A system for monitoring animal static behavior and metabolites in synchronization

Abstract

This invention relates to a system for synchronously monitoring animal static behavior and metabolites in the field of animal experimental monitoring technology. The system includes a behavior perception module, a behavior recognition module, an event triggering and verification module, a metabolite monitoring and control module, and a data fusion module. The system uses the perception and recognition module to identify specific static behaviors of animals in real time and generate event signals. The event triggering and verification module performs multi-dimensional cross-verification on these signals, and upon successful verification, generates trigger control commands based on pre-stored behavior-metabolism association rules. The metabolite monitoring and control module then controls the metabolite monitoring unit to adjust its sampling parameters accordingly. The data fusion module synchronously fuses behavior and metabolite data with unified timestamps to generate an associated dataset. This invention overcomes the shortcomings of existing technologies, such as the fragmentation of behavior and metabolic data and the difficulty in monitoring static behavior, achieving accurate, automatic, and highly reliable synchronous capture of changes in metabolites in the body during static behavior.

Description

Technical Field

[0001] This invention relates to the field of animal experimental monitoring technology, and more specifically, to a system for simultaneous monitoring of animal static behavior and metabolites. Background Technology

[0002] In pharmacology, toxicology, neuroscience, and metabolic disease research, simultaneous monitoring of the behavior and in vivo biochemical indicators of laboratory animals (such as mice and rats) is a key method for revealing their underlying mechanisms. However, existing monitoring technologies suffer from two fundamentally related shortcomings:

[0003] 1. Behavioral and metabolic data are severely disconnected in terms of source and time: On the one hand, macroscopic monitoring systems based on metabolic cages (such as the CLAMS system) can record animal activity, food intake, and indirect energy metabolism rate, but cannot provide real-time concentration data of specific chemical metabolites (such as neurotransmitters and stress hormones) in the animal's body. On the other hand, while emerging implantable or wearable biosensors can achieve real-time monitoring of specific metabolites, their data streams are independent of behavioral monitoring systems. These two types of systems usually operate independently, with asynchronous data acquisition clocks, making it difficult for researchers to precisely align observed animal behavior with the precise biochemical reactions occurring in the body in time, thus losing crucial behavioral-metabolic correlation information.

[0004] 2. Lack of effective identification capability for key static behaviors: Traditional animal behavior analysis mainly relies on video tracking or infrared beam arrays. These technologies are relatively sensitive to "dynamic behaviors" such as animal displacement and running, but their ability to automatically identify and quantify "static behaviors" that occur in situ and have subtle amplitudes (such as grooming, scratching, rigidity, and sleep micro-movements) is severely inadequate. These static behaviors are often the core observation indicators for assessing an animal's emotional state, pain response, neurological function, and drug efficacy.

[0005] The two aforementioned shortcomings together lead to a long-standing unresolved technical challenge: how to automatically, accurately, and timely capture the instantaneous dynamic changes of relevant metabolites in animals during specific static behaviors. Existing technical solutions mostly involve post-hoc manual correlation of data or simple parallel recording, which cannot meet the needs of precision biomedical research for high spatiotemporal resolution causal data. Summary of the Invention

[0006] To address the shortcomings of the existing technologies, the primary objective of this invention is to resolve the issue of the separation between behavioral and metabolic data, achieving source synchronization and fusion of the two types of data. Secondly, it aims to solve the challenge of static behavior recognition, enabling the automatic capture of fine behavioral patterns. Finally, by achieving the above objectives, it provides researchers with a high-quality dataset that directly reflects the causal relationship between behavior and metabolism.

[0007] To achieve the above objectives, the present invention provides a system for simultaneous monitoring of animal static behavior and metabolites, comprising:

[0008] The behavior perception module is used to collect animal behavior data in real time;

[0009] The behavior recognition module communicates with the behavior perception module to receive behavior data and identify specific static behavior patterns based on a predefined static behavior feature library, generating preliminary behavior event signals.

[0010] The event triggering and verification module communicates with the behavior recognition module. The event triggering and verification module includes a behavior-metabolism association knowledge base, which stores rules that associate static behavior patterns with metabolite monitoring parameter adjustment strategies. The event triggering and verification module is used to receive preliminary behavior event signals and perform multi-dimensional verification. After the verification is passed, it generates trigger control instructions based on the rules that match the specific static behavior pattern.

[0011] The metabolite monitoring and control module is connected in communication with the event triggering and verification module. It is used to control a metabolite monitoring unit to adjust its monitoring parameters according to the triggering control command.

[0012] The data fusion module is connected to the behavior perception module and the metabolite monitoring unit respectively. It is used to receive and synchronously fuse time-stamped behavior data and metabolite data to generate a related dataset.

[0013] As a further improvement of the present invention, multi-dimensional verification includes at least one of the following:

[0014] (1) Time window verification: Determine whether the identified specific static behavior pattern is stable within a preset duration;

[0015] (2) Type consistency verification: Obtain data from the auxiliary perception module that is different from the behavior perception module, and determine whether the behavior type represented by the data of the auxiliary perception module is consistent with the behavior type represented by the initial behavior event signal;

[0016] (3) Verification of the rationality of physiological state: Based on the background physiological data or metabolite baseline trend provided by the metabolite monitoring unit, determine whether it is consistent with the physiological state expected by a specific static behavior pattern.

[0017] As a further improvement of the present invention, the trigger control command includes a command for adjusting at least one of the following parameters of the metabolite monitoring unit: sampling frequency, operating mode, or target analyte type.

[0018] As a further improvement of the present invention, the behavior perception module includes a millimeter-wave radar sensor for acquiring micro-motion signals of animals; the behavior recognition module is configured to identify specific static behavior patterns by analyzing the micro-Doppler feature spectrum of the micro-motion signals; and the auxiliary perception module includes a visual sensor for providing image data in type consistency verification.

[0019] As a further improvement of the present invention, time window verification, type consistency verification and physiological state rationality verification are executed according to preset priority or sequence logic; the event triggering and verification module is configured to: only start the next level verification after the previous level verification is passed, otherwise terminate the verification process of the current preliminary behavioral event signal.

[0020] As a further improvement to the present invention, the verification of the rationality of the physiological state includes:

[0021] The event triggering and verification module is configured to build a short-term prediction model for metabolite concentration or trends based on background data provided by the metabolite monitoring unit.

[0022] When an initial behavioral event signal occurs, real-time monitored metabolite data is compared with a short-term predictive model;

[0023] If the real-time data deviates from the predicted value by more than a preset threshold, the physiological state rationality verification is deemed unsuccessful.

[0024] As a further improvement of the present invention, the system also includes a model optimization module;

[0025] The model optimization module is configured to: incrementally learn the static behavioral feature library in the behavior recognition module based on the validated static behavioral pattern fragments in the associated dataset and their synchronized metabolite change data, and / or evaluate and adjust the confidence of the rules in the behavior-metabolism association knowledge base.

[0026] As a further improvement of the present invention, the data fusion module is also configured to automatically generate structured annotation information, including experimental stage, behavior category, and target metabolite, for the synchronously fused data segment based on the behavior type and metabolite target associated with the trigger control command.

[0027] As a further improvement of the present invention, the metabolite monitoring unit is a biosensor that can be worn or implanted in an animal.

[0028] As a further improvement of the present invention, the data fusion module is also configured to mark the corresponding verified static behavior type and event time information for each metabolite data segment driven by the trigger control command in the associated dataset.

[0029] Compared with the prior art, the present invention has the following beneficial effects:

[0030] 1. By constructing an integrated hardware system and a unified time reference, this invention achieves millisecond-level synchronization of behavioral data streams and metabolite data streams at the acquisition end. The "association dataset" output by the system inherently possesses time alignment, fundamentally solving the technical obstacles to subsequent data fusion and providing the possibility for accurate causal analysis.

[0031] 2. By employing specialized sensors sensitive to micro-motions, such as millimeter-wave radar, and combining them with targeted feature analysis algorithms (such as micro-Doppler feature spectrum analysis), the system in this invention can effectively identify and quantify static behaviors that are difficult to capture by traditional methods, greatly expanding the dimensions and depth of behavioral observation.

[0032] 3. The core "identification-verification-triggering" closed-loop control logic of this invention changes the passive mode of continuous parallel recording in traditional systems. The system in this invention can actively identify target behaviors and, after high-confidence verification, intelligently trigger targeted monitoring of metabolites. This not only greatly improves the purposefulness, effectiveness, and system efficiency of data acquisition and avoids the generation of massive amounts of invalid data, but also makes it possible to capture transient biochemical events.

[0033] 4. Further optimization features in this invention, such as multi-dimensional verification sequences, rationality verification based on prediction models, self-optimization learning of models, and structured automatic annotation of data, jointly improve the reliability, adaptability, and direct usability of the output data of the system, enabling it to evolve from a monitoring tool into an intelligent research support platform. Attached Figure Description

[0034] Figure 1 This is a structural block diagram of the animal static behavior and metabolite synchronous monitoring system in Embodiment 1 of the present invention;

[0035] Figure 2 This is a flowchart of the cascading verification logic executed by the event triggering and verification module in Embodiment 2 of the present invention;

[0036] Figure 3 This is a flowchart of the model optimization module in Embodiment 3 of the present invention. 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 the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of protection of this invention.

[0038] Example 1:

[0039] This embodiment describes the basic hardware structure of the system of the present invention and the core mechanism for achieving high-precision data synchronization.

[0040] like Figure 1 As shown, an animal static behavior and metabolite synchronous monitoring system includes the following components:

[0041] Behavior perception module: A millimeter-wave radar sensor (e.g., a frequency-modulated continuous wave (FMCW) radar) is used as the main sensor and is installed above the laboratory mouse cage. A low-light infrared camera is also installed on the side of the cage as an auxiliary perception module.

[0042] Metabolite monitoring unit: A flexible biosensor patch that can be implanted under the skin is used as a specific example of a metabolite monitoring unit. Its integrated electrochemical working electrode can monitor the concentration of two metabolites, glucose and corticosterone, in real time.

[0043] Central Processing and Control System: Centered on an embedded processing unit (such as a high-performance microprocessor based on the ARM architecture), its internal structure is divided into a behavior recognition module, an event triggering and verification module, a metabolite monitoring and control module, and a data fusion module via software logic. To achieve millisecond-level synchronization, this processing unit is equipped with a high-precision clock source (e.g., a synchronous clock chip based on a temperature-controlled crystal oscillator) as the absolute time reference for the entire system. All external data input ports are configured for hardware interrupt triggering. Whenever a data packet arrives, the hardware interrupt service routine adds a unified microsecond-level precision timestamp generated by the current clock source.

[0044] Data storage: The system's predefined static behavioral feature library and behavior-metabolic association knowledge base are stored in the processing unit's memory as database files. The initial feature library contains radar micro-Doppler feature templates for behaviors such as "stereotypical grooming" and "exploratory rigidity," and this feature spectrum is obtained by performing a short-time Fourier transform (STFT) on the original radar I / Q signal.

[0045] The basic workflow of the system is as follows:

[0046] (1) Continuous sensing and synchronous stamping: The millimeter-wave radar continuously collects raw point cloud data, and the metabolite monitoring unit continuously collects concentration data at a base frequency of 1Hz. All data packets are appended with the above-mentioned unified timestamp when they enter the system.

[0047] (2) Behavior recognition: The behavior recognition module processes the radar data stream in real time, calculates the micro-Doppler feature spectrum of the signal within the current time window (e.g., 2 seconds), and calculates the cosine similarity with the feature library template. When the matching degree exceeds the preset threshold (e.g., 70%), a "preliminary behavior event signal" containing behavior type and confidence level is generated.

[0048] (3) Basic verification and triggering: After receiving the signal, the event triggering and verification module performs two levels of verification: First, it performs time window verification (to determine whether the behavioral feature is stable within the next 3 seconds); if it passes, it performs type consistency verification (triggering the infrared camera to capture the image and confirming the animal's posture through a lightweight image recognition algorithm). After the verification passes, the module queries the rules in the knowledge base (e.g., rule 1: IF behavior = "stereotypical grooming" THEN trigger {target: corticosterone, sampling rate: increased from 1Hz to 5Hz, duration: 120 seconds}) and generates the corresponding trigger control command.

[0049] (4) Control execution: The metabolite monitoring and control module parses the instruction and immediately adjusts the sampling rate of the corticosterone sensing channel to 5Hz through the control circuit.

[0050] (5) Data Fusion: Throughout the process, the data fusion module continuously receives data streams. After the process is completed, the module performs time alignment and interpolation on the behavioral event data, image snapshots and metabolite concentration time series data based on the unified timestamp of each data stream, and generates a standard HDF5 format file with the "Sticky Grooming_Trigger" event tag, namely the "Associated Dataset".

[0051] Example 2:

[0052] This embodiment, based on the system architecture described in Embodiment 1, focuses on refining the intelligent decision-making logic of the event triggering and verification module. The process is as follows: Figure 2 As shown, this module is configured to execute a three-level cascading verification sequence. The next level can only be started after the previous level has passed verification; if any step fails, the process terminates immediately.

[0053] Level 1: Fast Time Window Validation. This level of validation is more stringent, requiring the identified behavioral features to appear continuously and stably within a shorter time window (e.g., 1 second).

[0054] Level 2: Type consistency verification. Similar to Example 1, a visual sensor is used for auxiliary cross-validation.

[0055] Level 3: Physiological rationality verification (the core of this embodiment). This level of verification aims to filter out pseudo-events from a physiological logic perspective. One specific implementation method is as follows: The system uses recent background data (such as the corticosterone concentration time series of the most recent 5 minutes) provided by the metabolite monitoring unit to train a simple first-order autoregressive (AR(1)) prediction model. Specifically, the model is fitted using the historical sequence through the least squares method. To obtain model parameters and For the latest data points The single-step prediction value of the metabolite concentration at the next time step is The prediction interval is set to... ,in This represents the standard deviation of the recent forecast residuals. This is the preset confidence coefficient (usually set to 2).

[0056] When the initial behavioral event signal occurs, the system reads the concentration value reported in real time by the metabolite monitoring unit. .like If the change falls outside the predicted range, the current physiological state is determined to be inconsistent with the expectation based on historical stable trends, and the verification fails. This mechanism can effectively filter out sudden changes in physiological signals that are inconsistent with the current behavior recognition results and are caused by occasional strong external interference, ensuring that the final triggering event has a very high confidence level of physiological relevance.

[0057] Example 3:

[0058] This embodiment, based on the integration of the functions of Embodiment 1 and Embodiment 2, further adds a model optimization module, enabling the system to have self-evolution capabilities. Its workflow is as follows: Figure 3 As shown.

[0059] Feature library optimization (incremental learning): The model optimization module periodically (e.g., every 24 hours) scans all high-confidence "association datasets" generated by the data fusion module. For verified event fragments, the corresponding radar micro-Doppler feature spectra are extracted and added as new positive samples to a dynamic incremental training set. Whenever the training set accumulates to a certain size (e.g., 50 new samples), the module uses stochastic gradient descent to incrementally learn the classifier (e.g., Support Vector Machine, SVM) used in the behavior recognition module, updating its decision function weights so that the classifier can adapt to the individual behavioral characteristics of experimental animals and continuously optimize the recognition accuracy over time.

[0060] Knowledge Base Optimization (Rule Evaluation): This module performs a quantitative posterior evaluation of the execution effect of each triggering rule in the knowledge base. For a given rule, its confidence score is calculated. One calculation formula is:

[0061] ;

[0062] in, This represents the total number of times the rule is triggered. After triggering, the synchronously monitored metabolite concentration change exceeded twice the standard deviation of the baseline level. The number of times () is defined by a certain rule. If the value remains below the set threshold (e.g., 0.3) for an extended period, the system will automatically flag the rule for review by experimenters, or temporarily lower its trigger priority in the automatic decision-making logic.

[0063] Meanwhile, the data fusion module in this embodiment has been enhanced. When generating associated datasets, it automatically generates structured annotations for data segments, such as [Behavior: grooming, Target metabolite: corticosterone, Experimental stage: post-drug administration], based on the behavior type and metabolite target associated with the trigger control command. This greatly improves the readability of the output data and the import efficiency of subsequent analysis.

[0064] Example 4:

[0065] This embodiment aims to demonstrate the flexible application of the system of the present invention in different research scenarios by expanding the content of the static behavioral feature library and the behavior-metabolism association knowledge base, based on the aforementioned system architecture (Embodiment 1), intelligent verification (Embodiment 2), and self-optimization function (Embodiment 3).

[0066] Scenario A: Monitoring the correlation between micro-movements and metabolism during sleep

[0067] Behavior recognition extension: Add radar micro-Doppler feature templates corresponding to high-frequency, low-amplitude body micro-movements unique to "REM sleep" to the static behavior feature library.

[0068] Rules and Triggers: Pre-defined rules in the behavior-metabolism association knowledge base:

[0069] Rule 2: IF behavior = "REM sleep micro-movement" THEN trigger {target: lactic acid & glucose, sampling rate: 10Hz, duration: 180 seconds}.

[0070] Execution and Fusion: Once the system identifies the pattern and passes multi-dimensional verification, it will simultaneously control the metabolite monitoring unit to perform high-frequency sampling on both lactate and glucose channels. The data fusion module ultimately outputs strictly time-synchronized micro-motion signals, dual-channel metabolite concentration curves, and corresponding structured annotations.

[0071] Scenario B: Dynamic switching of sensor operating modes based on behavioral context

[0072] Behavior recognition extension: The feature library supports the recognition of "exploratory rigidity" behavior.

[0073] Rules and Triggers: Pre-defined rules in the knowledge base:

[0074] Rule 3: IF Behavior = "Exploratory Stiffness" THEN Trigger {Target: Dopamine, Working Mode: "High Sensitivity Scanning Mode"}.

[0075] Execution: After parsing the instructions, the metabolite monitoring and control module sends a mode switching instruction to the metabolite monitoring unit, controlling it to switch from the basic monitoring mode to a preset advanced detection mode with higher sensitivity or specificity to capture weaker or more specific chemical signals. The operating mode switching can be achieved by changing the detection parameters or excitation signal of the metabolite monitoring unit. For example, in one embodiment, the detection sensitivity and specificity can be optimized for different target metabolites by changing the voltage excitation waveform applied to the electrochemical sensor (from a constant voltage to a scanning voltage) or switching the excitation wavelength of the optical sensor.

[0076] The above embodiments demonstrate that the system of the present invention, through its modular design and configurable knowledge base, can flexibly adapt to various research scenarios. In such multi-scenario applications, the model optimization module described in Embodiment 3 can simultaneously perform continuous learning and effect evaluation on newly added behavioral features and their corresponding triggering rules, making the system an intelligent research platform with good scalability and adaptability.

[0077] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Various changes made within the scope of knowledge possessed by those skilled in the art without departing from the concept of the present invention still fall within the scope of protection of the present invention.

Claims

1. An animal static behavior and metabolite synchronization monitoring system, characterized in that, include: The behavior perception module is used to collect animal behavior data in real time; The behavior recognition module is communicatively connected to the behavior perception module. It is used to receive the behavior data and identify specific static behavior patterns based on a predefined static behavior feature library to generate preliminary behavior event signals. An event triggering and verification module is communicatively connected to the behavior recognition module. The event triggering and verification module includes a behavior-metabolism association knowledge base, which stores rules that associate static behavior patterns with metabolite monitoring parameter adjustment strategies. The event triggering and verification module is used to receive the preliminary behavior event signal and perform multi-dimensional verification. After the verification is passed, a trigger control command is generated based on the rules that match the specific static behavior pattern. The metabolite monitoring and control module is communicatively connected to the event triggering and verification module and is used to control a metabolite monitoring unit to adjust its monitoring parameters according to the triggering control command. The data fusion module is communicatively connected to the behavior perception module and the metabolite monitoring unit, respectively, and is used to receive and synchronously fuse time-stamped behavior data and metabolite data to generate a related dataset.

2. The animal static behavior and metabolite synchronization monitoring system of claim 1, wherein, The multi-dimensional verification includes at least one of the following: (1) Time window verification: Determine whether the identified specific static behavior pattern exists stably within a preset duration; (2) Type consistency verification: Obtain data from an auxiliary perception module that is different from the behavior perception module, and determine whether the behavior type represented by the data of the auxiliary perception module is consistent with the behavior type represented by the preliminary behavior event signal; (3) Physiological state rationality verification: Based on the background physiological data or metabolite baseline trend provided by the metabolite monitoring unit, determine whether it is consistent with the physiological state expected by the specific static behavior pattern.

3. The animal static behavior and metabolite synchronization monitoring system of claim 1, wherein, The trigger control command includes a command for adjusting at least one of the following parameters of the metabolite monitoring unit: sampling frequency, operating mode, or target analyte type.

4. The animal static behavior and metabolite synchronization monitoring system of claim 2, wherein, The behavior perception module includes a millimeter-wave radar sensor for acquiring the animal's micro-motion signals; the behavior recognition module is configured to identify the specific static behavior pattern by analyzing the micro-Doppler feature spectrum of the micro-motion signals; and the auxiliary perception module includes a visual sensor for providing image data in the type consistency verification.

5. The animal ambulatory behavior and metabolite synchronization monitoring system of claim 2, wherein, The time window verification, type consistency verification, and physiological state rationality verification are executed according to a preset priority or sequence logic; the event triggering and verification module is configured to: only start the next level verification after the previous level verification is passed, otherwise terminate the verification process of the current preliminary behavioral event signal.

6. The animal ambulatory behavior and metabolite synchronization monitoring system of claim 2, wherein, The verification of the rationality of the physiological state includes: The event triggering and verification module is configured to establish a short-term prediction model for metabolite concentration or trend based on the background data provided by the metabolite monitoring unit. When the initial behavioral event signal occurs, the real-time monitored metabolite data is compared with the short-term prediction model; If the real-time data deviates from the predicted value by more than a preset threshold, the validity of the physiological state is deemed to have failed the verification.

7. The animal ambulatory behavior and metabolite synchronization monitoring system of claim 1, wherein, It also includes a model optimization module; The model optimization module is configured to: incrementally learn the static behavioral feature library in the behavior recognition module based on the verified static behavioral pattern fragments and their synchronized metabolite change data in the associated dataset, and / or evaluate and adjust the confidence of the rules in the behavior-metabolism association knowledge base.

8. The animal ambulatory behavior and metabolite synchronization monitoring system of claim 1, wherein, The data fusion module is also configured to automatically generate structured annotation information, including experimental stage, behavior category, and target metabolite, for the synchronously fused data segment based on the behavior type and metabolite target associated with the trigger control command.

9. The animal ambulatory behavior and metabolite synchronization monitoring system of claim 1, wherein, The metabolite monitoring unit is a wearable or implantable biosensor in an animal.

10. The animal ambulatory behavior and metabolite synchronization monitoring system of claim 1, wherein, The data fusion module is also configured to mark the corresponding verified static behavior type and event time information for each metabolite data segment driven by the trigger control command in the associated dataset.

Images