Method and system for preparing an ai-based dynamic mouse craniocerebral thoracoabdominal combined injury model

Abstract

The application discloses a kind of based on AI mouse cranium brain thoracic and abdominal complex injury dynamic model preparation method and system, belong to the technical field of biomechanics injury.The method comprises: impact is applied to mouse skull, the impact is conducted along organ synergistic conduction chain, to induce the cranium brain thoracic and abdominal complex injury of mouse;In the process of applying impact, multi-modal physiological data are synchronously collected;The multi-modal physiological data collected are input to AI analysis model, the injury mechanism of the cranium brain thoracic and abdominal complex injury is analyzed, and dynamic model is established based on analysis result.The technical scheme of the present application realizes the high simulation modeling of multiple organ cascade injury by constructing organ synergistic conduction chain from cranium to thoracic cavity;Combined with multi-modal physiological data synchronous acquisition and AI analysis model, the injury timing characteristics can be accurately extracted, the causal path of local injury triggering system response is established, and the accuracy and practicality of complex injury mechanism prediction are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of biomechanical injury, and in particular to a mouse craniocerebral thoracoabdominal combined injury dynamic model preparation method and system based on AI. BACKGROUND

[0002] In the field of biomechanical injury research, constructing a high-precision combined injury model is an important basis for studying the interaction mechanism of multiple organ injuries. Among them, craniocerebral thoracoabdominal combined injury, as a typical type of systemic trauma, its physiological process involves autonomic nervous dysfunction caused by brain tissue injury, respiratory regulation disorder, and cascade reaction caused by inflammatory factor release. The widely used models such as explosion shock wave method, pneumatic impact method, free fall method have significant controllability and repeatability problems. For example, if the impact angle deviation exceeds 2°, the difference in organ injury degree of mice can be as high as more than 30%, and the impact force error of the conventional mechanical driving method such as the Harvard PAPM model is as high as ± 15%, with poor repeatability and a coefficient of variation of up to 28%. At the same time, the traditional model is mostly a single-point sensing structure, which can only collect local pressure or motion signals, and it is difficult to realize the synchronous monitoring of cross-organ parameters such as electroencephalogram, blood oxygen, and thoracic pressure.

[0003] In terms of injury mechanism analysis, although the existing finite element simulation (such as the Johns Hopkins FEA model) can numerically simulate the tissue deformation caused by impact, it relies on simplified geometric modeling and cannot accurately predict microscopic injury processes such as axon rupture, and it cannot effectively integrate biochemical reaction information (such as changes in inflammatory factor IL-6). Although the multi-modal data fusion platform introduces a synchronous acquisition idea to realize the cross-modal alignment of electroencephalogram, pressure, blood oxygen, and other data, and preliminarily establishes the injury causal chain (such as "craniocerebral injury-autonomic nervous disorder-thoracic pressure change"), it still has the problem of insufficient biological adaptation. For example, the industrial-level sampling period cannot meet the millisecond-level requirement of physiological signals, and there is a lack of dynamic calibration mechanism for sensor drift, physiological interference, and other problems. Therefore, an integrated small animal combined injury modeling method with impact controllability, multi-organ coupling response capability, cross-modal data synchronization, and AI mechanism analysis capability is urgently needed to improve the physiological authenticity and prediction ability of the experimental model. SUMMARY

[0004] The present application provides a mouse craniocerebral thoracoabdominal combined injury dynamic model preparation method and system based on AI, which realizes multi-organ collaborative injury modeling and AI mechanism prediction, improves injury reproduction accuracy and physiological reasoning ability, and enhances model extrapolation reliability.

[0005] According to a first aspect of the present application, a mouse craniocerebral thoracoabdominal combined injury dynamic model preparation method based on AI is provided, which comprises:

[0006] applying an impact to a skull of a mouse, and conducting the impact along an organ synergistic conduction chain to induce a craniocerebral thoracoabdominal combined injury of the mouse, the organ synergistic conduction chain being conducted to a thoracic cavity in sequence through a skull, cerebrospinal fluid, a spinal cord and a diaphragm;

[0007] synchronously collecting multi-modal physiological data including any one or more of electroencephalogram signals, thoracic cavity pressure signals and blood oxygen signals during the impact application;

[0008] inputting the collected multi-modal physiological data into an AI analysis model, analyzing an injury mechanism of the craniocerebral thoracoabdominal combined injury, and establishing a dynamic model based on an analysis result.

[0009] In an embodiment, the applying an impact to a skull of a mouse, and conducting the impact along an organ synergistic conduction chain includes:

[0010] controlling an electromagnetic coil to excite a pulsed magnetic field, and synchronously driving a pneumatic cavity to release compressed gas to generate a directional impact;

[0011] adjusting an output direction of the impact in real time according to a gyroscope feedback signal to make the impact conducted along the organ synergistic conduction chain.

[0012] In an embodiment, the synchronously collecting multi-modal physiological data including any one or more of electroencephalogram signals, thoracic cavity pressure signals and blood oxygen signals includes:

[0013] acquiring, in real time, intracerebral pressure changes, thoracic cavity gas pressure fluctuations and blood oxygen concentration signals through a flexible implantable sensor array;

[0014] performing cross-channel synchronous processing on the brain, thoracic cavity and blood data using a millisecond-level timestamp protocol to form multi-modal physiological data with a unified time reference.

[0015] In an embodiment, the acquiring, in real time, intracerebral pressure changes, thoracic cavity gas pressure fluctuations through a flexible implantable sensor array includes:

[0016] implanting a nano probe based on a principle of plasma resonance into a mouse brain parenchyma to acquire intracerebral pressure signals reflecting local tissue elasticity changes;

[0017] deploying a fiber-optic sensing network in a thoracoabdominal region to collect stress responses and thoracic cavity gas pressure fluctuations in a three-dimensional space.

[0018] In an embodiment, the inputting the collected multi-modal physiological data into an AI analysis model includes:

[0019] extracting timing change features in electroencephalogram signals, thoracic cavity pressure signals and blood oxygen signals.

[0020] According to the timing variation feature, a multi-dimensional feature vector representing an inter-organ response relationship is constructed;

[0021] Based on the multi-dimensional feature vector, an AI analysis model is used to generate a causal correlation output for predicting the evolution of the injury mechanism.

[0022] In one embodiment, further comprising:

[0023] The causal correlation output includes a dynamic path diagram starting from local shear stress changes in the craniocerebral region, and sequentially relating inflammatory factor release, autonomic nervous dysfunction, and thoracic pressure regulation disorder, and is used to assist in establishing a mathematical model of the cross-organ injury propagation mechanism.

[0024] According to a second aspect of the present application, a mouse craniocerebral thoraco-abdominal combined injury dynamic model preparation system based on AI is provided, comprising:

[0025] An impact module is configured to apply an impact to the skull of a mouse, so that the impact is conducted along an organ synergistic conduction chain to induce a craniocerebral thoraco-abdominal combined injury in the mouse, and the organ synergistic conduction chain is sequentially through the skull, cerebrospinal fluid, spinal cord, and diaphragm to the thoracic cavity;

[0026] A collection module is configured to synchronously collect multi-modal physiological data during the application of the impact, and the multi-modal physiological data includes any one or more of electroencephalogram signals, thoracic cavity pressure signals, and blood oxygen signals;

[0027] An analysis module is configured to input the collected multi-modal physiological data into an AI analysis model, analyze the injury mechanism of the craniocerebral thoraco-abdominal combined injury, and establish a dynamic model based on the analysis result.

[0028] In one embodiment, the impact module, the collection module, and the analysis module are controlled to implement any of the above-described mouse craniocerebral thoraco-abdominal combined injury dynamic model preparation methods based on AI.

[0029] According to a third aspect of the present application, an electronic device is provided, comprising a communication interface, a processor, and a memory;

[0030] The memory is configured to store program instructions, and the program instructions, when executed by the processor in communication connection with the memory through the communication interface, implement any of the above-described mouse craniocerebral thoraco-abdominal combined injury dynamic model preparation methods based on AI.

[0031] According to a fourth aspect of the present application, a computer readable storage medium is provided, and the computer readable storage medium stores computer program instructions, and the computer program instructions are executed by a computer (for example, a processor in the computer) to implement any of the above-mentioned AI-based mouse cranial thoraco-abdominal combined injury dynamic model preparation methods.

[0032] To sum up, the present application provides an AI-based mouse cranial thoraco-abdominal combined injury dynamic model preparation method and system, which comprises: applying an impact to the skull of a mouse, and making the impact conduct along an organ synergistic conduction chain to induce cranial thoraco-abdominal combined injury of the mouse, the organ synergistic conduction chain being sequentially conducted through the skull, cerebrospinal fluid, spinal cord and diaphragm to the thoracic cavity; in the process of applying the impact, synchronously collecting multi-modal physiological data, the multi-modal physiological data comprising any one or more of electroencephalogram signals, thoracic cavity pressure signals and blood oxygen signals; inputting the collected multi-modal physiological data into an AI analysis model, analyzing the injury mechanism of the cranial thoraco-abdominal combined injury, and establishing a dynamic model based on the analysis result. The technical solution of the present application realizes the organ synergistic injury path from the skull to the thoracic cavity through the explosion three-dimensional vector impact mechanism, solves the problems of uncontrollable impact and organ response fragmentation in the existing combined injury model; introduces a multi-modal physiological sensing array and a synchronous collection mechanism, improves the alignment accuracy of cross-organ data; fuses an AI analysis model, can realize dynamic modeling of the causal chain between local injury and systemic physiological response, significantly enhances the accuracy and interpretability of injury mechanism prediction, and improves the extrapolation value of the animal model in clinical mechanism research.

[0033] Other features and advantages of the present application will be set forth in the following description, and in part will become apparent to those skilled in the art from the description, or can be learned by practice of the present application. The objects and other advantages of the present application will be realized and attained by the structure particularly pointed out in the written description and claims.

[0034] The technical solutions of the present application will be further described in detail below with the aid of the accompanying drawings and examples. BRIEF DESCRIPTION OF DRAWINGS

[0035] In order to more clearly illustrate the specific embodiments of the present application or the technical solutions in the prior art, the drawings needed in the specific embodiments or prior art description will be briefly introduced below. Obviously, the drawings in the following description are some embodiments of the present application, and those skilled in the art can also obtain other drawings according to these drawings without creative labor.

[0036] Figure 1 A flowchart of an AI-based mouse cranial thoraco-abdominal combined injury dynamic model preparation method provided for an embodiment of the present application;

[0037] Figure 2 A flow chart of another AI-based preparation method of a mouse craniocerebral thoracoabdominal combined injury dynamic model is provided for an embodiment of the present application.

[0038] Figure 3 A flow chart of another AI-based preparation method of a mouse craniocerebral thoracoabdominal combined injury dynamic model is provided for an embodiment of the present application.

[0039] Figure 4 A flow chart of another AI-based preparation method of a mouse craniocerebral thoracoabdominal combined injury dynamic model is provided for an embodiment of the present application.

[0040] Figure 5 A flow chart of another AI-based preparation method of a mouse craniocerebral thoracoabdominal combined injury dynamic model is provided for an embodiment of the present application.

[0041] Figure 6 A structural diagram of an AI-based preparation system of a mouse craniocerebral thoracoabdominal combined injury dynamic model is provided for an embodiment of the present application.

[0042] Figure 7 A structural diagram of an electronic device is provided for an embodiment of the present application. DETAILED DESCRIPTION

[0043] The features and exemplary embodiments of various aspects of the present application will be described in detail below with reference to the drawings. The following detailed description is merely intended to explain the present application, and is not intended to limit the present application. The present application can be implemented without some of the specific details, which are well known to those skilled in the art. The following description of the embodiments is merely intended to provide a better understanding of the present application by showing examples of the present application.

[0044] It should be noted that, in this document, relational terms such as first and second, and the like, are used solely to distinguish one entity or action from another entity or action, without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises", "comprising", or any other variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without more limitations, an element preceded by "comprises... " does not, without more limitations, foreclose the existence of

[0045] As Figure 1As shown, the application provides a preparation method of an AI-based mouse craniocerebral thoracoabdominal combined injury dynamic model, which comprises the following steps:

[0046] In step S11, an impact is applied to the mouse skull, and the impact is conducted along an organ synergistic conduction chain to induce a craniocerebral thoracoabdominal combined injury of the mouse, the organ synergistic conduction chain being sequentially conducted through the skull, cerebrospinal fluid, spinal cord and diaphragm to the thoracic cavity;

[0047] In step S12, during the application of the impact, multi-modal physiological data are synchronously collected, the multi-modal physiological data including any one or more of electroencephalogram signals, thoracic cavity pressure signals and blood oxygen signals;

[0048] In step S13, the collected multi-modal physiological data are input into an AI analysis model, an injury mechanism of the craniocerebral thoracoabdominal combined injury is analyzed, and a dynamic model is established based on the analysis result.

[0049] In one embodiment, the existing small animal combined injury model has significant limitations in preparation accuracy and system synergy, mainly in three aspects. First, in the impact parameter control aspect, traditional models such as free-fall or explosion impact devices generally rely on mechanical open-loop control and cannot achieve accurate adjustment of impact direction and energy. Experiments have shown that an impact angle deviation of more than 2° may cause a difference in multi-organ injury of more than 30%, seriously affecting the consistency of experimental results. Taking the typical Harvard PAPM model as an example, the impact force error is as high as ± 15%, resulting in poor repeatability of organ injury (coefficient of variation CV is 28%). Second, the existing models generally lack the integrity of the organ synergistic conduction chain, often taking a single site as the impact target (such as the CCI craniocerebral model or the thoracoabdominal compression method), and failing to build a linkage path from the skull-cerebrospinal fluid-spinal cord-diaphragm-thoracic cavity, making it difficult to simulate the brain impact-induced autonomic nervous disorder and then trigger the cascade reaction of thoracic pressure changes. Third, the existing system generally uses static single-point setting for animal anesthesia control, without building an anesthesia depth regulation mechanism based on electroencephalogram feedback, resulting in distorted physiological responses of key organs (such as myocardium), and the inhibition rate may even exceed 40%.

[0050] At the mechanism modeling level, existing prediction models are still dominated by static geometric modeling and single-modal data-driven, resulting in limited prediction accuracy across scales and species. For example, the typical finite element analysis (FEA) method often uses simplified geometric bodies such as spheres or cylinders to replace complex brain tissue structures, making it difficult to capture the axon breakage induced by shear stress in high gradient areas such as sulci and gyri, resulting in an error of more than 40% in micro-damage prediction. At the same time, existing platforms do not time-series fuse impact mechanics parameters with biochemical reactions (such as the expression of inflammatory factors IL-6, TNF-α, etc.), resulting in the interruption of the causal chain of "mechanical impact-cellular pathology-organ disorder". In addition, existing systems have not established a parameter adaptive mechanism for species extrapolation. There are significant differences between mice and humans in key indicators such as brain blood flow regulation and physiological threshold (such as the mouse brain blood flow regulation threshold is 22% higher than that of humans). If there is a lack of migration calibration and dynamic calibration, the failure rate of model clinical transformation will increase significantly, and more than 80% of extrapolation tests do not work well in actual human scenarios.

[0051] The main technical shortcomings of existing models can be summarized in four dimensions. First, the impact parameter control accuracy is insufficient, which is due to the lack of closed-loop calibration capability of the driving system; second, the organ collaborative modeling capability is weak, which is due to the lack of structure in the conduction path; third, the AI damage recognition accuracy is low, which is limited by the traditional single-modal CNN model which is difficult to identify deep organ damage; fourth, the mechanism prediction accuracy is biased, which is rooted in the simplification of biomechanical modeling and the fragmentation of multi-omics information.

[0052] In current experimental animal model research, it is of great significance to construct a complex injury model that can controllably simulate the mechanism of combined injury of craniocerebral, thoracic and abdominal organs. By applying impact to the mouse skull, the conduction of impact along the organ collaborative conduction chain is established as the starting point, establishing the path dependence and structural collaboration of injury application. Traditional models rely on a single impact method (such as pneumatic or free fall), resulting in a lack of controllability in the force conduction process. Deviation in impact direction or angle may cause inconsistent responses between organs, affecting experimental repeatability and the effectiveness of mechanism derivation. By limiting the initial point of impact to the skull and clearly defining the conduction chain as including cerebrospinal fluid, spinal cord, diaphragm, and thoracic cavity.

[0053] Furthermore, unlike the physical separation of "thoracic and abdominal / craniocerebral separation impact" in the PAPM model of Harvard Medical School, the three-dimensional vector impact mechanism simulates the complete process of actual external force transmission from the head to the thorax and abdomen. For example, concussions are often accompanied by sudden changes in intrathoracic pressure, and there is a chain linkage in time and response between the two. By setting specific impact directions and structural constraints, this linkage is truly reproduced in the model. For example, when impact is applied from the skull, cerebrospinal fluid acts as a buffer medium to transmit pressure to the spinal cord, which in turn affects the thoracic volume through the diaphragm. This path can trigger abnormal autonomic nervous regulation mechanisms, leading to respiratory dysfunction.

[0054] In this structured impact process, multi-modal physiological data is synchronously collected. The multi-modal signals, including electroencephalogram (EEG), thoracic pressure fluctuation, and blood oxygen concentration, are the main indicators reflecting the interaction between the central nervous system, respiratory system, and oxygenation state. In contrast, traditional physiological monitoring often uses single-channel recording methods (such as single-point chest pressure gauges), which cannot reveal the dynamic correlation across organs. By implanting flexible sensor arrays in mice and simultaneously collecting data from different parts and physiological dimensions, and using millisecond-level timestamp protocols for cross-channel alignment, various types of time series signals can be mapped onto a standardized response time axis. Taking one impact experiment as an example, the system can accurately determine that the EEG high-frequency oscillation occurs at 120 ms after impact, the thoracic pressure rises at about 350 ms, and the blood oxygen decreases at 600 ms, thereby establishing a three-stage physiological cascade chain of "nervous system, respiratory system, and oxygenation."

[0055] The multi-modal data is processed through an artificial intelligence analysis model to convert into causal outputs at the mechanism level. The introduction of AI analysis models compensates for the limitations of traditional biostatistical methods in dealing with multi-variable, non-linear, and time-coupled problems. AI analysis models not only extract feature parameters from single-source signals such as EEG, thoracic pressure, and blood oxygen, but also fuse them into multi-dimensional interaction vectors between organs, thereby identifying the potential pathways of thoracic pressure changes induced by brain injury. For example, the model can learn the chain-like causal structure of "shear stress peak - release of inflammatory factors (such as IL-6) - autonomic nervous instability - diaphragmatic muscle reflex abnormalities - thoracic pressure changes." This structure not only comes from data fitting but also integrates known physiological knowledge graphs, making it highly interpretable. In addition, through AI analysis models, quantifiable injury risk scores, organ function deterioration trend prediction curves, and other clinically convertible indicators can be output.

[0056] The causal chain output generated based on the AI analysis model can also be further converted into a complete dynamic model. Unlike static injury determination, dynamic models focus on the development process of injury, the response timing between organs, and feedback loops. In practical applications, the model can predict whether a certain type of shear stress input will induce a surge in serum inflammatory factors after 6 hours or cause a decrease in lung interstitial elasticity within 12 hours. In addition, this dynamic model also provides a basis for cross-species extrapolation, and subsequent human trauma databases can be accessed to modify model parameters through transfer learning, improving its interpretability and reliability in clinical research.

[0057] This embodiment proposes a complex injury dynamic simulation system that integrates structure control, data-driven and intelligent modeling. It mainly includes a high-precision repeatable injury preparation module, a multi-modal driven mechanism prediction platform, and a cross-species adaptive mechanism migration framework. In terms of model construction, a bionic multi-directional force generator driven by electromagnetic-pneumatic composite is used to realize directional impact on the mouse skull. The impact realizes angle closed-loop control through a gyroscope feedback system, with an accuracy of ±0.1°, which can effectively avoid the problem of organ injury difference (more than 30%) caused by impact deviation exceeding 2° in traditional free impact devices. To simulate the multi-organ coupling injury path, an organ cooperative conduction chain structure from skull-cerebrospinal fluid-spinal cord-diaphragm-thoracic cavity is designed, and the head-thorax linkage type cascade injury is realized by referring to the mixed respiratory mechanics mechanism of chest and abdomen. In addition, an electroencephalogram (EEG) signal monitoring module is integrated in the animal experiment to dynamically adjust the concentration of anesthetics (such as isoflurane), forming a closed loop of anesthesia-physiological state coupling, thereby avoiding the interference of anesthesia suppression on the physiological response of key organs (such as myocardium), and improving the stability and biological authenticity of the model output.

[0058] In terms of mechanism modeling, the platform realizes high-resolution tracking of systemic physiological evolution after local injury through a multi-modal signal acquisition system (including EEG, thoracic pressure, blood oxygen, etc.), and constructs a cross-scale biomechanics simulation engine. The simulation system reconstructs complex structures such as brain gyri and lung alveoli based on μCT images at the macro level, avoiding the simplification error of organ geometry in traditional finite element models; at the micro level, molecular dynamics modeling technology is introduced to simulate the process of axon microtubule fracture induced by impact stress. To establish the dynamic relationship between physical stress and biochemical reaction, the platform introduces multi-omics coupling mechanism to map the shock wave parameters and the concentration changes of real-time acquired inflammatory factors (such as IL-6), forming a causal mathematical model of "mechanical stress-mitochondrial rupture-inflammation factor storm-multiple organ failure", which improves the prediction depth and biological explanation of injury mechanism.

[0059] To solve the problem of physiological differences between species in the clinical extrapolation of existing animal models, a cross-species mechanism transfer framework is further designed. By importing human trauma databases (such as cases of craniocerebral injury combined with chest and abdominal injury), the system compares and analyzes the structural differences between mice and humans in key physiological parameters such as brain blood flow regulation threshold and chest elasticity. The transfer learning algorithm is used to dynamically adjust the mouse model parameters, which can effectively eliminate the bias between species in response threshold, inflammation speed, and organ compensatory capacity, making the model prediction results have higher clinical translation reliability. According to experimental verification, the mechanism prediction error of this framework has been compressed from more than 40% of the traditional method to within 15%, which has a significant advantage in the clinical mechanism extrapolation of animal experiment results. In summary, the present application constructs a complete "controllable impact-organ response-AI mechanism modeling-clinical extrapolation" technical path from mechanical structure control, signal linkage modeling to cross-species correction, breaking through the key technical bottlenecks of existing models in precision, synergy and conversion rate.

[0060] By integrating multi-modal technology, cross-organ collaborative mechanism and dynamic species adaptation, the present application has achieved a leap in injury preparation accuracy, mechanism prediction depth and clinical translation reliability. The horizontal comparison with the prior art is shown in Table 1 below.

[0061] Table 1

[0062]

[0063] To verify the feasibility and performance advantage of the AI-based mouse craniocerebral-thoracoabdominal complex injury dynamic model preparation system proposed by the present application, systematic animal experiments and model control tests were carried out. The experimental subjects were 50 C57BL / 6 experimental mice. The experimental group used the electromagnetic-pneumatic composite driving impact system proposed by the present application to construct a complex injury model with an organ collaborative conduction chain of "skull→cerebrospinal fluid→spinal cord→diaphragm→thoracic cavity". The control group used the Harvard PAPM model (pure mechanical impact). In the experimental group, a three-dimensional vector impact was applied, with a craniocerebral lateral impact angle of 30°±0.1° and a thoracoabdominal area impact pressure of 50 kPa. A multi-modal sensor array was implanted in each mouse to record the changes in brain parenchymal shear force, thoracic pressure response and physiological signal synchronous data in real time. The data were quantitatively analyzed by referring to the evaluation method of the improved compression injury model, and the specific data are shown in Table 2.

[0064] Table 2

[0065]

[0066] To ensure the physiological stability of experimental animals during the injury application process, an anesthesia-physiological coupling control module is integrated. Based on the dynamic changes of mouse electroencephalogram (EEG) signals, the concentration of inhaled anesthetic isoflurane is adjusted in real time to maintain a stable anesthesia plane and minimize interference with myocardial and circulatory systems. Experimental results show that after using the anesthesia control strategy of the application, the inhibition rate of myocardial contractile function of mice is significantly reduced from >40% commonly seen in traditional models to <8% (p<0.01), effectively eliminating the physiological response distortion caused by excessive or insufficient anesthesia, and providing a guarantee for the stability and reproducibility of injury results.

[0067] In terms of multi-modal sensor array performance, the experiment uses nano-probes based on plasmonic resonance principles to monitor the change in elastic modulus of pulmonary interstitial tissue, with an accuracy of 0.1 kPa; and combines micro-biochips to capture the concentration change of inflammatory factor IL-6 in blood, with a sampling frequency of 10 Hz. Experimental results show that the system has an accuracy of 96.3% in identifying intracranial hemorrhage (hematoma), which is significantly better than the 65% of traditional MIT systems; in terms of lung contusion identification, the minimum blood lesion size that can be detected is 0.2 mm, while the resolution limit of traditional MRI equipment under similar conditions is 2 mm, showing the significant advantage of the system in deep injury identification accuracy.

[0068] In terms of artificial intelligence analysis, a dual-channel AI analysis model is constructed, which inputs the CT image data of mice and structured medical text semantics (such as "axon rupture → IL-6 increase") at the same time, fusing image features and physiological semantics. Through the CrossAttention mechanism, multiple source features are fused, and the composite injury score and injury grade prediction results are output. Test results show that the accuracy of deep injury grading of this model reaches 92.7%, while the accuracy of traditional CNN model without semantic information fusion is only 78.4%. In addition, the model successfully associates the peak value of craniocerebral shear force (>15 kPa) with the rising trend of serum IL-6 concentration after 6 hours, with a correlation coefficient as high as r=0.91 (p<0.001), verifying the modeling ability of the physiological causal chain "local shear injury → systemic inflammatory response".

[0069] For cross-species extrapolation, the system introduces a dynamic transfer learning mechanism to correct the mouse model parameters. In the experiment, the case data with "craniocerebral injury combined with abdominal cavity hemorrhage" in the human traffic accident database is called to extract the physiological response characteristics for model calibration. The model automatically completes the structure mapping and response adjustment according to the differences between humans and mice in the brain blood flow regulation threshold (mice are relatively higher). Taking the brain blood flow regulation threshold as an example, the system reduces the related parameters of the mouse model by about 22%, and the key indicators such as the inflammatory response time window and the change of carbon monoxide hemoglobin level verify the feasibility and accuracy of cross-species mechanism transfer. The specific data is shown in Table 3.

[0070] Table 3

[0071]

[0072] The technical solution in this embodiment realizes the organ coordination injury path from the skull to the chest through the bionic three-dimensional impact mechanism, solves the problem of uncontrollable impact and organ response fragmentation in the existing composite injury model; introduces a multi-modal physiological sensing array and a synchronous acquisition mechanism to improve the alignment accuracy of cross-organ data; and fuses an AI analysis model to dynamically model the causal chain between local injury and systemic physiological response, significantly enhancing the accuracy and interpretability of injury mechanism prediction and improving the extrapolation value of animal models in clinical mechanism research.

[0073] In one embodiment, as shown in Figure 2 Step S11 includes steps S21-S22 as follows:

[0074] In step S21, the electromagnetic coil is controlled to generate a pulsed magnetic field, and the pneumatic cavity is driven to release compressed gas synchronously to generate a directional impact;

[0075] In step S22, the output direction of the impact is adjusted in real time according to the gyroscope feedback signal, so that the impact is conducted along the organ coordination conduction chain.

[0076] In one embodiment, the impact is applied to the mouse skull, and the process of conducting the impact along the organ synergistic conduction chain substantially embodies a multi-source mechanical excitation mechanism with direction control capability, the core of which is to realize the cascade mechanical conduction from the skull to the downstream multiple organs under the premise of ensuring the precise consistency of the impact angle, through a structure-controllable energy input method. A composite mechanical impact source is constructed by controlling the electromagnetic coil to excite the pulsed magnetic field and synchronously driving the pneumatic cavity to release compressed gas, that is, the directional shock wave with three-dimensional vector characteristics is generated by using electromagnetic driving to realize fast response and initial direction positioning, and combining pneumatic release to enhance the impact amplitude and effective energy transmission range. Compared with the traditional single force source mechanical impact method, the design significantly improves the flexibility and simulation of impact output through the dual-channel cooperative driving mechanism of electromagnetic and pneumatic combination, and is especially suitable for simulating the damage process of multiple directions and multiple organs in complex injury scenes such as explosion or collision.

[0077] The closed-loop calibration mechanism of adjusting the output direction of the impact in real time according to the gyroscope feedback signal solves the problems of poor experimental repeatability and large damage error caused by unstable impact direction or uncontrollable deviation in the prior art. In actual operation, the size of the mouse individual is small and the structure is precise, and even if the impact angle deviates by 1-2 degrees, it may cause the impact path to deviate from the originally set organ synergistic conduction chain of “skull-cerebrospinal fluid-spinal cord-diaphragm-thoracic cavity”, ultimately leading to significant differences in the type and severity of damage between different experimental individuals. Through the gyroscope feedback device, the angular deviation between the current driving device and the spatial posture of the mouse head can be measured in real time before the impact is applied, and the system adjusts accordingly to ensure that the impact is applied along the set conduction chain direction, effectively improving the consistency and reproducibility of the model preparation.

[0078] For example, when constructing a model of brain injury linked thoracic pressure change, if the impact direction deviates slightly downward, the neck muscle may be injured, inducing non-target injury response; if it deviates upward, the brain tissue may not be stressed enough to effectively trigger the subsequent physiological causal chain (such as abnormal thoracic pressure caused by abnormal autonomic nervous system). The design can control the impact direction to within ±0.1° by introducing a gyroscope in real time, thereby ensuring that the impact force is always conducted along the established organ synergistic path from the skull, accurately reproducing the cascade injury process of “skull-brain-spinal cord-thorax”. The controllability of the mechanical path not only improves the biological authenticity of the complex injury model, but also provides a stable and consistent basis for subsequent multi-modal data acquisition, AI mechanism modeling, and physiological causal chain analysis.

[0079] In one embodiment, as shown in Figure 3 Steps S12 include steps S31-S32 as follows:

[0080] In step S31, the brain pressure changes, chest cavity gas pressure fluctuations and blood oxygen concentration signals are acquired in real time by a flexible implantable sensor array;

[0081] In step S32, the brain, chest and blood data are cross-channel synchronized using a millisecond-level timestamp protocol to form multi-modal physiological data with a unified time reference.

[0082] In one embodiment, for data acquisition in a mouse craniocerebral thoracic abdominal complex injury model, multi-modal physiological data acquisition integrating flexible implantable sensing and high-precision time synchronization mechanism can ensure the timing consistency of spatial multi-source signals and the high-precision restoration of physiological signals. By arranging a flexible implantable sensor array, brain pressure, chest cavity pressure and blood oxygen concentration signals are collected to realize structured, continuous and real-time perception of the physiological response of the injured individual among the "craniocerebral-thoracic-blood" multi-system. Among them, the brain pressure reflects the intracranial stress level, the chest cavity gas pressure reflects the respiratory mechanics change, and the blood oxygen concentration directly reflects the systemic oxygen supply imbalance. Compared with the traditional independent channel recording method, this sensor array layout improves the spatiotemporal coverage capability of the signal source, especially in the dynamic injury induction stage, the fine differences in the coordinated response of each organ can be captured at the millisecond level. By introducing a millisecond-level timestamp protocol, the data streams of cross-site and cross-sensor channels are accurately aligned to construct a multi-modal data set under a unified time reference. For example, if the brain pressure peak appears earlier than the chest cavity pressure change and blood oxygen decline, it may indicate that the injury chain starts from the central nervous system; on the contrary, if the chest cavity pressure fluctuation precedes the brain electrical abnormality, it may point to the reverse impact mechanism. Due to the sampling clock drift or physical isolation of data channels, the traditional acquisition system often has signal misplacement, which causes incomplete or even misinterpretation of the causal chain in mechanism modeling. Through the millisecond unified time reference, the time logical relationship between various physiological signals is ensured to be comparable, greatly improving the AI analysis model's ability to distinguish complex physiological pathways. Not only does it achieve concurrent perception of multiple organs / multiple indicators in data acquisition means, but it also embodies the engineering design advantages of high precision, high stability and high integration in time sequence processing strategy, providing a data foundation for subsequent AI-assisted modeling, physiological mechanism reconstruction and cross-species migration applications.

[0083] In one embodiment, as shown in Figure 4 Step S31 includes steps S41-S42 as follows:

[0084] In step S41, a nano probe based on plasma resonance principle is implanted in the mouse brain parenchyma to acquire brain pressure signals reflecting local tissue elasticity changes;

[0085] In step S42, a fiber-optic sensing network is arranged in the thoracic and abdominal regions to collect stress response and chest cavity gas pressure fluctuations in three-dimensional space.

[0086] In one embodiment, to solve the problems of precision and spatial resolution of physiological signal acquisition in mouse complex injury modeling, a dual-region monitoring mechanism based on implantable flexible sensor array is proposed, covering the brain parenchyma and the chest and abdominal regions. Through the combination of micro-scale resolution and three-dimensional stress detection, the quantification accuracy and spatial resolution of the injury state are improved. Specifically, by implanting a nano probe based on the principle of plasmonic resonance in the mouse brain parenchyma, the microstructure elastic modulus change of the local brain tissue under the impact can be monitored in real time, which is particularly sensitive in the early stage of nerve injury and can reflect early micro-injury states such as loose cell structure, fluid leakage or axon rupture. The plasmonic resonance nano probe has ultra-high sensitivity and tissue compatibility, and can obtain high-precision intracerebral pressure dynamic curve under very low energy intervention, avoiding secondary injury or signal distortion caused by the mechanical rigidity of traditional sensors.

[0087] The optical fiber sensor network arranged in the chest and abdominal region is used to sense the three-dimensional stress field distribution and the gas pressure fluctuation in the chest cavity in real time. The optical fiber sensor has high spatial resolution and anti-electromagnetic interference capability, and is suitable for arrangement in physiological regions with dynamic strain. Especially when simulating the phenomena of chest cavity respiratory motion disorder, lung tissue collapse or abdominal cavity pressure rise induced by impact, the optical fiber network can accurately feedback the physical characteristics such as stress concentration area, fluctuation frequency and amplitude change. The collected data can form a spatial distribution vector field, which, combined with the intracerebral sensing data, realizes the cross-regional synchronous tracking from local neural stress to trunk stress response, so as to analyze the spatial and temporal correlation characteristics of the injury force conduction path and organ response law. The integrated sensing system solves the problems of brain signal ambiguity, chest cavity data loss and cross-regional linkage in traditional models.

[0088] In one embodiment, as shown in Figure 5 Step S13 includes steps S51-S53 as follows:

[0089] In step S51, the time sequence change characteristics in the electroencephalogram signal, the chest cavity pressure signal and the blood oxygen signal are extracted;

[0090] In step S52, a multi-dimensional feature vector representing the response relationship between organs is constructed according to the time sequence change characteristics;

[0091] In step S53, a causal association output for predicting the evolution of injury mechanism is generated by an AI analysis model based on the multi-dimensional feature vector.

[0092] In one embodiment, the key to AI-driven pathological mechanism facing mouse complex injury research lies in constructing response relationship vectors across organs through the temporal characteristics of multi-modal physiological data, and realizing the causal reconstruction of pathological process based on this. Through the synchronous acquisition of electroencephalogram signals, thoracic pressure signals and blood oxygen signals, the trends of various physiological indicators over time after impact injury induction are obtained, and then the sequence and correlation strength of the dynamic response between organs are revealed. For example, if the electroencephalogram signal rate decreases synchronously first, followed by an increase in the amplitude of thoracic pressure fluctuations, and the blood oxygen signal decreases suddenly after the two, it can be inferred that brain function impairment may be an upstream factor leading to thoracic pressure regulation disorders and systemic hypoxia. A multi-dimensional feature vector representing the functional synergy between organs is also constructed. By differentiating the time series data, converting it to the frequency domain, fitting the trend, and other operations, the synergy parameters (such as time delay, response amplitude ratio, phase synchronization coefficient, etc.) between the brain-thorax-blood systems are extracted, and these physiological covariates are combined into a vectorized structure and input into the AI analysis model. The model can use the graph attention mechanism (Graph Attention) or the Transformer structure in neural networks to highlight the relative weight and interaction mode of each organ in the overall injury response chain. Taking the brain-thorax as an example, if multiple experiments show that there is a stable delay relationship (such as 300 ms) between the peak shear stress and the increase in thoracic pressure, the model can adaptively strengthen the weight of this causal pathway and embed it into the overall mechanism map.

[0093] The final output of the AI analysis model is not only the prediction of the injury trend (such as the future change of a certain indicator value), but also the generation of a cross-organ causal relationship chain, expressed in the form of a path diagram. For example, the model can identify a chain reaction from "local increase in cranial shear stress" to "increase in IL-6 and other inflammatory factor concentrations", "decrease in HRV parameters of the autonomic nervous system", and "imbalance in thoracic pressure regulation", and express its dynamic evolution path in a graph structure. Thus, the complex injury research is transformed from the traditional "injury-outcome" research paradigm to the "force-response-intervention" mechanism reasoning framework, improving the interpretability and intervenability of injury prediction.

[0094] By combining the path diagram output by the AI analysis model with the physiological coupling equations between organs, a differential expression model with time / space as variables is formed, which theoretically describes how the injury spreads from the nervous system to the respiratory system and further affects the function of the circulatory system. For example, based on the rate of change of brain shear stress, an equation for the release rate of inflammatory factors can be constructed, and then a model of thoracic gas exchange parameter changes is established to realize mechanism cascade simulation. This mathematical model not only helps to simulate and optimize under animal experiment conditions, but also can adjust the parameter structure in future cross-species migration, supporting the extrapolation modeling of trauma mechanisms in other species.

[0095] In one embodiment, Figure 6 is a system block diagram of an AI-based mouse cranial thoracic and abdominal complex injury dynamic model preparation system according to an exemplary embodiment. As shown in the figure, Figure 6 The AI-based mouse cranial thoracic and abdominal complex injury dynamic model preparation system includes an impact module 61, a collection module 62 and an analysis module 63.

[0096] The impact module 61 is configured to apply an impact to the mouse skull, and the impact is conducted along an organ synergistic conduction chain to induce cranial thoracic and abdominal complex injury in the mouse, wherein the organ synergistic conduction chain is sequentially conducted through the skull, cerebrospinal fluid, spinal cord and diaphragm to the thoracic cavity.

[0097] The collection module 62 is configured to synchronously collect multi-modal physiological data during the impact, wherein the multi-modal physiological data includes any one or more of electroencephalogram signals, thoracic cavity pressure signals and blood oxygen signals.

[0098] The analysis module 63 is configured to input the collected multi-modal physiological data into an AI analysis model, analyze the injury mechanism of the cranial thoracic and abdominal complex injury, and establish a dynamic model based on the analysis result.

[0099] The impact module 61, the collection module 62 and the analysis module 63 included in the system block diagram of the AI-based mouse cranial thoracic and abdominal complex injury dynamic model preparation system are controlled to perform the AI-based mouse cranial thoracic and abdominal complex injury dynamic model preparation method described in any of the above embodiments.

[0100] As shown in the figure, Figure 7 The present application provides an electronic device 700, which comprises a communication interface, a processor 701 and a memory 702.

[0101] The memory 702 is configured to store program instructions, and when the program instructions are executed by the processor 701 which is in communication connection with the memory 702 through the communication interface, an impact is applied to the mouse skull, and the impact is conducted along an organ synergistic conduction chain to induce cranial thoracic and abdominal complex injury in the mouse, wherein the organ synergistic conduction chain is sequentially conducted through the skull, cerebrospinal fluid, spinal cord and diaphragm to the thoracic cavity; during the impact, multi-modal physiological data is synchronously collected, wherein the multi-modal physiological data includes any one or more of electroencephalogram signals, thoracic cavity pressure signals and blood oxygen signals; the collected multi-modal physiological data is input into an AI analysis model, the injury mechanism of the cranial thoracic and abdominal complex injury is analyzed, and a dynamic model is established based on the analysis result.

[0102] The application provides a computer readable storage medium, and computer program instructions are stored on the computer readable storage medium. When the computer program instructions are executed by a processor, an impact is applied to a skull of a mouse, and the impact is conducted along an organ synergistic conduction chain to induce a craniocerebral thoracoabdominal combined injury of the mouse, and the organ synergistic conduction chain is conducted to a thoracic cavity through the skull, cerebrospinal fluid, spinal cord and diaphragm in sequence. In the process of applying the impact, multi-modal physiological data are synchronously collected, the multi-modal physiological data include any one or more of electroencephalogram signals, thoracic cavity pressure signals and blood oxygen signals, the collected multi-modal physiological data are input into an AI analysis model, an injury mechanism of the craniocerebral thoracoabdominal combined injury is analyzed, and a dynamic model is established based on an analysis result.

[0103] It should be understood that the specific features, operations and details described above with respect to the method of the application can be similarly applied to the system of the application, or vice versa. In addition, each step of the method of the application described above can be performed by a corresponding component or unit of the system of the application.

[0104] It should be understood that each module / unit of the system of the application can be implemented in whole or in part by software, hardware, firmware or a combination thereof. Each module / unit can be embedded in a processor of a computer device in hardware or firmware form or independent of the processor, or can be stored in a memory of the computer device in software form to be invoked by the processor to perform the operations of each module / unit. Each module / unit can be implemented as an independent component or module, or two or more modules / units can be implemented as a single component or module.

[0105] In one embodiment, a computer device is provided, which includes a memory and a processor, and the memory stores computer instructions executable by the processor, and the computer instructions, when executed by the processor, instruct the processor to perform each step of the method of the embodiment of the application. The computer device can be a server, a terminal or any other electronic device with necessary computing and / or processing capabilities in a broad sense. In one embodiment, the computer device can include a processor, a memory, a network interface, a communication interface and the like connected by a system bus. The processor of the computer device can be used to provide necessary computing, processing and / or control capabilities. The memory of the computer device can include a non-volatile storage medium and an internal memory. The non-volatile storage medium can store an operating system, a computer program and the like therein or thereon. The internal memory can provide an environment for running of the operating system and the computer program in the non-volatile storage medium. The network interface and the communication interface of the computer device can be used to connect and communicate with external devices through a network. The computer program, when executed by the processor, performs the steps of the method of the application.

[0106] The present application can be implemented as a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, causes the steps of the method of an embodiment of the present application to be performed. In one embodiment, the computer program is distributed over a network of network-coupled computer devices or processors such that the computer program is stored, accessed and executed by one or more computer devices or processors in a distributed manner. A single method step / operation, or two or more method steps / operations, can be performed by a single computer device or processor or by two or more computer devices or processors. One or more method steps / operations can be performed by one or more computer devices or processors and one or more other method steps / operations can be performed by one or more other computer devices or processors. One or more computer devices or processors can perform a single method step / operation, or perform two or more method steps / operations.

[0107] As will be appreciated by one of ordinary skill in the art, the method steps of the present application can be directed to relevant hardware, such as computer devices or processors, by way of computer programs that can be stored in non-transitory computer-readable storage media, which, when executed, cause the steps of the present application to be performed. Any reference to memory, storage, databases, or other media herein can include non-volatile and / or volatile memory storage. Examples of non-volatile memory include read-only memory (ROM), programmable ROM (PROM), erasable programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), flash memory, magnetic tapes, hard disks, optical data storage devices, and the like. Examples of volatile memory include random access memory (RAM), external cache memory, and the like.

[0108] The various technical features described above can be combined in any manner. Although not all possible combinations thereof are described, any combination of the technical features should be considered to be within the scope of the present specification, as long as such a combination does not result in a contradiction.

[0109] Finally, it should be noted that the above-described embodiments are merely intended to illustrate the technical solutions of the present application, not to limit the technical solutions of the present application; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still make modifications to the technical solutions recorded in the foregoing embodiments, or make equivalent replacements to some or all of the technical features; and such modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for analyzing a dynamic model of combined craniocerebral, thoracic, and abdominal injuries in mice based on AI, characterized in that... include: Based on mice that have completed sensor implantation, the sensor is implanted in the brain parenchyma and thoracic and abdominal regions, and is used to collect signals of changes in intracranial pressure, fluctuations in thoracic gas pressure, and blood oxygen concentration. Multimodal physiological data were simultaneously collected during the process of inducing craniocerebral, thoracic, and abdominal combined injury in mice by impacting their skulls. This multimodal physiological data included electroencephalogram (EEG) signals, intrathoracic pressure signals, and blood oxygenation signals. The impact was conducted along the organ-coordinated conduction chain, which sequentially transmitted through the skull, cerebrospinal fluid, spinal cord, and diaphragm to the thoracic cavity, including: The system uses a flexible implantable sensor array to acquire real-time signals of intracranial pressure changes, pleural gas pressure fluctuations, and blood oxygen concentration. The EEG signal, thoracic pressure signal, and blood oxygenation signal are processed synchronously across channels using a millisecond-level timestamp protocol to form multimodal physiological data with a unified time reference. The method of acquiring real-time changes in intracranial pressure and fluctuations in pleural gas pressure through a flexible implantable sensor array includes: Nanoprobes based on the principle of plasmon resonance, deployed in the brain parenchyma of mice, are used to acquire intracranial pressure signals that reflect changes in local tissue elasticity. A fiber optic sensor network deployed in the chest and abdomen region is used to collect stress response and pleural gas pressure fluctuations in three-dimensional space. The collected multimodal physiological data are input into a dual-channel AI analysis model to analyze the injury mechanism of the combined craniocerebral, thoracic and abdominal injuries, and a dynamic model is established based on the analysis results. The step of inputting the collected multimodal physiological data into a dual-channel AI analysis model includes: Extract temporal variation features from electroencephalogram (EEG), intrathoracic pressure (IP), and blood oxygenation (BO) signals; Based on the temporal variation characteristics, a multidimensional feature vector representing the response relationship between organs is constructed; Based on the multidimensional feature vector, a dual-channel AI analysis model is used to generate causal correlation outputs for predicting the evolution of damage mechanisms.

2. The AI-based dynamic model analysis method for combined craniocerebral, thoracic, and abdominal injuries in mice as described in claim 1, characterized in that, The method of impacting the mouse skull to induce combined craniocerebral, thoracic, and abdominal injuries in the mice includes: The electromagnetic coil is controlled to generate a pulsed magnetic field, which simultaneously drives the pneumatic cavity to release compressed gas, generating a directional impact. The output direction of the impact is adjusted in real time based on the feedback signal from the gyroscope, so that the impact is transmitted along the organ-coordinated conduction chain.

3. The AI-based dynamic model analysis method for combined craniocerebral, thoracic, and abdominal injuries in mice as described in claim 1, characterized in that... Also includes: The causal correlation output includes a dynamic path diagram that starts from changes in local shear stress in the cranium and sequentially correlates the release of inflammatory factors, abnormal autonomic nerve function, and disordered regulation of intrathoracic pressure. This path is used to assist in establishing a mathematical model of the mechanism of cross-organ injury propagation.

4. An AI-based dynamic model analysis system for combined craniocerebral, thoracic, and abdominal injuries in mice, characterized in that, The method for implementing the AI-based dynamic model analysis of combined craniocerebral, thoracic, and abdominal injuries in mice as described in claim 1 includes: An impact module for impacting the skull of mice to induce combined craniocerebral, thoracic, and abdominal injuries in the mice; The acquisition module is used to simultaneously acquire multimodal physiological data during the process of impacting the skull of a mouse to induce a combined craniocerebral, thoracic and abdominal injury in the mouse. The multimodal physiological data includes electroencephalogram (EEG) signals, intrathoracic pressure signals and blood oxygenation signals. The impact is conducted along the organ-coordinated conduction chain, which is conducted sequentially through the skull, cerebrospinal fluid, spinal cord and diaphragm to the thoracic cavity. The analysis module is used to input the collected multimodal physiological data into a dual-channel AI analysis model to analyze the injury mechanism of the combined craniocerebral, thoracic and abdominal injuries, and to establish a dynamic model based on the analysis results.

5. An electronic device, characterized in that, include: Communication interface, processor, memory; The memory is used to store program instructions, which, when executed by the processor connected to the memory via the communication interface, enable the electronic device to implement the AI-based dynamic model analysis method for combined craniocerebral, thoracic and abdominal injuries in mice as described in any one of claims 1 to 3.

6. A computer-readable storage medium having program instructions stored thereon, characterized in that, When the program instructions are executed by a computer, the computer implements the AI-based dynamic model analysis method for combined craniocerebral, thoracic and abdominal injuries in mice as described in any one of claims 1 to 3.

Images