A social behavior detection system for laboratory animals

Abstract

This invention discloses a system for detecting social behavior in laboratory animals, comprising implantable sensors, multiple receivers, and a host computer. The system, through the collaborative data acquisition of the implantable sensors and multiple receivers, simultaneously obtains co-occurrence information and behavioral category information of multiple observed laboratory animals, and judges their social behavior accordingly. This invention employs implantable telemetry technology to achieve long-term, continuous, and synchronous monitoring of physiological signals such as body temperature and movement in a multi-animal environment under low-interference conditions. It cleverly utilizes the multi-receiver deployment and receiver selection mechanism during wireless data transmission to achieve large-scale co-occurrence judgment, and further combines this with animal behavior category recognition to detect social behaviors in multiple observed animals.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of biological detection, and more particularly relates to an experimental animal social behavior detection system. BACKGROUND

[0002] Monitoring of animal behavior and physiological parameters is an important part of biomedical and behavioral science research, and the social behavior of animals is particularly complex due to the synchronous interaction of multiple individuals.

[0003] The behavior detection method of experimental animals can be roughly divided into two types: non-invasive and invasive. Non-invasive methods (such as infrared thermal imaging temperature measurement, radio frequency identification) are convenient to operate and have less disturbance to animals, but are easily disturbed by environmental factors, and the measurement accuracy is difficult to guarantee; while invasive methods (such as rectal temperature measurement) have higher accuracy, but frequent intervention during operation can easily cause stress reactions in animals, which is not conducive to long-term and multi-animal continuous monitoring. In order to balance data accuracy and low disturbance to animal daily behavior, implantable telemetry technology has emerged. This technology realizes automatic and real-time wireless transmission of important physiological signals (such as body temperature, motion data) through implantation of micro sensors in animals, significantly reducing human intervention and behavior disturbance, and showing great application potential in the fields of neural activity, blood pressure measurement and metabolism research.

[0004] However, the core problem of the widely used commercial implantable telemetry equipment in multi-animal social behavior research is to accurately judge the social behavior of experimental animals or independent activities. These technical bottlenecks significantly limit the in-depth study of animal social behavior. SUMMARY

[0005] In view of the above defects or improvement needs of the prior art, the present application provides an experimental animal social behavior detection system, which aims to obtain co-occurrence information and behavior category information of multiple experimental animals as observation objects through the cooperative data collection of implantable sensors and their multiple receivers, and to make social behavior judgments accordingly, to realize social behavior detection without affecting the activities of the observation objects, thereby solving the technical problem that the existing implantable telemetry technology cannot accurately judge the social behavior of experimental animals.

[0006] To achieve the above purpose, according to one aspect of the present application, an experimental animal social behavior detection system is provided, comprising an implantable sensor implanted in the body of an observation experimental animal, multiple receivers, and an upper computer;

[0007] The implantable sensor is used to collect physiological data of the implanted observation object, and to send the data to the upper computer through the physically closest receiver by wireless communication;

[0008] The receiver is fixedly arranged in the observation range of the experimental animal at a preset distance, and is used for receiving physiological data sent by the implantable sensor within a preset physical distance or observation data sent by other receivers, and forwarding the received data to the upper computer; the observation data is packaged by the receiver receiving the physiological data sent by the implantable sensor to form an identifier of the receiver and the physiological data;

[0009] The upper computer is provided with a social behavior detection module, including a co-occurrence judgment submodule, a behavior category judgment submodule, and a social association detection submodule; the co-occurrence judgment submodule is used for judging whether multiple observation objects are within a social distance; when the receivers identified by the receiver identification codes of the observation objects at the same sampling time are within the coverage range of the social distance, it is judged that the observation objects are within the social distance, and the judgment result is sent to the social association detection submodule; the behavior category judgment submodule is used for classifying the observation objects according to the physiological data to obtain the behavior category of the observation objects at the sampling time, and the classification result is sent to the social association detection submodule; the social association detection submodule is used for detecting multiple observation objects that have social behaviors within a preset time; when multiple observation objects are within the social distance at the same sampling time and have the same behavior category, it is judged that the multiple observation objects have social behaviors.

[0010] Preferably, the implantable sensor of the experimental animal social behavior detection system comprises a temperature sensor and a three-axis accelerometer.

[0011] Preferably, the physiological data of the experimental animal social behavior detection system comprises an identifier for identifying the implantable sensor, a sampling time, a temperature, and a three-axis acceleration; and the observation data specifically comprises an implantable sensor identifier, a sampling time, a receiver identifier, a temperature, and a three-axis acceleration.

[0012] Preferably, the experimental animal social behavior detection system comprises multiple receivers that communicate cooperatively through a wireless ad hoc network and forward data to the upper computer.

[0013] Preferably, the co-occurrence judgment submodule of the experimental animal social behavior detection system judges whether multiple observation objects are within a social distance according to the following method: the observation data of the multiple observation objects at the same sampling time has the same receiver identification code, and it is judged that the multiple observation objects are within the social distance; or

[0014] The observation data of the multiple observation objects at the same sampling time has the same receiver identified by the receiver identification code or the coverage range is adjacent, and it is judged that the multiple observation objects are within the social distance.

[0015] Preferably, in the experimental animal social behavior detection system, the behavior category determination submodule determines the behavior category of the observed object according to the following method:

[0016] (1) Reconstruct the temperature change curve and acceleration change curve of the observed object with time based on the physiological data of continuous sampling time, and decompose the movement pattern to identify the temperature data characteristics and activity data characteristics of the animal within a preset unit time period.

[0017] (2) Based on the temperature data features and activity data features obtained in step (1), a classifier is used to identify the behavior category at each observation time; the behavior category includes rest, exercise, and stress.

[0018] Preferably, the temperature data features of the experimental animal social behavior detection system include temperature mean, temperature standard deviation, and temperature change rate.

[0019] Preferably, the activity data features of the experimental animal social behavior detection system include: average acceleration in each direction, variance of acceleration in each direction, activity intensity, skewness of activity intensity, and kurtosis of activity intensity.

[0020] Preferably, the experimental animal social behavior detection system assesses activity intensity according to the following method:

[0021] (1-1) Input the acceleration-time curve into the median filter to reduce short-term noise and obtain a smooth acceleration curve;

[0022] (1-2) Based on the smooth acceleration change curve obtained in step (1-1), calculate the acceleration in each direction at each moment. , ,as well as , average value in all directions , ,as well as Assuming the effects of gravity are subtracted, the remaining motion is used to assess the intensity of the activity. for:

[0023] ;

[0024] Skewness Calculate using the following method:

[0025] ;

[0026] in, For exercise intensity, This represents the average exercise intensity. The standard deviation of exercise intensity This is a mathematical expectation operation.

[0027] cliff Calculate using the following method:

[0028] ;

[0029] in, For exercise intensity, This represents the average exercise intensity. The standard deviation of exercise intensity This is a mathematical expectation operation.

[0030] Preferably, in the experimental animal social behavior detection system, the classifier is a support vector machine.

[0031] Overall, the above-described technical solutions conceived by this invention can achieve the following beneficial effects compared with the prior art.

[0032] The experimental animal social behavior detection system provided by this invention adopts implantable telemetry technology to achieve long-term, continuous and synchronous monitoring of physiological signals such as body temperature and movement in a multi-animal environment under low interference conditions. It cleverly utilizes the multi-receiver deployment and receiver selection mechanism during wireless data transmission to achieve large-scale co-occurrence judgment, and further combines animal behavior category recognition to detect social behaviors of multiple observed objects. Attached Figure Description

[0033] Figure 1 This is a schematic diagram of the experimental animal social behavior detection system provided by the present invention;

[0034] Figure 2 This is a schematic diagram of an implantable sensor provided in an embodiment of the present invention;

[0035] Figure 3 This is a schematic diagram of the experimental animal social behavior detection module deployed on the host computer in an embodiment of the present invention.

[0036] In all the accompanying drawings, the same reference numerals are used to denote the same elements or structures, where: 1 is a triaxial accelerometer, and 2 is a temperature sensor. Detailed Implementation

[0037] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0038] The experimental animal social behavior detection system provided by this invention, such asFigure 1 As shown, it includes implantable sensors, multiple receivers, and a host computer implanted in the experimental animals.

[0039] The implantable sensor is used to collect physiological data of the implanted observation object and transmits it to the host computer via wireless communication through the physically nearest receiver; it includes a temperature sensor and a triaxial accelerometer; the physiological data includes an identification code for identifying the implantable sensor, sampling time, temperature, and triaxial acceleration;

[0040] The receiver is fixedly positioned within the observation range of the experimental animal at a preset distance. It is used to receive physiological data sent by the implanted sensor or observation data sent by other receivers within the preset physical distance, and forward the received data to the host computer. The observation data is formed by the receiver that receives the physiological data sent by the implanted sensor and packages its own identifier with the physiological data. Preferably, the observation data specifically includes the implanted sensor identification code, sampling time, receiver identification code, temperature, and triaxial acceleration. In a preferred embodiment, multiple receivers communicate collaboratively through a wireless ad hoc network to forward the data to the host computer.

[0041] The host computer is equipped with a social behavior detection module, including a co-occurrence judgment submodule, a behavior category judgment submodule, and a social association detection submodule. The co-occurrence judgment submodule determines whether multiple observed objects are within social distance. Specifically, when the receivers identified by the receiver identifiers of the multiple observed objects at the same sampling time are within the social distance coverage area, it is determined that the multiple observed objects are within social distance, and the judgment result is sent to the social association detection submodule. The behavior category judgment submodule classifies the observed objects based on their physiological data to obtain their behavior category at the sampling time, and sends the classification result to the social association detection submodule. The social association detection submodule detects multiple observed objects engaging in social behavior at a preset time. Specifically, when multiple observed objects are within social distance and have the same behavior category at the same sampling time, it is determined that the multiple observed objects have engaged in social behavior.

[0042] The co-occurrence determination submodule determines whether multiple observation objects are within social distancing using the following method: if the receiver identification codes of the observation data of the multiple observation objects at the same sampling time are the same, it is determined that the multiple observation objects are within social distancing; or

[0043] If the receivers identified by the receiver identification codes of the multiple observation objects at the same sampling time are the same or have adjacent coverage areas, it is determined that the multiple observation objects are within the social distance range.

[0044] Co-occurrence determination is a key indicator for observing social behavior. However, implanted telemetry systems often cannot determine the physical distance between observed subjects, making it impossible to determine whether social behavior has occurred. This invention creatively utilizes multiple receivers for physical distance positioning. Multiple receivers network collaboratively to cover a larger observation range. Within this range, the experimental animals with implanted sensors can move freely. When social behavior occurs, the multiple observed subjects should be within a reasonable distance. The collected data is relayed through the nearest receiver, and marking the receiver's location allows for determination of whether the observed subjects are within the social distance. The denser the receiver network, the more flexible and accurate the determination of whether observed subjects are within the social range can be achieved by setting a social distance, thus providing a reliable basis for judging the social behavior of experimental animals. Generally, the distance covered by the receivers is comparable to the social distance, satisfying social distance detection while reducing judgment redundancy and saving costs. Therefore, strictly speaking, if the data sampled at the same sampling time is received by the same receiver, it is determined that the observed object is within the social distance. Considering that when social behavior occurs, the observed object may be in the adjacent area of ​​adjacent receivers, the accuracy can be improved by setting up more dense receivers. It is considered that if the observed object's data at the same sampling time is identified by the same receiver or has an adjacent coverage area, it is considered to be within the social distance.

[0045] The behavior category determination submodule determines the behavior category of the observed object according to the following method:

[0046] (1) Based on the physiological data of continuous sampling time, the temperature change curve and acceleration change curve of the observed object are reconstructed, and the movement pattern is decomposed to identify the temperature data characteristics and activity data characteristics of the animal within a preset unit time period. The temperature data characteristics include the temperature mean, temperature standard deviation, and temperature change rate. The activity data characteristics include the average acceleration in each direction, the variance of acceleration in each direction, activity intensity, skewness of activity intensity, and kurtosis of activity intensity. Among them, the average acceleration in each direction, activity intensity, skewness, and kurtosis reflect more of the large movement state such as running and jumping, while the variance of acceleration in each direction is more caused by short-distance high-frequency movement.

[0047] Activity intensity, used to characterize large-scale motion information, is assessed as follows: Since the acceleration-time curve measured by implanted sensors exhibits abrupt changes, including extremely short displacements and high-frequency motion information, short-term noise needs to be filtered out.

[0048] (1-1) Input the acceleration-time curve into the median filter to reduce short-term noise and obtain a smooth acceleration curve;

[0049] (1-2) Based on the smooth acceleration change curve obtained in step (1-1), calculate the acceleration in each direction at each moment. , ,as well as , average value in all directions , ,as well as Assuming the effects of gravity are subtracted, the remaining motion is used to assess the intensity of the activity. for:

[0050] ;

[0051] Skewness Calculate using the following method:

[0052] ;

[0053] in, For exercise intensity, This represents the average exercise intensity. The standard deviation of exercise intensity This is a mathematical expectation operation.

[0054] cliff Calculate using the following method:

[0055] ;

[0056] in, For exercise intensity, This represents the average exercise intensity. The standard deviation of exercise intensity This is a mathematical expectation operation.

[0057] (2) Based on the temperature data features and activity data features obtained in step (1), a classifier is used to identify the behavior category at each observation time; the behavior category includes rest, exercise and stress; the classifier is preferably a support vector machine, which has good generalization ability for small sample and low-dimensional data.

[0058] The following is an example:

[0059] The experimental animal social behavior detection system provided by this invention, such as Figure 1 As shown, the system includes implantable sensors inserted into the experimental animals, multiple receivers, and a host computer. Each animal has a small implantable sensor that collects physiological data and transmits signals wirelessly. Several receivers are placed around the experimental area. The entire hardware system also includes a host computer, which serves as a platform for centralized data processing and monitoring.

[0060] The implantable sensor is used to collect physiological data of the implanted observation object and transmits it to the host computer via wireless communication through the physically nearest receiver; it includes a temperature sensor and a triaxial accelerometer; the physiological data includes an identification code for identifying the implantable sensor, sampling time, temperature, and triaxial acceleration.

[0061] The implantable sensor used in this embodiment, such as Figure 2 As shown, the implantable sensor features a highly integrated, miniaturized design, measuring approximately 16.3 mm × 13.2 mm × 8.2 mm and weighing only 1.69 grams. This small size facilitates implantation in small animals without affecting their normal activities. The module is encapsulated in biocompatible materials, such as a polychlorotrifluoroethylene (PCTFE) shell, ensuring safe and reliable long-term implantation and compliance with ISO 10993 biocompatibility standards. Internally, the implantable sensor contains an ultra-low-power microcontroller unit (MCU) responsible for controlling sensor readings and wireless data transmission. Figure 2 The internal structure of the implantable sensor module is illustrated, with its core comprising three main functional components: a precision temperature sensor, a triaxial accelerometer, and a short-pitch reed switch. The temperature sensor, for example, is the M601 model, with a high accuracy of ±0.1℃ (within the range of -70℃ to +150℃), capable of real-time monitoring of animal body temperature changes. The accelerometer, for example, is the LIS2DW12 model, capable of highly sensitively capturing subtle movements of the animal along the X, Y, and Z axes. The reed switch is used for energy-saving control; when the implantable sensor is near a strong magnetic field, the switch cuts off the circuitry, thus maintaining a dormant state during device storage and transportation. This design ensures zero power consumption before formal implantation and extends battery life, and allows researchers to control the device's power supply via a magnetic field before and after implantation, facilitating experimental operations.

[0062] The implantable sensor firmware runs directly on the microcontroller of the implantable sensor module. Its main function is to read temperature and acceleration data according to a set sampling period and broadcast the data wirelessly. The firmware employs an ultra-low-power programming model: during the interval between two sampling periods, the MCU and sensor enter sleep or low-power mode, only waking up momentarily when data acquisition and transmission are needed, thus maximizing energy savings. The sampling interval can be configured according to experimental needs, typically between 10 seconds and 30 minutes. At high-frequency sampling (e.g., once every 10 seconds), the implantable sensor can still operate continuously for about one month without battery replacement; if the sampling frequency is reduced to once every 30 minutes, the device's battery life can be extended to about two years. Researchers can flexibly adjust the sampling strategy in the firmware according to different experimental requirements for time resolution and battery life.

[0063] The receiver is fixedly positioned within the observation range of the experimental animal at a preset distance. It is used to receive physiological data sent by the implanted sensor or observation data sent by other receivers within the preset physical distance, and forward the received data to the host computer. The observation data is formed by the receiver that receives the physiological data sent by the implanted sensor and packages its own identifier with the physiological data. Preferably, the observation data specifically includes the implanted sensor identification code, sampling time, receiver identification code, temperature, and triaxial acceleration. In this embodiment, multiple receivers communicate collaboratively through a wireless ad hoc network to forward data to the host computer.

[0064] The receivers used in this embodiment are wireless data collection nodes deployed around the experimental space. Each receiver contains dual processing modules, including an RF receiving module and a relay transmission module. The RF receiving module, for example, uses a Nordic RF52832 chip, responsible for listening to and capturing wireless data broadcast by surrounding implanted sensors. Considering the limitations of the implanted sensor's transmission power and transmission distance, multiple receivers are needed to cover the entire experimental site. After capturing the signal, the receiver transmits the data to the relay transmission module (e.g., an ESP32 processor) via its internal serial interface. The ESP32 module has dual WiFi / Bluetooth communication capabilities, which can package the received data with an identifier and send it to the host computer via a wired or wireless network. Multiple receiver units can coordinate through a wireless self-organizing network (Mesh network): each receiver can communicate directly with the host computer or with adjacent receivers, hop-by-hop transmitting data to the node closest to the host computer, thereby achieving flexible expansion of coverage and reliable signal forwarding. This mesh network communication architecture significantly enhances the system's resilience to packet loss and its coverage radius. When animals carrying implanted sensors move around the site, at least one receiver maintains a constant communication connection, and the collaboration between receivers ensures that data is ultimately aggregated to the host computer. In practical testing, this system, through multi-receiver collaboration, can extend the effective monitoring range to a space with a diameter exceeding 40 meters. Furthermore, the number of simultaneously online receivers is positively correlated with the number of implants, allowing for linear expansion of system capacity. The receiver hardware employs a low-power design and supports modular expansion, enabling the number of nodes to be added or removed according to the scale of the experiment, ensuring seamless signal coverage throughout the multi-animal interaction area.

[0065] The receiver firmware runs on the nRF52832 and ESP32 modules of the receiver unit. The former's firmware is responsible for wirelessly monitoring signals from multiple implanted sensors and transmitting the received raw data to the ESP32 module via interfaces such as UART. The firmware on the ESP32 is responsible for networking and relay transmission: it aggregates data from the local node and other neighboring nodes and sends it to the host computer via WiFi or wired connection. If there are multiple receiver nodes in the network, the ESP32 modules discover each other and maintain communication through a pre-configured Mesh network protocol, enabling data received by any node to be transmitted to the master node or host computer within milliseconds. This firmware also has a simple data caching and retransmission mechanism: when the host computer is temporarily unavailable or the network is congested, the node will briefly store the data and forward it immediately after the link is restored to avoid data loss.

[0066] The host computer is equipped with a social behavior detection module, such as... Figure 3 The system includes a co-occurrence judgment submodule, a behavior category judgment submodule, and a social association detection submodule. The co-occurrence judgment submodule determines whether multiple observed objects are within social distance. Specifically, if the receivers identified by the receiver identifiers of the multiple observed objects at the same sampling time are within the social distance coverage area, the system determines that the multiple observed objects are within social distance and sends the judgment result to the social association detection submodule. The behavior category judgment submodule classifies the observed objects based on their physiological data to obtain their behavior category at the sampling time. The social association detection submodule detects multiple observed objects that have engaged in social behavior at a preset time. Specifically, if multiple observed objects are within social distance at the same sampling time and have the same behavior category, the system determines that the multiple observed objects have engaged in social behavior.

[0067] In this embodiment, the host computer application is a set of data receiving and analysis software running on a computer. One end is connected to the receiver ESP32 module via a serial port and network interface, continuously receiving real-time data streams from all implants. The host computer software distinguishes data from different implants (e.g., based on device ID tags) and performs preliminary processing and visualization. The data processing flow of the host computer software includes modules for data recording, preprocessing, storage, and display. Specifically, the host computer application first calibrates and formats the temperature and acceleration data: converting raw temperature values ​​into actual Celsius temperature data and reconstructing a temperature change curve using timestamps; and reconstructing the acceleration change curve of the observed object using triaxial accelerometer data and timestamps. The host computer software also provides a real-time monitoring interface, allowing researchers to view the curve changes of indicators such as current body temperature and activity level, as well as the receiver identification code for each animal. When multiple animals are tested simultaneously, the software can display multi-channel data simultaneously to compare the status of different individuals. This lays the data foundation for subsequent social behavior analysis.

[0068] The co-occurrence judgment submodule of the social behavior detection system determines whether multiple observation objects are within social distance according to the following method: if the observation data of the multiple observation objects at the same sampling time have the same receiver identification code, it is determined that the multiple observation objects are within social distance.

[0069] The behavior category determination submodule determines the behavior category of the observed object according to the following method:

[0070] (1) Based on the physiological data of continuous sampling time, the temperature change curve and acceleration change curve of the observed object are reconstructed, and the movement pattern is decomposed to identify the temperature data characteristics and activity data characteristics of the animal within a preset unit time period. The temperature data characteristics include the temperature mean, temperature standard deviation, and temperature change rate. The activity data characteristics include the average acceleration in each direction, the variance of acceleration in each direction, activity intensity, skewness of activity intensity, and kurtosis of activity intensity. Among them, the average acceleration in each direction, activity intensity, skewness, and kurtosis reflect more of the large movement state such as running and jumping, while the variance of acceleration in each direction is more caused by short-distance high-frequency movement.

[0071] Activity intensity, used to characterize large-scale motion information, is assessed as follows: Since the acceleration-time curve measured by implanted sensors exhibits abrupt changes, including extremely short displacements and high-frequency motion information, short-term noise needs to be filtered out.

[0072] (1-1) Input the acceleration-time curve into the median filter to reduce short-term noise and obtain a smooth acceleration curve;

[0073] (1-2) Based on the smooth acceleration change curve obtained in step (1-1), calculate the acceleration in each direction at each moment. , ,as well as , average value in all directions , ,as well as Assuming the effects of gravity are subtracted, the remaining motion is used to assess the intensity of the activity. for:

[0074] ;

[0075] Skewness Calculate using the following method:

[0076] ;

[0077] in, For exercise intensity, This represents the average exercise intensity. The standard deviation of exercise intensity This is a mathematical expectation operation.

[0078] cliff Calculate using the following method:

[0079] ;

[0080] in, For exercise intensity, This represents the average exercise intensity. The standard deviation of exercise intensity This is a mathematical expectation operation.

[0081] Low-pass filters (such as Butterworth filters) are used to remove high-frequency noise from temperature and acceleration data, improving data quality and stability. The data is divided into fixed time windows (e.g., 10 seconds or 30 seconds) to extract temperature and activity features. The data is then normalized to standardize the data range to the same scale, facilitating model training.

[0082] (2) Based on the temperature data features and activity data features obtained in step (1), a classifier is used to identify the behavior category at each observation time; the behavior category includes rest, exercise and stress; the classifier is preferably a support vector machine, which has good generalization ability for small sample and low-dimensional data.

[0083] A classification model is used to identify the behavior category at each observation time, statistically analyze the animal's behavior categories within the observation period, and calculate the biological activity level. The behavior categories include rest, movement, and stress. The classification model employs a Support Vector Machine (SVM). SVM is a powerful supervised learning algorithm widely used in classification and regression analysis. SVM classifies data by finding an optimal hyperplane in a high-dimensional space. The selection criterion for this optimal hyperplane is to maximize the margin between support vectors (data points closest to the decision boundary), thereby achieving efficient classification of unknown samples. The core ideas of SVM are: mapping: mapping input data to a higher-dimensional space; optimization: finding the decision boundary in this space that can maximally distinguish different categories of data; and kernel function: effectively handling nonlinear data through kernel functions (such as linear, polynomial, and Gaussian radial basis functions (RBF)). This embodiment uses the RBF kernel function for classification.

[0084] Constructing the training dataset: The preprocessed temperature and activity data features are organized into a feature matrix, corresponding to behavioral labels. The behavioral labels are derived from the continuous free movement of mice observed in computer vision studies, and are in the following format:

[0085]

[0086] The results showed that the implanted sensors and computer vision measurements exhibited a high degree of synchronicity and similar movement patterns, with a correlation coefficient as high as r=0.95 (p<0.001) for the calm period, the running wheel phase, and the subsequent recovery period.

[0087] When multiple observed objects are within social distance and exhibit the same behavior category at the same sampling time, it is determined that the multiple observed objects have engaged in social behavior. For example, social behaviors such as grooming each other's hair or fighting both meet the characteristics of co-occurrence and the same behavior category.

[0088] The technical challenges of detecting social behavior in multiple subjects lie in: 1. Achieving high-density, simultaneous monitoring of multiple animals while avoiding signal conflicts and data loss; 2. Extending wireless communication distance and improving signal transmission stability to ensure continuous data acquisition in large-scale free-movement environments; 3. Minimizing and reducing the intrusiveness of sensor modules to ensure long-term implantation of the device in animals without significantly affecting their natural behavior; 4. Minimizing device power consumption and extending device lifespan to meet the data acquisition requirements of different frequencies; 5. Achieving fusion analysis of multimodal data (such as triaxial acceleration and body temperature signals) to further improve the ability to identify and interpret social interaction behaviors. The animal social behavior monitoring system provided in this embodiment aims to overcome the above technical challenges, achieving a high-precision, low-interference, and wide-coverage monitoring solution, and promoting the in-depth development of animal behavior research.

[0089] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A system for detecting social behavior in laboratory animals, characterized in that, This includes implantable sensors, multiple receivers, and a host computer that are implanted into experimental animals for observation. The implantable sensor is used to collect physiological data of the implanted observation object and transmits it to the host computer via wireless communication through the physically closest receiver. The receiver is fixedly positioned within the observation range of the experimental animal at a preset distance. It is used to receive physiological data sent by the implanted sensor or observation data sent by other receivers within the preset physical distance, and forward the received data to the host computer. The observation data is formed by the receiver that receives the physiological data sent by the implanted sensor and packages its own identifier with the physiological data. The host computer is equipped with a social behavior detection module, including a co-occurrence judgment submodule, a behavior category judgment submodule, and a social association detection submodule; The co-occurrence determination submodule is used to determine whether multiple observed objects are within social distance, and to send the determination result to the social association detection submodule; The behavior category judgment submodule is used to classify the behavior category of the observed object at the sampling time based on the physiological data of the observed object, and send the classification result to the social association detection submodule. The social association detection submodule is used to detect multiple observation objects that have engaged in social behavior at a preset time. When multiple observation objects are within social distance and have the same behavior category at the same sampling time, it is determined that the multiple observation objects have engaged in social behavior. The co-occurrence determination submodule determines whether multiple observation objects are within social distance using the following method: if the receiver identification codes of the observation data of the multiple observation objects at the same sampling time are the same, it is determined that the multiple observation objects are within social distance; or; If the receivers identified by the receiver identification codes of the multiple observation objects at the same sampling time are the same or have adjacent coverage areas, it is determined that the multiple observation objects are within the social distance range.

2. The experimental animal social behavior detection system as described in claim 1, characterized in that, The implantable sensors include a temperature sensor and a triaxial accelerometer.

3. The experimental animal social behavior detection system as described in claim 1, characterized in that, The physiological data includes an identification code for identifying the implantable sensor, sampling time, temperature, and triaxial acceleration; the observation data specifically includes the implantable sensor identification code, sampling time, receiver identification code, temperature, and triaxial acceleration.

4. The experimental animal social behavior detection system as described in claim 1, characterized in that, Multiple receivers communicate collaboratively via a wireless ad hoc network to forward data to the host computer.

5. The experimental animal social behavior detection system as described in claim 1, characterized in that, The behavior category determination submodule determines the behavior category of the observed object according to the following method: (1) Reconstruct the temperature change curve and acceleration change curve of the observed object with time based on the physiological data of continuous sampling time, and decompose the movement pattern to identify the temperature data characteristics and activity data characteristics of the animal within a preset unit time period. (2) Based on the temperature data features and activity data features obtained in step (1), a classifier is used to identify the behavior category at each observation time; the behavior category includes rest, exercise, and stress.

6. The experimental animal social behavior detection system as described in claim 5, characterized in that, The temperature data features include temperature mean, temperature standard deviation, and temperature change rate.

7. The experimental animal social behavior detection system as described in claim 5, characterized in that, The activity data features include: the average acceleration in each direction, the variance of acceleration in each direction, the activity intensity, the skewness of the activity intensity, and the kurtosis of the activity intensity.

8. The experimental animal social behavior detection system as described in claim 7, characterized in that, Assess activity intensity using the following methods: (1-1) Input the acceleration-time curve into the median filter to reduce short-term noise and obtain a smooth acceleration curve; (1-2) Based on the smooth acceleration change curve obtained in step (1-1), calculate the acceleration in each direction at each moment. , ,as well as , average value in all directions , ,as well as Assuming the effects of gravity are subtracted, the remaining motion is used to assess the intensity of the activity. for: ； Skewness Calculate using the following method: ； in, For exercise intensity, This represents the average exercise intensity. The standard deviation of exercise intensity For mathematical expectation calculation; cliff Calculate using the following method: ； in, For exercise intensity, This represents the average exercise intensity. The standard deviation of exercise intensity This is a mathematical expectation operation.

9. The experimental animal social behavior detection system as described in claim 5, characterized in that, The classifier is a support vector machine.

Images