An adaptive chest compression system and control method based on fuzzy neural network

Abstract

This invention discloses an adaptive chest compression system and control method based on a fuzzy neural network, belonging to the field of medical device technology. The system includes a compression device and a control system. The compression device consists of a compression head and a contractile band encircling the chest, simultaneously providing chest compression force and circumferential compression force. This disperses the significant impact of a single chest compression on the ribs, reducing the risk of injury. This invention utilizes a control system to control the parameters of the compression device and incorporates a fuzzy neural network into the control system to learn the nonlinear mapping between motor control and system output, providing more precise adaptive control for the compression system.

Description

Technical Field

[0001] This invention relates to the field of medical device technology, and in particular to an adaptive chest compression system and control method based on a fuzzy neural network. Background Technology

[0002] Cardiac arrest poses a serious threat to human life and health. Patients experiencing cardiac arrest will die within a short period due to ischemia and hypoxia in vital organs such as the heart and brain. Timely and effective cardiopulmonary resuscitation (CPR) is a necessary measure to save lives. Effective chest compressions play a crucial role in promoting blood circulation, maintaining blood supply to vital organs such as the heart and brain, preventing organ necrosis, and promoting the patient's spontaneous circulation recovery, thus being key to successful CPR. During chest compressions, attention should be paid to controlling parameters such as compression depth and frequency. Compressions that are too shallow may not achieve the desired effect and fail to maintain blood circulation, while compressions that are too deep may lead to adverse events such as pneumothorax, damage to the chest wall and internal organs, and rib fractures, causing secondary injury to the patient. Currently, single-point chest compressions are the primary method because the force area of ​​a single-point chest compression is small, requiring a higher compression depth to achieve sufficient effect. In this case, the compression will generate a large impact on the chest, and prolonged high pressure can cause significant damage to the ribs, muscles, and soft tissues.

[0003] According to the AHA CPR guidelines, the chest compression depth for adults should be no less than 5 cm but no more than 6 cm, with a compression rate of 100–120 compressions per minute. The chest should fully recoil after each compression, with a compression-to-relaxation ratio that is approximately equal. These are compression parameters suitable for most people, derived from large datasets. However, each patient's chest anteroposterior diameter, shape, and elasticity differ, leading to variations in the appropriate compression force and depth. Furthermore, the biomechanical characteristics of the chest change during chest compressions. Additionally, due to the significant uncertainty and time-varying nature of the actual chest environment during compressions, the transition from motor control to the system's actual output exhibits strong nonlinearity, making traditional control methods such as PID control difficult to apply.

[0004] In summary, there is an urgent need for a well-designed chest compression system that can adaptively adjust compression parameters according to patient differences, thereby reducing the risk of injury during chest compressions. Summary of the Invention

[0005] To address the shortcomings of existing technologies, the purpose of this invention is to provide an adaptive chest compression system and control method based on a fuzzy neural network. The system comprises a compression head and a contractile band encircling the chest, simultaneously providing chest compression force and circumferential compression force. This disperses the significant impact of a single chest compression on the ribs and reduces the risk of injury. Furthermore, this invention incorporates an adaptive fuzzy neural network into the control system to learn the nonlinear mapping between motor control and system output, providing more precise control of the compression system.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solution:

[0007] The first aspect of this invention provides an adaptive chest compression system based on a fuzzy neural network, comprising a compression device and a control system:

[0008] The compression device includes a compression head, a chest band, and sensors. Both ends of the chest band are provided with connectors, which connect the two ends of the compression head and fix the compression head in the middle of the chest band. The compression head and the chest band together form a closed ring. The binding space formed by the closed ring is used to accommodate the patient's chest. Multiple sensors are provided on the inner side of the chest band and the compression head.

[0009] Control system, including:

[0010] The data acquisition module is used to acquire the pressure data collected by the sensor in real time and to preprocess the pressure data.

[0011] The model building module combines the influence parameters during the compression process to model the chest, resulting in a chest model. A fuzzy neural network is then introduced into the chest model to obtain a state evaluation model.

[0012] The condition assessment module uses a condition assessment model to assess the damage risk of preprocessed real-time pressing data and adjusts the pressing parameters applied to the pressing device based on the damage risk assessment results.

[0013] Furthermore, the pressing head is a single-point pressing head, which includes a suction cup structure and a controller. The controller is installed above the suction cup structure and has a display screen and control buttons.

[0014] Furthermore, the chest band includes an inner band and an outer band. The outer band is a rigid structure that provides support for the pressing device and does not participate in the chest compression operation. The inner band is composed of inelastic nylon tape and memory foam. Driven by motors on both sides, it provides chest compression force, and the memory foam provides corresponding cushioning to reduce chest injury.

[0015] Furthermore, it also includes a motor for providing driving force to the pressing device, with the pressing head driven by a single motor and the chest band driven by dual motors on both sides.

[0016] Furthermore, the motor parameters on both sides of the chest band are set identically, allowing for synchronized tightening and loosening on both sides.

[0017] Furthermore, the sensors include pressure sensors and acceleration sensors, which are installed on both the compression head and the chest band.

[0018] Furthermore, the data acquisition module collects compression data including acceleration measured by an accelerometer and compression force and chest compression force measured by a pressure sensor.

[0019] Furthermore, in the model building module, the influencing parameters during the pressing process include the pressing force, pressing displacement, velocity, and acceleration.

[0020] Furthermore, in the model building module, the input of the fuzzy neural network is the system's desired force and displacement, and the output is the motor acceleration parameters. The fuzzy rules adopt the TS fuzzy model.

[0021] A second aspect of the present invention provides a control method for the adaptive chest compression system based on a fuzzy neural network as described in the first aspect, comprising the following steps:

[0022] Real-time acquisition of pressure data collected by sensors, and preprocessing of the pressure data;

[0023] By combining the influencing parameters during the compression process, a chest model is obtained. A fuzzy neural network is then introduced into the chest model to obtain a state evaluation model.

[0024] A condition assessment model is used to assess the damage risk of preprocessed real-time compression data, and the compression parameters applied to the compression device are adjusted based on the damage risk assessment results.

[0025] The above one or more technical solutions have the following beneficial effects:

[0026] This invention discloses an adaptive chest compression system and control method based on a fuzzy neural network. The compression system consists of a compression head and a contraction band around the chest. It combines single-point chest compression and circumferential compression to provide both chest compression force and circumferential compression force, thereby dispersing the large impact of a single chest compression on the ribs and reducing the risk of injury.

[0027] This invention incorporates an adaptive fuzzy neural network into the control system to learn the nonlinear mapping between motor control and system output in the chest compression system, providing more precise control. To better adapt to the differences in chest biomechanics among different patients and changes in the chest during compression, this system collects chest force and displacement information from various directions during compression, models the chest biomechanical state, and feeds back the risk of injury to the neural network to adaptively guide motor parameter adjustment, thereby improving the quality of chest compression while ensuring safety and avoiding secondary injury.

[0028] Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0029] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0030] Figure 1 This is a structural diagram of the pressing device in Embodiment 1 of the present invention;

[0031] Figure 2 This is a control flowchart of the control system in Embodiment 1 of the present invention;

[0032] Figure 3 This is a structural diagram of the state evaluation model in Embodiment 1 of the present invention;

[0033] 1. Controller; 2. Outer ring belt; 3. Inner ring belt; 4. Suction cup; 5. Display screen; 6. Control button; 7. Sensor; 8. Motor; 9. First buckle; 10. Second buckle. Detailed Implementation

[0034] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0035] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0036] Example 1:

[0037] Embodiment 1 of the present invention provides an adaptive chest compression system based on a fuzzy neural network, including a compression device and a control system.

[0038] Currently, chest compressions primarily use single-point chest compressions because the area of ​​force application is small, requiring a greater compression depth to achieve sufficient effect. In this case, the compression can cause significant impact on the chest, and prolonged high pressure can lead to considerable damage to the ribs, muscles, and soft tissues. To address these issues, this invention employs a chest compression method that combines single-point chest compressions with circumferential chest compressions.

[0039] Specifically, such as Figure 1 As shown, the compression device includes a compression head, a chest band, a motor 8, and sensors 7. Connectors are provided at both ends of the chest band, connecting the two ends of the compression head. The connectors are first buckles 9, which fix the compression head in the middle of the chest band. The compression head and the chest band together form a closed loop, and the binding space formed by the closed loop is used to accommodate the patient's chest. Multiple sensors 7 are provided on the inner side of the chest band and on the compression head for real-time collection of compression data. The motor 8 provides driving force for the compression device; the compression head is driven by a single motor, and the chest band is driven by dual motors on both sides.

[0040] In this embodiment, the pressing head is a single-point pressing head, comprising a suction cup structure 4 and a controller 1. The suction cup structure 4 is a rubber suction cup, which, in actual application, is located at the midpoint of the line connecting the patient's two nipples, thereby increasing the suction force with the chest. During the relaxation phase, it lifts the chest cavity, creating negative pressure in the chest, which is beneficial for chest rebound. During the pressing process, the rubber suction cup cushions the impact on the chest cavity through elastic deformation. The controller 1 is installed above the suction cup structure and is equipped with a display screen 5 and control buttons 6. The controller controls the pressing device by adjusting the motor acceleration parameters.

[0041] The chest band consists of an inner band 3 and an outer band 2. The outer band 2 is a rigid structure used to provide support for the compression device and does not participate in the chest compression operation. The inner band 3 is composed of a non-elastic nylon band and memory foam. Driven by motors on both sides, it provides chest compression force. The memory foam forms a hollow, bag-like structure that wraps around the non-elastic nylon band, and the outer surface of the memory foam is bonded to the outer band. The memory foam provides cushioning to reduce chest injury. The motors are fixed to the outer band and directly connected to the non-elastic nylon band.

[0042] This embodiment uses non-elastic nylon tape because it has high tensile strength and durability. Furthermore, since it lacks elastic components, it will not deform due to stretching during use, providing sufficient contractile force and further enhancing the adaptability of the pressing device.

[0043] The chest band consists of two side bands, left and right. Each side band has two types of buckles at both ends: a first buckle 9 at one end and a second buckle 10 at the other. The second buckles 10 can be connected to each other to form a complete chest band. The first buckle 9 can be connected to the compression head, specifically to both sides of the controller. It should be noted that the chest band and compression head can operate together or independently, suitable for various patient injuries. The chest band is formed by connecting the second buckles 10 at both ends of the side bands to the patient's back, facilitating installation and operation.

[0044] To ensure even force distribution on the chest, the motor parameters on both sides of the chest band are set identically, allowing for synchronized tightening and loosening on both sides.

[0045] Sensor 7 includes a pressure sensor and an acceleration sensor, with one pressure sensor and one acceleration sensor forming a group. Pressure sensors and acceleration sensors are also installed on both the compression head and the chest band. Specifically, one group of sensors is located at the bottom of the compression head, and three groups of sensors are symmetrically arranged on each side band inside the chest band (a total of six groups). During compression, the system's acceleration and pressure information are measured in real time, providing approximate displacement and pressure information for the entire chest cavity, facilitating comprehensive analysis of chest stress and injury.

[0046] The operating procedure for the pressing device is as follows: Figure 2 As shown, the chest compression device is first installed: the straps are connected to the compression head, and the two straps are fastened to the patient's back to complete the installation. After installation, the compression head and inner strap automatically conform to the chest cavity based on pressure sensor signals. The operating mode is selected according to the patient's local chest injury. Operating modes include single-point compression, circumferential compression, and combined operation. Compression begins. In the initial stage of compression without learning chest biomechanical information, the compression device operates according to preset parameters, which are customized based on experience. During operation, acceleration signals and pressure information are collected in real time. The chest biomechanical characteristics are then fitted again based on the patient's local chest injury, and the desired displacement, desired force, and the participation of single-point compression and circumferential compression are adjusted in stages. Then, a fuzzy neural network in the control system is used for evaluation. Based on the evaluation results, the motor parameters are adjusted, and then cyclic compression is performed until the compression ends.

[0047] The control system includes a data acquisition module, a model building module, and a status assessment module.

[0048] The data acquisition module is used to acquire the pressure data collected by the sensor in real time and to preprocess the pressure data.

[0049] Specifically, the collected compression data includes acceleration measured by an accelerometer and compression force and chest compression force measured by a pressure sensor.

[0050] The process of preprocessing the press data includes:

[0051] The velocity and displacement signals are obtained by integrating the acceleration during the pressing process.

[0052] v=∫a(t)dt (1),

[0053] s=∫v(t)dt=∫(∫a(t)dt)dt (2).

[0054] Where s, v, and a represent the chest displacement during the compression process and the overall velocity and acceleration at the corresponding displacement, respectively, and t represents time.

[0055] The sensor collects discrete data, which is then integrated using the trapezoidal rule:

[0056]

[0057] Where f(τ) is the sampled value of the τth (τ=1,2,3…n) sampling point, n is the total number of sampling points, and T is the sampling interval. When the acceleration sampling frequency is large enough, the error caused by the trapezoidal formula can be ignored.

[0058] Actual measured acceleration data contains many interferences and differs from the ideal acceleration waveform, leading to deviations in the calculated displacement data. Due to the influence of gravitational acceleration, environmental interference, and other factors, the acceleration collected by the sensor contains a DC component. Therefore, this system uses the average value of the raw acceleration data. Instead of the DC component, the DC component is removed by subtracting the mean value from the original acceleration data.

[0059]

[0060] a τ Let a be the acceleration value at the τth sampling point. τ ′ represents the corrected acceleration value at the τth sampling point.

[0061] The model building module models the chest by incorporating influencing parameters during the compression process, resulting in a chest model. A fuzzy neural network is then introduced into this chest model to obtain a state assessment model. The overall structure of the state assessment model is as follows: Figure 3 As shown in the figure. The influencing parameters during the pressing process include pressing force, pressing displacement, velocity, and acceleration. The fuzzy neural network inputs are the system's desired force and displacement, and the output is the motor acceleration parameters. The fuzzy rule adopts the TS fuzzy model.

[0062] Because the actual chest environment during chest compressions exhibits significant uncertainty and time-varying characteristics, the transition from motor control to the system's actual output is highly nonlinear, making traditional control methods such as PID control difficult to apply. Therefore, this system employs a fuzzy neural network as the controller. The neural network learns the complex nonlinear relationship between the system's input and output, constructing the inverse dynamics between the system's output and input, thereby building an accurate control model. During training, the model parameters are adjusted based on the output error to improve the system's fitting capability.

[0063] Specifically, during chest compressions, the chest is modeled based on the compression force, displacement, velocity, and acceleration to assess the risk level of chest injury. Improving compression quality while avoiding secondary injury helps improve blood and oxygen supply to the patient. The chest modeling formula is as follows:

[0064] F = K0s + Ks 2 +Dv+Ma (6).

[0065] Where F is the chest pressure, s, v and a are the chest displacement during the compression process and the overall velocity and acceleration under the corresponding displacement, respectively, and K0, K, D and M are the corresponding parameters for chest modeling. The force, displacement, velocity and acceleration data of the two-minute compression process are used to fit the formula (6) to obtain the personalized values ​​of parameters K0, K, D and M, and then obtain the injury risk of each part of the chest. Based on this, the expected force and displacement of the system and the participation of compression and squeezing are set to realize the adaptive control of the system based on the mechanical characteristics of the chest. When the actual pressure is less than the expected pressure range, it indicates that the chest is softened and the compression depth and the amount of ring contraction can be appropriately increased; when the actual pressure is greater than the expected pressure range, it indicates that the risk of fracture is increased and the compression depth and the amount of ring contraction should be appropriately reduced. When the ratio of actual pressure to displacement in one direction is suddenly too large or too small, it indicates that the mechanical state of the chest in the current direction is abnormal and there may be fracture injury. When this situation continues for more than 3 compression cycles, the participation of the corresponding compression mode is reduced and the compression mode is changed.

[0066] The system control objective is to determine the motor parameters to control the output displacement and force within the desired range. This system is nonlinear, so a fuzzy neural network is introduced to learn the system's inverse dynamics. The neural network inputs are the desired force and displacement, and the outputs are the motor acceleration parameters. The fuzzy rules adopt the Takagi-Sugeno (TS) fuzzy model, and the fuzzy rules are described as follows:

[0067]

[0068] Where, x i Input to the model, Let yj be the j-th (j=1,2,…N) linguistic variable value of the i-th (i=1,2,…N) input, where N is the total number of inputs, K is the total number of linguistic variable values ​​(i.e., the number of nodes), and yj is the output of the corresponding rule. The system contains a 5-layer structure:

[0069] The specific structure of the state assessment model is as follows:

[0070] First layer: Input layer, which imports system input into the network.

[0071] The second layer is the input fuzzification layer, where the membership degree of the input to each fuzzy level is calculated at each node. In this system, the membership function adopts a Gaussian distribution:

[0072]

[0073] Among them, c ij and σ ij These are the center and width of the membership function, respectively.

[0074] Third layer: Calculate the applicability of each rule a j The applicability of each rule is determined by the membership degree of each input, and there are K nodes in this layer.

[0075]

[0076] Fourth layer: Applicability normalization:

[0077]

[0078] in, This indicates the applicability after normalization.

[0079] Fifth layer: Output layer:

[0080]

[0081] The condition assessment module uses a condition assessment model to assess the damage risk of preprocessed real-time pressing data and adjusts the pressing parameters applied to the pressing device based on the damage risk assessment results.

[0082] Due to differences in external injuries and patients' underlying physiological characteristics, the biomechanical state of the chest varies significantly among patients, and this biomechanical state changes with chest compressions during emergency treatment. Adjusting compression parameters based on the patient's chest physiological characteristics is crucial for improving the quality of chest compressions and reducing the risk of chest wall injury. This invention utilizes real-time acceleration and chest wall pressure data to model the chest biomechanical state, assess injury risk, set personalized desired displacement and force, and adjust the involvement of compression and squeezing based on the risk of injury in each direction, achieving adaptive adjustment of motor parameters based on the chest biomechanical state.

[0083] Because the patient's chest condition changes over time during emergency treatment, increasing the nonlinearity and uncertainty of the system, assessing the chest's biomechanical state is necessary during control. Therefore, a state assessment network is incorporated into the system. The output of this network is recursively fed back to the control network, enabling adaptive control based on real-time chest characteristics. The state assessment network's input parameters are measurements from various pressure and acceleration sensors, and its output is the fracture risk at different locations within the chest.

[0084] The specific structure of the state evaluation network is as follows:

[0085] The first layer is the input layer. The mechanical data in each direction is not processed. Based on the acceleration data, the real-time displacement and velocity in each direction during operation are calculated by formulas (1) and (2).

[0086] The second layer is the fitting layer, which fits the chest biomechanical characteristics according to (6) to obtain the chest biomechanical parameters K in each direction. i,0 ,K i D i M i .

[0087] The third layer is the classification layer, which classifies the risk level of chest injury based on the chest biomechanical parameters output from the second layer and assesses the possible location of injury.

[0088] In this embodiment, four levels of chest injury risk are set: K1 (no risk), K2 (low risk), K3 (high risk), and K4 (extremely high risk). The Random Forest algorithm is used for risk level classification in this system. The expected force and expected displacement matrix X for chest compressions are obtained based on the injury risk experience database. Furthermore, the participation matrix α of the compression motor and the squeezing motor is obtained based on the estimated injury location.

[0089] The unknown parameter of the system is the weight parameter ω of each state rule. ij Let α be the motor participation matrix. The parameters are optimized by minimizing the following objective function E.

[0090]

[0091] in, This represents the actual output value of the control system.

[0092] The objective function described above is constrained by maintaining the risk of chest injury at K1 and K2.

[0093] The parameter optimization employs the quantum particle swarm optimization algorithm. The candidate solutions to the optimization problem are called particles, which in this system are the input weight parameters ω. ij The particle search process is affected by the individual optimal position of each particle. and the optimal position of the particle swarm The attraction of particle r is defined by the attractor p. r Its position in space is defined as follows:

[0094]

[0095]

[0096] Where k is the number of iterations of the optimization algorithm, and σ is a random vector in the interval (0,1). r γ is the adaptive adjustment factor for particle r. dω is the width of the Gaussian membership function. In the early stages of evolution, maintaining a larger dω value is expected to increase population diversity; in the later stages of evolution, maintaining a smaller dω value is expected to improve the convergence of the algorithm. The nonlinear decreasing trend of dω can be adjusted by changing γ, thus providing a more flexible nonlinear adjustment strategy for the dω value.

[0097] The position of particle r at the (k+1)th iteration can be obtained. renew:

[0098]

[0099] Where u is a random number uniformly distributed in the range (0,1). When u>0.5, a negative sign is chosen; otherwise, a positive sign is chosen. It is the standard deviation of particle distribution, which determines the search range of particles during the evolution process. It can be evaluated by the following equation:

[0100]

[0101] Where, p mbestLet β represent the average of the individual best positions of all particles, and let β represent the expansion-contraction coefficient. The larger the value of β, the larger the search space of the particles, which can prevent the particles from getting trapped in local optima; conversely, the smaller the value of β, the stronger the local search ability of the particles, which can effectively improve the convergence accuracy. During the iteration process, β decreases linearly according to the following formula, so that the particles have a large search space in the early stage of iteration and a strong local search ability in the later stage of iteration.

[0102]

[0103] The optimal position of each particle at the (k+1)th iteration and the optimal position of the particle swarm The update is as follows:

[0104]

[0105] Where f(·) is the fitting function, used to determine the quality of the solution. The optimal positions of individual particles and the optimal positions of the particle swarm move in directions that optimize the system, attracting the actual positions of particles to converge towards these directions. Ultimately, the entire particle swarm reaches the optimal position of the system, thus yielding the system parameter ω. ij The optimal solution.

[0106] This invention provides multiple chest compression options for patients with different chest injuries. The combined use of single-point compressions and circumferential chest compressions disperses chest pressure and reduces the risk of injury. Simultaneously, the device adjusts chest compression parameters in real time based on pressure and acceleration information, meeting personalized cardiopulmonary resuscitation requirements and improving resuscitation efficiency.

[0107] Example 2:

[0108] Embodiment 2 of the present invention provides a control method for the adaptive chest compression system based on a fuzzy neural network as described in Embodiment 1, comprising the following steps:

[0109] Step 1: Acquire the pressure data collected by the sensor in real time and preprocess the pressure data.

[0110] Step 2: Model the chest by combining the influence parameters during the compression process to obtain a chest model. Introduce a fuzzy neural network into the chest model to obtain a state evaluation model.

[0111] Step 3: Use the condition assessment model to assess the damage risk of the preprocessed real-time pressing data, and adjust the pressing parameters implemented on the pressing device based on the damage risk assessment results.

[0112] The steps involved in the above embodiment two correspond to those in embodiment one. For specific implementation details, please refer to the relevant description section of embodiment one.

[0113] Those skilled in the art will understand that the modules or steps of the present invention described above can be implemented using general-purpose computer devices. Optionally, they can be implemented using computer-executable program code, thereby allowing them to be stored in a storage device for execution by a computer device, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. The present invention is not limited to any particular combination of hardware and software.

[0114] While the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.

[0115] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. An adaptive chest compression system based on a fuzzy neural network, characterized in that, Includes pressing device and control system: The compression device includes a compression head, a chest band, and sensors. Both ends of the chest band are provided with connectors, which connect the two ends of the compression head and fix the compression head in the middle of the chest band. The compression head and the chest band together form a closed ring. The binding space formed by the closed ring is used to accommodate the patient's chest. Multiple sensors are provided on the inner side of the chest band and the compression head. Control system, including: The data acquisition module is used to acquire the pressure data collected by the sensor in real time and to preprocess the pressure data. The model building module combines the influence parameters during the compression process to model the chest, resulting in a chest model. A fuzzy neural network is then introduced into the chest model to obtain a state evaluation model. The status assessment module uses a status assessment model to assess the damage risk of preprocessed real-time compression data and adjusts the compression parameters applied to the compression device based on the damage risk assessment results. The chest band includes an inner band and an outer band. The outer band is a rigid structure. The inner band is composed of a non-elastic nylon strip and memory foam. Driven by motors on both sides, it provides chest compression force. The motors are fixed to the outer band and directly connected to the non-elastic nylon strip. Multiple sensors are provided on the inner side of the inner band.

2. The adaptive chest compression system based on a fuzzy neural network as described in claim 1, characterized in that, The pressing head is a single-point pressing head, which includes a suction cup structure and a controller. The controller is installed above the suction cup structure and has a display screen and control buttons.

3. The adaptive chest compression system based on a fuzzy neural network as described in claim 1, characterized in that, It also includes a motor for providing driving force to the compression device, with the compression head driven by a single motor and the chest band driven by dual motors on both sides.

4. The adaptive chest compression system based on a fuzzy neural network as described in claim 3, characterized in that, The motor parameters on both sides of the chest band are set identically, allowing for synchronized tightening and loosening on both sides.

5. The adaptive chest compression system based on a fuzzy neural network as described in claim 1, characterized in that, The sensors include pressure sensors and acceleration sensors, which are installed on both the compression head and the chest band.

6. The adaptive chest compression system based on a fuzzy neural network as described in claim 1, characterized in that, The data acquisition module collects compression data including acceleration measured by the accelerometer and compression force and chest compression force measured by the pressure sensor.

7. The adaptive chest compression system based on a fuzzy neural network as described in claim 1, characterized in that, In the model building module, the influencing parameters during the pressing process include pressing force, pressing displacement, velocity, and acceleration.

8. The adaptive chest compression system based on a fuzzy neural network as described in claim 1, characterized in that, In the model building module, the input of the fuzzy neural network is the expected force and displacement of the system, and the output is the motor acceleration parameters. The fuzzy rules adopt the TS fuzzy model.

Images