to the technical field of animal behavior analysis, and in particular to a method and system for quickly distinguishing animal states based on hierarchical behavior information. Background Art

[0002] Spontaneous behavior is a behavior that occurs naturally in animals without any external stimulation, clues, or task guidance. It includes action modules such as spontaneous movement, spontaneous exploration, and grooming, and represents the behavioral habits that a species has solidified in the long-term evolution process. When animals enter an unfamiliar open field environment, they will instinctively explore the surrounding environment and exhibit specific behavioral patterns. This behavioral paradigm, which allows animals to lead rather than experimenters, is simple and practical, and can more objectively reflect the internal state of animals. Changes in spontaneous behavior can indicate deviations in the internal state of an individual. These deviations may be related to potential health problems, diseases, and acute stress. Therefore, spontaneous behavior tests are widely used to evaluate animal movement status, anxiety-like behavior levels, etc.

[0003] However, due to the limitation of technical means, the interpretation of animal behavior status is inevitably biased, mainly based on motion trajectory analysis and position preference, lacking analysis of relevant information about posture and action. In the past decade, with the development of artificial intelligence technology, algorithms based on computer vision and machine learning can achieve multi-point synchronous tracking, posture estimation, three-dimensional skeleton reconstruction, and action segmentation and recognition of animal body parts in videos, greatly improving the accuracy of animal behavior analysis. The general method involves extracting the time series of animal posture changes from the data set to identify specific actions. The development of machine learning algorithms in quantitative posture estimation and behavior segmentation has promoted the process of decomposing continuous behaviors into quantifiable motion modules. Advanced tools such as DeepLabCut, Moseq, DANNCE, MotionMapper, LEAP, B-SOiD, and BehaviorAtlas Analyzer enable researchers to objectively quantify animal behavior. These tools can dissect the microstructure of behavior and have far-reaching impacts on fields such as behavioral science, neuroscience, ethics, and pharmacology. However, these emerging behavioral analysis methods currently lack a mature and sophisticated behavioral quantitative analysis framework, and computational methods for extracting core behavioral parameters from massive posture data to achieve high-throughput rapid screening still need to be proposed and improved.

[0004] The spontaneous behavior of rodents includes action modules such as movement, self-grooming and exploration. Grooming is an instinctive behavior shared by many species and is a particularly important part of rodent behavior. Rodent grooming activities usually increase under high stress and relaxation conditions. Sniffing is an innate behavior observed in rodents during exploration, which helps rodents understand their surroundings, find resources, communicate socially and assess potential threats. Specific behavior is a continuous process, and its changes can be identified by changes and repetitions captured by the reproduction of dynamic nervous systems. In order to more accurately judge the behavioral state of animals, it is necessary to use advanced artificial intelligence technology to conduct a refined analysis of animal behavior from the aspects of posture, movement, and behavior patterns; on the other hand, screening out meaningful behavioral parameters from massive high-dimensional data requires a lot of computing resources and time and labor costs. In order to overcome these shortcomings, this application proposes a method and system for rapid animal state discrimination based on hierarchical behavioral information based on the components and basic organizational principles of spontaneous behavior. Summary of the invention

[0005] The purpose of this application is to provide a method and system for quickly distinguishing animal status based on hierarchical behavioral information, aiming to solve the above-mentioned problems.

[0006] To achieve the above objectives, this application provides the following technical solutions:

[0007] The present application provides a method for quickly distinguishing animal states based on hierarchical behavioral information, including:

[0008] Obtaining spontaneous behavior of animals;

[0009] Performing behavioral analysis on the spontaneous behavior of the animal, and decomposing the spontaneous behavior of the animal into action modules;

[0010] The frequency of animal action transitions within a preset time and the proportion of sniffing and grooming actions are calculated, and the animal's behavioral state is determined based on the data in the original database.

[0011] Furthermore, in the step of performing behavioral analysis on the spontaneous behavior of the animal and decomposing the spontaneous behavior of the animal into action modules, the following steps are specifically included:

[0012] Behavioral analysis software was used to analyze the spontaneous behavioral data of animals;

[0013] Identify body parts of the animal, including but not limited to: nose, left ear, right ear, neck, left forelimb, right forelimb, left hindlimb, right hindlimb, left front paw, right front paw, left hind paw, right hind paw, back, base of tail, middle of tail, and tip of tail;

[0014] Constructing a key point three-dimensional skeleton of the animal according to the acquired calibration information and the body position information of the animal; and normalizing, scaling and aligning the key point three-dimensional skeleton;

[0015] The similarity and speed value of the animal's posture dynamics within a preset time are calculated, and the animal's spontaneous behavior is decomposed into a plurality of action modules based on the similarity and the speed value.

[0016] Furthermore, after the step of performing behavioral analysis on the spontaneous behavior of the animal and decomposing the spontaneous behavior of the animal into action modules, the method specifically includes the following steps:

[0017] Through tree-like hierarchical clustering, several action modules obtained by decomposition are clustered;

[0018] The action modules are assigned corresponding digital labels, and the digital labels are used to annotate the actions.

[0019] Furthermore, the step of calculating the animal's action switching frequency within a preset time specifically includes:

[0020] Determine the first behavior state of the animal by comparing the action conversion frequency of the animal within a preset time with the original database;

[0021] When the action forwarding frequency is higher than a preset threshold, the animal is judged to be in an active state, otherwise the animal is judged to be in an inactive state.

[0022] Furthermore, in the step of calculating the animal's action conversion frequency and the proportion of sniffing actions and grooming actions within a preset time, and judging the animal's behavior state based on the data in the original database, the following steps are specifically included:

[0023] Calculate the behavioral ratio of the animal's sniffing action and grooming action within a preset time and a preset area, including the ratio of time proportion and frequency of occurrence;

[0024] By calculating the behavioral ratio of sniffing actions to grooming actions and comparing them with the emotional state in the original database and the data in the pre-built disease model, the animal's final behavioral state can be determined.

[0025] The present application provides a system for quickly distinguishing animal states based on hierarchical behavior information, comprising:

[0026] Acquisition module: obtain the spontaneous behavior of animals;

[0027] Analysis module: performing behavioral analysis on the spontaneous behavior of the animal and decomposing the spontaneous behavior of the animal into action modules;

[0028] Calculation and judgment module: Calculates the animal's action conversion frequency and the proportion of sniffing and grooming actions within a preset time, and judges the animal's behavioral status based on the data in the original database.

[0029] The present application provides a device, which includes a processor and a memory coupled to the processor, wherein the memory stores program instructions for implementing a method for quickly distinguishing the state of an animal based on hierarchical behavioral information; the processor is used to execute the program instructions stored in the memory to achieve rapid distinction of the state of an animal based on hierarchical behavioral information.

[0030] The present application provides a storage medium storing program instructions executable by a processor, wherein the program instructions are used to execute a method for quickly distinguishing an animal state based on hierarchical behavior information.

[0031] The present application provides a method and system for quickly distinguishing animal states based on hierarchical behavior information, which has the following beneficial effects:

[0032] (1) This application records and analyzes the spontaneous behavior of animals, decomposing the behavior into quantifiable and comparable action modules, thereby achieving fine quantification of animal behavior; more accurately reflecting the true state of the animals and improving the accuracy of behavioral analysis;

[0033] (2) By calculating the ratio of action conversion frequency to sniffing action and grooming action and comparing it with the original database, the animal's behavioral state can be quickly determined; this not only simplifies the analysis process but also improves the judgment efficiency;

[0034] (3) During drug screening and development, this application can quickly evaluate the effects of drugs on animal behavior, thereby accelerating the development of new drugs and providing new means for the treatment of mental illness and mood disorders. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 This is a flow chart of a method for quickly distinguishing animal states based on hierarchical behavior information according to Example 1 of the present application;

[0036] Figure 2 This is a schematic diagram of obtaining and analyzing the spontaneous behavior of animals in Example 1 of the present application;

[0037] Figure 3 This is a schematic diagram of the change in the frequency of animal movement conversion in Example 1 of the present application;

[0038] Figure 4 This is a schematic diagram of the time ratio and the frequency ratio of the sniffing action and the grooming action in Example 1 of the present application;

[0039] Figure 5This is a schematic diagram of predicting the behavior state of an animal based on the ratio of sniffing action to grooming action according to Example 1 of the present application;

[0040] Figure 6 This is a schematic diagram of the structure of a system for quickly distinguishing animal states based on hierarchical behavior information according to Example 2 of the present application;

[0041] Figure 7 This is a schematic diagram of the device structure of Example 3 of the present application;

[0042] Figure 8 This is a schematic diagram of the storage medium structure of Example 4 of the present application. DETAILED DESCRIPTION

[0043] It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0044] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0045] Example 1

[0046] See also Figure 1 , is a flow chart of a method for quickly distinguishing animal states based on hierarchical behavior information according to Example 1 of the present application; the steps include:

[0047] S1: Obtain the spontaneous behavior of animals.

[0048] In this embodiment, the spontaneous behavior of the animal is acquired from multiple angles for subsequent reconstruction of its three-dimensional skeleton.

[0049] S2: Performing behavioral analysis on the spontaneous behavior of the animal, and decomposing the spontaneous behavior of the animal into action modules.

[0050] In this embodiment, the behavior analysis software is used to analyze the spontaneous behavior data of the animals. The body parts of the animals are identified, including but not limited to: nose, left ear, right ear, neck, left forelimb, right forelimb, left hindlimb, right hindlimb, left front paw, right front paw, left hind paw, right hind paw, back, tail base, tail middle and tail tip.

[0051] According to the obtained calibration information and the animal's body position information, the key point three-dimensional skeleton of the animal is constructed; and the key point three-dimensional skeleton is normalized, scaled and aligned. The similarity and speed value of the animal's posture dynamics within a preset time are calculated, and the animal's spontaneous behavior is decomposed into several action modules based on the similarity and the speed value.

[0052] Finally, several action modules obtained by decomposition are clustered through tree-like hierarchical clustering; corresponding digital labels are assigned to the action modules, and the digital labels are used to annotate the actions.

[0053] See also Figure 2 , is a schematic diagram of obtaining and analyzing the spontaneous behavior of animals in Example 1 of the present application; Figure 2 In the figure, a is the multi-angle animal posture acquisition device, and b is the detailed behavior analysis process and output results.

[0054] Taking BehaviorAtlas animal behavior analysis software as an example, the process is as follows:

[0055] First, extract a number of frames, such as 2000 frames, of random field of view images from four cameras, and manually mark 16 body parts of the animal as a training set. Use Deeplabcut academic open source code and ResNet-101 deep convolutional neural network to train the manually marked data set, and then use the trained model for subsequent animal body part tracking, where the loss function is less than 0.0005. Deeplabcut tracks the 16 body parts of the animal corresponding to each frame of the four cameras based on the trained neural network model.

[0056] Before the experiment, the positions of the four cameras were calibrated using the Stereo Camera Calibrator GUI module of Matlab. Specifically, the four cameras were used to shoot the black and white chessboard with a certain regular tilt and rotation at the center; the boundary positions of the black and white chessboard detected by two cameras were calculated, the error was controlled within 0.5 pixels, the positions of the two cameras were determined one by one, and finally the position of the main camera was determined.

[0057] According to the above camera calibration information, the three-dimensional space captured by the four cameras is constructed, and then each frame of the image, that is, the two-dimensional animal body position, is reconstructed into a three-dimensional skeleton. The body orientation is unified and normalized and scaled, and the back point is fixed to align the animal skeleton. The normalized skeleton and the three-dimensional coordinates of 16 coordinates are then used to form a 48-dimensional time series, which is reduced to 2 dimensions using umap. Based on the dynamic time kernel alignment algorithm (DATK), similar time series are cut, and several short action clips are obtained, which are action modules.

[0058] Taking MOSeq motion sequence analysis as an example, the workflow is as follows:

[0059] First, the animals' spontaneous behavior in an open field was recorded at 30 Hz using a Microsoft Kinect depth sensor.

[0060] Based on the depth information of the image, the body outline of the animal and the depth information of the outline are extracted through image processing and computer vision methods. The image of the animal's location is cropped, and the center of the animal outline in each frame is aligned and the body direction is unified. Motion features such as speed, acceleration, direction change, etc. are extracted from each frame of data. These features will be used as input to the model to help identify the animal's behavior pattern.

[0061] Subsequently, these features were modeled using a probabilistic autoregressive hidden Markov model (AR-HMM) to decompose the animal's behavior into multiple short-term motion modules. Similar to the action modules automatically identified in Behavior Atlas, AR-HMM automatically identifies the boundaries between different motion modules and calculates the temporal data of these action modules and the frequency of their mutual transfer. Based on the above module features, each action module was clustered unsupervisedly and action labels were obtained. Finally, experienced experimenters proofread the video and assigned biologically meaningful action annotations to each action module label.

[0062] S3: Calculate the animal's action switching frequency and the proportion of sniffing and grooming actions within a preset time, and determine the animal's behavioral status based on the data in the original database.

[0063] In this embodiment, the first behavior state of the animal is determined by comparing the action conversion frequency of the animal within a preset time with the original database. When the action forwarding frequency is higher than the preset threshold, the animal is determined to be in an active state, otherwise the animal is determined to be in an inactive state. The original database in this embodiment is an animal behavior phenotype database constructed by the long-term animal behavior research institute of the classroom group.

[0064] Calculate the behavioral ratio of the animal's sniffing and grooming actions within a preset time and area, including the ratio of time proportion and frequency of occurrence. When the animal enters the open field, it will spontaneously sniff and groom, and the duration, frequency, and sequence logic of the occurrence are stable within a certain range. However, when the animal is in an abnormal emotional state or in a certain disease state, this balance will be broken. By calculating the behavioral ratio of sniffing and grooming actions, and comparing it with the emotional state in the original database and the data in the pre-built disease model, the final behavioral state of the animal can be quickly and accurately determined.

[0065] See also Figure 5, which is a schematic diagram of predicting the behavioral state of an animal based on the ratio of sniffing action and grooming action in Example 1 of the present application. This embodiment can more accurately identify sniffing and grooming actions through the sophisticated behavior analysis system BehaviorAtlas. On this basis, the spontaneous behavior of animals in a square open field for one hour was collected under the conditions of turning on the lights in the morning, turning on the lights in the afternoon, turning on the lights in the evening, turning off the lights at night, and experiencing acute restraint stress. It was found that the ratio of its sniffing action and grooming action, both in duration and number of occurrences, was significantly smaller than that of the other animals that did not experience stress stimulation. For animals that did not receive stress stimulation, although the ratio fluctuated under different environmental conditions, it was generally stable and maintained within a certain range. The above experimental results show that the stability proposed in this application is high and has good generalization.

[0066] In summary, Example 1 of the present application applies a hierarchical and refined behavior analysis framework to calculate the core data of spontaneous behavior characteristics from high-dimensional posture data, and establishes a cross-species, humanized, and genetically diverse model animal behavior phenotype database; by analyzing the important parameters in "posture-action-action set-behavior logic", high-throughput, rapid, and accurate judgment of animal behavior status is achieved, thereby serving scientific research and industry needs such as model animal identification and pharmacodynamic evaluation.

[0067] Example 2

[0068] See also Figure 6 , is a schematic diagram of the structure of the animal state rapid identification system based on hierarchical behavior information of Example 2 of the present application; the specific contents include:

[0069] Acquisition module: obtain the spontaneous behavior of animals;

[0070] Analysis module: performing behavioral analysis on the spontaneous behavior of the animal and decomposing the spontaneous behavior of the animal into action modules;

[0071] Calculation and judgment module: Calculates the animal's action conversion frequency and the proportion of sniffing and grooming actions within a preset time, and judges the animal's behavioral status based on the data in the original database.

[0072] In this embodiment, an experimental simulation is conducted on mice, and the following table data is the action conversion frequency of mice under different experimental conditions.

[0073]

[0074] Example 3

[0075] See also Figure 7 , is a schematic diagram of the device structure of Embodiment 3 of the present application. The device 50 includes a processor 51 and a memory 52 coupled to the processor 51 .

[0076] The memory 52 stores program instructions for implementing the above-mentioned method for quickly distinguishing the state of an animal based on hierarchical behavior information.

[0077] The processor 51 is used to execute program instructions stored in the memory 52 to achieve rapid determination of the animal state based on hierarchical behavior information.

[0078] The processor 51 may also be referred to as a CPU (Central Processing Unit).

[0079] The processor 51 may be an integrated circuit chip with signal processing capabilities. The processor 51 may also be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0080] Example 4

[0081] See also Figure 8 , which is a schematic diagram of the structure of the storage medium of Example 4 of the present application. The storage medium of the embodiment of the present application stores a program file 61 that can implement all the above methods, wherein the program file 61 can be stored in the above storage medium in the form of a software product, including a number of instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) or a processor (processor) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, or a computer, a server, a mobile phone, a tablet, and other devices.

[0082] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "includes a ..." does not exclude the presence of other identical elements in the process, device, article or method including the element.

[0083] The above description is only a preferred embodiment of the present application, and does not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly used in other related technical fields, are also included in the patent protection scope of the present application.

[0084] Although the embodiments of the present application have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present application, and that the scope of the present application is defined by the appended claims and their equivalents.

[0085] Of course, the present invention may have many other implementations. Based on this implementation, other implementations obtained by ordinary technicians in this field without any creative work are all within the scope of protection of the present invention.

Claims

1. A method for quickly distinguishing animal states based on hierarchical behavioral information, characterized in that: include: Obtaining spontaneous behavior of animals; Performing behavioral analysis on the spontaneous behavior of the animal, and decomposing the spontaneous behavior of the animal into action modules; The frequency of animal action transitions within a preset time and the proportion of sniffing and grooming actions are calculated, and the animal's behavioral state is determined based on the data in the original database.

2. The method for quickly distinguishing animal states based on hierarchical behavior information according to claim 1, characterized in that: The step of performing behavioral analysis on the spontaneous behavior of the animal and decomposing the spontaneous behavior of the animal into action modules specifically includes the following steps: Behavioral analysis software was used to analyze the spontaneous behavioral data of animals; Identify body parts of the animal, including but not limited to: nose, left ear, right ear, neck, left forelimb, right forelimb, left hindlimb, right hindlimb, left front paw, right front paw, left hind paw, right hind paw, back, base of tail, middle of tail, and tip of tail; Constructing a key point three-dimensional skeleton of the animal according to the acquired calibration information and the body position information of the animal; and normalizing, scaling and aligning the key point three-dimensional skeleton; The similarity and speed value of the animal's posture dynamics within a preset time are calculated, and the animal's spontaneous behavior is decomposed into a plurality of action modules based on the similarity and the speed value.

3. The method for quickly distinguishing animal states based on hierarchical behavior information according to claim 1, characterized in that: After the step of performing behavioral analysis on the spontaneous behavior of the animal and decomposing the spontaneous behavior of the animal into action modules, the method specifically includes the following steps: Through tree-like hierarchical clustering, several action modules obtained by decomposition are clustered; The action modules are assigned corresponding digital labels, and the digital labels are used to annotate the actions.

4. The method for quickly distinguishing animal states based on hierarchical behavior information according to claim 1, characterized in that: The step of calculating the animal's action conversion frequency within a preset time specifically includes: Determine the first behavior state of the animal by comparing the action conversion frequency of the animal within a preset time with the original database; When the action forwarding frequency is higher than a preset threshold, the animal is judged to be in an active state, otherwise the animal is judged to be in an inactive state.

5. The method for quickly distinguishing animal states based on hierarchical behavior information according to claim 1, characterized in that: The step of calculating the animal's action conversion frequency and the proportion of sniffing actions and grooming actions within a preset time, and judging the animal's behavior state based on the data in the original database specifically includes the following steps: Calculate the behavioral ratio of the animal's sniffing action and grooming action within a preset time and a preset area, including the ratio of time proportion and frequency of occurrence; By calculating the behavioral ratio of sniffing actions to grooming actions and comparing them with the emotional state in the original database and the data in the pre-built disease model, the animal's final behavioral state can be determined.

6. A system for quickly distinguishing animal states based on hierarchical behavior information according to any one of claims 1 to 5, characterized in that: include: Acquisition module: obtain the spontaneous behavior of animals; Analysis module: performing behavioral analysis on the spontaneous behavior of the animal and decomposing the spontaneous behavior of the animal into action modules; Calculation and judgment module: Calculates the animal's action conversion frequency and the proportion of sniffing and grooming actions within a preset time, and judges the animal's behavioral status based on the data in the original database.

7. A device, characterized in that: The device includes a processor and a memory coupled to the processor, wherein the memory stores program instructions for implementing the method for quickly determining the state of an animal based on hierarchical behavioral information as described in any one of claims 1 to 5; and the processor is used to execute the program instructions stored in the memory to realize rapid determination of the state of an animal based on hierarchical behavioral information.

8. A storage medium, characterized in that: The method stores program instructions executable by a processor, wherein the program instructions are used to execute the method for quickly distinguishing the state of an animal based on hierarchical behavior information as described in any one of claims 1 to 5.

Images