An intra-abdominal pressure control system and method based on a shallow neural network model

Abstract

The present application relates to the technical field of pressure control, in particular to a kind of abdominal cavity pressure control system and method based on shallow neural network model, and the initial intake flow of sensor module acquisition, the intake time of pre-set and the pressure difference calculated before and after intake are combined with the physiological parameters and operation type code of target object according to the shallow neural network model built in control unit, and individualized correction coefficient is output;Control unit accurately calculates the required air supplement and the basic volume of abdominal cavity based on the correction coefficient, and controls the execution module to complete air supplement one-time, to establish pneumoperitoneum.Compared with the existing trial-and-error mode of repeated air intake, stop and pressure measurement, the present application only needs two air intakes to establish pneumoperitoneum, avoids the risk of overpressure caused by the accumulation of pressure collection deviation, overcomes the defects of the existing technology, such as the lack of abdominal cavity volume estimation, the lack of basis for air intake control, the long process of pneumoperitoneum establishment and frequent trial-and-error, and realizes safe, efficient and individualized pneumoperitoneum establishment.

Description

Technical Field

[0001] This invention relates to the field of pressure control technology, and in particular to an intra-abdominal pressure control system and method based on a shallow neural network model. Background Technology

[0002] An insufflator is a key piece of equipment in laparoscopic surgery, used to inject medical CO2 gas into the abdominal cavity to establish and maintain a stable pneumoperitoneum environment. Current insufflators typically include an inlet tubing, a pressure monitoring tubing, and an exhaust tubing. During the insufflation phase, because only the inlet tubing is connected and the pressure monitoring tubing is not yet connected to the abdominal cavity, it is impossible to directly measure the actual pressure within the abdominal cavity.

[0003] To address this, existing technologies employ an alternating operating mode: first, a certain amount of gas is injected into the abdominal cavity; then, the air intake is stopped and a preset time is waited for the airflow to stabilize; next, a static pressure sensor on the intake line approximates the abdominal pressure to determine if the preset target pressure has been reached; if not, the above steps are repeated. To avoid the risk of overpressure, existing technologies typically employ two methods: one is to rely on the operator's experience to reduce the airflow rate; the other is to set a fixed, small-step air intake mode in the program, with short single air intake times, and adjust the air intake time in conjunction with the pressure difference.

[0004] However, the aforementioned existing technologies have significant drawbacks. On the one hand, the method of approximating the actual abdominal cavity pressure by using the static pressure of the air intake pipeline has a large deviation. If the waiting time is too short, the pressure acquisition will be inaccurate; if the waiting time is too long, the pneumoperitoneum establishment efficiency will be low. On the other hand, the approach of reducing the flow rate or using small-step air intake either relies excessively on the operator's experience and is difficult to control precisely, or requires multiple repeated air intakes, which is time-consuming and prone to causing abdominal overpressure due to the cumulative pressure acquisition deviation, posing a safety hazard of damaging the target tissue. In addition, the existing technologies lack a precise calculation mechanism for the abdominal cavity volume, relying solely on air intake volume estimation and pressure feedback for air intake control. This fails to consider the impact of individual body size, tissue elasticity, and different operation types on the abdominal cavity volume, resulting in blind air intake control and further exacerbating the problems of overpressure and frequent air intake. Summary of the Invention

[0005] In view of this, the purpose of the present invention is to provide a peritoneal pressure control system and method based on a shallow neural network model.

[0006] In a first aspect, embodiments of the present invention provide an intra-abdominal pressure control system based on a shallow neural network model, comprising: The sensor module includes a pressure sensor and a flow sensor; the pressure sensor is used to monitor changes in pressure within the abdominal cavity in real time; the flow sensor is used to measure the gas flow rate during intake and exhaust. The execution module includes a proportional valve and a switching valve located on the intake pipe, and a vent valve located on the exhaust pipe; The control unit communicates with the sensor module and the execution module. The control unit includes a correction module, a cavity volume calculation unit, a fixed leakage detection unit, and a volume compensation module. The correction module has a built-in shallow neural network model. The control unit outputs correction coefficients based on the physiological parameters of the target object, the operation type code, and the initial air intake parameters. It calculates the basic abdominal cavity volume based on the correction coefficients and the initial air intake parameters, and updates the actual abdominal cavity volume based on the leakage volume. During the maintenance phase, it determines whether a fixed leakage exists based on the air intake and exhaust flow rates and the real-time abdominal cavity pressure, and determines the leakage flow rate. It calculates the leakage volume based on the leakage flow rate and leakage duration, and generates compensation commands. The control unit is also used to control the execution module to supplement air during the pneumoperitoneum establishment phase based on the basic abdominal cavity volume and target pressure to complete the pneumoperitoneum establishment, and to dynamically control the execution module to compensate for the leakage volume during the maintenance phase based on the actual abdominal cavity volume and pressure deviation.

[0007] In conjunction with the first aspect, shallow neural network models include: The input layer includes nodes for the target object's physiological parameters, operation type encoding, single air intake flow rate, single air intake time, and pressure difference before and after air intake. The hidden layer is a single-layer structure. The hidden layer consists of multiple nodes, and each node is connected to all nodes in the input layer through trainable weights. A non-linear activation function is used to transform the weighted sum. The output layer contains one node. The node in the output layer is connected to all nodes in the hidden layer through trainable weights and uses an activation function with output limiting to constrain the output value within a specified range.

[0008] In conjunction with the first aspect, shallow neural network models are obtained through offline training in the following manner: Multiple samples were collected. Each sample contained an input feature set and a corresponding true correction coefficient. The true correction coefficient was calculated from the actual abdominal cavity volume, air intake flow rate, air intake time, and pressure difference before and after air intake in the actual scenario of the sample. The input features are normalized, and the samples are divided into training set, validation set and test set; Iterative training is performed using optimization algorithms and loss functions until the loss value on the validation set converges and the prediction accuracy on the test set reaches the preset requirements.

[0009] In conjunction with the first aspect, the correction module is used for: After the initial inhalation during the pneumoperitoneum establishment phase, physiological parameters, operation type codes, single inhalation flow rate, single inhalation time, and pressure difference before and after inhalation of the target subject are collected. The physiological parameters, operation type encoding, single air intake flow rate, single air intake time, and pressure difference before and after air intake of the target object are subjected to Min-Max normalization. The minimum and maximum values ​​used for normalization are the same as the minimum and maximum values ​​of the corresponding features used during offline training. The normalized physiological parameters of the target object, operation type encoding, single air intake flow rate, single air intake time, and pressure difference before and after air intake are used as the input values ​​of the corresponding nodes in the input layer of the shallow neural network model. The input values ​​of each node in the input layer are input into the shallow neural network model. The weighted sum of the input layer and the hidden layer is calculated in sequence and the hidden layer bias is superimposed. The hidden layer output is obtained by transforming through a nonlinear activation function. Then the weighted sum of the hidden layer and the output layer is calculated and the output layer bias is superimposed. The prediction correction coefficient is obtained by transforming through an activation function with output limiting. Apply output constraints to the predicted correction coefficients: if they exceed a preset range, prune them to the nearest boundary value; The constrained correction coefficients are transmitted to the cavity volume calculation unit.

[0010] In conjunction with the first aspect, the cavity volume calculation unit is used to receive the correction coefficients output by the shallow neural network model; the correction coefficients are calculated by multiplying the single air intake flow rate by the single air intake time, and then divided by the pressure difference before and after air intake to obtain the basic volume of the abdominal cavity. The cavity volume calculation unit is also used to add the basic volume of the abdominal cavity to the leakage volume when the leakage volume is obtained, and update the actual volume of the abdominal cavity.

[0011] In conjunction with the first aspect, the fixed leak detection unit is used for: During the pneumoperitoneum maintenance phase, the intake flow rate of the intake pipe, the exhaust flow rate of the exhaust pipe, and the real-time abdominal pressure are obtained in real time. Calculate the flow difference between the intake flow rate and the exhaust flow rate, as well as the pressure difference between the preset target pressure and the real-time abdominal pressure; When the fluctuation range of the flow rate difference is less than or equal to the preset first fluctuation threshold, the fluctuation range of the pressure difference is less than or equal to the preset second fluctuation threshold, and the duration of the fluctuation range is greater than or equal to the preset duration, it is determined that there is a fixed air leak. The average leakage flow rate is calculated based on the flow rate difference over a continuous period of time, and then transmitted to the volume compensation module.

[0012] In conjunction with the first aspect, the volume compensation module is used for: Receive the average leakage flow rate transmitted by the fixed leakage detection unit; Calculate the leakage volume based on the average leakage flow rate and leakage duration; The leak volume is transmitted to the cavity volume calculation unit to update the true volume of the abdominal cavity; It also generates compensation instructions based on the updated actual abdominal cavity volume and pressure difference, and sends the compensation instructions to the execution module to dynamically adjust the intake flow rate of the intake pipeline.

[0013] Secondly, the present invention provides a method for controlling intra-abdominal pressure based on a shallow neural network model, applied to the system described above; the method includes: Acquire the target object's physiological parameters, operation type encoding, and preset target pressure; The intake pipeline is controlled to perform an initial intake at a preset flow rate and time. After the intake is completed, the static pressure of the intake pipeline is collected, and the pressure difference before and after the intake is calculated. The physiological parameters, operation type encoding, preset air intake flow rate, preset air intake time, and pressure difference before and after air intake are input into the shallow neural network model in the correction module, and the correction coefficient is output. The basic volume of the abdominal cavity is calculated using the cavity volume calculation unit based on the correction coefficient, the preset air intake flow rate, the preset air intake time, and the pressure difference before and after air intake. The control unit calculates the required amount of gas supplementation based on the basic volume of the abdominal cavity and the target pressure, and controls the execution module to perform gas supplementation once according to the required amount, thus completing the establishment of pneumoperitoneum.

[0014] In conjunction with the second aspect, the step of the control execution module performing a gas replenishment according to the replenishment amount also includes: Obtain real-time abdominal pressure; When the real-time abdominal pressure exceeds the preset safety threshold, the pressure relief valve is opened to release pressure.

[0015] In conjunction with the second aspect, the method also includes: Real-time acquisition of air intake flow rate in the intake pipe, exhaust flow rate in the exhaust pipe, and real-time abdominal pressure; Calculate the flow difference between the intake flow rate and the exhaust flow rate, as well as the pressure difference between the target pressure and the real-time abdominal pressure; When the fluctuation range of the flow rate difference meets the preset first stability condition, the fluctuation range of the pressure difference meets the preset second stability condition, and the duration reaches the preset duration, it is determined that there is a fixed air leak, and the average leakage flow rate is determined based on the flow rate difference within the duration. The leakage volume is calculated based on the average leakage flow rate and leakage duration, and then the leakage volume is added to the baseline abdominal cavity volume to update the true abdominal cavity volume. The intake flow rate of the intake pipeline is dynamically adjusted based on the updated actual volume of the abdominal cavity and the pressure difference to compensate for the leakage volume.

[0016] The present invention provides the following beneficial effects: An intra-abdominal pressure control system and method based on a shallow neural network model is provided. The system includes: a sensor module comprising a pressure sensor and a flow sensor; the pressure sensor is used to monitor real-time pressure changes within the intra-abdominal cavity; the flow sensor is used to measure the gas flow rate during intake and exhaust; an execution module including a proportional valve and a switching valve on the intake pipe, and a vent valve on the exhaust pipe; and a control unit communicatively connected to the sensor module and the execution module; the control unit includes a correction module, a cavity volume calculation unit, a fixed leakage detection unit, and a volume compensation module; the correction module incorporates a shallow neural network model; the control unit... The control unit is used to output correction coefficients based on the physiological parameters of the target object, operation type code, and initial air intake parameters; calculate the basic abdominal cavity volume based on the correction coefficients and initial air intake parameters, and update the actual abdominal cavity volume based on the leakage volume; determine whether there is a fixed air leak during the maintenance phase based on the air intake and exhaust flow rates and real-time abdominal cavity pressure, and determine the leakage flow rate; calculate the leakage volume based on the leakage flow rate and leakage duration, and generate compensation instructions; the control unit is also used to control the execution module to supplement air to complete the pneumoperitoneum establishment phase based on the basic abdominal cavity volume and target pressure, and to dynamically control the execution module to compensate for the leakage volume during the maintenance phase based on the actual abdominal cavity volume and pressure deviation.

[0017] This invention utilizes a shallow neural network model built into the control unit. Based on the initial airflow rate collected by the sensor module, the preset airflow time, and the calculated pressure difference before and after airflow, combined with the physiological parameters of the target subject and the operation type encoding, it outputs an individualized correction coefficient. The control unit then accurately calculates the basic abdominal cavity volume and the required air supply based on this correction coefficient, and controls the execution module to complete the air supply in one go, thereby establishing pneumoperitoneum. Compared to the existing trial-and-error mode of repeated airflow, stopping, and pressure measurement, this invention only requires two airflows to establish pneumoperitoneum, avoiding the risk of overpressure caused by the cumulative pressure acquisition deviation. It overcomes the shortcomings of existing technologies, such as lack of abdominal cavity volume estimation, lack of basis for airflow control, and lengthy and frequent trial-and-error processes in establishing pneumoperitoneum, achieving safe, efficient, and individualized pneumoperitoneum establishment.

[0018] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention are realized and obtained in accordance with the structures particularly pointed out in the description, claims and drawings.

[0019] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0020] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0021] Figure 1 A schematic diagram illustrating the composition of an intra-abdominal pressure control system based on a shallow neural network model provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of a conventional pneumoperitoneum machine in an embodiment of the present invention; Figure 3 for Figure 2 The diagram shows the connection of the pneumoperitoneum machine; Figure 4 A flowchart illustrating the intra-abdominal pressure control method based on a shallow neural network model provided in an embodiment of the present invention; Figure 5 This is a schematic diagram of the electronic device structure provided in an embodiment of the present invention.

[0022] Figure label: 130 - Processor, 131 - Memory, 132 - Bus, 133 - Communication interface. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] To facilitate understanding of this embodiment, the technical terms used in this invention will be briefly introduced below.

[0025] The pneumoperitoneum establishment phase refers to the process from the initial injection of gas into the abdominal cavity until the abdominal pressure first reaches the preset target pressure. During this phase, the pneumoperitoneum machine is only connected to the air inlet line; the pressure measurement line and the exhaust line are not yet connected to the abdominal cavity.

[0026] The pneumoperitoneum maintenance phase refers to the process where, after the abdominal pressure reaches the target pressure, the pneumoperitoneum machine continues to operate to maintain stable pressure. During this phase, the air intake line, pressure monitoring line, and smoke exhaust line are all connected to the abdominal cavity. The pressure monitoring line monitors the abdominal pressure in real time, and the smoke exhaust line continuously draws in smoke.

[0027] Fixed air leakage: refers to a stable air leakage phenomenon caused by continuous minor leakage at the gaps between instruments and the abdominal wall or at the connection of tubing during the pneumoperitoneum maintenance phase.

[0028] After introducing the technical terms involved in this invention, the application scenarios and design concepts of the embodiments of this invention will be briefly described below.

[0029] Existing pneumoperitoneum machines cannot directly measure intra-abdominal pressure during the insufflation phase. They can only gradually inflate by repeatedly inflating, stopping, and measuring pressure, resulting in low insufficiency, easy overpressure, lack of intra-abdominal volume calculation mechanism, blind insufflation control, and inability to adapt to individual differences.

[0030] Based on this, embodiments of the present invention provide an intra-abdominal pressure control system and method based on a shallow neural network model.

[0031] Example 1 This invention provides an intra-abdominal pressure control system based on a shallow neural network model, such as... Figure 1 As shown, the system includes a sensor module, an execution module, and a control unit.

[0032] The sensor module includes a pressure sensor and a flow sensor; the pressure sensor is installed on the pressure measuring line and the air intake line of the insufflator to monitor the pressure changes in the abdominal cavity in real time; the flow sensor is installed on the air intake line and the exhaust line of the insufflator to measure the gas flow rate of intake and exhaust.

[0033] The execution module includes a proportional valve and a switching valve installed on the intake pipe, and a vent valve installed on the exhaust pipe. The execution module automatically adjusts the opening degree of the proportional valve to control the intake flow, controls the on / off state of the switching valve, and controls the opening and closing of the vent valve according to the control signal from the central processing module.

[0034] The control unit communicates with the sensor module and the execution module. The control unit includes a correction module, a cavity volume calculation unit, a fixed leakage detection unit, and a volume compensation module. The correction module incorporates a shallow neural network model. The control unit outputs correction coefficients based on the physiological parameters of the target object, the operation type code, and the initial air intake parameters. It calculates the basic abdominal cavity volume based on the correction coefficients and the initial air intake parameters, and updates the actual abdominal cavity volume based on the leakage volume. During the maintenance phase, it determines whether a fixed leakage exists based on the air intake and exhaust flow rates and the real-time abdominal cavity pressure, and determines the leakage flow rate. It calculates the leakage volume based on the leakage flow rate and leakage duration, and generates a compensation command. The control unit also controls the execution module to supplement air during the pneumoperitoneum establishment phase based on the basic abdominal cavity volume and the target pressure to complete the pneumoperitoneum establishment, and dynamically controls the execution module to compensate for the leakage volume during the maintenance phase based on the actual abdominal cavity volume and pressure deviation.

[0035] The shallow neural network model built into the control unit of this invention outputs individualized correction coefficients based on the initial data collected by the sensor module, combined with the physiological parameters and operation type encoding of the target object. The control unit accurately calculates the basic volume of the abdominal cavity and the required air supplementation volume based on the correction coefficients, and controls the execution module to complete the air supplementation in one go, thereby establishing pneumoperitoneum. Compared with the existing trial mode of repeated air intake, stop, and pressure measurement, this invention only requires two air intakes to establish pneumoperitoneum, avoiding the risk of overpressure caused by the superposition of pressure acquisition deviations, and avoiding the defects of lengthy and frequent trial-and-error process in establishing pneumoperitoneum, thus achieving safe, efficient, and individualized establishment of pneumoperitoneum.

[0036] like Figure 2 , Figure 3 As shown, the existing pneumoperitoneum machine has three parallel pipelines inside: an air inlet pipeline, a pressure measuring pipeline, and a smoke exhaust pipeline.

[0037] The intake pipeline connects sequentially from the gas source end to the intake pressure reducing and filtering module, the on / off valve, the proportional valve, the flow sensor, and the pressure sensor. The intake pressure reducing and filtering module is located at the gas source outlet and is used to reduce and filter the input medical CO2 gas to ensure gas cleanliness and stable pressure. The on / off valve serves as the main on / off control and is usually located at the beginning of the pipeline. The proportional valve is located after the on / off valve and is used to precisely regulate the intake flow rate. The flow sensor and pressure sensor are located downstream of the proportional valve and are used to monitor the instantaneous intake flow rate and the static pressure in the pipeline, respectively.

[0038] The pressure measurement line has a separate branch line with a pressure sensor. This line is connected to the abdominal cavity during the pneumoperitoneum maintenance phase to collect the real abdominal pressure in real time, and is closed or disconnected during the establishment phase.

[0039] The exhaust duct is equipped with an air pump and a pressure sensor. The air pump is located in the middle of the duct and is used to actively draw out the smoke and exhaust gas generated in the abdominal cavity during the maintenance phase. The pressure sensor is used to monitor the pressure on the exhaust side.

[0040] The control unit, located inside the insufflator, serves as the overall control center. It is electrically connected to all the aforementioned sensors, actuators (switching valves, proportional valves, air pumps, and venting valves), and the display interface. Located on the front panel of the insufflator, the interface allows operators to input parameters such as target pressure and airflow rate, and displays real-time pressure and flow data, as well as initiating / stopping insufflation. The control unit receives sensor signals, runs the control program, and sends control commands to each actuator.

[0041] The vent valve is usually located on a branch of the air intake line or pressure measuring line and is used to automatically open and release pressure when the abdominal pressure exceeds a safe threshold.

[0042] like Figure 3As shown, during the pneumoperitoneum establishment phase, only the air intake line is connected; the pressure measurement line and the smoke exhaust line are not yet connected to the target's abdominal cavity. Therefore, the control board cannot obtain the actual abdominal cavity pressure through the pressure measurement line and can only use an alternating pattern of air intake, stop, wait, collecting the static pressure of the air intake line as an approximation of the abdominal cavity pressure, judgment, and circulation. After pneumoperitoneum establishment is completed, the air intake line, pressure measurement line, and smoke exhaust line are simultaneously connected to the abdominal cavity through a trocar, entering the maintenance phase: the control unit controls the proportional valve to adjust the air intake flow rate based on the pressure value fed back from the pressure measurement line, the air pump continuously draws in and exhausts smoke, and the vent valve opens when there is overpressure, thereby achieving basic pressure stabilization.

[0043] This invention uses a shallow neural network model built into the control unit to output individualized correction coefficients based on parameters such as the physiological parameters and operation type encoding of the input object. This allows for the accurate calculation of the basic abdominal cavity volume and the required air supply. The execution module completes the air supply in one go, requiring only two air intakes to establish pneumoperitoneum. This avoids the risk of overpressure caused by pressure acquisition deviations and solves the problems of inability to calculate abdominal cavity volume, blind air intake control, and low establishment efficiency.

[0044] In conjunction with the first aspect, a shallow neural network model includes: an input layer, a hidden layer, and an output layer.

[0045] The input layer includes nodes for the physiological parameters of the target object, operation type encoding, single intake flow rate, single intake time, and pressure difference before and after intake.

[0046] The hidden layer is a single-layer structure, consisting of multiple nodes. Each node is connected to all nodes in the input layer through trainable weights, and a non-linear activation function is used to transform the weighted sum.

[0047] The output layer contains one node. The node in the output layer is connected to all nodes in the hidden layer through trainable weights and uses an activation function with output limiting to constrain the output value within a specified range.

[0048] In this embodiment, the shallow neural network model adopts a three-layer structure, including an input layer, a single hidden layer, and an output layer. The overall design is lightweight and adaptable to the embedded hardware environment of the pneumoperitoneum control unit.

[0049] The input layer contains five nodes, corresponding to: the target's physiological parameters BMI, operation type code, and single intake flow rate. Single intake time and the pressure difference before and after intake. Among them, physiological parameters are used to quantify the body shape of the target subject; operation type coding is used to distinguish the impact of different surgical procedures on abdominal cavity volume; single inhalation flow rate and time jointly determine the initial injected gas volume; and the pressure difference before and after inhalation reflects the pressure response characteristics of the abdominal cavity after the initial inhalation. These five parameters serve as input features of the model and jointly participate in the generation of correction coefficients.

[0050] The hidden layer is a single-layer structure containing multiple nodes (preferably 10 in this embodiment). Each hidden layer node is fully connected to all 5 nodes of the input layer through trainable weights. The calculation process is as follows: first, the weighted sum of the input feature vectors is calculated and a bias parameter is added; then, the result is input into a non-linear activation function for transformation.

[0051] In this embodiment, the ReLU activation function is specifically used, as follows: .

[0052] The weight parameters are initialized using Xavier, and their value range is [value range missing]. ,in The number of nodes in the input layer is 5. The number of hidden layer nodes is set to 10; the bias parameter is initialized to 0.1 to accelerate model training convergence. The introduction of a nonlinear activation function enables the model to learn the complex nonlinear mapping relationship between input features and output correction coefficients, overcoming the limitations of insufficient expressive power of linear models.

[0053] The output layer contains one node for outputting correction coefficients. This output layer node is fully connected to all 10 nodes in the hidden layer via trainable weights. The calculation process is as follows: first, the weighted sum of the hidden layer outputs is calculated and a bias is added; then, the result is input into an activation function with output limiting functionality. In this embodiment, a custom... The function, with the formula: The system also includes output constraint logic: if the original output value of the model exceeds the range of 0.9 to 1.1, it will be automatically cropped to the nearest boundary value (for example, an output of 0.85 will be cropped to 0.9, and an output of 1.12 will be cropped to 1.1). The purpose of output limiting is to ensure the rationality of the correction coefficients, adapt to the compensation needs under different target objects' physiological parameters and operation types, and avoid distortion in subsequent abdominal cavity volume calculations due to extreme values.

[0054] The entire model has approximately 71 parameters, specifically including: 50 weights (5×10) from the input layer to the hidden layer, 10 hidden layer biases, 10 weights (10×1) from the hidden layer to the output layer, and 1 output layer bias. In this embodiment, the lightweight model architecture described above can run efficiently on the microprocessor of the insufflator control motherboard, enabling real-time inference.

[0055] In conjunction with the first aspect, shallow neural network models are obtained through offline training in the following manner: Multiple samples were collected, each of which included multiple input features and their corresponding true correction coefficients. The true correction coefficients were calculated from the actual abdominal cavity volume, air intake flow rate, air intake time, and pressure difference before and after air intake in the actual scenario of the sample. Multiple input features are normalized, and the samples are divided into training set, validation set and test set; Iterative training is performed using optimization algorithms and loss functions until the loss value on the validation set converges and the prediction accuracy on the test set reaches the preset requirements.

[0056] In this embodiment, the shallow neural network model is pre-trained offline. The training process is independent of the pneumoperitoneum establishment process. After training is completed, the parameters are burned into the storage unit of the control unit. Subsequent background updates can be performed via USB or other interfaces without modifying the hardware.

[0057] Training data acquisition: Collect 300 to 500 samples from different scenarios. Each sample contains input feature sets and corresponding true label values, and near-true correction coefficients. .in, The method for obtaining the data is as follows: under the corresponding input feature scenario, obtain the actual abdominal cavity volume when the target pressure is reached through the actual scenario. Then, based on the ideal gas law, using the formula: By reverse deduction, that is This process ensures the authenticity and accuracy of the label values.

[0058] To improve accuracy, data preprocessing is typically performed before actual training: outlier samples are removed, such as those with zero intake flow rate or pressure difference, physiological parameters of the target object exceeding the range of 15-35, or true correction coefficients exceeding the range of 0.8-1.2. Min-Max normalization is applied to the five input features, mapping each feature value to the [0,1] interval to eliminate the biased influence of different units (such as the unit difference between the physiological parameters of the target object and the flow rate) on model training.

[0059] The normalization formula is: ;in, These are the original eigenvalues. The maximum value of the feature in the sample; This represents the minimum value of the feature in the sample.

[0060] The preprocessed samples were divided into training set (70%), validation set (20%) and test set (10%) in a ratio of 7:2:1.

[0061] Next, the training parameters were set: a stochastic gradient descent (SGD) optimizer was used, with a learning rate of 0.03 and momentum of 0.9 to reduce oscillations during training and accelerate convergence. Mean squared error (MSE) was used as the loss function, calculated as follows: ;in, For sample size The first output of the model Predicted values ​​for each sample For the first The true label value of each sample. The training objective is to minimize this loss value. The training epochs are set to 150. After each epoch, the model accuracy is verified using a validation set. If the loss value on the validation set does not decrease for 10 consecutive epochs, training is stopped early to prevent overfitting. In addition, L2 regularization is added, with the weight decay coefficient λ set to 0.001 to avoid the model overlearning noise in the training samples and improve generalization ability.

[0062] Training convergence criterion: The model is considered to have completed training when the validation set loss is less than 0.0001 and the proportion of samples in the test set whose absolute error between the predicted and true correction coefficients does not exceed 0.02 is not less than 95%. After training, the model's weights and bias parameters are saved as binary files and burned to the control unit for storage. For example, specific parameters include: Input layer to hidden layer weights: Dimensions 5×10; Hidden layer bias: Dimension 10; Hidden layer to output layer weights: 10×1 dimensions; Output layer bias: Dimension 1.

[0063] The model parameters can be updated subsequently via the back-end interface of the pneumoperitoneum machine (such as USB upgrade) without any hardware modifications.

[0064] In conjunction with the first aspect, the correction module is used to collect the target object's physiological parameters, operation type code, single inhalation flow rate, single inhalation time, and pressure difference before and after inhalation after the initial inhalation is completed during the pneumoperitoneum establishment phase. The physiological parameters, operation type encoding, single air intake flow rate, single air intake time, and pressure difference before and after air intake of the target object are subjected to Min-Max normalization. The minimum and maximum values ​​used for normalization are the same as the minimum and maximum values ​​of the corresponding features used during offline training. The normalized physiological parameters of the target object, operation type encoding, single air intake flow rate, single air intake time, and pressure difference before and after air intake are used as the input values ​​of the corresponding nodes in the input layer of the shallow neural network model. The input values ​​of each node in the input layer are input into the shallow neural network model. The weighted sum of the input layer and the hidden layer is calculated in sequence and the hidden layer bias is superimposed. The hidden layer output is obtained by transforming through a nonlinear activation function. Then the weighted sum of the hidden layer and the output layer is calculated and the output layer bias is superimposed. The prediction correction coefficient is obtained by transforming through an activation function with output limiting. Output constraints are applied to the predicted correction coefficients: if they exceed the range of 0.9 to 1.1, they are pruned to the nearest boundary value; The constrained correction coefficients are transmitted to the cavity volume calculation unit.

[0065] In this embodiment, the correction module (i.e., the correction module) performs the following real-time inference steps during the pneumoperitoneum establishment phase: First, after the initial small-step air intake is completed, the correction module simultaneously collects five input features: the target's physiological parameters (BMI), operation type code, single air intake flow rate (BMI, B ... ), single intake time ( ) and the pressure difference before and after intake ( ).

[0066] Among them, single intake flow rate and single intake time The flow sensor on the intake manifold monitors the airflow in real time and transmits the data to the correction module via the intake control module; the pressure difference before and after intake... The abdominal pressure before initial air intake is measured by the static pressure sensor on the intake pipe. Assuming a pressure of 0 mmHg, after the air intake is complete, wait briefly to allow the airflow to stabilize, and then collect the static pressure. ,calculate The five features mentioned above together form the input basis for model inference.

[0067] Secondly, the correction module performs Min-Max normalization on the five collected features. The minimum and maximum values ​​used for normalization are exactly the same as the minimum and maximum values ​​of the corresponding features used during offline training to ensure the consistency of the input data distribution. The purpose of normalization is to eliminate the influence of different units (such as physiological parameters of the target object in kg / m², and flow rate in L / min) on the model calculation, so that each feature is numerically on the same order of magnitude (usually mapped to the [0,1] interval), thereby avoiding the model bias towards features with larger values.

[0068] Then, the correction module uses the five normalized feature values ​​as the input values ​​for the corresponding nodes in the input layer of the shallow neural network model. Specifically, the input layer has a total of 5 nodes, which correspond to the physiological parameters of the target object, the operation type code, the single air intake flow rate, the single air intake time, and the pressure difference before and after air intake, respectively. The input value of each node is the normalized corresponding feature value.

[0069] Next, the correction module feeds the input values ​​from each node of the input layer into the shallow neural network model and performs forward propagation calculations. First, it calculates the weighted sum of the input layer and the hidden layer, and then adds the hidden layer biases. ,Right now: ,in, is the weight matrix from the input layer to the hidden layer (dimension 5×10), and x is the normalized input feature vector (dimension 5×1). This is the hidden layer bias vector (10×1 dimension). The weighted sum is then passed through a non-linear activation function.

[0070] Then, the weighted sum of the hidden layer and the output layer is calculated and the output layer bias is added, i.e.: ,in, This is the weight matrix from the hidden layer to the output layer (dimension 10×1). This is the output layer bias (scalar). This result is then transformed using an activation function with output limiting to obtain the prediction correction coefficients. .

[0071] To ensure the physical reasonableness of the output values, the correction module applies output constraints to the predicted correction coefficients: if If the value exceeds the preset range (specifically 0.9~1.1 in this embodiment), it will be automatically cropped to the nearest boundary value (for example, an output of 0.85 will be cropped to 0.9, and an output of 1.12 will be cropped to 1.1). This constraint avoids distortion in subsequent abdominal cavity volume calculations due to extreme outliers, while ensuring that the model's compensation requirements for physiological parameters and operation types of different target objects are always within a safe range.

[0072] Finally, the correction module transmits the constrained correction coefficients to the cavity volume calculation unit. Through this individualized correction, the abdominal cavity volume can be accurately estimated based on a single initial air intake, thus providing a basis for precise air replenishment and avoiding the problems of repeated trial and error, easy overpressure, and low efficiency of existing technologies.

[0073] In summary, the real-time inference process of the correction module fully realizes the automated process from feature acquisition, normalization, forward propagation to output limiting and coefficient transmission. The entire calculation is lightweight (only 71 parameters) and can run in real time on the pneumoperitoneum control unit, thereby achieving individualized and precise gas control.

[0074] In conjunction with the first aspect, the cavity volume calculation unit is used to receive the correction coefficients output by the shallow neural network model; the correction coefficients are calculated by multiplying the single air intake flow rate by the single air intake time, and then divided by the pressure difference before and after air intake to obtain the basic volume of the abdominal cavity.

[0075] The cavity volume calculation unit is also used to add the basic volume of the abdominal cavity to the leakage volume when the leakage volume is obtained, and update the actual volume of the abdominal cavity.

[0076] In this embodiment, the cavity volume calculation unit is signal-connected to the correction module, and is used to receive the correction coefficients output by the shallow neural network model in the correction module. At the same time, it also obtains initial intake operating condition parameters from the sensor module or other parts of the control unit, including: single intake flow rate. Single intake time and the pressure difference before and after intake. Based on the ideal gas law, the basic volume of the peritoneal cavity. The calculation formula is: .

[0077] In this way, during the establishment phase, based on the correction coefficients output by the shallow neural network, the basic volume of the abdominal cavity can be accurately calculated with only one initial inhalation, thereby achieving the establishment of pneumoperitoneum with one inhalation, avoiding overpressure and greatly improving efficiency.

[0078] During the pneumoperitoneum maintenance phase, the volume of gas in the abdominal cavity continuously decreases due to a fixed gap between the trocar and the abdominal wall or minor leaks at tubing connections, leading to a drop in pressure. To maintain pressure stability, dynamic compensation for the leaked volume is necessary.

[0079] The fixed air leakage detection module monitors the air intake flow rate in the intake pipeline in real time. Exhaust flow rate of flue gas duct and real-time abdominal pressure When a fixed leak is determined to exist, calculate the average leak flow rate. The volume compensation module further adjusts the calculation based on the duration of the leak. Calculate the leakage volume The leak volume is then transmitted to the cavity volume calculation unit.

[0080] The cavity volume calculation unit receives the leakage volume. Then, the previously preserved basic volume of the abdominal cavity was... Add this to the current leakage volume to obtain the real-time true volume of the abdominal cavity. : The actual volume of the abdominal cavity. The data is transmitted to the intake control module (i.e., the proportional valve control section within the execution module). Based on the deviation between the current actual abdominal cavity volume and the target pressure, the intake control module dynamically adjusts the opening of the proportional valve to precisely compensate for any leaked gas, thereby stabilizing the abdominal cavity pressure near the target value and ensuring a stable field of vision during operation.

[0081] During the maintenance phase, by updating the actual abdominal cavity volume in real time, a data basis for dynamic volume compensation is provided, which reduces the pressure fluctuation range from the existing ±1~2 mmHg to within ±0.5 mmHg, thereby improving the stability of pneumoperitoneum.

[0082] In conjunction with the first aspect, the fixed leak detection unit is used for: During the pneumoperitoneum maintenance phase, the intake flow rate of the intake pipe, the exhaust flow rate of the exhaust pipe, and the real-time abdominal pressure are obtained in real time. Calculate the flow difference between the intake flow rate and the exhaust flow rate, as well as the pressure difference between the preset target pressure and the real-time abdominal pressure; When the fluctuation range of the flow rate difference is less than or equal to the preset first fluctuation threshold, the fluctuation range of the pressure difference is less than or equal to the preset second fluctuation threshold, and the duration of the fluctuation range is greater than or equal to the preset duration, it is determined that there is a fixed air leak. The average leakage flow rate is calculated based on the flow rate difference over a continuous period of time, and then transmitted to the volume compensation module.

[0083] During the pneumoperitoneum maintenance phase, the air intake line, pressure monitoring line, and smoke exhaust line are all connected to the target's abdominal cavity. A fixed leak detection unit acquires the following three data streams in real time via signal connection: Intake flow rate of intake pipe The source is the flow sensor installed on the intake manifold; Exhaust flow rate of the flue gas duct The flow rate is derived from the flow sensor installed on the exhaust pipe (or the flow monitoring device built into the air pump). Real-time abdominal pressure The pressure is derived from a pressure sensor installed on the pressure measuring line. This sensor is directly connected to the abdominal cavity and can reflect the pressure changes within the abdominal cavity in real time.

[0084] Flow difference This difference reflects the net increase in gas volume within the abdominal cavity per unit time. If there is no leakage, then under steady-state conditions... It should be close to zero (intake and exhaust are basically balanced); if there is a persistent, fixed leak, then It will stabilize within a positive range, and its value will be equal to the leakage flow rate.

[0085] Pressure difference ,in, The preset target pressure (e.g., 12 mmHg). This represents the current real-time abdominal pressure. This difference reflects the deviation between the current pressure and the target pressure. In the presence of a fixed air leak, the air intake control module will actively replenish air to maintain pressure. It will stabilize within a relatively small fluctuation range.

[0086] In this embodiment, calculation is performed within a certain time window. The difference between the maximum and minimum values ​​(or statistical measures such as standard deviation) indicates that the leakage flow rate is relatively constant when the fluctuation amplitude is less than or equal to the preset first fluctuation threshold (in this embodiment, the first fluctuation threshold is 1L / min).

[0087] Monitor pressure difference Fluctuation amplitude: Similarly, calculate the difference between its maximum and minimum values. When the fluctuation amplitude is less than or equal to the preset second fluctuation threshold (in this embodiment, the first fluctuation threshold is 0.5 mmHg), it indicates that the pressure has basically stabilized.

[0088] The duration of the above fluctuation state is monitored: when the two fluctuation amplitudes meet the conditions at the same time, and the duration is greater than or equal to the preset duration (in this embodiment, the preset duration is 3 seconds), it is determined that there is a fixed air leak, rather than a momentary air leak (such as a brief leak caused by the insertion or removal of a device).

[0089] Once a fixed leak is identified, the fixed leak identification unit will use the flow rate difference sampled during this period. Calculate the average value as the average leakage flow rate. Specifically, within 3 seconds The average value. This average leakage flow rate represents the stable gas leakage rate per unit time.

[0090] The fixed leak detection unit will calculate the... The signal is transmitted to the volume compensation module for further calculation of the leakage volume, which is then used to update the true volume of the abdominal cavity and adjust the air intake flow rate.

[0091] The fixed leak detection unit can accurately distinguish between fixed leaks and transient interference, avoiding misjudgments caused by brief pressure fluctuations or flow rate jumps, thus providing reliable basic data for volume compensation. Compared to existing technologies that can only passively replenish gas and cannot identify the type of leak, it can achieve accurate identification and quantitative measurement of fixed leaks.

[0092] In conjunction with the first aspect, the volume compensation module is used for: Receive the average leakage flow rate transmitted by the fixed leakage detection unit; Calculate the leakage volume based on the average leakage flow rate and leakage duration; The leak volume is transmitted to the cavity volume calculation unit to update the true volume of the abdominal cavity; It also generates compensation instructions based on the updated actual abdominal cavity volume and pressure difference, and sends the compensation instructions to the execution module to dynamically adjust the intake flow rate of the intake pipeline.

[0093] The volume compensation module simultaneously acquires the leakage duration, i.e., the cumulative time from the start of the fixed leakage detection to the current moment. Based on the product of the cumulative time and the average leakage flow rate, it calculates the total volume of gas lost from the abdominal cavity due to the fixed leakage during the leakage duration. The volume compensation module transmits the calculated leakage volume to the cavity volume calculation unit via a signal connection. After receiving the leakage volume, the cavity volume calculation unit adds the previously saved basic abdominal cavity volume to the current leakage volume to obtain the updated true abdominal cavity volume. It then generates an intake flow rate compensation command based on the updated true abdominal cavity volume and the current pressure difference (the difference between the preset target pressure and the real-time abdominal cavity pressure). This command includes the intake flow rate adjustment amount required to compensate for the leaked gas volume and return the abdominal cavity pressure to the target value. The compensation command is sent to the execution module via a signal connection, specifically acting on the proportional valve on the intake pipeline. The execution module dynamically adjusts the opening of the proportional valve according to the command, precisely increasing the intake flow rate, thereby compensating for the gas lost due to leakage and maintaining the abdominal cavity pressure near the target pressure. Compared with passive pneumoperitoneum that relies solely on pressure feedback, this module significantly reduces the pressure fluctuation range (from ±1~2 mmHg to within ±0.5 mmHg) through feedforward volume calculation, improving the stability of the pneumoperitoneum maintenance phase and providing a more stable visual environment for laparoscopic operations.

[0094] Understandably, this update process is dynamic. As the duration of the leak increases, the leak volume gradually accumulates, and the actual volume of the abdominal cavity is adjusted accordingly, providing an accurate volume reference for subsequent air intake flow control.

[0095] Secondly, this invention also provides a method for controlling intra-abdominal pressure based on a shallow neural network model, applied to the system described above. Combined with... Figure 4 As shown, the method includes: S110, acquire the physiological parameters of the target object, the operation type code, and the preset target pressure.

[0096] S120 controls the intake pipeline to perform an initial intake at a preset flow rate and time. After the intake is completed, the static pressure of the intake pipeline is collected, and the pressure difference before and after the intake is calculated.

[0097] S130 inputs physiological parameters, operation type encoding, preset air intake flow rate, preset air intake time, and pressure difference before and after air intake into the shallow neural network model in the correction module, and outputs correction coefficients.

[0098] S140 calculates the basic volume of the abdominal cavity using the cavity volume calculation unit, based on the correction coefficient, preset air intake flow rate, preset air intake time, and pressure difference before and after air intake.

[0099] S150, through the control unit, calculates the required amount of gas supplementation based on the basic volume of the abdominal cavity and the target pressure, and controls the execution module to perform gas supplementation once according to the amount of gas supplementation to complete the establishment of pneumoperitoneum.

[0100] In this embodiment, real-time operating parameters are acquired through an initial air intake, and combined with the physiological parameters of the target object and the operation type encoding, input into a shallow neural network model to quickly output individualized correction coefficients, thereby calculating the basic abdominal cavity volume. Based on this, the amount of supplementary air required to reach the target pressure is accurately calculated, and the execution module completes the supplementary air intake in one go. Compared to existing repeated air intake methods... Pressure measurement The trial-and-error mode of judgment allows this method to establish pneumoperitoneum with only two air inhalations, avoiding the risk of overpressure caused by the superposition of pressure acquisition deviations and the dependence on operator experience. At the same time, it solves the problems of existing technologies not being able to know the abdominal cavity volume and blind air inhalation control, improving the efficiency of pneumoperitoneum establishment and eliminating the risk of tissue damage.

[0101] In conjunction with the second aspect, step S150, in which the control execution module performs a gas replenishment according to the replenishment amount, also includes: S151, obtain real-time abdominal pressure.

[0102] S152, when the real-time abdominal pressure exceeds the preset safety threshold, control the pressure relief valve to open and relieve pressure.

[0103] During the gas replenishment process, the real-time abdominal pressure is continuously acquired and compared with a preset safety threshold. Once the real-time pressure exceeds the safety threshold, the pressure relief valve is immediately opened to release gas, allowing the abdominal pressure to drop back to a safe range. In this way, pressure can be actively relieved in the event of abnormalities such as sensor noise or external disturbances, avoiding damage to the target object due to abnormal pressure increases, thus forming a dual protection combining active overpressure prevention and passive pressure relief.

[0104] In conjunction with the second aspect, the method also includes: S210 can acquire the intake flow rate of the intake pipe, the exhaust flow rate of the exhaust pipe, and the real-time abdominal pressure in real time.

[0105] S220 calculates the flow difference between the intake flow rate and the exhaust flow rate, as well as the pressure difference between the target pressure and the real-time abdominal pressure.

[0106] S230, when the fluctuation range of the flow rate difference meets the preset first stability condition, the fluctuation range of the pressure difference meets the preset second stability condition, and the duration reaches the preset duration, it is determined that there is a fixed air leak, and the average leakage flow rate is determined based on the flow rate difference within the duration.

[0107] S240, calculate the leakage volume based on the average leakage flow rate and leakage duration, and add the leakage volume to the basic abdominal cavity volume to update the true abdominal cavity volume.

[0108] S250 dynamically adjusts the intake flow rate of the intake pipeline based on the updated actual volume of the abdominal cavity and the pressure difference to compensate for the leakage volume.

[0109] In this embodiment, the intake flow rate of the intake pipe, the exhaust flow rate of the exhaust pipe, and the real-time abdominal cavity pressure are continuously acquired from the sensor module. The net intake flow rate difference and the target pressure deviation are calculated. When the fluctuation range of the flow rate difference and the pressure deviation are consistently stable within a preset range for a set duration, it is determined to be a fixed leak rather than an instantaneous disturbance, and the average leakage flow rate is calculated accordingly. Then, the average leakage flow rate is multiplied by the leakage duration to obtain the leakage volume, which is added to the basic abdominal cavity volume to update the current true abdominal cavity volume. Finally, the intake flow rate of the intake pipe is dynamically adjusted based on the updated true volume and pressure deviation to achieve equal compensation for the leakage volume. This process can identify and quantify fixed leaks, avoid misjudgments caused by instantaneous airflow disturbances (such as interface on / off), and significantly narrow the range of abdominal cavity pressure fluctuations through feedforward volume compensation, thereby stabilizing the target air pressure environment and solving the problem of large pressure fluctuations and unstable air pressure caused by passive air replenishment relying solely on pressure feedback.

[0110] Thirdly, embodiments of the present invention provide an electronic device, combined with Figure 5 As shown, the electronic device includes a memory 131 and a processor 130. The memory 131 stores a computer program, and the processor 130 runs the computer program to make the electronic device perform the above-described method.

[0111] Furthermore, combined Figure 5 The electronic device shown also includes a bus 132 and a communication interface 133, with the processor 130, the communication interface 133 and the memory 131 connected via the bus 132.

[0112] The memory 131 may include high-speed random access memory (RAM) and may also include non-volatile memory, such as at least one disk storage device. Communication between this system network element and at least one other network element is achieved through at least one communication interface 133 (which can be wired or wireless), such as the Internet, wide area network, local area network, metropolitan area network, etc. The bus 132 may be an ISA bus, PCI bus, or EISA bus, etc. The bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 5 The symbol is represented by a single double-headed arrow, but this does not mean that there is only one bus or one type of bus.

[0113] Processor 130 may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method can be completed by the integrated logic circuitry in the hardware of processor 130 or by instructions in software form. Processor 130 may be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc.; it may also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this invention. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this invention can be directly manifested as execution by a hardware decoding processor, or execution by a combination of hardware and software modules in the decoding processor. The software module can reside in a mature storage medium in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, or registers. This storage medium is located in memory 131, and processor 130 reads the information in memory 131 and, in conjunction with its hardware, completes the steps of the method described in the foregoing embodiments.

[0114] Fourthly, embodiments of the present invention provide a readable storage medium storing computer program instructions, which are read and executed by a processor to perform the above-described method.

[0115] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the system and apparatus described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0116] Furthermore, in the description of the embodiments of the present invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in the present invention based on the specific circumstances.

[0117] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0118] In the description of this invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0119] Finally, it should be noted that the above embodiments are merely specific implementations of the present invention, used to illustrate the technical solutions of the present invention, and not to limit it. The scope of protection of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments within the technical scope disclosed in the present invention, or make equivalent substitutions for some of the technical features; and these modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A peritoneal pressure control system based on a shallow neural network model, characterized in that, include: The sensor module includes a pressure sensor and a flow sensor; the pressure sensor is used to monitor pressure changes within the abdominal cavity in real time; the flow sensor is used to measure the gas flow rate during intake and exhaust. The execution module includes a proportional valve and a switching valve located on the intake pipe, and a vent valve located on the exhaust pipe; The control unit is communicatively connected to the sensor module and the execution module; the control unit includes a correction module, a cavity volume calculation unit, a fixed leakage identification unit, and a volume compensation module; the correction module has a built-in shallow neural network model; the control unit is used to output correction coefficients based on the physiological parameters of the input target object, the operation type code, and the initial air intake condition parameters; calculate the basic abdominal cavity volume based on the correction coefficients and the initial air intake condition parameters, and update the true abdominal cavity volume based on the leakage volume; During the maintenance phase, the presence of a fixed leak is determined based on the intake and exhaust flow rates and the real-time abdominal pressure, and the leak flow rate is determined; the leak volume is calculated based on the leak flow rate and the leak duration, and a compensation command is generated. The control unit is also used to control the execution module to supplement air to complete the pneumoperitoneum establishment stage based on the basic volume of the abdominal cavity and the target pressure, and to dynamically control the execution module to compensate for the leakage volume based on the actual volume of the abdominal cavity and the pressure deviation during the maintenance stage.

2. The system according to claim 1, characterized in that, The shallow neural network model includes: The input layer includes nodes for the target object's physiological parameters, operation type encoding, single air intake flow rate, single air intake time, and pressure difference before and after air intake. The hidden layer is a single-layer structure. The hidden layer includes multiple nodes, and each node is connected to all nodes of the input layer through trainable weights. A non-linear activation function is used to transform the weighted sum. The output layer contains one node. The node in the output layer is connected to all nodes in the hidden layer through trainable weights and uses an activation function with output limiting to constrain the output value within a specified range.

3. The system according to claim 2, characterized in that, The shallow neural network model was obtained through offline training in the following manner: Multiple samples are collected, each of which contains an input feature set and a corresponding true correction coefficient. The true correction coefficient is calculated from the actual abdominal cavity volume, air intake flow rate, air intake time, and pressure difference before and after air intake in the actual scenario of the sample. The input features are normalized, and the samples are divided into a training set, a validation set, and a test set. Iterative training is performed using optimization algorithms and loss functions until the loss value on the validation set converges and the prediction accuracy on the test set reaches the preset requirements.

4. The system according to claim 1, characterized in that, The correction module is used for: After the initial inhalation during the pneumoperitoneum establishment phase, physiological parameters, operation type codes, single inhalation flow rate, single inhalation time, and pressure difference before and after inhalation of the target subject are collected. The physiological parameters, operation type encoding, single air intake flow rate, single air intake time, and pressure difference before and after air intake of the target object are subjected to Min-Max normalization processing, wherein the minimum and maximum values ​​used for normalization are the same as the minimum and maximum values ​​of the corresponding features used during offline training. The normalized physiological parameters, operation type code, single air intake flow rate, single air intake time, and pressure difference before and after air intake of the target object are used as the input values ​​of the corresponding nodes in the input layer of the shallow neural network model. The input values ​​of each node in the input layer are input into the shallow neural network model. The weighted sum of the input layer and the hidden layer is calculated sequentially and the hidden layer bias is superimposed. The hidden layer output is obtained by transformation through a nonlinear activation function. Then, the weighted sum of the hidden layer and the output layer is calculated and the output layer bias is superimposed. The prediction correction coefficient is obtained by transformation through an activation function with output limiting. Output constraints are applied to the predicted correction coefficients: if they exceed a preset range, they are pruned to the nearest boundary value; The constrained correction coefficients are transmitted to the cavity volume calculation unit.

5. The system according to claim 1, characterized in that, The cavity volume calculation unit is used to receive the correction coefficient output by the shallow neural network model; calculate the correction coefficient by multiplying the single air intake flow rate by the single air intake time, and then divide it by the pressure difference before and after air intake to obtain the basic volume of the abdominal cavity. The cavity volume calculation unit is also used to add the basic abdominal cavity volume to the leakage volume when the leakage volume is obtained, and update the actual abdominal cavity volume.

6. The system according to claim 1, characterized in that, The fixed leak detection unit is used for: During the pneumoperitoneum maintenance phase, the intake flow rate of the intake pipe, the exhaust flow rate of the exhaust pipe, and the real-time abdominal pressure are obtained in real time. Calculate the flow difference between the intake flow rate and the exhaust flow rate, and the pressure difference between the preset target pressure and the real-time abdominal pressure; When the fluctuation range of the flow rate difference is less than or equal to a preset first fluctuation threshold, the fluctuation range of the pressure difference is less than or equal to a preset second fluctuation threshold, and the duration of the fluctuation range is greater than or equal to a preset duration, it is determined that there is a fixed air leak. The average leakage flow rate is calculated based on the flow rate difference within the duration, and the average leakage flow rate is transmitted to the volume compensation module.

7. The system according to claim 5, characterized in that, The volume compensation module is used for: Receive the average leakage flow rate transmitted by the fixed leakage detection unit; Calculate the leakage volume based on the average leakage flow rate and leakage duration; The leak volume is transmitted to the cavity volume calculation unit to update the true volume of the abdominal cavity; The system generates a compensation command based on the updated actual abdominal cavity volume and pressure difference, and sends the compensation command to the execution module to dynamically adjust the air intake flow rate of the air intake pipeline.

8. A method for controlling intra-abdominal pressure based on a shallow neural network model, characterized in that, Applied to the system as described in any one of claims 1-7; the method comprises: Acquire the target object's physiological parameters, operation type encoding, and preset target pressure; The intake pipeline is controlled to perform an initial intake at a preset flow rate and time. After the intake is completed, the static pressure of the intake pipeline is collected, and the pressure difference before and after the intake is calculated. The physiological parameters, the operation type code, the preset air intake flow rate, the preset air intake time, and the pressure difference before and after air intake are input into the shallow neural network model in the correction module, and the correction coefficient is output. The basic volume of the abdominal cavity is calculated by the cavity volume calculation unit based on the correction coefficient, the preset air intake flow rate, the preset air intake time, and the pressure difference before and after air intake. The control unit calculates the required amount of air replenishment based on the basic abdominal cavity volume and the target pressure, and controls the execution module to replenish air once according to the required amount, thus completing the establishment of pneumoperitoneum.

9. The method according to claim 8, characterized in that, The step of controlling the execution module to perform one air replenishment according to the air replenishment amount further includes: Obtain real-time abdominal pressure; When the real-time pressure in the abdominal cavity exceeds a preset safety threshold, the pressure relief valve is opened to release pressure.

10. The method according to claim 8, characterized in that, The method further includes: Real-time acquisition of air intake flow rate in the intake pipe, exhaust flow rate in the exhaust pipe, and real-time abdominal pressure; Calculate the flow rate difference between the intake flow rate and the exhaust flow rate, and the pressure difference between the target pressure and the real-time abdominal pressure; When the fluctuation range of the flow rate difference meets the preset first stability condition, the fluctuation range of the pressure difference meets the preset second stability condition, and the duration reaches the preset duration, it is determined that there is a fixed air leak, and the average air leak flow rate is determined based on the flow rate difference within the duration. The leakage volume is calculated based on the average leakage flow rate and leakage duration, and the leakage volume is added to the basic abdominal cavity volume to update the true abdominal cavity volume. The intake flow rate of the intake pipeline is dynamically adjusted based on the updated actual volume of the abdominal cavity and the pressure difference to compensate for the leakage volume.

Images