Sputum suction pressure optimization method and system combined with multi-parameter cooperative feedback

Abstract

The application discloses a sputum suction pressure optimization method and system combining multi-parameter cooperative feedback, and relates to the technical field of intelligent control of medical equipment. The method comprises the following steps: predicting and obtaining a physiological state parameter sequence and a sputum viscosity sequence of a wounded person in a preset time zone; based on the body parameters of the wounded person and the transfer scene parameters, a predicted body position feature sequence is constructed through simulation; a body position fluctuation coefficient is obtained according to the body position feature sequence, and an adaptive adjustment time window is dynamically set; taking the maximization of sputum removal efficiency and the minimization of airway mucosa injury risk as double objectives, an adaptive sputum suction negative pressure control scheme is optimized and generated in the time window by combining a generative adversarial network and a genetic algorithm; and finally, the sputum suction operation is performed according to the scheme. Through multi-parameter cooperative prediction and intelligent optimization, the application solves the poor adaptability problem of traditional sputum suction pressure control in a dynamic transfer environment, and significantly improves the accuracy, safety and reliability of sputum suction operation.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology for medical devices, specifically to a method and system for optimizing suction pressure by combining multi-parameter collaborative feedback. Background Technology

[0002] Maintaining a patent airway is a crucial first aid step during patient transport. Suctioning is a key method for clearing airway secretions and ensuring effective ventilation. Traditional suctioning pressure control methods typically rely on the experience of caregivers to set a fixed negative pressure value, or make minor adjustments based on limited static physiological parameters. However, transport scenarios are significantly dynamic and uncertain. Vehicle bumps, accelerations, decelerations, and turns cause continuous changes in the patient's position, which in turn alters the airway anatomy, affecting the position and flow of sputum within the airway, and interfering with the contact between the suction catheter and the airway mucosa.

[0003] Fixed or delayed suction pressures are ill-suited to the complex dynamic environment, leading to two adverse consequences: first, when sputum is viscous or the patient is in an unfavorable position, the preset pressure is insufficient to effectively clear sputum, increasing the risk of airway obstruction; second, when the pressure is set too high or changes in position cause mucosal vulnerability, the probability of mechanical damage to the airway mucosa increases. Current technology lacks a scheme that can proactively predict multi-dimensional changes in parameters throughout the transport process and dynamically optimize suction pressure accordingly, resulting in significant deficiencies in the accuracy, safety, and reliability of suctioning procedures in real-world transport scenarios. Summary of the Invention

[0004] This invention addresses the technical problems in existing technologies where suction pressure control is poorly adaptable to dynamic transport environments and cannot be proactively optimized in conjunction with multi-dimensional dynamic parameters, resulting in low sputum clearance efficiency and high risk of airway mucosal damage. It provides a suction pressure optimization method and system that combines multi-parameter collaborative feedback.

[0005] The technical solution of the present invention to solve the above-mentioned technical problems is as follows:

[0006] In a first aspect, the present invention provides a method for optimizing suction pressure by combining multi-parameter synergistic feedback, comprising:

[0007] Predict and obtain the target user's predicted user state parameter sequence and predicted sputum viscosity sequence within a preset time zone;

[0008] Based on the target user's basic physical parameters and the transfer scenario parameters within the preset time zone, the target user's transfer posture feature changes are analyzed, and a predicted posture feature sequence is output.

[0009] The body position change status is evaluated based on the predicted body position feature sequence, the body position fluctuation coefficient is output, and an adaptation adjustment time window is set based on the body position fluctuation coefficient.

[0010] Based on the aforementioned adaptive adjustment time window, with the dual optimization objectives of maximizing the sputum clearance efficiency index and minimizing the airway mucosal damage risk index, the suction negative pressure control scheme within the preset time zone is optimized according to the predicted user state parameter sequence, predicted sputum viscosity sequence, and predicted body position feature sequence, and an adaptive suction negative pressure control scheme is output.

[0011] The target user is subjected to optimized suctioning within the preset time zone according to the adapted suctioning negative pressure control scheme.

[0012] Secondly, the present invention provides a suction pressure optimization system that combines multi-parameter collaborative feedback, comprising:

[0013] The prediction parameter acquisition module is used to predict and acquire the target user's state parameter sequence and sputum viscosity sequence within a preset time zone.

[0014] The body position feature analysis module is used to analyze the changes in body position features of the target user during transit based on the target user's basic body parameters and the transit scenario parameters within the preset time zone, and output a predicted body position feature sequence.

[0015] The time window setting module is used to evaluate the state of postural change based on the predicted postural feature sequence, output the postural fluctuation coefficient, and set an adaptation adjustment time window based on the postural fluctuation coefficient.

[0016] The negative pressure scheme optimization module is used to optimize the suction negative pressure control scheme within the preset time zone based on the adaptive adjustment time window, with the dual optimization objectives of maximizing the sputum clearance efficiency index and minimizing the airway mucosal damage risk index, according to the predicted user state parameter sequence, predicted sputum viscosity sequence and predicted body position feature sequence, and output an adapted suction negative pressure control scheme.

[0017] The suctioning operation execution module is used to perform optimized suctioning operations for the target user within the preset time zone according to the adapted suctioning negative pressure control scheme.

[0018] The beneficial effects of this invention are:

[0019] Compared to existing technologies, this invention first constructs a parameter system that comprehensively reflects the dynamic environment of suctioning operations by acquiring and collaboratively analyzing the user's physiological state sequence, sputum viscosity sequence, and body position change sequence during transport, overcoming the limitations of traditional methods that rely on static or lagging parameters. Secondly, it innovatively introduces a body position fluctuation coefficient to quantify the impact of transport disturbances and dynamically sets an adaptive adjustment time window accordingly, enabling the adjustment rhythm of suction pressure to precisely match the dynamic characteristics of body position changes. Thirdly, with the dual optimization objectives of maximizing sputum clearance efficiency and minimizing the risk of airway mucosal damage, it comprehensively utilizes intelligent prediction and optimization algorithms to generate a globally optimal suction negative pressure control scheme, improving clearance efficiency while ensuring operational safety. Finally, by directly applying the optimized scheme to the execution of suctioning operations, a complete closed-loop control system is formed, from multi-parameter prediction and intelligent optimization to precise execution, improving the accuracy, adaptability, and reliability of suctioning operations in complex transport environments. Attached Figure Description

[0020] Figure 1 A flowchart illustrating the suction pressure optimization method combining multi-parameter collaborative feedback provided by the present invention;

[0021] Figure 2 This is a schematic diagram of the suction pressure optimization system that combines multi-parameter collaborative feedback provided by the present invention.

[0022] In the attached diagram, the components represented by each number are as follows:

[0023] Predictive parameter acquisition module 11, body position feature analysis module 12, time window setting module 13, negative pressure scheme optimization module 14, suctioning operation execution module 15. Detailed Implementation

[0024] The technical solutions of the embodiments 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, and 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.

[0025] In the description of this invention, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0026] In the description of this invention, the term "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this invention is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use the invention. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that the invention can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid obscuring the description of the invention with unnecessary detail. Therefore, the invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed herein.

[0027] Example 1, as Figure 1 As shown, embodiments of the present invention provide a method for optimizing suction pressure by combining multi-parameter collaborative feedback, including:

[0028] S10: Predict and obtain the target user's predicted user state parameter sequence and predicted sputum viscosity sequence within a preset time zone;

[0029] Specifically, the system predicts and obtains the target user's state parameter sequence and sputum viscosity sequence within a preset time zone, including:

[0030] The system monitors and acquires the user status parameter sequence and sputum viscosity sequence of the target user within a historical time zone. The target user is an injured person, and the user status parameters include the target user's blood oxygen saturation, heart rate, and respiratory rate.

[0031] Using the target user's basic physical parameters and current condition data as constraints, a sample training set is collected to train a long short-term memory network until convergence, and a state parameter predictor and a sputum viscosity predictor are constructed.

[0032] Using the state parameter predictor and sputum viscosity predictor, a predicted user state parameter sequence and a predicted sputum viscosity sequence within a preset time zone are predicted based on the user state parameter sequence and the sputum viscosity sequence.

[0033] The target users are transported patients who require airway management and suctioning. By obtaining the predicted user status parameter sequence and predicted sputum viscosity sequence within a preset time zone, the physiological state and sputum characteristics of the patients can be dynamically and proactively perceived. This provides a crucial data foundation and decision-making basis for the precise optimization of subsequent suctioning negative pressure control schemes.

[0034] First, the system monitors and acquires the target user's status parameter sequence and sputum viscosity sequence within a historical time zone. The user status parameter sequence consists of key indicators reflecting the patient's physiological state, collected continuously or periodically, specifically including blood oxygen saturation, heart rate, and respiratory rate. The sputum viscosity sequence characterizes the temporal changes in the physical properties of the patient's respiratory secretions. The historical time zone refers to a continuous monitoring period immediately preceding the current moment. Its length is determined based on the dynamic rate of change in the patient's condition and the frequency of clinical monitoring data collection; for example, it can be set to a monitoring duration of 15 to 30 minutes before the start of transport.

[0035] Specifically, the sputum viscosity sequence was obtained through indirect measurement and data fusion. Since direct online real-time sensing of airway secretions presents technical challenges, a method combining physical models and manual annotation was employed for estimation.

[0036] The core method involves establishing a model relating negative pressure and flow rate. During each suctioning procedure, the suctioning equipment simultaneously monitors and records two key physical parameters: the negative pressure applied to the airway and the real-time volumetric flow rate of the sputum in the drainage tubing. The fundamental principle is that the rheological properties of sputum determine its different flow velocities under the same negative pressure. Thin sputum has lower internal viscous resistance, resulting in a faster flow rate; while viscous sputum has higher internal viscous resistance, resulting in a slower flow rate. By collecting negative pressure and flow rate data points from multiple suctioning procedures, a pressure-flow characteristic curve model specific to the patient or type of condition can be constructed. Using this model, the relative viscosity of the sputum can be estimated by reverse calculation based on the real-time monitored negative pressure and flow rate data.

[0037] To improve the accuracy and clinical relevance of the estimation model, a manual annotation mechanism should be introduced as a model calibration method. Specifically, after each suctioning procedure, healthcare professionals are prompted to select from preset levels or manually input viscosity assessment results based on visual observation of the physical properties of the sputum in the actual collection container, such as fluidity, transparency, and whether it is stringy. These levels typically include thin, medium, and viscous. Manually annotated data validated through clinical experience can serve as supervisory signals for continuous calibration and training of the aforementioned indirect measurement model based on the negative pressure-flow relationship, thereby gradually improving its inference accuracy.

[0038] Finally, the estimated viscosity or labeled level corresponding to each suctioning operation was sorted and recorded according to the operation timestamp, forming a time-related sputum viscosity sequence. This sputum viscosity sequence dynamically reflects the changing trend of the physical properties of the patient's respiratory secretions over time, and can provide a data basis for predicting sputum viscosity in future time zones.

[0039] Secondly, dedicated prediction models are constructed, namely a state parameter predictor and a sputum viscosity predictor. First, using the target user's basic physical parameters and current condition data as constraints, sample data similar to the target user is collected and acquired to form the sample dataset required for model training. Using this training set, the Long Short-Term Memory network is trained until the model performance converges, thereby constructing the dedicated state parameter predictor and sputum viscosity predictor respectively.

[0040] Specifically, since the physiological state parameters of the wounded during the transport process have significant time-dependent and nonlinear dynamic characteristics with changes in sputum viscosity, and long short-term memory networks have unique advantages in capturing long-term dependencies and complex dynamic patterns in time series, the long short-term memory network architecture was chosen to construct the state parameter predictor and the sputum viscosity predictor, respectively.

[0041] Specifically, both the state parameter predictor and the sputum viscosity predictor employ a standard structure comprising an input layer, a long short-term memory (LSTM) network layer, a fully connected layer, and an output layer. The input layer receives a standardized temporal feature vector containing historical user state parameter sequences and sputum viscosity sequences. The LSM network layer uses a single or multi-layer structure, with the number of hidden units in each layer adaptively configured based on the input sequence length and feature dimension. Its unique gating mechanism effectively captures long-term dependencies in the temporal data. The fully connected layer transforms the dimension of the feature representation output by the LSM network layer. The output layer employs linear activation functions to map the final features to predicted user state parameter sequences and predicted sputum viscosity sequences within a preset time zone.

[0042] During training, key hyperparameters included a learning rate of 0.001, 200 training epochs, and a batch size of 32. The learning rate was set to balance training stability and prediction accuracy; the number of training epochs ensured the model fully learned the dynamic evolution of physiological parameters and sputum characteristics; and the batch size balanced the specificities of time-series modeling with training efficiency. Specifically, a supervised learning approach was used. Time-series samples containing user state parameter sequences and sputum viscosity sequences were collected from historical patient monitoring data as the input sample set. Simultaneously, corresponding future time-zone observations were acquired to form a sample label set. The input sample set and the corresponding label sample set were divided into training, validation, and test sets in an 8:1:1 ratio.

[0043] Furthermore, using time-series samples from the training set as input and corresponding future observation sequences as supervisory signals, the network weight parameters are iteratively optimized using a backpropagation algorithm combined with the Adam optimizer. The mean squared error loss function is used to measure the deviation between the predicted sequence and the actual observed sequence. The training process is monitored using a validation set. Training is terminated when the validation set loss function value no longer decreases for several consecutive rounds and the model prediction accuracy reaches a predetermined threshold, such as 95%, resulting in converged state parameter predictors and sputum viscosity predictors. This set of predictors can effectively capture the dynamic evolution of the patient's physiological state and sputum characteristics over time, achieving accurate prediction of future time zone state changes.

[0044] Furthermore, using the trained state parameter predictor and sputum viscosity predictor, based on the acquired user state parameter sequence and sputum viscosity sequence, the system performs future time zone projection calculations, ultimately outputting the predicted user state parameter sequence and predicted sputum viscosity sequence within a preset time zone. This preset time zone refers to a future time range preceding the current moment. Its length is comprehensively set based on the expected duration of the transport scenario, the stability of the patient's condition, and the planned intervention frequency of the suctioning operation. For example, it can be set to a predicted duration of 5 to 10 minutes in the future, providing crucial future state information input for generating a forward-looking negative pressure suctioning control scheme.

[0045] S20: Based on the target user's basic physical parameters and the transfer scenario parameters within the preset time zone, perform a transfer posture feature change analysis on the target user and output a predicted posture feature sequence;

[0046] Specifically, based on the target user's basic physical parameters and the transportation scenario parameters within the preset time zone, the system analyzes the changes in the target user's body position characteristics during transportation and outputs a predicted body position feature sequence, including:

[0047] The basic physical parameters of the target user and the transfer scenario parameters within the preset time zone are obtained. The basic physical parameters include at least age, weight and height, and the transfer scenario parameters include transfer route information, vehicle parameters and environmental parameters.

[0048] A user transfer simulation space is constructed, and several transfer simulations are performed within the user transfer simulation space based on the basic body parameters and transfer scenario parameters. High-frequency body position features during the transfer process are selected to construct a predicted body position feature sequence, wherein the body position features include head and neck body position features and chest and back body position features.

[0049] Based on the target user's basic physical parameters and the transport scenario parameters within the preset time zone, the system analyzes the changes in the target user's transport position characteristics. This enables quantitative prediction and feature extraction of the dynamic changes in the patient's position caused by factors such as vehicle movement and path bumps during transport. This provides key kinematic input parameters for subsequent assessment of the impact of positional fluctuations on suctioning operations and for developing appropriate suctioning pressure regulation strategies.

[0050] Specifically, firstly, the basic physical parameters of the target user and the transfer scenario parameters within the preset time zone are obtained. The basic physical parameters are characteristic data describing the user's basic physiological structure, including at least age, weight, and height. The transfer scenario parameters describe the planned or anticipated transfer environmental conditions within the preset time zone, specifically including transfer route information, vehicle parameters, and environmental parameters. Transfer route information may cover route length, road surface smoothness, and the distribution of curves and slopes; vehicle parameters involve stretcher type and the shock absorption performance of the vehicle or aircraft; environmental parameters include parameters such as temperature and air pressure that may affect the evaporation and viscosity trends of sputum. Specifically, the preset time zone is a continuous period of time extending from the current moment to a specific future point in time. Its time span is comprehensively set based on the expected total duration of the transfer task, the complexity of the route, and the clinical nursing requirements for predictive timeliness. For example, it can be set to cover a period of 30 to 60 minutes from the start of the transfer to arrival at the destination, to ensure that the prediction of body position characteristics can fully cover the key dynamic stages of the entire transfer process.

[0051] Secondly, a user transfer simulation space is constructed based on the aforementioned parameters. This user transfer simulation space is a digital virtual environment used to simulate the physical dynamics during a real transfer process. Within this user transfer simulation space, multiple independently run transfer simulations are executed, combining the acquired basic body parameters and transfer scenario parameters. Each simulation generates continuous positional change data of the user throughout the entire transfer process.

[0052] Finally, statistical analysis was performed on the postural change data generated from multiple transport simulations to identify and extract representative high-frequency postural features. These high-frequency postural features were then organized and arranged chronologically to construct a predicted postural feature sequence. This predicted postural feature sequence specifically distinguishes between head and neck postural features and chest and back postural features, representing the posture and positional changes of the user's head and neck and chest and back in space, respectively.

[0053] S30: Evaluate the state of postural change based on the predicted postural feature sequence, output the postural fluctuation coefficient, and set an adaptation adjustment time window based on the postural fluctuation coefficient;

[0054] First, the postural change state is assessed based on the predicted postural feature sequence, and the postural fluctuation coefficient is output, including:

[0055] The predicted head and neck positional feature sequence and the predicted chest and back positional feature sequence are evaluated for positional variability, and the head and neck positional variability and chest and back positional variability are output. The positional variability is the ratio of the standard deviation of the positional parameter to the mean of the positional parameter in the positional feature sequence.

[0056] The head and neck positional fluctuation and the chest and back positional fluctuation are weighted and fitted according to preset site weights to obtain the positional fluctuation coefficient.

[0057] Specifically, firstly, the postural variability of the predicted head and neck postural feature sequences and the predicted chest and back postural feature sequences, contained within the predicted postural feature sequences, is assessed independently. This assessment is achieved by calculating postural variability, which is the ratio of the standard deviation of the positional parameters in the postural feature sequences to the mean of the positional parameters. This ratio effectively eliminates the influence of absolute numerical scaling, thus objectively reflecting the relative dispersion of postural changes. Through this calculation process, two indicators characterizing the dynamic stability of specific body parts—head and neck postural variability and chest and back postural variability—are output separately.

[0058] Secondly, according to pre-set site weights, a weighted fitting calculation is performed on the head and neck positional fluctuations and the chest and back positional fluctuations. These pre-set weights typically assign higher values ​​to the head and neck to emphasize their direct impact on airway patency and suctioning procedures; for example, the head and neck weight is set to 0.6 to 0.8, and the chest and back weight to 0.4 to 0.2, with the sum of the two being 1. Through weighted calculation, the fluctuation indices of the two independent sites are merged into a comprehensive positional fluctuation coefficient. This positional fluctuation coefficient is a normalized scalar value that comprehensively characterizes the severity and instability of the patient's overall positional changes within a pre-set transport time zone.

[0059] Furthermore, setting an adaptation adjustment time window based on the postural fluctuation coefficient includes:

[0060] The ratio of the preset standard body position fluctuation coefficient to the body position fluctuation coefficient is used as the time window compensation coefficient.

[0061] The product of the time window compensation coefficient and the preset fixed adjustment time window is used as the adaptation adjustment time window.

[0062] The adaptation adjustment time window refers to the update cycle of the suction negative pressure parameter, which is dynamically adjusted according to the body position fluctuation coefficient. This is used to ensure that the suction pressure control can keep pace with the dynamic changes in body position during transport, thereby achieving precise pressure regulation.

[0063] First, the preset standard postural fluctuation coefficient is compared with the actually calculated postural fluctuation coefficient, and the ratio is used as the time window compensation coefficient. This time window compensation coefficient directly reflects the ratio of the severity of postural fluctuation in the actual transport environment to the baseline level. The preset standard postural fluctuation coefficient is a benchmark value characterizing postural fluctuation in patients under typical stable transport conditions, derived from statistical analysis of a large amount of clinical transport data. It serves as an objective reference for assessing the degree of turbulence and postural instability in the actual transport environment.

[0064] The calculated time window compensation coefficient is multiplied by the preset fixed adjustment time window, and the product is the final determined adaptive adjustment time window. When the body position fluctuation coefficient increases, it indicates that the body position change is more drastic. At this time, the time window compensation coefficient decreases accordingly, resulting in a shorter adaptive adjustment time window. This conforms to the control logic that the greater the fluctuation, the higher the adjustment frequency should be. By shortening the adjustment time window, the frequency and response speed of pressure adjustment can be increased, thereby responding more promptly to the impact of drastic changes in body position and effectively improving the accuracy and adaptability of suction pressure control.

[0065] S40: Based on the adaptive adjustment time window, with the dual optimization objectives of maximizing the sputum clearance efficiency index and minimizing the airway mucosal damage risk index, the suction negative pressure control scheme within the preset time zone is optimized according to the predicted user state parameter sequence, predicted sputum viscosity sequence and predicted body position feature sequence, and the adaptive suction negative pressure control scheme is output.

[0066] Specifically, based on the adapted adjustment time window, with the dual optimization objectives of maximizing sputum clearance efficiency index and minimizing airway mucosal damage risk index, the suction negative pressure control scheme within the preset time zone is optimized according to the predicted user state parameter sequence, predicted sputum viscosity sequence, and predicted body position feature sequence, outputting an adapted suction negative pressure control scheme, including:

[0067] The preset time zone is divided based on the adaptive adjustment time window to generate an adjustment time window sequence;

[0068] Obtain the suction negative pressure threshold of the suction device, randomly select any suction negative pressure within the suction negative pressure threshold to fill the parameters of any window in the adjustment time window sequence, obtain the first suction negative pressure control scheme, and sequentially select several suction negative pressure control schemes.

[0069] With the dual optimization objectives of maximizing the sputum clearance efficiency index and minimizing the airway mucosal damage risk index, a scheme quality evaluation function is constructed by combining the first and second weights.

[0070] The predicted user state parameter sequence, predicted sputum viscosity sequence, and predicted body position feature sequence are combined with the several suction negative pressure control schemes to obtain several suction operation schemes.

[0071] A suctioning state predictor is constructed based on a generative adversarial network. Using the suctioning state predictor, several suctioning operation results are predicted according to several suctioning operation schemes. Each suctioning operation result includes a predicted sputum clearance efficiency index and a predicted airway mucosal damage risk index.

[0072] Using the aforementioned scheme quality evaluation function, several scheme quality indices are obtained by evaluating the results of the several suctioning operations respectively;

[0073] Based on a genetic algorithm, the suction negative pressure control scheme is optimized and searched according to several suctioning operation schemes and several scheme quality indices until the preset convergence condition is met, and the adapted suction negative pressure control scheme is output.

[0074] First, based on the adaptation adjustment time window obtained in the preceding steps, the entire preset time zone is divided into time periods, generating an adjustment time window sequence composed of continuous time windows. Each time window represents an independent suction pressure adjustment unit, and this adjustment time window sequence serves as the basic time framework for subsequent fine-tuning and optimization of the suction negative pressure control scheme.

[0075] Next, the allowable negative pressure threshold for the suctioning device is obtained. This threshold, defined by medical device safety regulations and clinical operating guidelines, represents the pressure boundary that the suctioning device can safely apply without causing mechanical injury to the patient. Within this threshold range, any suctioning negative pressure is randomly selected to fill parameters in any window of the adjustment time window sequence, thereby generating a first suctioning negative pressure control scheme. This first scheme is an initial pressure configuration scheme built based on a random initialization strategy. Its pressure parameters are randomly distributed within the safety threshold and cover the entire preset time zone, effectively preventing the optimization process from getting trapped in local optima and providing diverse initial search starting points for subsequent intelligent optimization algorithms. This random allocation process is repeated multiple times to generate several different suctioning negative pressure control schemes, forming an initial scheme set.

[0076] Secondly, with the dual optimization objectives of maximizing sputum clearance efficiency and minimizing airway mucosal damage risk, a comprehensive protocol quality evaluation function is constructed by combining pre-defined first and second weights. This function integrates the two optimization objectives into a single evaluation index.

[0077] Specifically, maximizing the sputum clearance efficiency index is the primary goal of the optimization process, characterizing the efficiency and thoroughness of sputum suctioning in clearing respiratory secretions, used to ensure airway patency and maintain effective gas exchange in the injured person; minimizing the airway mucosal damage risk index is the binding goal of the optimization process, characterizing the probability and severity of mechanical damage to airway mucosal tissue caused by suctioning negative pressure and operation methods, ensuring operational safety and avoiding secondary tissue damage.

[0078] The method for setting the first weight and the second weight includes:

[0079] The product of the preset standard body position fluctuation coefficient and the ratio of the body position fluctuation coefficient to the constant K is set as the damage compensation coefficient.

[0080] The second initial weight is updated by using the damage compensation coefficient to obtain the second weight, wherein the second weight is the weight of the airway mucosal damage risk index, the second initial weight is 0.35, and the second weight is not less than 0.25 and not greater than 0.5;

[0081] The first weight is obtained by subtracting the second weight from 1, where the first weight is the weight of the sputum clearance efficiency index.

[0082] The first weight is a quantitative value representing the relative importance of the sputum clearance efficacy index in the overall evaluation, indicating the degree of emphasis placed on achieving the goal of efficient sputum clearance during the optimization process. The second weight is a quantitative value representing the relative importance of the airway mucosal damage risk index in the overall evaluation, indicating the degree of emphasis placed on controlling operational risks and ensuring safety during the optimization process. Specifically, the methods for setting the first and second weights are as follows:

[0083] First, the ratio of the preset standard postural fluctuation coefficient to the actual postural fluctuation coefficient is multiplied by a constant K, and the product is defined as the injury compensation coefficient. This constant K is a weighted gain coefficient determined based on clinical expert experience and historical data statistical analysis; typically, K is set within the range of 0.8 to 1.2, for example, it can be set to 1.0. The resulting injury compensation coefficient is a dimensionless adjustment factor used to quantify the degree to which the risk of injury requires extra attention due to increased postural fluctuations in the current transport environment.

[0084] Secondly, the calculated damage compensation coefficient is used to compensate for and update the second initial weight, thereby determining the final second weight. The second weight is specifically used to quantify the importance of the airway mucosal injury risk index in the overall assessment. Specifically, the second initial weight is set to a baseline value of 0.35. After compensation and updating, the value of the second weight is constrained within a closed interval of 0.25 to 0.5, ensuring that it is neither too low to ignore safety nor too high to be overly conservative.

[0085] Specifically, when the actual positional fluctuation coefficient equals the standard positional fluctuation coefficient, the injury compensation coefficient is the K value itself, which is 1. In this case, the second weight should maintain its initial value of 0.35. Under this condition, taking K=1.0 can achieve the optimal balance in weight allocation. If the actual fluctuation coefficient increases, the injury compensation coefficient will be amplified proportionally, thereby appropriately increasing the weight ratio of injury risk. However, by setting upper and lower limits for the value of the second weight, the rationality of weight allocation can still be maintained.

[0086] Furthermore, the first weight is obtained by subtracting the predetermined second weight from the numerical value of 1. This first weight is specifically used to quantify the importance of the sputum clearance efficacy index in the overall evaluation. This setting ensures that the sum of the two weights is always 1, while dynamically adjusting the relative importance of clearance efficacy and injury risk in the protocol evaluation based on actual postural fluctuations. When postural fluctuations intensify, the injury compensation coefficient increases, and the second weight increases accordingly, indicating that the optimization process will prioritize operational safety.

[0087] Furthermore, the predicted user state parameter sequence, predicted sputum viscosity sequence, and predicted body position feature sequence are combined with several suctioning negative pressure control schemes to form several complete suctioning operation schemes. Each suctioning operation scheme includes environmental prediction parameters and corresponding negative pressure control parameters for a specific time window.

[0088] Specifically, the formation process of each suctioning procedure is as follows: For the i-th time window in the adjustment time window sequence, the corresponding parameters such as blood oxygen saturation, heart rate, and respiratory rate are extracted from the predicted user state parameter sequence; the viscosity estimate for the window is extracted from the predicted sputum viscosity sequence; the head and neck position features and chest and back position features for the window are extracted from the predicted body position feature sequence; and the specific negative pressure value configured for the window is obtained from the suctioning negative pressure control scheme. Each parameter is structured and encapsulated according to the time window to form a complete data unit containing environmental prediction parameters and negative pressure control parameters.

[0089] By iterating through all time windows and repeating this process, several complete suctioning procedures are ultimately constructed. This combination method ensures that each suctioning procedure can fully describe the matching relationship between the dynamically changing environmental conditions and the corresponding negative pressure control strategies over the entire preset time zone.

[0090] Furthermore, a suctioning state predictor is constructed based on a generative adversarial network architecture. The suctioning state predictor is a multivariate temporal prediction model trained adversarially. Based on the sequence of environmental prediction parameters and negative pressure control parameters included in the input suctioning procedure, it can accurately predict the clinical effects that may be produced after the procedure is executed. It is used to quantitatively evaluate the expected performance of different negative pressure control procedures in terms of sputum clearance efficacy and airway mucosal safety.

[0091] Specifically, the suctioning status predictor consists of two core components: a generator and a discriminator. It learns the complex mapping relationship between suctioning procedures and their results through an adversarial training mechanism. The generator takes the aforementioned complete suctioning procedure as input data, which includes environmental prediction parameters and negative pressure control parameters arranged in chronological order. The generator employs an encoder-decoder structure for feature extraction and result prediction. The encoder uses a multi-layer long short-term memory network to deeply encode the input sequence, capturing its inherent temporal dependencies. The decoder uses a fully connected neural network to map the extracted temporal features into two consecutive output values, corresponding to the predicted sputum clearance efficiency index and the predicted airway mucosal damage risk index of the suctioning procedure, respectively. The discriminator receives the suctioning procedure and its corresponding result data as input. The discriminator uses a deep neural network architecture, and its core function is to determine whether the input result data originates from actual clinical records or the generator's predicted output, thereby verifying the authenticity of the predicted results.

[0092] During model training, the generator continuously optimizes its network parameters to produce predictions that closely approximate the distribution of real clinical data, thus enabling the discriminator to distinguish between real and generated data. Simultaneously, the discriminator optimizes its discrimination capabilities to accurately differentiate between real and generated data. This adversarial training mechanism continuously drives the generator to improve its prediction accuracy, causing the predicted results to increasingly approximate real clinical outcomes.

[0093] The training data comes from a database of historical suctioning operation records. The input features are the actual environmental parameters and negative pressure control schemes used, while the supervision labels are the sputum clearance efficiency index and airway mucosal damage risk index, both confirmed through clinical evaluation. After sufficient adversarial training, the generator is able to accurately predict the performance of any suctioning procedure, providing reliable predictive data support for subsequent scheme quality evaluation and optimization search.

[0094] Using this suctioning status predictor, the corresponding suctioning results can be predicted based on the aforementioned generated suctioning operation plans. Each result includes a predicted sputum clearance efficiency index and a predicted airway mucosal damage risk index.

[0095] Furthermore, using the constructed scheme quality evaluation function, the results of several suctioning procedures are comprehensively evaluated to obtain corresponding scheme quality indices. Each scheme quality index is equal to the product of the first weight and the standardized sputum clearance efficiency index, minus the product of the second weight and the standardized airway mucosal damage risk index, and is used to quantify the overall quality of each suctioning procedure.

[0096] Finally, based on a genetic algorithm, an iterative optimization search for a negative pressure control scheme for suctioning is performed according to several suctioning operation schemes and their corresponding scheme quality indices. Through genetic operations such as selection, crossover, and mutation, the negative pressure control scheme for suctioning is continuously evolved and improved until the preset convergence condition is reached, and the optimal adapted negative pressure control scheme for suctioning is output.

[0097] Specifically, the genetic algorithm is a metaheuristic optimization algorithm that simulates the natural evolutionary process. It iteratively improves and globally optimizes suction negative pressure control schemes by simulating mechanisms such as selection, crossover, and mutation in biological evolution. The algorithm encodes each suction negative pressure control scheme as a chromosome representing a sequence of negative pressure values. It uses a scheme quality index as a standard to measure chromosome fitness, prioritizes high-quality schemes through roulette wheel selection, fuses superior gene fragments from different schemes using single-point or multi-point crossover operations, and introduces new gene features through low-probability mutation operations. After multiple generations of evolution, the algorithm gradually converges from an initially randomly generated population of schemes to the optimal suction negative pressure control scheme with the best overall performance. The process of optimizing and searching for suction negative pressure control schemes based on the genetic algorithm is as follows:

[0098] This process treats the initially generated suctioning procedures as an initial population, with each procedure's quality index serving as its fitness value. The quality of the procedures is iteratively improved by simulating a natural evolutionary mechanism. Specifically, a roulette wheel selection algorithm is used, calculating the probability of each procedure being selected based on its fitness value. Procedures with higher fitness have a greater probability of being selected as parents, thus ensuring that superior traits are preserved and passed on.

[0099] Crossover operations perform gene exchange on selected parent schemes. For example, if two parent schemes A and B are selected, with negative pressure control sequences of [-120, -150, -180, -160] and [-140, -130, -170, -190] respectively, a offspring scheme [-120, -150, -170, -190] may be generated through single-point or multi-point crossover. This offspring inherits the favorable pressure configurations of different time windows from the parent schemes.

[0100] The mutation operation randomly changes the negative pressure value of a certain time window in the offspring scheme with a preset low probability. For example, the offspring scheme [-120, -150, -170, -190] may mutate into [-120, -145, -170, -190]. By introducing random perturbation, the diversity of the population is enhanced, and the optimization process is prevented from getting trapped in local optima too early.

[0101] The iterative process repeatedly performs fitness evaluation, selection, crossover, and mutation operations, continuously generating new generations of the population. In each generation, the fitness value of each option is recalculated, and a new round of genetic operations is performed accordingly. Finally, the termination conditions include two aspects: first, reaching the maximum number of iterations, such as 100 generations; second, the fitness of the optimal option in the population no longer significantly improves over multiple generations, such as a change rate of less than 2%. The optimization process terminates when either condition is met.

[0102] Finally, the scheme with the highest fitness was selected from the last generation of the population as the output adaptive suction negative pressure control scheme. This adaptive suction negative pressure control scheme is the optimal negative pressure control strategy obtained through multiple generations of evolution under given constraints, which can achieve the best balance between sputum clearance efficiency and airway mucosal safety within a preset time zone.

[0103] S50: Perform optimized suctioning operations for the target user within the preset time zone according to the adapted suctioning negative pressure control scheme.

[0104] The adaptive suction negative pressure control scheme, as the final output of the aforementioned optimization process, is a sequence of negative pressure control commands arranged in a time sequence. It specifies the exact negative pressure value to be applied within each adjustment time window in the preset time zone, as well as the duration of each negative pressure value. During the execution of the scheme, the core controller of the suction device receives the control scheme and dynamically adjusts its output negative pressure according to the time sequence set in the scheme. The suction controller will automatically switch to the specified negative pressure value at the corresponding time point according to the preset adjustment time window sequence in the scheme, thereby achieving precise pressure control throughout the entire preset time zone.

[0105] At the start of the preset time zone, the suction device initiates the suction operation according to the negative pressure parameters set in the first adjustment time window of the plan. When the end of the first adjustment time window is reached, the device automatically switches to the negative pressure value specified in the second adjustment time window, and so on, until all suction operations within the preset time zone are completed.

[0106] In summary, this suctioning procedure based on an adaptive negative pressure control scheme ensures that the suctioning operation remains consistent with predicted changes in user status, evolution of sputum characteristics, and dynamic postural features. By matching the optimal negative pressure parameters in real time, it not only guarantees sputum clearance efficiency but also effectively controls the risk of airway mucosal damage, ultimately achieving a comprehensive optimization of the safety and effectiveness of suctioning.

[0107] In summary, the embodiments of this application have at least the following technical effects:

[0108] Compared to existing technologies, this invention firstly achieves forward-looking perception and collaborative analysis of multi-dimensional dynamic parameters of the suctioning environment by predicting user state parameter sequences, sputum viscosity sequences, and body position characteristic sequences during transport, laying a data foundation for precise control. Secondly, it innovatively introduces a body position fluctuation coefficient and sets an adaptive adjustment time window accordingly, enabling the adjustment rhythm of suction pressure to automatically adapt to the dynamic changes in body position during transport, improving the timeliness and adaptability of control. Thirdly, it collaboratively optimizes the dual objectives of maximizing sputum clearance efficiency and minimizing the risk of mucosal damage, and uses intelligent algorithms to generate an adaptive negative pressure control scheme for suctioning, achieving a balance between clearance efficiency and operational safety in terms of control strategy.

[0109] Finally, by directly applying the optimization scheme to the suctioning operation, a closed-loop control process from multi-parameter prediction and intelligent optimization to precise execution was constructed. This effectively solved the problems of poor adaptability and insufficient reliability of traditional methods in dynamic transport environments, and significantly improved the accuracy, safety and overall efficiency of suctioning operations.

[0110] Example 2, as Figure 2 As shown, based on the same inventive concept as the suction pressure optimization method combining multi-parameter collaborative feedback provided in Embodiment 1, this embodiment of the invention also provides a suction pressure optimization system combining multi-parameter collaborative feedback, including:

[0111] Prediction parameter acquisition module 11 is used to predict and acquire the target user's predicted user state parameter sequence and predicted sputum viscosity sequence within a preset time zone;

[0112] The body position feature analysis module 12 is used to perform a change analysis of the body position feature of the target user during transit based on the target user's basic body parameters and the transit scenario parameters within the preset time zone, and output a predicted body position feature sequence.

[0113] The time window setting module 13 is used to evaluate the state of postural change based on the predicted postural feature sequence, output the postural fluctuation coefficient, and set an adaptation adjustment time window based on the postural fluctuation coefficient.

[0114] The negative pressure scheme optimization module 14 is used to optimize the suction negative pressure control scheme within the preset time zone based on the adaptive adjustment time window, with the dual optimization objectives of maximizing the sputum clearance efficiency index and minimizing the airway mucosal damage risk index, according to the predicted user state parameter sequence, predicted sputum viscosity sequence and predicted body position feature sequence, and output an adapted suction negative pressure control scheme.

[0115] The suctioning operation execution module 15 is used to perform the optimized suctioning operation within the preset time zone for the target user in accordance with the adapted suctioning negative pressure control scheme.

[0116] The prediction parameter acquisition module 11 is specifically used for:

[0117] The system predicts and obtains the target user's state parameter sequence and sputum viscosity sequence within a preset time zone, including:

[0118] The system monitors and acquires the user status parameter sequence and sputum viscosity sequence of the target user within a historical time zone. The target user is an injured person, and the user status parameters include the target user's blood oxygen saturation, heart rate, and respiratory rate.

[0119] Using the target user's basic physical parameters and current condition data as constraints, a sample training set is collected to train a long short-term memory network until convergence, and a state parameter predictor and a sputum viscosity predictor are constructed.

[0120] Using the state parameter predictor and sputum viscosity predictor, a predicted user state parameter sequence and a predicted sputum viscosity sequence within a preset time zone are predicted based on the user state parameter sequence and the sputum viscosity sequence.

[0121] Specifically, the body position feature analysis module 12 is used for:

[0122] Based on the target user's basic physical parameters and the transportation scenario parameters within the preset time zone, the system analyzes the changes in the target user's body position characteristics during transportation and outputs a predicted body position characteristic sequence, including:

[0123] The basic physical parameters of the target user and the transfer scenario parameters within the preset time zone are obtained. The basic physical parameters include at least age, weight and height, and the transfer scenario parameters include transfer route information, vehicle parameters and environmental parameters.

[0124] A user transfer simulation space is constructed, and several transfer simulations are performed within the user transfer simulation space based on the basic body parameters and transfer scenario parameters. High-frequency body position features during the transfer process are selected to construct a predicted body position feature sequence, wherein the body position features include head and neck body position features and chest and back body position features.

[0125] The time window setting module 13 is specifically used for:

[0126] Based on the predicted postural feature sequence, the postural change state is assessed, and the postural fluctuation coefficient is output, including:

[0127] The predicted head and neck positional feature sequence and the predicted chest and back positional feature sequence are evaluated for positional variability, and the head and neck positional variability and chest and back positional variability are output. The positional variability is the ratio of the standard deviation of the positional parameter to the mean of the positional parameter in the positional feature sequence.

[0128] The head and neck positional fluctuation and the chest and back positional fluctuation are weighted and fitted according to preset site weights to obtain the positional fluctuation coefficient.

[0129] Furthermore, setting an adaptation adjustment time window based on the postural fluctuation coefficient includes:

[0130] The ratio of the preset standard body position fluctuation coefficient to the body position fluctuation coefficient is used as the time window compensation coefficient.

[0131] The product of the time window compensation coefficient and the preset fixed adjustment time window is used as the adaptation adjustment time window.

[0132] The negative pressure scheme optimization module 14 is specifically used for:

[0133] Based on the aforementioned adaptive adjustment time window, with the dual optimization objectives of maximizing sputum clearance efficiency and minimizing airway mucosal damage risk, the suction negative pressure control scheme within the preset time zone is optimized according to the predicted user state parameter sequence, predicted sputum viscosity sequence, and predicted body position feature sequence, outputting an adapted suction negative pressure control scheme, including:

[0134] The preset time zone is divided based on the adaptive adjustment time window to generate an adjustment time window sequence;

[0135] Obtain the suction negative pressure threshold of the suction device, randomly select any suction negative pressure within the suction negative pressure threshold to fill the parameters of any window in the adjustment time window sequence, obtain the first suction negative pressure control scheme, and sequentially select several suction negative pressure control schemes.

[0136] With the dual optimization objectives of maximizing the sputum clearance efficiency index and minimizing the airway mucosal damage risk index, a scheme quality evaluation function is constructed by combining the first and second weights.

[0137] The predicted user state parameter sequence, predicted sputum viscosity sequence, and predicted body position feature sequence are combined with the several suction negative pressure control schemes to obtain several suction operation schemes.

[0138] A suctioning state predictor is constructed based on a generative adversarial network. Using the suctioning state predictor, several suctioning operation results are predicted according to several suctioning operation schemes. Each suctioning operation result includes a predicted sputum clearance efficiency index and a predicted airway mucosal damage risk index.

[0139] Using the aforementioned scheme quality evaluation function, several scheme quality indices are obtained by evaluating the results of the several suctioning operations respectively;

[0140] Based on a genetic algorithm, the suction negative pressure control scheme is optimized and searched according to several suctioning operation schemes and several scheme quality indices until the preset convergence condition is met, and the adapted suction negative pressure control scheme is output.

[0141] Specifically, the methods for setting the first weight and the second weight include:

[0142] The product of the preset standard body position fluctuation coefficient and the ratio of the body position fluctuation coefficient to the constant K is set as the damage compensation coefficient.

[0143] The second initial weight is updated by using the damage compensation coefficient to obtain the second weight, wherein the second weight is the weight of the airway mucosal damage risk index, the second initial weight is 0.35, and the second weight is not less than 0.25 and not greater than 0.5;

[0144] The first weight is obtained by subtracting the second weight from 1, where the first weight is the weight of the sputum clearance efficiency index.

[0145] The suctioning operation execution module 15 is specifically used for:

[0146] The target user is subjected to optimized suctioning within the preset time zone according to the adapted suctioning negative pressure control scheme.

[0147] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0148] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0149] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.

Claims

1. A method for optimizing suction pressure by combining multi-parameter collaborative feedback, characterized in that, The methods include: Predict and obtain the target user's predicted user state parameter sequence and predicted sputum viscosity sequence within a preset time zone; Based on the target user's basic physical parameters and the transfer scenario parameters within the preset time zone, the target user's transfer posture feature changes are analyzed, and a predicted posture feature sequence is output. The body position change status is evaluated based on the predicted body position feature sequence, the body position fluctuation coefficient is output, and an adaptation adjustment time window is set based on the body position fluctuation coefficient. Based on the aforementioned adaptive adjustment time window, with the dual optimization objectives of maximizing sputum clearance efficiency and minimizing airway mucosal damage risk, the suction negative pressure control scheme within the preset time zone is optimized according to the predicted user state parameter sequence, predicted sputum viscosity sequence, and predicted body position feature sequence, outputting an adapted suction negative pressure control scheme, including: The preset time zone is divided based on the adaptive adjustment time window to generate an adjustment time window sequence; Obtain the suction negative pressure threshold of the suction device, randomly select any suction negative pressure within the suction negative pressure threshold to fill the parameters of any window in the adjustment time window sequence, obtain the first suction negative pressure control scheme, and sequentially select several suction negative pressure control schemes. With the dual optimization objectives of maximizing the sputum clearance efficiency index and minimizing the airway mucosal damage risk index, a scheme quality evaluation function is constructed by combining the first and second weights. The predicted user state parameter sequence, predicted sputum viscosity sequence, and predicted body position feature sequence are combined with the several suction negative pressure control schemes to obtain several suction operation schemes. A suctioning state predictor is constructed based on a generative adversarial network. Using the suctioning state predictor, several suctioning operation results are predicted according to several suctioning operation schemes. Each suctioning operation result includes a predicted sputum clearance efficiency index and a predicted airway mucosal damage risk index. Using the aforementioned scheme quality evaluation function, several scheme quality indices are obtained by evaluating the results of the several suctioning operations respectively; Based on the genetic algorithm, the suction negative pressure control scheme is optimized and searched according to the several suctioning operation schemes and several scheme quality indices until the preset convergence condition is reached, and the adapted suction negative pressure control scheme is output. The target user is subjected to optimized suctioning within the preset time zone according to the adapted suctioning negative pressure control scheme.

2. The suction pressure optimization method combining multi-parameter collaborative feedback as described in claim 1, characterized in that, The system predicts and obtains the target user's state parameter sequence and sputum viscosity sequence within a preset time zone, including: The system monitors and acquires the user status parameter sequence and sputum viscosity sequence of the target user within a historical time zone. The target user is an injured person, and the user status parameters include the target user's blood oxygen saturation, heart rate, and respiratory rate. Using the target user's basic physical parameters and current condition data as constraints, a sample training set is collected to train a long short-term memory network until convergence, and a state parameter predictor and a sputum viscosity predictor are constructed. Using the state parameter predictor and sputum viscosity predictor, a predicted user state parameter sequence and a predicted sputum viscosity sequence within a preset time zone are predicted based on the user state parameter sequence and the sputum viscosity sequence.

3. The suction pressure optimization method combining multi-parameter collaborative feedback as described in claim 1, characterized in that, Based on the target user's basic physical parameters and the transportation scenario parameters within the preset time zone, the system analyzes the changes in the target user's body position characteristics during transportation and outputs a predicted body position characteristic sequence, including: The basic physical parameters of the target user and the transfer scenario parameters within the preset time zone are obtained. The basic physical parameters include at least age, weight and height, and the transfer scenario parameters include transfer route information, vehicle parameters and environmental parameters. A user transfer simulation space is constructed, and several transfer simulations are performed within the user transfer simulation space based on the basic body parameters and transfer scenario parameters. High-frequency body position features during the transfer process are selected to construct a predicted body position feature sequence, wherein the body position features include head and neck body position features and chest and back body position features.

4. The suction pressure optimization method combining multi-parameter collaborative feedback according to claim 3, characterized in that, Based on the predicted postural feature sequence, the postural change state is assessed, and the postural fluctuation coefficient is output, including: The predicted head and neck positional feature sequence and the predicted chest and back positional feature sequence are evaluated for positional variability, and the head and neck positional variability and chest and back positional variability are output. The positional variability is the ratio of the standard deviation of the positional parameter to the mean of the positional parameter in the positional feature sequence. The head and neck positional fluctuation and the chest and back positional fluctuation are weighted and fitted according to preset site weights to obtain the positional fluctuation coefficient.

5. The method for optimizing suction pressure by combining multi-parameter collaborative feedback according to claim 1, characterized in that, Based on the aforementioned postural fluctuation coefficient, an adaptation adjustment time window is set, including: The ratio of the preset standard body position fluctuation coefficient to the body position fluctuation coefficient is used as the time window compensation coefficient. The product of the time window compensation coefficient and the preset fixed adjustment time window is used as the adaptation adjustment time window.

6. The method for optimizing suction pressure by combining multi-parameter collaborative feedback according to claim 1, characterized in that, The methods for setting the first weight and the second weight include: The product of the preset standard body position fluctuation coefficient and the ratio of the body position fluctuation coefficient to the constant K is set as the damage compensation coefficient. The second initial weight is updated by using the damage compensation coefficient to obtain the second weight, wherein the second weight is the weight of the airway mucosal damage risk index, the second initial weight is 0.35, and the second weight is not less than 0.25 and not greater than 0.5; The first weight is obtained by subtracting the second weight from 1, where the first weight is the weight of the sputum clearance efficiency index.

7. A suction pressure optimization system combining multi-parameter collaborative feedback, characterized in that, The method for optimizing suction pressure by combining multi-parameter collaborative feedback as described in any one of claims 1-6 includes: The prediction parameter acquisition module is used to predict and acquire the target user's state parameter sequence and sputum viscosity sequence within a preset time zone. The body position feature analysis module is used to analyze the changes in body position features of the target user during transit based on the target user's basic body parameters and the transit scenario parameters within the preset time zone, and output a predicted body position feature sequence. The time window setting module is used to evaluate the state of postural change based on the predicted postural feature sequence, output the postural fluctuation coefficient, and set an adaptation adjustment time window based on the postural fluctuation coefficient. The negative pressure scheme optimization module is used to optimize the suction negative pressure control scheme within the preset time zone based on the adapted adjustment time window, with the dual optimization objectives of maximizing the sputum clearance efficiency index and minimizing the airway mucosal damage risk index. This optimization is based on the predicted user state parameter sequence, predicted sputum viscosity sequence, and predicted body position feature sequence. The resulting optimized suction negative pressure control scheme includes: The preset time zone is divided based on the adaptive adjustment time window to generate an adjustment time window sequence; Obtain the suction negative pressure threshold of the suction device, randomly select any suction negative pressure within the suction negative pressure threshold to fill the parameters of any window in the adjustment time window sequence, obtain the first suction negative pressure control scheme, and sequentially select several suction negative pressure control schemes. With the dual optimization objectives of maximizing the sputum clearance efficiency index and minimizing the airway mucosal damage risk index, a scheme quality evaluation function is constructed by combining the first and second weights. The predicted user state parameter sequence, predicted sputum viscosity sequence, and predicted body position feature sequence are combined with the several suction negative pressure control schemes to obtain several suction operation schemes. A suctioning state predictor is constructed based on a generative adversarial network. Using the suctioning state predictor, several suctioning operation results are predicted according to several suctioning operation schemes. Each suctioning operation result includes a predicted sputum clearance efficiency index and a predicted airway mucosal damage risk index. Using the aforementioned scheme quality evaluation function, several scheme quality indices are obtained by evaluating the results of the several suctioning operations respectively; Based on the genetic algorithm, the suction negative pressure control scheme is optimized and searched according to the several suctioning operation schemes and several scheme quality indices until the preset convergence condition is reached, and the adapted suction negative pressure control scheme is output. The suctioning operation execution module is used to perform optimized suctioning operations for the target user within the preset time zone according to the adapted suctioning negative pressure control scheme.

Images