A medical oxygen flow metering and control system

Abstract

This invention belongs to the field of medical oxygen control, and particularly relates to a medical oxygen flow metering and control system. This system monitors the oxygen supply pressure in real time and initiates personalized oxygen supply settings within a safe range. It matches a stepped oxygen supply flow rate mapping rule based on the user's symptom characteristics and simultaneously generates oxygen supply valve opening commands, achieving precise oxygen supply in stages. Simultaneously, it integrates quality detection and segmented point-based billing, adjusts costs in real time based on reimbursement ratios, and uses blockchain technology for data storage and sharing on the blockchain. The system also supports oxygen humidification comfort adjustment, bidirectional data synchronization with electronic medical record systems, and electronic annual inspection and equipment locking functions based on the PDF / A standard. Therefore, while improving oxygen supply safety, personalization, and treatment effectiveness, it achieves integrated intelligent control with transparent billing, reliable data, and standardized operation and maintenance.

Description

Technical Field

[0001] This invention belongs to the field of medical oxygen control, and particularly relates to a medical oxygen flow metering and control system. Background Technology

[0002] Currently, the traditional float-type oxygen inhalers widely used in hospitals suffer from measurement defects and technical bottlenecks. Not only do they suffer from significant instantaneous flow rate indication errors due to equipment aging, contamination, and limited capacity, but more importantly, they lack the function of measuring cumulative flow, forcing billing to rely on manual recording of duration and rough estimations. This inefficient approach leads to substantial hidden losses of hospital oxygen resources and distorted financial accounting, while also posing extremely high audit and compliance risks. Against this backdrop, developing an oxygen flow control system capable of accurate measurement, automatic accumulation, and data transparency has become an urgent technical need to address the challenges of operational efficiency, cost control, and compliance.

[0003] Existing technologies have the following problems: Traditional oxygen supply control methods often use fixed or simple flow rate settings, which cannot provide precise oxygen supply in stages according to the patient's real-time symptom characteristics. At the same time, there is a lack of coordinated management and closed-loop control of pressure safety, oxygen quality, billing accuracy and abnormal conditions during the oxygen supply process, resulting in low oxygen supply safety, difficulty in optimizing treatment effects, opaque billing and slow system response. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention proposes a medical oxygen flow metering and control system. This system monitors oxygen supply pressure in real time and initiates personalized oxygen supply settings within a safe range. It matches a stepped oxygen supply flow rate mapping rule based on the user's symptom characteristics and simultaneously generates oxygen supply valve opening commands, achieving precise oxygen supply in stages. Simultaneously, it integrates quality detection and segmented point-based billing, adjusting costs in real time based on reimbursement ratios, and uses blockchain technology for data storage and sharing on the blockchain. The system also supports oxygen humidification comfort adjustment, bidirectional data synchronization with electronic medical record systems, and electronic annual inspection and equipment locking functions based on the PDF / A standard. Thus, while improving oxygen supply safety, personalization, and treatment effectiveness, it achieves integrated intelligent control with transparent billing, reliable data, and standardized operation and maintenance.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] A medical oxygen flow metering and control system includes: a block monitoring node chain, wherein the block monitoring node chain includes:

[0007] A safety node is used to perform safety checks based on the pressure change curve of the real-time monitored oxygen supply subsystem or oxygen cylinder and a preset safety pressure range. When the check result meets the safety pressure range, it responds to the oxygen supply setting node. The oxygen supply setting node obtains the current user's staged oxygen supply flow rate mapping rule based on the input user's symptom characteristics and a preset stepped oxygen supply flow rate mapping rule library, and simultaneously responds to the proportional setting node. The proportional setting node generates staged oxygen supply valve opening commands based on the current user's staged oxygen supply flow rate mapping rule and a reinforcement learning-optimized PID control algorithm, and simultaneously responds to the oxygen supply quality detection node and the oxygen supply billing node. The oxygen supply billing node is used to determine the current user's oxygen supply flow rate mapping rule based on the current user's oxygen supply flow rate mapping rule and a reinforcement learning-optimized PID control algorithm, and simultaneously responds to the oxygen supply quality detection node and the oxygen supply billing node. The user's phased oxygen supply flow rate mapping rule, combined with a phased standardized metering model, obtains the phased integral oxygen supply amount. Based on the phased integral oxygen supply amount and a preset phased billing rule, real-time oxygen supply billing is performed, and the billing result is fed back to the display interface and to the block record node for on-chain storage. The oxygen supply quality detection node is used to detect the oxygen concentration and oxygen supply flow rate of each oxygen supply stage in real time during the oxygen supply process, and to perform oxygen supply detection by combining exponential moving average and preset anomaly detection thresholds. The detection results are fed back to the oxygen supply billing node to adjust the billing status and trigger an early warning. The preset anomaly detection thresholds include an oxygen concentration anomaly threshold and the standard phased instantaneous oxygen supply flow rate of the corresponding stage.

[0008] Specifically, the block monitoring node chain also includes a transmission synchronization node; the response oxygen supply setting node includes a symptom assessment module and a mapping module;

[0009] The symptom assessment module is used to obtain a severity score for the current type of symptom of the corresponding user based on the user's symptom characteristics and symptom type information synchronized from the electronic medical record subsystem by the transmission synchronization node, combined with a fuzzy comprehensive evaluation algorithm.

[0010] The mapping module is used to generate a phased oxygen supply flow rate mapping rule for the current user based on the user's current symptom severity score, combined with the symptom type-severity-phased oxygen supply flow rate mapping rule in the phased oxygen supply flow rate mapping rule library and the duration of each phase, and combined with preset matching mapping rules. The phased oxygen supply flow rate mapping rule for the current user includes the valve opening size corresponding to each oxygen supply subsystem or oxygen cylinder output port, the oxygen flow rate and concentration corresponding to each phase, and the duration corresponding to the valve opening.

[0011] Specifically, the construction and training process of the piecewise standardized econometric model includes:

[0012] Based on symptom type, symptom severity score, filtered temperature change gradient value and corresponding compensation power value, constant temperature difference, standard staged instantaneous oxygen supply flow rate, and combined with the Apriori algorithm, a staged hierarchical mapping chain is established.

[0013] The temperature change gradient value and corresponding compensation power value of the heating element at different oxygen supply stages are collected in real time by the configured heating element and temperature sensor. The data is filtered by the filtering algorithm to obtain the filtered temperature change gradient value and corresponding compensation power value.

[0014] The phased hierarchical mapping chain, the filtered temperature change gradient value, and the compensation power value are used as training parameters and input into the preset phased nonlinear correction algorithm to calculate the real-time phased instantaneous oxygen supply flow rate corresponding to the current phase.

[0015] Specifically, the construction and training process of the piecewise standardized econometric model also includes:

[0016] Based on the real-time phased instantaneous oxygen supply flow rate corresponding to the current stage and the standard phased instantaneous oxygen supply flow rate of the corresponding stage, the deviation of the standard oxygen supply flow rate is obtained, and the first oxygen supply metering loss is constructed.

[0017] Based on the standard oxygen supply flow rate deviation, the standard temperature gradient deviation value is obtained through the backpropagation algorithm along the staged hierarchical mapping chain, and a staged temperature damage compensation matrix is ​​constructed.

[0018] Based on the real-time phased instantaneous oxygen supply flow rate and the corresponding oxygen supply phase duration, a piecewise integral kernel function is constructed using a trapezoidal integral algorithm to calculate the cumulative oxygen supply amount for each oxygen supply phase.

[0019] The second standard loss is obtained by combining the cumulative oxygen supply of each oxygen supply stage with the standard cumulative oxygen supply of the corresponding stage.

[0020] Specifically, the construction and training process of the piecewise standardized econometric model also includes:

[0021] The staged nonlinear correction algorithm is fine-tuned by constructing a weighted combined loss function based on the first oxygen supply metering loss, the staged temperature damage compensation matrix and the second standard loss; when the weighted combined loss function meets the corresponding comprehensive loss threshold and the first oxygen supply metering loss is less than the first loss threshold under the preset training period, the trained segmented standardized metering model is obtained.

[0022] When either the weighted combined loss function or the first oxygen supply metering loss fails to meet the corresponding loss threshold under the preset training period, an abnormal oxygen supply valve opening warning is issued, and the oxygen supply valve opening is readjusted N times. The temperature change gradient value and the corresponding compensation power value after the adjustment are collected to perform iterative adjustment and training on the phased nonlinear correction algorithm until the weighted combined loss function meets the corresponding comprehensive loss threshold and the first oxygen supply metering loss is less than the first loss threshold. At the same time, the compensation value of the oxygen supply valve opening for the corresponding oxygen supply stage is output.

[0023] The compensation value of the oxygen supply valve opening degree of the corresponding oxygen supply stage is mapped to the hierarchical mapping chain corresponding to the stage-by-stage hierarchical mapping chain to obtain the stage-by-stage hierarchical compensation mapping chain and perform opening degree early warning compensation.

[0024] Specifically, the stepped oxygen supply flow rate mapping rule base is constructed by combining a hash algorithm and a tree database, based on the disease type, the severity score corresponding to each disease type, the number of oxygen supply stages corresponding to the corresponding severity score, the interval duration and average oxygen supply flow rate corresponding to each oxygen supply stage, and is used to match the stepped oxygen supply flow rate mapping rules corresponding to the severity score of the disease type; the segmented billing rule is constructed by the average oxygen supply flow rate of the corresponding stage and the billing unit price corresponding to the average oxygen supply flow rate.

[0025] Specifically, the detection results are fed back to the oxygen supply billing node to adjust the billing status and trigger an early warning, including:

[0026] The symptoms of the current user are extracted from the electronic medical record subsystem, and the severity score of the symptoms is obtained by combining the fuzzy comprehensive evaluation algorithm.

[0027] Simultaneously, based on the current user's symptom characteristics and extracted basic user information, combined with information gain and mutual information algorithms, the corresponding user's access sensitivity is obtained;

[0028] Based on the user's access sensitivity, combined with hierarchical permission mapping and access role access level, the current user's symptom characteristics and symptom severity score are transmitted to the oxygen supply end through the transmission synchronization node.

[0029] Specifically, feeding back the test results to the oxygen supply billing node to adjust the billing status and trigger an early warning also includes:

[0030] Based on the current user's symptom characteristics and symptom severity score, the step-by-step oxygen supply flow rate mapping rules are obtained from the step-by-step oxygen supply flow rate mapping rule library using a depth-first search algorithm.

[0031] The system responds to the stepped oxygen supply flow rate mapping rule and simultaneously obtains the cumulative oxygen supply corresponding to each stage through a segmented standardized metering model.

[0032] Based on the cumulative oxygen supply for each stage and the segmented billing rules, the cumulative billing result and total billing result for each stage are obtained.

[0033] Specifically, feeding back the test results to the oxygen supply billing node to adjust the billing status and trigger an early warning also includes:

[0034] While responding to the step-type oxygen supply flow rate mapping rule, the oxygen concentration value is monitored in real time by the configured oxygen concentration measurement sensor, and the oxygen concentration change prediction function is fitted by combining the radial basis kernel function.

[0035] Based on real-time monitoring of oxygen concentration values, combined with an oxygen concentration change prediction function and a preset oxygen concentration anomaly threshold, the time points corresponding to the oxygen concentration anomalies during the oxygen supply process are obtained.

[0036] The time point of the abnormal oxygen concentration is fed back to the oxygen supply billing node, generating a pre-oxygen supply metering stop command. When the oxygen concentration value is less than the abnormal oxygen concentration threshold, the cumulative oxygen flow metering and cumulative billing metering are stopped through the pre-oxygen supply metering stop command, and the abnormal result is fed back to the oxygen metering display interface.

[0037] Specifically, feeding back the test results to the oxygen supply billing node to adjust the billing status and trigger an early warning also includes:

[0038] Based on the current user’s disease type and disease severity score, combined with the disease type-reimbursement ratio rules preset by the reimbursement node, the cumulative reimbursement ratio for the current user is obtained.

[0039] Based on the cumulative reimbursement ratio corresponding to the current user and the cumulative oxygen supply measured by the segmented standardized metering model, the corresponding cumulative billing is reimbursed and allocated. The allocation result is transmitted to the information storage node corresponding to the current user in the electronic medical record subsystem through the transmission synchronization node and displayed in real time.

[0040] Compared with the prior art, the beneficial effects of the present invention are:

[0041] This invention addresses the shortcomings of existing technologies by constructing a multi-node collaborative intelligent oxygen supply control system, achieving precise, personalized, and intelligent management of medical oxygen flow metering. In particular, a tiered oxygen flow rate mapping rule base based on symptom characteristics and severity scores, combined with a reinforcement learning-optimized PID control algorithm, enables precise on-demand oxygen supply, significantly improving treatment efficacy. A segmented standardized metering model and temperature damage compensation matrix effectively eliminate environmental interference, ensuring flow measurement accuracy. A real-time linkage mechanism between oxygen supply quality detection and billing nodes automatically adjusts the billing status when abnormal oxygen concentrations are detected, protecting patient rights. Blockchain-based data storage and hierarchical access control ensure the security and traceability of medical data. Intelligent stage transition control and parameter preloading mechanisms guarantee a smooth transition during the oxygen supply process. Through closed-loop control and self-learning optimization, this application improves oxygen supply safety and treatment efficacy while achieving transparent billing. Attached Figure Description

[0042] Figure 1 This is a block monitoring node chain module diagram provided in an embodiment of the present invention;

[0043] Figure 2 A three-dimensional view of the oxygen flow metering device provided in an embodiment of the present invention;

[0044] Figure 3 This is a plan view of the oxygen flow metering device provided in an embodiment of the present invention. Detailed Implementation

[0045] Please see Figure 1 The present invention provides an embodiment of a medical oxygen flow metering and control system, comprising: a block monitoring node chain, wherein the block monitoring node chain of this embodiment includes:

[0046] A safety node is used to perform safety checks based on the pressure change curve of the real-time monitored oxygen supply subsystem or oxygen cylinder and a preset safety pressure range. When the check result meets the safety pressure range, it responds to the oxygen supply setting node. The oxygen supply setting node obtains the current user's staged oxygen supply flow rate mapping rule based on the input user's symptom characteristics and a preset stepped oxygen supply flow rate mapping rule library, and simultaneously responds to the proportional setting node. The proportional setting node generates staged oxygen supply valve opening commands based on the current user's staged oxygen supply flow rate mapping rule and a reinforcement learning-optimized PID control algorithm, and simultaneously responds to the oxygen supply quality detection node and the oxygen supply billing node. In this embodiment, the responding oxygen supply setting node includes a symptom assessment module and a mapping module.

[0047] The symptom assessment module is used to obtain a severity score for the current type of symptom of the corresponding user based on the user's symptom characteristics and symptom type information synchronized from the electronic medical record subsystem by the transmission synchronization node, combined with a fuzzy comprehensive evaluation algorithm.

[0048] The mapping module is used to generate a phased oxygen supply flow rate mapping rule for the current user based on the user's current symptom severity score, combined with the symptom type-severity-phased oxygen supply flow rate mapping rule in the phased oxygen supply flow rate mapping rule library and the duration of each phase, and in conjunction with preset matching mapping rules. The phased oxygen supply flow rate mapping rule for the current user includes the valve opening size corresponding to each oxygen supply subsystem or oxygen cylinder output port, the oxygen flow rate and concentration corresponding to each phase, and the duration corresponding to the valve opening. The matching mapping rule is specifically set by those skilled in the art; for example, a score of 8-10 points is mapped to a high flow rate range of 5-10 L / min and a 1-hour phased duration; a score of 4-7 points is mapped to a medium flow rate range of 3-4 L / min and a 6-hour phased duration.

[0049] It should be further explained that the design motivation for the safety node, oxygen supply setting node, and proportion setting node in this embodiment stems from the core problem of the disconnect between safety monitoring and treatment execution in traditional oxygen supply subsystems. Addressing the shortcomings of existing technologies where pressure safety detection and personalized oxygen supply control are independent and lack a linkage mechanism, this embodiment constructs a closed-loop control system covering the entire process of pressure monitoring, symptom assessment, and oxygen supply execution, achieving seamless integration from safety warning to personalized treatment. This embodiment ensures the oxygen supply subsystem operates within a safe range through real-time pressure curve analysis, achieves precise treatment plan matching based on a step-wise oxygen supply flow rate mapping rule library according to patient symptom characteristics, and dynamically adjusts oxygen supply parameters using an intelligent PID control algorithm. Ultimately, this forms an intelligent oxygen supply solution integrating safety protection, individualized treatment, and precise control, effectively solving the technical challenge of balancing safety and treatment effectiveness in traditional systems.

[0050] It should be further explained that the specific process of security detection in this embodiment includes:

[0051] Step 1: Based on the redundant pressure sensor array deployed at the main pipeline, branch pipelines, and terminal interfaces of the oxygen supply pipeline, multiple raw pressure data are synchronously collected at a frequency no lower than the preset sampling frequency through a hardware synchronous triggering mechanism to ensure the spatiotemporal consistency of the data. The preset sampling frequency is determined based on the pressure dynamic characteristics of the oxygen supply subsystem and the Nyquist sampling theorem to ensure complete capture of the system pressure fluctuation characteristics. The determination of the preset sampling frequency comprehensively considers the maximum frequency component of the system pressure fluctuation and the signal reconstruction requirements. The minimum sampling frequency is determined through spectrum analysis, and a preset safety margin is retained. The safety margin is set by those skilled in the art according to actual safety needs. For example, in this embodiment, the safety margin is usually set to 20%-50% of the theoretical minimum sampling frequency.

[0052] Step 2: Construct a pressure state estimation model based on the Kalman filter algorithm. Multiple raw pressure data collected by redundant pressure sensor arrays deployed at the main oxygen supply pipeline, branch pipelines, and terminal interfaces are used as model input. The Kalman filter is used to perform prediction, update, and iterative fusion operations on the multiple raw pressure data using a pre-configured process noise covariance matrix and observation noise covariance matrix. The result is the output of the optimal pressure estimate for the corresponding oxygen supply pipeline. The observation noise covariance matrix is ​​pre-set based on the factory calibration results of each pressure sensor, on-site calibration data, and long-term measurement error statistics. The matrix elements are negatively correlated with the measurement accuracy of the corresponding sensor; the higher the sensor's measurement accuracy, the smaller the value of the corresponding matrix element. The process noise... The covariance matrix is ​​dynamically adjusted based on the real-time pressure change characteristics of the oxygen supply system. By monitoring the rate of change, fluctuation amplitude, and frequency of abrupt changes in pressure data in real time, a correlation mapping relationship between pressure change characteristics and matrix element values ​​is established. When the rate of change in pressure increases, the fluctuation amplitude rises, or a pressure abrupt change occurs, the corresponding element values ​​of the process noise covariance matrix are increased synchronously to adapt to the state estimation requirements of dynamic pressure changes. When the pressure is in a stable state, the corresponding element values ​​of the process noise covariance matrix are decreased to improve the accuracy of pressure state estimation. Through the dynamic adjustment of the process noise covariance matrix and the preset configuration of the observation noise covariance matrix, the Kalman filter algorithm can accurately iteratively fuse multiple raw pressure data, ensuring the real-time performance and accuracy of the optimal pressure estimate.

[0053] Step 3: A sliding window mean filtering algorithm based on a preset window size is used to perform secondary smoothing on the fused pressure data, effectively filtering out high-frequency noise interference and obtaining stable pressure values. The preset window size is dynamically adjusted according to the pressure fluctuation characteristics. Its preset range is determined based on the balance between system response speed and filtering effect. The preset window size is determined by analyzing the fluctuation characteristics of historical pressure data, ensuring the filtering effect while maintaining the system response speed, and dynamically adjusting the window size according to different working conditions.

[0054] Step 4: Based on the real-time temperature data collected by the ambient temperature sensor, the reference safe pressure range is dynamically calibrated using a preset temperature compensation coefficient to obtain a dynamic safe pressure range that adapts to environmental changes. The temperature compensation coefficient is obtained through the following calibration process: multiple uniformly distributed calibration temperature points are set in a constant-temperature experimental environment. The number and distribution range of these calibration temperature points are determined based on the system's operating temperature range and temperature sensitivity analysis to ensure good compensation performance across the entire operating temperature range. Standard pressure values ​​of different gradients are applied to the pressure sensor using a standard pressure generator. Measured sensor data under various temperature-pressure combinations are collected. The temperature-pressure relationship curve is fitted using the least squares method to obtain the preliminary compensation coefficient. The deviation between the measured data and the model output is compared using a gradient descent algorithm in conjunction with the physical model to iteratively correct the preliminary compensation coefficient. The accuracy and stability of the corrected compensation coefficient are verified using field operation data. The physical model in this embodiment characterizes the inherent physical correlation between the measured values ​​of the pressure sensor in the oxygen supply system and the ambient temperature and actual pressure. The physical model is constructed based on the laws of gas thermodynamics and the thermal conductivity characteristics of the pressure sensor, incorporating the linear relationship between gas pressure and temperature, the temperature drift characteristics of the sensor element, and the thermal expansion and contraction physical characteristics of the gas in the oxygen supply pipeline. It comprehensively depicts the influence mechanism of ambient temperature changes on the measurement accuracy of the pressure sensor. It is used to iteratively correct the initial compensation coefficients obtained by least squares fitting using the gradient descent algorithm. The model is obtained by inputting the actual sensor data under various temperature-pressure combinations into the physical model. The theoretical output value is compared with the deviation between the measured data and the theoretical output value of the model. Using this deviation as the optimization target, the value of the initial compensation coefficient is adjusted successively through the gradient descent algorithm until the deviation converges to the preset deviation threshold. This corrects the possible deviation of the least squares method based solely on experimental data fitting, so that the corrected temperature compensation coefficient not only fits the actual measurement data, but also conforms to the objective physical laws of gas pressure and temperature changes. This improves the scientificity and accuracy of temperature compensation, and ensures that the dynamic safety pressure range after dynamic calibration based on this coefficient can accurately adapt to the pressure monitoring needs of the oxygen supply system under different ambient temperatures, ensuring the accuracy and reliability of pressure safety detection.

[0055] Step 5: Based on the continuous multi-cycle pressure monitoring mechanism, the stable pressure value is compared with the dynamic safe pressure range in real time. When the pressure value exceeds the safe pressure range for a number of consecutive preset sampling cycles, it is determined to be a pressure anomaly. The number of consecutive preset sampling cycles is determined based on the balance between system stability requirements and false alarm rate. The preset number of consecutive sampling cycles in the pressure anomaly determination is based on the system inertial characteristics and false alarm suppression requirements. The optimal threshold is determined by statistical analysis of historical abnormal data patterns.

[0056] Step 6: Based on the preset multi-level pressure threshold judgment logic, set the trigger conditions for fault codes for low pressure and high pressure respectively. The trigger conditions for the fault codes are determined based on the system's safe operating pressure range and the severity level of the fault. The preset trigger conditions for the fault codes adopt a hierarchical setting strategy, setting different duration thresholds according to different severity levels of faults to ensure timely response to major faults while avoiding false triggers.

[0057] Step 7: Based on the high pressure fault handling process, determine the release conditions of the proportional valve lockout state by monitoring the pressure drop trend and stability indicators; wherein the preset release conditions are determined based on the system safety recovery characteristics and equipment protection requirements, and the preset release conditions of the proportional valve lockout state comprehensively consider the system recovery safety and operating efficiency, and determine the corresponding parameters by analyzing the dynamic characteristics of the pressure drop process and the system stability requirements.

[0058] Step 8: Based on the timestamp of the block record node, use a hash algorithm to perform encryption operations on the entire process event sequence to generate an immutable secure log record; wherein the selection of the hash algorithm is determined based on the system security level and anti-tampering requirements.

[0059] The oxygen supply billing node is used to obtain the segmented integral oxygen supply amount based on the current user's segmented oxygen supply flow rate mapping rules and the segmented standardized metering model. It then performs real-time oxygen supply billing based on the segmented integral oxygen supply amount and preset segmented billing rules, simultaneously feeding the billing results back to the display interface and to the block record node for on-chain storage. The oxygen supply quality detection node is used to detect the oxygen concentration and oxygen supply flow rate in real time during the oxygen supply process, and performs oxygen supply detection by combining exponential moving average and preset anomaly detection thresholds. The detection results are fed back to the oxygen supply billing node to adjust the billing status and trigger warnings. The preset anomaly detection thresholds include an abnormal oxygen concentration threshold and the standard segmented instantaneous oxygen supply flow rate for the corresponding stage.

[0060] It should be further explained that the stepped oxygen supply flow rate mapping rule base in this embodiment is constructed by combining a hash algorithm and a tree-structured database, based on the disease type, the severity score corresponding to each disease type, the number of oxygen supply stages corresponding to the severity score, the interval duration and average oxygen supply flow rate corresponding to each oxygen supply stage, and the severity score. This rule is used to quickly match the stepped oxygen supply flow rate mapping rules corresponding to the severity score of each disease type. The segmented billing rules are constructed by the average oxygen supply flow rate of the corresponding stage and the billing unit price corresponding to the average oxygen supply flow rate. It should also be noted that in this embodiment, oxygen measurement is performed using the segmented standardized metering model built into the oxygen flow metering device. Please refer to [link to relevant documentation]. Figure 2 and Figure 3The diagram shows an oxygen flow metering device. It should be further explained that the design motivation for the oxygen supply billing node and oxygen supply quality detection node in this embodiment stems from core problems in traditional oxygen therapy, such as inaccurate billing and lack of quality supervision. Addressing the issues of existing technologies being unable to dynamically adjust billing standards based on patient conditions and lacking linkage between oxygen quality and cost, this design achieves precise matching between patient condition and oxygen supply plan by constructing a tiered oxygen supply flow rate mapping rule library. It employs a segmented standardized metering model to ensure accurate measurement of oxygen usage, establishes a real-time linkage mechanism between quality detection and billing to guarantee oxygen safety and cost fairness, and utilizes blockchain technology to ensure the immutable storage of billing data. This integrated design not only solves the problem of mismatch between traditional time-based billing and actual oxygen consumption but also avoids billing disputes related to ineffective oxygen supply through quality monitoring. Ultimately, it achieves a technological leap in medical oxygen supply from extensive management to precise metering and quality-price correspondence, providing medical institutions with a transparent and reliable basis for charging while protecting the legitimate rights and interests of patients.

[0061] The humidification control node is used to humidify the supplied oxygen simultaneously with a preset oxygen humidification ratio. It also collects comfort data and assesses the comfort of the humidified oxygen supply through a questionnaire, and adjusts the oxygen humidification ratio in real time based on the assessment results. It should be further noted that the core motivation of the humidification control node in this embodiment is to solve the problem that traditional oxygen humidification systems, which use a fixed humidification ratio, cannot adapt to individual patient differences and environmental changes. Its specific implementation process includes:

[0062] Step 1: Using high-precision temperature and humidity sensors deployed in the oxygen supply pipeline, the temperature and relative humidity of the humidified oxygen are monitored in real time, with a sampling frequency of no less than once per second. Simultaneously, a standardized electronic questionnaire on the patient's terminal device is used to collect the patient's comfort rating for oxygen humidity and temperature, as well as subjective feelings such as nasal dryness and throat irritation. All data collection processes are synchronized using a unified timestamp.

[0063] Step 2: The physical parameters collected by the sensors are fused with the patient's subjective evaluation data at the feature level. First, the temperature and humidity sensor data is smoothed by a sliding window to eliminate the interference of instantaneous fluctuations. Then, the questionnaire data is normalized and coded to convert the verbal descriptions into numerical scores. Finally, the correspondence between physical parameters and subjective feelings is established through a dynamic time warping algorithm to form a data sample containing multi-dimensional features such as temperature, humidity, comfort score, and dryness index.

[0064] Step 3: Construct a comfort assessment model. The training and construction process of this comfort assessment model is as follows: First, collect physiological parameters and subjective evaluation data of patients with different disease types under ideal oxygen supply conditions from historical data. Physiological parameters include the temperature and relative humidity of oxygen after humidification. Subjective evaluation data includes comfort scores for oxygen humidity and temperature, nasal dryness scores, and throat irritation scores. Use the K-means clustering algorithm to perform cluster analysis on the above data to identify the optimal comfort range for oxygen humidification for each type of patient. Based on the optimal comfort range obtained from clustering, use the random forest regression algorithm to construct a comfort assessment model. The input of this comfort assessment model is the temperature and relative humidity values ​​of the current patient after oxygen supply and humidification, collected in real time, as well as the multi-dimensional feature vector formed by feature-level fusion processing of the current patient's comfort score, nasal dryness score, and throat irritation score. The output is a quantified comfort score, with the comfort score ranged from 0 to 10 points, used to accurately reflect the current patient's comfort level with oxygen supply and humidification, providing a quantitative basis for the dynamic adjustment of the subsequent humidification ratio.

[0065] Step 4: When the system detects that the comfort score is consistently lower than the preset comfort score threshold, it automatically triggers the humidification ratio adjustment algorithm. The humidification ratio adjustment algorithm in this embodiment is based on the fuzzy PID control principle, dynamically calculating the required output power and humidification water volume of the humidifier according to the magnitude and trend of the comfort deviation. The control strategy comprehensively considers the patient's condition type, current environmental conditions, and historical adjustment effects to ensure a smooth and gradual adjustment process.

[0066] Step 5: The system creates a unique personalized configuration file for each patient, recording parameter settings, adjustment effects, and patient feedback during each intervention process. Reinforcement learning algorithms are used to analyze historical data, continuously optimizing the patient's comfort assessment model parameters and intervention strategies to achieve precise adaptation based on individual differences.

[0067] Step Six: After executing the control command, the system continuously monitors changes in the patient's comfort level and verifies the control effect by comparing the score data before and after the adjustment. If the effect does not meet expectations, the system will automatically reassess and adjust the control parameters, forming a complete closed-loop control cycle of monitoring-assessment-control-verification. All control process and effect data are recorded to provide data support for subsequent treatment.

[0068] The reimbursement node is used to obtain the current user's reimbursement ratio based on the input user's symptom characteristics and a preset symptom type-reimbursement ratio rule. It then maps the current user's reimbursement ratio to the oxygen supply billing node to adjust the billing results proportionally and displays the adjusted results in real time. When the reimbursement ratio changes, the preset symptom type-reimbursement ratio rule is updated and stored on the blockchain. In this embodiment, the preset symptom type-reimbursement ratio rule is constructed by those skilled in the art based on symptom types and medical reimbursement ratios.

[0069] It should be further explained that the design motivation for the reimbursement node in this embodiment is to solve the problems of cumbersome manual operation, error-prone reimbursement ratio calculation, and delayed adjustment response in the traditional medical expense reimbursement process. By establishing intelligent mapping rules between disease types and reimbursement ratios, automated execution is achieved, integrating reimbursement calculation into the real-time billing process, effectively avoiding errors and delays caused by manual intervention. Blockchain technology is used to ensure the immutability and traceability of the reimbursement ratio rules, guaranteeing both the standardization of use and improving the transparency and efficiency of patient expense settlement, ultimately building a precise, efficient, and reliable intelligent medical expense reimbursement management system.

[0070] The transmission synchronization node is used to transmit real-time bidirectional data between the oxygen supply terminal and the corresponding user in the electronic medical record subsystem, and to feed back the synchronized data to the response oxygen supply setting node and the reimbursement node.

[0071] A hierarchical permission node is used to establish a hierarchical permission mapping based on the sensitivity of the corresponding user's symptom characteristics or basic information and the corresponding access role permission level. This hierarchical permission mapping is then synchronized to the transmission synchronization node. When an application for access to the corresponding user's symptom characteristics or basic information is submitted to the electronic medical record subsystem from the oxygen supply end, permission verification is performed based on the hierarchical permission mapping. If verification passes, access is allowed; if verification fails, access is denied, and a cross-permission access warning is issued. It should be further noted that the access role permission level in this embodiment is specifically set by those skilled in the art based on the corresponding role type, which includes at least medical staff, engineers, and administrators. The sensitivity of the corresponding user's symptom characteristics or basic information in this embodiment is preferably obtained through a comprehensive fuzzy evaluation algorithm. It should also be noted that the design motivation of the hierarchical permission node in this embodiment is to solve the problem of rigid and unresponsive data access permission settings in traditional medical systems. To address the diverse sensitivity levels and access roles of medical data, a dynamic permission mapping mechanism based on symptom characteristics and basic information sensitivity is established. This mechanism, combined with permission levels for multiple roles such as medical staff, engineers, and administrators, enables a shift from extensive permission management to intensive access control. The system uses a comprehensive fuzzy evaluation algorithm to quantitatively assess data sensitivity and achieves real-time synchronization of permission policies through transmission synchronization nodes. When the oxygen supply end requests access to electronic medical records, automatic permission verification is performed. This ensures the security and privacy of medical data while meeting the differentiated needs of different roles in diagnosis, equipment maintenance, and system management. It effectively prevents unauthorized access and data leakage risks, complying with medical information system security standards.

[0072] The block record node is used to collect the data and corresponding monitoring timestamps monitored by each node in the block monitoring node chain, combine them with a hash algorithm to obtain the on-chain hash index of the corresponding node, and store the data of the corresponding node and the on-chain hash index on the chain to form a fast traceability index log. Simultaneously, it combines with smart contracts in the blockchain for data sharing between nodes. In this embodiment, the block record node is also configured with a data disaster recovery early warning mechanism. The specific implementation process includes: a monitoring agent program deployed in the primary and backup active-active data centers performs bidirectional heartbeat detection and data integrity verification at a second-level frequency of once per second. Simultaneously, it uses the Gossip protocol to synchronize the status information of the storage nodes in the active-active data centers in real time, establishing a health assessment index including three core parameters: network latency, storage availability, and data verification pass rate. The preset network latency threshold is 50ms, the preset storage availability standard is 99.9%, and the preset data verification pass rate standard is 99.99%. A three-level early warning response mechanism triggers different levels of early warning based on the degree of non-compliance of the above health assessment indicators. Specifically, when... If any health assessment indicator slightly fails to meet the standard, i.e., does not reach the moderate anomaly threshold (e.g., network latency of 51-100ms, storage availability of 99.5%-99.89%, data verification pass rate of 99.90%-99.98%), a Level 1 warning is triggered, immediately initiating RAID reconstruction and dynamic bandwidth allocation of the storage array to prioritize the read / write performance of core medical data and alleviate storage pressure. If any health assessment indicator moderately fails to meet the standard, i.e., reaches the moderate anomaly threshold (e.g., network latency of 101-200ms, storage availability of 99.0%-99.49%, data verification pass rate of 99.80%-99.89%), a Level 2 warning is triggered, activating a block-level fast migration and consistency verification algorithm to migrate critical data blocks from the abnormal node to a backup storage node and verify data integrity block by block to ensure no data loss. If any health assessment indicator severely fails to meet the standard, i.e., reaches the severe anomaly threshold (e.g., network latency > 200ms, storage availability < 99.0%, data verification pass rate < 99.98%), a Level 1 warning is triggered.When 80% of the time, or when a single point of hardware failure or network interruption occurs in the active-active dual-data center, a level-three warning is triggered, executing a traffic switching and data repair process based on the BGP routing protocol. This completes the service traffic switching between the active-active dual-data center within seconds, while simultaneously initiating a full data repair process to synchronously restore abnormal data. Furthermore, the system uses the SHA-256 algorithm to perform daily periodic hash comparisons of the data in the active-active dual-data center, automatically building a retransmission queue for differing data blocks and completing data completion incrementally, significantly improving synchronization efficiency. When an extreme failure is detected in the primary data center, the system can complete session persistence transfer and transaction consistency maintenance within seconds, ensuring... This system ensures the integrity and traceability of medical data within a ten-year retention period, fully meeting the stringent requirements of JCI certification for medical device data management. By integrating blockchain technology, all medical operation records are generated into an immutable on-chain hash index using a hash algorithm, forming a rapid traceability index log. Smart contracts enable data sharing between nodes, while health assessment indicators are continuously collected to monitor the operational status of the dual-active data center. A three-tiered early warning response mechanism and incremental synchronization process work together to ensure data security, ultimately achieving the management goals of zero data loss, traceability, and high availability. This addresses the core issues of traditional medical systems, such as single points of failure, insufficient data traceability capabilities, and inability to meet international certification standards.

[0073] The annual inspection lock point automatically generates an electronic verification certificate containing a QR code based on the PDF / A standard, directly connecting the control terminal of each oxygen flow meter to the quality supervision department's subsystem to provide annual inspection reminders. When the annual inspection is not carried out within the time limit, the corresponding oxygen flow meter is automatically locked.

[0074] In this embodiment, the block monitoring node chain integrates security nodes, oxygen supply setting nodes, proportion setting nodes, oxygen supply quality detection nodes, oxygen supply billing nodes, humidification control nodes, reimbursement nodes, transmission synchronization nodes, hierarchical permission nodes, block record nodes, and annual inspection lock nodes to construct an intelligent supervision system covering the entire oxygen supply process. Security nodes implement pressure safety interlock control; oxygen supply setting nodes generate personalized oxygen supply plans based on electronic medical records; proportion setting nodes precisely execute flow control through PID control algorithms; oxygen supply quality detection nodes monitor oxygen concentration and flow rate in real time; oxygen supply billing nodes achieve accurate billing based on segmented standardized metering models; humidification control nodes dynamically adjust oxygen supply comfort; reimbursement nodes automatically complete cost allocation; transmission synchronization nodes ensure real-time synchronization of medical data; hierarchical permission nodes establish a tiered access mechanism; block record nodes achieve trusted data storage through blockchain; and annual inspection lock nodes achieve equipment lifecycle management. Ultimately, this forms a closed-loop control system integrating safety early warning, precise oxygen supply, intelligent billing, quality supervision, and data security.

[0075] It should be further explained that the construction and training process of the segmented standardized econometric model in this embodiment includes:

[0076] Based on symptom type, symptom severity score, filtered temperature change gradient value and corresponding compensation power value, constant temperature difference, and standard staged instantaneous oxygen supply flow rate, a staged hierarchical mapping chain is established using the Apriori algorithm. The implementation logic of the Apriori algorithm in this embodiment is as follows: Input parameters include symptom type, symptom severity score, filtered temperature change gradient value, filtered compensation power value, constant temperature difference, and standard staged instantaneous oxygen supply flow rate. Two core parameters are preset: minimum support of 0.15 and minimum confidence of 0.85. Simultaneously, the itemset iteration threshold is set to 5, and the minimum itemset length is set to 2. First, all input parameters are discretized, dividing continuous parameters into reasonable intervals and converting them into discrete terms. An initial 1-itemset is generated, followed by iterative scanning of the dataset to select frequent 1-itemsets that meet the minimum support. Candidate 2-itemsets are then generated based on these frequent 1-itemsets. This process of selecting frequent itemsets and generating higher-order candidate itemsets is repeated until no new frequent itemsets can be generated. Strong association rules that meet the minimum confidence are then extracted from all frequent itemsets. The core association rules between symptom type, symptom severity score, and standard phased instantaneous oxygen supply flow rate are identified. Simultaneously, the abnormal association logic between filtered temperature change gradient value, filtered compensation power value, constant temperature difference, and standard phased instantaneous oxygen supply flow rate is also identified. Finally, all strong association rules that meet the conditions are output. Based on this, a phased hierarchical mapping chain is constructed to clarify the hierarchical correspondence between each parameter. In this embodiment, the input items of the Apriori algorithm are disease type, disease severity score, filtered temperature change gradient value, filtered compensation power value, constant temperature difference, and standard phased instantaneous oxygen supply flow rate. After discretizing these input items, a single discretized parameter is the initial 1-itemset: for example, the disease severity score is discretized into mild (0-3 points), moderate (4-7 points), and severe (8-10 points), and the standard phased instantaneous oxygen supply flow rate is discretized into low flow rate (1-2 L / min), medium flow rate (3-4 L / min), and high flow rate (5-10 L / min); if the support of the severe score in all samples is greater than or equal to the preset minimum support, then the "severe score" is a frequent 1-itemset; if the support of the high flow rate is greater than or equal to the preset minimum support, then the "high flow rate" is also a frequent 1-itemset.In the application scenario of constructing a phased hierarchical mapping chain based on the Apriori algorithm in this embodiment, the candidate 2-itemset refers to a set containing only two discretized parameter items generated by combining two different frequent 1-itemsets based on the frequent 1-itemsets obtained after discretization. It is a preliminary intermediate product for mining frequent 2-itemsets. It does not naturally meet the frequency requirement. It needs to be further analyzed by scanning medical oxygen supply-related datasets, calculating the support of the set in the dataset, and comparing the calculation results with the preset minimum support to select those that meet the support standard as frequent 2-itemsets. For example, "severe score + high flow rate" formed by combining "severe score" and "high flow rate", "medium score + medium flow rate" formed by combining "medium score" and "medium flow rate", and "high flow rate + compensation power 8-12W" formed by combining "high flow rate" and "compensation power 8-12W" are all candidate 2-itemsets in this embodiment.

[0077] Specifically, this embodiment uses the disease type and severity score as the core input for clinical needs, constructs a phased and hierarchical mapping chain to achieve precise matching between clinical needs and physical oxygen supply, and builds a reverse diagnosis mechanism based on three types of physical measurement parameters: temperature change gradient value, constant temperature difference, and compensation power value, to ensure the accuracy and reliability of oxygen supply control. The "symptom type" refers to the specific category of the patient's disease, such as chronic obstructive pulmonary disease (COPD) or acute respiratory distress syndrome (ARDS). This determines the basic direction of the oxygenation plan. For example, COPD patients require continuous low-flow oxygen therapy, while ARDS patients require high-flow, high-concentration oxygen therapy. The "symptom severity score" is a quantitative assessment of the severity of the patient's current condition, using a 0-10 scale: 0-3 for mild, 4-7 for moderate, and 8-10 for severe. This score refines the intensity and stages of the oxygenation plan. Both criteria work together to determine the appropriate theoretical oxygenation plan for the patient, including key parameters such as the oxygen flow rate and duration at each stage. For example, for severe ARDS patients with a severity score of 8-10, the theoretical oxygenation plan is a high-flow rate range, specifically 5-10 L / min, with a step-up duration of 1 hour; for moderate ARDS patients with a score of 4-7, the plan is a medium-flow rate range, specifically 3-4 L / min, with a step-up duration of 6 hours. Temperature gradient, constant temperature difference, and compensation power are used as physical measurement parameters to reflect the actual oxygen supply status in real time. Temperature gradient is the rate of change of the temperature difference between the sensing element and the reference element over time, measured in °C / s. It reflects the rate at which oxygen flow carries away heat, indirectly reflecting the instantaneous fluctuation of the oxygen supply flow rate. For example, when the oxygen supply flow rate suddenly increases from 3 L / min to 5 L / min, the temperature gradient will rapidly increase from 0.2 °C / s to 0.5 °C / s. Constant temperature difference is a fixed temperature difference that the system presets to be maintained between the sensing element and the reference element. In this embodiment, it is set to 5 °C, which serves as a benchmark to ensure the accuracy of flow measurement and is used to stabilize the thermal equilibrium state. Compensation power is the electrical power supplied to the sensing element by the circuit system to maintain a constant temperature difference, measured in W. Its magnitude is proportional to the mass of oxygen and the oxygen supply flow rate, and can directly reflect the actual oxygen supply flow rate. For example, when the oxygen supply flow rate is 5 L / min, the compensation power is stable at 10 W, and the compensation power increases or decreases synchronously when the flow rate deviates. The standard phased instantaneous oxygen supply flow rate is used as the benchmark connecting clinical needs and physical implementation. This standard value is determined from a pre-set step-by-step oxygen supply flow rate mapping rule library based on the type and severity of the disease. It is both the core execution parameter of the theoretical oxygen supply plan and a reference for measuring whether the actual oxygen supply meets clinical needs.When the actual measured temperature change gradient, constant temperature difference, or compensation power deviates from the standard staged instantaneous oxygen supply flow rate of the corresponding phase, the system uses a staged hierarchical mapping chain to perform a layer-by-layer reverse deduction to quickly locate the problem and implement targeted compensation: If the compensation power is abnormal, for example, the normal range of compensation power for a patient with severe acute respiratory distress syndrome at a standard flow rate of 5-10 L / min is 8-12 W, but the actual monitored value is 15 W, which deviates from the normal range. Since the compensation power is directly determined by the heat loss rate caused by oxygen flow, and the heat loss rate is closely related to the temperature change gradient, the larger the temperature change gradient, the faster the heat loss rate, and the higher the required compensation power. Therefore, the system first checks whether the measurement of the temperature change gradient is affected by environmental interference. For example, external airflow disturbance causes the temperature change gradient to rise sharply from 0.3℃ / s to 0.8℃ / s, which in turn causes the compensation power adjustment to deviate from the normal logic. At this time, the system automatically triggers the sensor calibration program to eliminate the influence of environmental interference. If the temperature change gradient does not match the standard staged instantaneous oxygen supply flow rate, that is, the temperature change is not normal at the same standard flow rate, the system performs a step-by-step reverse deduction based on the staged hierarchical mapping chain to quickly locate the problem and implement targeted compensation: The temperature gradient deviates from the normal range. For example, the normal range for the temperature gradient corresponding to the standard oxygen flow rate of 3-4 L / min for patients with moderate acute respiratory distress syndrome is 0.15-0.3℃ / s, while the actual monitored value is 0.4℃ / s. This needs further examination from two aspects: First, the standard staged instantaneous oxygen flow rate is determined based on the severity score of the disease. If the score is incorrect, a moderate patient may be misdiagnosed as a severe patient, resulting in a standard oxygen flow rate set at 5 L / min, while the actual flow rate is only 3 L / min. Naturally, the temperature gradient will deviate from the incorrect standard. If the oxygen supply flow rate is mismatched, the system will automatically correct the severity score of the illness and rematch the standard oxygen supply flow rate. On the other hand, the constant temperature difference is the benchmark for the system to maintain thermal balance. If the constant temperature difference fluctuates, for example, from 5℃ to 4.5℃, deviating from the preset fixed difference, even if the actual oxygen supply flow rate is consistent with the standard oxygen supply flow rate, the heat loss rate will change due to the change in the benchmark temperature difference. This will cause the normal correspondence between the temperature change gradient and the standard oxygen supply flow rate to be broken. At this time, the system will start the constant temperature difference stabilization control program to restore the constant temperature difference to 5℃. This embodiment, through this standard-link-based reverse diagnostic mechanism, can quickly locate problems such as measurement errors, parameter setting errors, and control instability, and implement targeted compensation operations such as sensor calibration, scoring correction, and temperature difference stabilization control. For example, to address temperature change gradient measurement deviations caused by sensor aging, the system automatically performs zero-point calibration; to address standard oxygen supply flow rate deviations caused by scoring errors, the system collaborates with the electronic medical record subsystem to reassess the patient's condition; and to address constant temperature difference fluctuations, the system adjusts the heating power in real time to restore thermal balance. This ensures the accuracy and reliability of oxygen supply control in complex clinical environments, effectively preventing insufficient oxygen supply from affecting treatment outcomes or excessive oxygen supply from causing patient discomfort.

[0078] This embodiment uses a configured heating element and temperature sensor to collect the temperature change gradient values ​​and corresponding compensation power values ​​of the heating element in real time at different oxygen supply stages. A filtering algorithm is then used to filter the data, obtaining the filtered temperature change gradient values ​​and corresponding compensation power values. Specifically, based on the real-time collected temperature change gradient values ​​and compensation power values, this embodiment first uses a sliding window mean filter for preliminary smoothing. The arithmetic mean is calculated based on a preset window size of eight sampling points to obtain the preliminary filtered signal. A Kalman filter state equation is established based on a heat conduction physical model. Iterative calculations of the prediction and update steps are used to optimally estimate the preliminary filtered signal, obtaining an optimized signal that removes process noise and measurement noise. Finally, a 32nd-order FIR low-pass filter is designed based on the Hanning window function, and a 10 Hz cutoff frequency is set to perform frequency domain filtering on the optimized signal, obtaining the final temperature change gradient values ​​and compensation power values. The heat conduction physical model in this embodiment is constructed based on the law of heat conduction and the heat exchange characteristics of heating elements and sensing elements in the oxygen supply system. This heat conduction physical model characterizes the heat transfer law of the heating element in different oxygen supply stages, the intrinsic relationship between temperature change gradient and compensation power, and the influence mechanism of oxygen flow on the heat conduction process of sensing elements. At the same time, it incorporates the interference characteristics of external factors such as ambient temperature and airflow disturbance on the heat conduction process. It is used to establish a state equation that fits the actual oxygen supply scenario for the Kalman filter algorithm, and provides accurate physical constraints and parameter support for the iterative calculation of the prediction and update steps of the Kalman filter. This enables the Kalman filter to make an optimal estimate of the temperature change gradient and compensation power preliminary filtered signal after sliding window mean filtering based on this model, effectively removing process noise and measurement noise, improving the accuracy and stability of the filtered signal, and laying the foundation for obtaining accurate temperature change gradient values ​​and compensation power values ​​in the future.

[0079] The phased hierarchical mapping chain, the filtered temperature change gradient value, and the compensation power value are used as training parameters and input into the preset phased nonlinear correction algorithm to calculate the real-time phased instantaneous oxygen supply flow rate corresponding to the current phase.

[0080] It should be further explained that the construction process of the staged nonlinear correction algorithm in this embodiment includes:

[0081] First, the oxygen supply flow rate is monitored in real time. When the oxygen supply flow rate is detected to enter the preset ultra-low flow rate operating range, the porous media laminar flow differential pressure detection device and the thermal diffusion measurement unit are immediately and synchronously activated. The differential pressure detection device senses the pressure difference signal generated by the gas flow through its precisely calibrated laminar flow element, and this signal is accurately acquired by a high-precision differential pressure sensor. At the same time, the thermal diffusion measurement unit acquires the temperature distribution data of the airflow channel in real time through a distributed temperature sensor array, and simultaneously records the compensation power value required to maintain a constant temperature difference. All data acquisition processes are strictly time-synchronized through a hardware clock.

[0082] Second, the collected multi-source signals are standardized and preprocessed. The differential pressure signal is filtered and denoised, the temperature distribution data is converted into gradient change, the compensation power value is normalized and calibrated, and a high-dimensional feature vector is constructed based on the three types of processed data.

[0083] Third, the constructed high-dimensional feature vector is input into a support vector regression model specifically designed based on laminar flow characteristics. In the support vector regression model, the pressure difference signal serves as the core parameter and undertakes the main function of flow velocity calculation, while the temperature gradient and compensation power serve as auxiliary parameters responsible for environmental compensation and signal verification. The nonlinear features are mapped to a high-dimensional space through a convolutional radial kernel function, and linear regression calculation is completed in this space. Finally, the accurate flow velocity value after environmental factor compensation and signal validity verification is output.

[0084] When the oxygen supply flow rate is detected to be within the normal flow rate range, the random forest regression algorithm is activated to achieve multi-source sensor data fusion and accurate flow rate correction through the following steps:

[0085] Construct an ensemble learning model consisting of 50-100 decision trees. Each decision tree is trained based on a randomly selected feature subset with a size equal to the square root of the total number of features. The maximum depth of the decision trees is limited to 5-15 layers. The training subset for each decision tree is generated using a bootstrap sampling method.

[0086] Simultaneously collect temperature change gradient data and initial oxygen supply flow rate data, and perform standardized preprocessing on multi-source sensor data: use Z-score standardization for temperature change gradient data and remove outliers exceeding three times the standard deviation; use minimum-maximum normalization for initial oxygen supply flow rate data and eliminate instantaneous fluctuations through moving average filtering to form a feature vector with uniform dimensions.

[0087] Each decision tree uses the Gini impurity criterion to split nodes, sets the minimum number of leaf node samples to 5, and recursively partitions and predicts feature vectors.

[0088] The prediction results of all decision trees are aggregated, and the corrected accurate flow value is calculated by weighted average. The weights are obtained by calculating the mean square error of each decision tree on the validation set, determining the relative accuracy weights, normalizing by the softmax function, and setting the weight update period and decay mechanism.

[0089] Using the oxygen flow rate data from the last 30 sampling points in the current oxygen supply phase, the ARIMA algorithm is used to predict the trend of cumulative oxygen supply in the next phase.

[0090] Based on the prediction results and the preset flow rate-valve opening mapping table, the appropriate oxygen supply flow rate parameters and corresponding valve opening adjustment instructions are generated in advance.

[0091] It should be further noted that this embodiment presets a time window to complete parameter preloading before triggering the conversion during the flow rate stage. The control of the flow rate stage conversion includes:

[0092] First, a basic sampling period is defined, and a parameter preloading time window is set based on system response time and control accuracy requirements. This window is set for a specific duration before the flow rate stage transition trigger, ensuring that parameters are in place in advance. The system also adopts a dual-buffer architecture, including an independently operating current parameter buffer and a preloaded parameter buffer. When entering the preloaded time window, the system obtains the oxygen supply flow rate parameters and valve opening adjustment commands for the next stage from the proportional setting node and writes these parameters into the preloaded parameter buffer. After the parameters are written, the system performs a strict parameter integrity check, verifying key fields such as the target flow rate value, valve opening adjustment step size, and duration. Only parameters that pass the check will trigger the buffer locking command. When the system receives the flow rate stage transition trigger signal, it immediately performs a buffer output channel switch, achieving parameter activation without delay.

[0093] Secondly, to maintain the accuracy of the mapping relationship between flow rate and valve opening, this embodiment establishes a periodic update mechanism. Specifically, the update cycle is set according to the system's operational stability and the frequency of changes in operating conditions. The system collects corrected instantaneous oxygen supply flow rate and actual valve opening feedback values ​​within historical periods to form a linear regression fitting dataset. The linear relationship expression between flow rate and valve opening is calculated using the least squares method. Based on this, a new mapping relationship table is generated. After the new table is generated, the system selects multiple sets of real-time sampled data for error verification. The verification criteria comprehensively consider the system's control accuracy requirements and equipment characteristics. If the error exceeds the allowable range, the system automatically re-executes the fitting process; if it meets the requirements, the new table is activated and the old table is retained for a specific period of time according to preset rules to ensure a smooth transition between the old and new tables.

[0094] Third, the system uses dual-signal verification logic to confirm the timing of stage transitions. Specifically, the first signal is generated by the ARIMA time series prediction model. When the predicted cumulative oxygen supply for the next stage reaches a specific proportion of the current stage threshold, an initial trigger signal is generated. The second signal is generated by continuously comparing the real-time corrected flow rate value with the current stage standard flow rate. When the real-time value reaches a specific proportion of the standard value within multiple consecutive sampling periods, a confirmation trigger signal is generated. In addition, the system sets an anti-jitter time window. Only when two valid signals are detected simultaneously within this window will the proportion setting node generate a stage transition confirmation command; otherwise, it is considered a false trigger and the signal is discarded.

[0095] Fourth, before the phase transition is executed, the system needs to perform three core condition verifications, specifically: First, oxygen supply completion verification, checking whether the cumulative oxygen supply in the current phase has reached the preset phase threshold and the error is within the allowable range; second, flow stability verification, analyzing the fluctuation range of the corrected flow rate value within multiple consecutive sampling cycles to ensure that the system is in a stable working state; and finally, sensor reliability verification, detecting the fault status and output signal amplitude of the temperature, pressure, and flow sensors to confirm that all sensors are working properly. Only when all three verifications are passed will the system output a transition permission signal; for example, if any condition is not met, a transition pause signal will be output and the specific abnormal information will be fed back to the display terminal.

[0096] Fifth, to ensure the computational accuracy of the random forest regression model, this embodiment establishes a weight stability detection mechanism. The system monitors the weight data of each decision tree within multiple consecutive weight update cycles and quantifies the weight stability by calculating the weight variation coefficient. The variation coefficient threshold is set according to the model accuracy requirements and system stability requirements. When the variation coefficient of a decision tree exceeds the variation coefficient threshold, the system immediately starts the weight reset procedure, retrieves the weight value of the previous stable cycle from the parameter history cache to overwrite it, and pauses the model weight update. At the same time, the data diagnosis module will investigate abnormal data sources that cause weight mutations, including sensor fault data and abnormal traffic data. After the abnormal data is processed, the system restarts the weight update process.

[0097] Sixth, this embodiment defines three types of abnormal conversion scenarios and monitors them in real time, specifically including: the first type is abnormal flow rate, which is manifested as the flow rate deviating from the standard value by a certain proportion for multiple consecutive sampling cycles after conversion; the second type is valve jamming, which is manifested as no change for multiple consecutive sampling cycles after the opening is adjusted; the third type is abnormal power, which is manifested as the compensation power exceeding the normal range by a preset proportion. When any abnormal scenario is detected, the block monitoring node chain immediately outputs a rollback command, the system closes the current parameter output channel, and retrieves the stable operating parameters of the previous stage from the parameter history buffer, including the flow rate setpoint, opening setpoint, and compensation power setpoint; at the same time, the system sends an alarm message containing the abnormality type and occurrence time to the control terminal, and completely records the parameters and status data of the rollback process to the fault log.

[0098] It should be further explained that the flow rate stage transition control in this embodiment also includes setting a performance evaluation mechanism, a parameter optimization mechanism, and a log recording function to ensure the accuracy of the stage transition and the long-term stability of the system. The performance evaluation mechanism, using a preset control cycle as a unit, collects real-time oxygen supply flow rate, valve opening feedback value, temperature change gradient value, compensation power value, and oxygen concentration value after the stage transition. It then performs a quantitative evaluation based on three core indicators: mean square error, flow rate stability, and response time. For example, in a certain oxygen supply stage, the flow rate transitions from a low flow rate of 3 L / min to a high flow rate. At a flow rate of 5 L / min, the evaluation indicators include a flow rate stability of over 95%, a response time of no more than 2 seconds, and a mean square error of less than 0.2 L / min. If any indicator fails to meet the threshold requirements, it is automatically marked as a conversion anomaly and a secondary calibration process is triggered. The parameter optimization mechanism, based on the performance evaluation results, combines particle swarm optimization with NSGA-III to construct a multi-objective optimization model. With flow rate control accuracy, system energy consumption, and oxygen supply comfort as optimization objectives, iteratively adjusts the oxygen valve opening step size, oxygen supply flow rate transition coefficient, and oxygen concentration compensation parameters. For example, this is applied to the high flow rate stage conversion. To address the issue of flow rate overshoot, the particle swarm optimization algorithm reduced the oxygen valve opening step size from 0.5% to 0.2% and the oxygen supply flow rate transition coefficient from 1.2 to 0.9, resulting in a 60% reduction in flow rate fluctuations during the transition process. Furthermore, the optimization process incorporated real-time constraints from a phased, hierarchical mapping chain to ensure that the adjusted parameters met medical safety standards. The logging function, using timestamps as the core index, comprehensively recorded key parameters and status data for each stage transition, including the transition trigger signal type, preloaded parameter buffer verification results, oxygen valve opening adjustment commands, and other relevant information. The log records the flow rate deviation, performance evaluation index score, anomaly type, and handling measures. For example, a certain stage of the conversion log records the conversion trigger time, the preload parameter verification passed, the oxygen valve opening was adjusted from 15% to 25%, the real-time flow rate deviation was 0.15L / min, the performance evaluation indicators met the standards, and no anomalies occurred. All log data is synchronously uploaded to the block record node for blockchain storage, which not only provides a traceable data source for subsequent system operation and maintenance and algorithm optimization, but also ensures that the entire stage conversion process is auditable, further improving the reliability and controllability of flow rate stage conversion control.

[0099] After the flow velocity abrupt transition phase ends, the system automatically performs stability detection. When the flow velocity change rate is lower than the set threshold for multiple consecutive sampling periods, the system is determined to have returned to a stable state. At this time, the system automatically switches back to the random forest regression algorithm to perform signal processing for the normal flow velocity phase, thus completing the smooth transition from the transition phase to the stable phase.

[0100] Based on the real-time phased instantaneous oxygen supply flow rate corresponding to the current stage and the standard phased instantaneous oxygen supply flow rate of the corresponding stage, the deviation of the standard oxygen supply flow rate is obtained, and the first oxygen supply metering loss is constructed.

[0101] Based on the standard oxygen supply flow rate deviation, the standard temperature gradient deviation value is obtained through backpropagation algorithm along the staged hierarchical mapping chain, and a staged temperature damage compensation matrix is ​​constructed. Specifically, in the medical oxygen flow metering control system of this embodiment, a stable and accurate oxygen supply flow rate directly determines the patient's treatment effectiveness. However, in actual operation, although the heat exchange between oxygen flow and sensing elements follows the law of conservation of energy, making the temperature change gradient of sensing elements and compensation power naturally linearly correlated, factors such as ambient temperature fluctuations, long-term sensor aging, and airflow disturbances can disrupt this physical correlation, causing the actual oxygen supply flow rate to deviate from the standard oxygen supply flow rate value determined by the disease type and severity score, thereby leading to insufficient oxygen supply, affecting the treatment effect or oxygen supply. To address the issue of patient discomfort caused by excessive oxygen supply, a phased temperature damage compensation matrix needs to be constructed for precise compensation. The matrix is ​​constructed using the standard oxygen supply flow rate deviation as the core error input. Following a phased, layered mapping chain, the error is derived layer by layer through backpropagation. Specifically, the deviation between the real-time collected standard phased instantaneous oxygen supply flow rate and the actual phased instantaneous oxygen supply flow rate is used as the loss function. The gradient contribution of the output flow rate layer to the hidden compensation power layer is calculated to determine the adjustment direction and magnitude of the compensation power. Then, combining the mapping relationship between the temperature gradient and the compensation power constructed using the convolution kernel function, the calculated power deviation is converted into the corresponding temperature gradient deviation, thus realizing the error transfer from the flow rate layer to the compensation power layer. Backpropagation; matrix construction uses disease type and severity score as the core stage division criteria, establishing independent mapping relationships for oxygen supply stages corresponding to different disease types and severity scores. For example, for the high-flow-rate stage with acute respiratory distress syndrome and a severity score of 8-10, based on the standard oxygen supply flow rate of 5-10 L / min and historical monitoring data, the normal fluctuation range of temperature gradient is determined to be 0.2-0.5℃ / s, the stable temperature difference range is 5±0.2℃, and the compensation power threshold is 8-12W. Then, the contribution weight of each parameter to the flow rate deviation is iteratively calculated through the backpropagation algorithm, ultimately forming a matrix containing temperature gradient correction coefficient, power compensation threshold, and temperature difference stable range. The system uses a three-dimensional matrix element. When the system detects in real time that the compensation power for a given stage exceeds the normal range of 8-12W, or the temperature gradient deviates from the range of 0.2-0.5℃ / s, it immediately calls the corresponding matrix element for that stage to automatically adjust the sensor heating power to correct the temperature measurement value, or adjust the compensation power to restore the linear relationship between the temperature gradient and the flow rate. Simultaneously, the matrix incorporates a continuous learning mechanism. After each deviation compensation, the system includes the deviation data, compensation effect, and parameter adjustment results in the historical database. It dynamically optimizes the parameter weights of each matrix element through a backpropagation algorithm. For example, if a stage experiences three consecutive instances of temperature gradient deviation exceeding the limit, the system will automatically adjust the temperature gradient correction coefficient for that stage from 0.8 to 1.1. The power compensation threshold has been optimized from 12W to 13W to adapt to long-term operating condition fluctuations such as sensor aging and environmental changes. This ensures that temperature-related errors are accurately compensated in real time under complex clinical environments, effectively eliminating flow rate deviations, maintaining system stability throughout its entire lifecycle, and guaranteeing the precise implementation of phased oxygen supply solutions.

[0102] Based on the real-time, phased instantaneous oxygen supply flow rate and the corresponding time length of the oxygen supply phase, a piecewise integral kernel function is constructed using the trapezoidal integral algorithm to calculate the cumulative oxygen supply amount for each oxygen supply phase. In this embodiment, the specific implementation of constructing the piecewise integral kernel function using the trapezoidal integral algorithm is as follows: taking the time length of the oxygen supply phase as the integration interval, the interval is discretized into several equidistant sub-intervals according to the sensor sampling frequency (not less than once per second), and the left and right endpoints of each sub-interval correspond to the sampled values ​​of the real-time, phased instantaneous oxygen supply flow rate, respectively; the integral kernel function is defined as the arithmetic square of the flow rate values ​​at the two endpoints of the sub-interval. The cumulative oxygen supply for a given oxygen supply phase is obtained by multiplying the mean by the duration of the sub-interval and summing the products of all sub-intervals. For example, if an oxygen supply phase lasts for 60 seconds, with a sampling frequency of 1 second / time, it is discretized into 60 sub-intervals. The instantaneous flow velocity sampling values ​​at the two ends of the i-th sub-interval are v(i) and v(i+1). The duration of the sub-interval is 1 second, and the corresponding kernel function term is [v(i)+v(i+1)] / 2×1. Summing 60 kernel function terms yields the cumulative oxygen supply for that phase. This method balances computational efficiency and measurement accuracy, and is suitable for the real-time cumulative requirements of phased oxygen supply.

[0103] The second standard loss is obtained by combining the cumulative oxygen supply of each oxygen supply stage with the standard cumulative oxygen supply of the corresponding stage.

[0104] The staged nonlinear correction algorithm is fine-tuned by constructing a weighted combined loss function based on the first oxygen supply metering loss, the staged temperature damage compensation matrix, and the second standard loss. When the weighted combined loss function satisfies the corresponding comprehensive loss threshold and the first oxygen supply metering loss is less than the first loss threshold under a preset training period, the trained staged standardized metering model is obtained. In this embodiment, the weighted combined loss function is constructed by weighted fusion of mean square error loss, smoothing L1 loss, and constraint penalty term. The mean square error loss takes the first oxygen supply metering loss and the second standard loss as inputs to accurately quantify the regression error between the real-time staged instantaneous oxygen supply flow rate and the standard staged instantaneous oxygen supply flow rate, and between the cumulative oxygen supply and the standard cumulative oxygen supply, ensuring the core accuracy of oxygen supply metering. The smoothing L1 loss takes the staged temperature damage compensation matrix as input to constrain the smoothness of the gradient change of the matrix and avoid the smoothness of the gradient change due to the first oxygen supply metering loss and the second standard loss. Sudden changes in temperature compensation parameters cause oscillations in oxygen supply control, thus improving the stability of the algorithm. The constraint penalty term is an L2 regularization penalty term that takes the parameters of the staged nonlinear correction algorithm as input. It is used to limit the range of model parameter values, preventing overfitting and ensuring that the oxygen supply flow rate and valve opening compensation values ​​output by the algorithm are within the medical safety threshold range. Finally, the three types of loss functions are weighted and summed by preset weights to obtain a weighted combined loss function. The weight ratios of mean squared error loss, smoothing L1 loss, and constraint penalty term are dynamically calibrated based on clinical oxygen supply measurement accuracy requirements, temperature compensation effect, and algorithm generalization ability, combined with historical training data. This weighted combined loss function is used to fine-tune the staged nonlinear correction algorithm, ensuring that the algorithm training process is quantifiable and executable, and that the output results can accurately match the actual needs of staged precise oxygen supply and measurement in medical scenarios.

[0105] When either the weighted combined loss function or the first oxygen supply metering loss fails to meet the corresponding loss threshold under the preset training period, an abnormal oxygen supply valve opening warning is issued, and the oxygen supply valve opening is readjusted N times. The temperature change gradient value and the corresponding compensation power value after the adjustment are collected to perform iterative adjustment and training on the phased nonlinear correction algorithm until the weighted combined loss function meets the corresponding comprehensive loss threshold and the first oxygen supply metering loss is less than the first loss threshold. At the same time, the compensation value of the oxygen supply valve opening for the corresponding oxygen supply stage is output.

[0106] The compensation value of the oxygen supply valve opening degree of the corresponding oxygen supply stage is mapped to the hierarchical mapping chain corresponding to the stage-by-stage hierarchical mapping chain to obtain the stage-by-stage hierarchical compensation mapping chain and perform opening degree early warning compensation.

[0107] This embodiment uses a pre-set weighted combined loss function and the first oxygen supply metering loss within a model training period as the system state evaluation benchmark. The weighted combined loss function integrates temperature, flow rate, and pressure deviations according to parameter influence weights, embedding the correlation between temperature change gradient and compensation power under the constraint of energy conservation principle, as well as the nonlinear mapping relationship between valve opening and flow rate. The first oxygen supply metering loss quantifies the absolute and relative differences between the actual flow rate and the standard flow rate. If any loss index exceeds the threshold, it is determined that the oxygen supply valve opening is abnormal and triggers a three-level early warning including on-site prompts, log recording, and remote notification. After the early warning, the system gradually adjusts according to the gradient. The valve opening is adjusted, and relevant data is synchronously collected and input into a staged nonlinear correction algorithm constructed using a PID control algorithm based on reinforcement learning. This algorithm divides the correction interval according to the severity of the disease, fits a three-dimensional relationship surface, and uses the reduction of weighted combined loss as a reward to train the model in a loop, dynamically compensating for deviations caused by sensor aging and environmental fluctuations. Simultaneously, a three-level hierarchical mapping chain is constructed: clinical demand layer - physical parameter layer - compensation execution layer. The clinical demand layer is based on ICD-10 encoded disease types and APACHE II scores, and determines the standard flow rate and theoretical valve opening value through a stepped oxygen supply flow rate mapping rule base. The physical... The parameter layer collects parameters such as temperature gradient, compensation power, and oxygen flow rate in real time and eliminates high-frequency noise through wavelet denoising. The compensation execution layer generates targeted compensation parameters through a reverse derivation mechanism. Multi-source data fusion verification is achieved by synchronously comparing measured values ​​from the flow sensor, pressure sensor feedback values ​​optimized by Kalman filtering, and predicted values ​​from a staged nonlinear correction algorithm. In case of anomalies, a backup sensor channel is activated to reduce the risk of single sensor failure. A dual-buffer architecture combined with cyclic redundancy verification ensures parameter integrity. Hardware-level channel switching enables parameter activation without delay to avoid the impact of sudden flow rate changes. A sliding window is introduced to accumulate new samples in conjunction with dynamic... The gradient descent method with a dynamically adjusted learning rate iteratively updates the compensation model weights, adaptively correcting systematic errors in long-term operation. Simultaneously, it calculates the coefficient of variation of the weight parameters over multiple consecutive update cycles for stability testing. When the threshold is exceeded, it automatically reverts to a stable version and pauses updates, restarting after troubleshooting. A conversion anomaly rollback mechanism is set up. When the flow rate deviation is abnormal after valve opening adjustment, it immediately switches to historical stable parameters and triggers an alarm. A daily self-check process is also set up to check the timeliness of sensor communication link response and the computational efficiency of data processing module, ultimately achieving high-precision and high-stability control of the oxygen supply subsystem under complex operating conditions.

[0108] It should be further explained that, in this embodiment, the detection results are fed back to the oxygen supply billing node to adjust the billing status and trigger an early warning, including:

[0109] The symptom characteristics of the current user are extracted from the electronic medical record subsystem, and a fuzzy comprehensive evaluation algorithm is used to obtain a symptom severity score. The specific implementation process of the fuzzy comprehensive evaluation algorithm in this embodiment is as follows: The symptom characteristics of the current user extracted from the electronic medical record subsystem are used as the core input variables. These symptom characteristics specifically include the specific type of the user's illness, clinical symptoms, blood oxygen saturation, respiratory rate, previous treatment response, and complications. A preset set of evaluation indicators and grading standards for the input variables are established, dividing each input variable into three evaluation levels: mild, moderate, and severe. Simultaneously, weight parameters for each input variable are set. For example, the symptom type has a weight of 0.3, and the clinical severity score is... The weights of clinical symptoms (0.25), blood oxygen saturation (0.15), respiratory rate (0.15), previous treatment response (0.1), and complications (0.05) are used to construct a fuzzy relation matrix. A preset membership function maps the actual values ​​of each input variable to the membership degree of the corresponding evaluation level. Combining the set weight parameters and the fuzzy relation matrix, a maximum-minimum synthesis method is used for comprehensive calculation. The final output variable is the severity score of the current user's condition, with a score range of 0-10. 0-3 points represent mild symptoms, 4-7 points represent moderate symptoms, and 8-10 points represent severe symptoms, achieving a precise quantitative evaluation of the user's condition severity.

[0110] Simultaneously, based on the current user's symptom characteristics and extracted basic user information, combined with information gain and mutual information algorithms, the corresponding user's access sensitivity is obtained. In specific implementation, symptom characteristics include clinical variables directly related to oxygen therapy, such as disease diagnosis type, current blood oxygen saturation, respiratory rate, clinical symptoms, and historical treatment records. The user's basic information includes personal attribute variables closely related to data security, such as patient age, allergy history, insurance type, special identity identifiers, and past privacy protection preferences. The specific implementation process of the information gain and mutual information algorithm in this embodiment is as follows: using the current user's symptom characteristics and basic user information as core input variables, and presetting three core parameters for the algorithm: feature threshold, information gain threshold, and mutual information threshold, where the feature threshold is set to 0.05, the information gain threshold is set to 0.1, and the mutual information threshold is set to 0.08. First, the information gain value and mutual information value of each input variable relative to the access sensitivity evaluation target are calculated. The information gain value is used to measure the contribution of a single input variable to the access sensitivity evaluation target, and the mutual information value is used to measure the degree of correlation between a single input variable and the access sensitivity evaluation target. Subsequently, a weighted fusion method is used to fuse the information gain and mutual information values ​​of each input variable. The information gain weight is set to 0.6 and the mutual information weight is set to 0.4. The comprehensive contribution value of each input variable is obtained by weighted summation. Then, effective input variables are selected by combining the preset feature thresholds and redundant variables with comprehensive contribution values ​​lower than the feature thresholds are removed. Finally, the comprehensive contribution values ​​of the effective input variables are accumulated and the output variable is the current user's access sensitivity score. The score range is set to 0-10 points, where 0-3 points are low sensitivity, 4-7 points are medium sensitivity, and 8-10 points are high sensitivity, so as to achieve accurate quantification of user access sensitivity.

[0111] Based on the user's access sensitivity combined with hierarchical permission mapping and access role access level, the current user's symptom characteristics and symptom severity score are transmitted to the oxygen supply end through the transmission synchronization node;

[0112] Based on the current user's symptom characteristics and symptom severity score, a depth-first search algorithm is used to retrieve the stepped oxygen supply flow rate mapping rules from the stepped oxygen supply flow rate mapping rule library. Specifically, the stepped oxygen supply flow rate mapping rule library is constructed as a tree-like retrieval structure with symptom type as the root node, symptom severity score as the child node, and stepped oxygen supply flow rate mapping rules as leaf nodes. The root node standardizes and classifies various symptom characteristics according to ICD-10 encoding; the child nodes are divided according to the severity score's grading range; and the leaf nodes store the complete stepped oxygen supply flow rate mapping rules for the corresponding symptom type and severity score grading, including the flow rate range and duration of each oxygen supply stage. The algorithm retrieves relevant parameters such as duration, oxygen concentration, and valve opening. Using the standardized symptom type corresponding to the current user's symptom characteristics as the root node and the grade interval corresponding to the symptom severity score as the child node, it initiates a depth-first search algorithm. Starting from the root node of the rule base tree structure, it traverses downwards along the branch corresponding to the current user's symptom type to the child node corresponding to the symptom severity score grade, and then traverses from that child node to the corresponding leaf node. It extracts the stepped oxygen supply flow rate mapping rules stored in the leaf node, terminates after the retrieval is complete, and outputs the rule. This achieves accurate and rapid retrieval of the corresponding user's stepped oxygen supply flow rate mapping rules from the stepped oxygen supply flow rate mapping rule base.

[0113] The system responds to the stepped oxygen supply flow rate mapping rule and simultaneously obtains the cumulative oxygen supply corresponding to each stage through a segmented standardized metering model.

[0114] Based on the cumulative oxygen supply corresponding to each stage and the segmented billing rules, the cumulative billing result and the total billing result for each stage are obtained.

[0115] While responding to the stepped oxygen supply flow rate mapping rule, the system monitors the output oxygen concentration value in real time through a configured oxygen concentration measurement sensor, and combines it with a radial basis function to fit an oxygen concentration change prediction function. For example, in this embodiment, the oxygen concentration time series data collected by the oxygen concentration measurement sensor at a sampling frequency of 1 time / second is used as input data. A Gaussian radial basis function is selected, and in the fitting stage, full-scene oxygen concentration sample data covering the normal range, fluctuation range and the range near the abnormal threshold of oxygen concentration within the past 30 minutes are selected for training and fitting to obtain the oxygen concentration change prediction function. This function can output the continuous predicted value of oxygen concentration within the next 5 minutes. When the predicted value approaches the preset abnormal oxygen concentration threshold, the system automatically triggers the oxygen supply quality early warning mechanism in advance to intervene and adjust in a timely manner to ensure the oxygen supply quality and the patient's oxygen use safety.

[0116] Based on real-time monitoring of oxygen concentration values, combined with an oxygen concentration change prediction function and a preset oxygen concentration anomaly threshold, the time points corresponding to the oxygen concentration anomalies during the oxygen supply process are obtained.

[0117] The time point of the abnormal oxygen concentration is fed back to the oxygen supply billing node, and a pre-oxygen supply metering stop command is generated. When the oxygen concentration value is less than the abnormal oxygen concentration threshold, the cumulative oxygen flow metering and cumulative billing metering are stopped through the pre-oxygen supply metering stop command, and the abnormal result is fed back to the oxygen metering display interface.

[0118] Based on the current user’s disease type and disease severity score, combined with the disease type-reimbursement ratio rules preset by the reimbursement node, the cumulative reimbursement ratio for the current user is obtained.

[0119] Based on the cumulative reimbursement ratio corresponding to the current user and the cumulative oxygen supply measured by the segmented standardized metering model, the corresponding cumulative billing is reimbursed and allocated. The allocation result is transmitted to the information storage node corresponding to the current user in the electronic medical record subsystem through the transmission synchronization node and displayed in real time.

[0120] This embodiment achieves several technological breakthroughs in the field of oxygen measurement and billing by constructing an intelligent node collaborative oxygen supply control system. The system adopts a phased and hierarchical mapping chain architecture, which transforms the disease type and dynamically updated disease severity score into a multi-stage oxygen supply scheme through a stepped oxygen supply flow rate mapping rule base. Each stage sets independent flow rate targets, durations, and concentration requirements, achieving accurate mapping from patient clinical characteristics to oxygen supply parameters. In the oxygen measurement stage, a segmented standardized measurement model is established. In the ultra-low flow rate stage, a dual-path fusion mechanism of porous medium laminar flow differential pressure detection and thermal diffusion measurement is used, combined with a support vector regression model based on laminar flow characteristics to achieve accurate capture of minute flow rates. In the conventional flow rate stage, a random forest regression model containing multiple decision trees is used to perform multi-sensor data fusion on the temperature change gradient and compensation power value after three-level filtering, effectively overcoming the measurement bias of traditional single sensors in complex clinical environments. To address the issue of equipment performance degradation during long-term operation, the system introduces a phased temperature damage compensation matrix. Using a backpropagation algorithm, it calculates the impact weight of temperature gradient deviation on flow rate measurement in real time, forming a dynamic compensation mechanism with self-learning capabilities, ensuring stable measurement accuracy throughout the entire lifecycle. In terms of billing management, the system deeply binds the segmented integral kernel function to the clinical oxygen supply stages. Based on a trapezoidal integral algorithm, it continuously accumulates the instantaneous flow rate for each oxygen supply stage, forming oxygen supply data that perfectly corresponds to the treatment stage. The billing node converts the accumulated oxygen supply for each stage into corresponding fees through preset segmented billing rules, achieving a precise billing model based on stage measurement and usage. Furthermore, a real-time response mechanism linking oxygen supply quality and billing is established: when the oxygen supply quality detection node identifies an abnormal concentration through the oxygen concentration measurement sensor and radial basis function, it immediately sends a timestamped abnormal signal to the oxygen supply billing node, triggering a pre-emptive oxygen supply metering stop command, ensuring patient safety. While ensuring the safety of patients using oxygen, the system automatically suspends the accumulation of costs for that period, effectively solving the problem of disconnect between quality anomalies and billing adjustments in traditional systems. Furthermore, this embodiment uses a blockchain recording node to store all measurement data and billing records, ensuring data immutability and traceability. A transmission synchronization node enables bidirectional data synchronization between the oxygen supply unit and the electronic medical record system, guaranteeing data consistency between clinical decision-making and cost calculation. The reimbursement node automatically calculates medical insurance ratios and allocates costs, significantly improving medical settlement efficiency. The synergistic effect of these technical features enables the system to maintain the accuracy of oxygen therapy while achieving real-time transparency of measurement data, automated and standardized billing processes, and end-to-end quality supervision. This provides medical institutions with a complete solution integrating precision treatment, intelligent billing, and quality assurance, demonstrating significant technical advantages and application value in improving the quality of medical services, optimizing resource allocation, and enhancing doctor-patient trust.

[0121] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments under the guidance of the present invention without departing from the spirit and scope of the claims. All of these variations are within the protection scope of the present invention.

[0122] If the technical solution disclosed herein involves personal information, the product using this technical solution has clearly informed the user of the personal information processing rules and obtained the user's voluntary consent before processing the personal information. If the technical solution disclosed herein involves sensitive personal information, the product using this technical solution has obtained the user's separate consent before processing the sensitive personal information, and also meets the requirement of "express consent". For example, at personal information collection devices such as cameras, clear and prominent signs are set up to inform users that they have entered the scope of personal information collection and that personal information will be collected. If an individual voluntarily enters the collection scope, it is deemed that they have agreed to the collection of their personal information; or on the personal information processing device, with clear signs / information informing users of the personal information processing rules, authorization is obtained from the individual through pop-up information or by asking the individual to upload their personal information; wherein, the personal information processing rules may include information such as the personal information processor, the purpose of personal information processing, the processing method, and the types of personal information processed.

Claims

1. A medical oxygen flow metering and control system, characterized in that, include: A block monitoring node chain, comprising: A safety node is used to perform safety checks based on the pressure change curve of the real-time monitored oxygen supply subsystem or oxygen cylinder and a preset safety pressure range. When the check result meets the safety pressure range, it responds to the oxygen supply setting node. The oxygen supply setting node obtains the current user's staged oxygen supply flow rate mapping rule based on the input user's symptom characteristics and a preset stepped oxygen supply flow rate mapping rule library, and simultaneously responds to the proportional setting node. The proportional setting node generates staged oxygen supply valve opening commands based on the current user's staged oxygen supply flow rate mapping rule and a reinforcement learning-optimized PID control algorithm, and simultaneously responds to the oxygen supply quality detection node and the oxygen supply... The billing node is used to obtain the segmented integral oxygen supply based on the current user's segmented oxygen supply flow rate mapping rules and a segmented standardized metering model. It then performs real-time oxygen supply billing based on the segmented integral oxygen supply and a preset segmented billing rule, while simultaneously feeding the billing results back to the display interface and the block record node for on-chain storage. The oxygen supply quality detection node is used to detect the oxygen concentration and oxygen supply flow rate in real-time during the oxygen supply process, and performs oxygen supply detection by combining exponential moving average and a preset anomaly detection threshold. The detection results are then fed back to the oxygen supply billing node to adjust the billing status and trigger an early warning. The preset anomaly detection thresholds include oxygen concentration anomaly thresholds; The block monitoring node chain also includes a transmission synchronization node; the response oxygen supply setting node includes a symptom assessment module and a mapping module. The symptom assessment module is used to obtain a severity score for the current type of symptom of the corresponding user based on the user's symptom characteristics and symptom type information synchronized from the electronic medical record subsystem by the transmission synchronization node, combined with a fuzzy comprehensive evaluation algorithm. The mapping module is used to generate a phased oxygen supply flow rate mapping rule for the current user based on the user's current symptom severity score, combined with the symptom type-severity-phased oxygen supply flow rate mapping rule in the phased oxygen supply flow rate mapping rule library and the duration of each phase, and combined with preset matching mapping rules. The phased oxygen supply flow rate mapping rule for the current user includes the valve opening size corresponding to each oxygen supply subsystem or oxygen cylinder output port, the oxygen flow rate and concentration corresponding to each phase, and the duration corresponding to the valve opening.

2. The medical oxygen flow metering and control system as described in claim 1, characterized in that, The construction and training process of the segmented standardized econometric model includes: Based on symptom type, symptom severity score, filtered temperature change gradient value and corresponding compensation power value, constant temperature difference, standard staged instantaneous oxygen supply flow rate, and combined with the Apriori algorithm, a staged hierarchical mapping chain is established. The temperature change gradient value and corresponding compensation power value of the heating element at different oxygen supply stages are collected in real time by the configured heating element and temperature sensor. The data is filtered by the filtering algorithm to obtain the filtered temperature change gradient value and corresponding compensation power value. The phased hierarchical mapping chain, the filtered temperature change gradient value, and the compensation power value are used as training parameters and input into the preset phased nonlinear correction algorithm to calculate the real-time phased instantaneous oxygen supply flow rate corresponding to the current phase.

3. The medical oxygen flow metering and control system as described in claim 2, characterized in that, The construction and training process of the segmented standardized econometric model also includes: Based on the real-time phased instantaneous oxygen supply flow rate corresponding to the current stage and the standard phased instantaneous oxygen supply flow rate of the corresponding stage, the deviation of the standard oxygen supply flow rate is obtained, and the first oxygen supply metering loss is constructed. Based on the standard oxygen supply flow rate deviation, the standard temperature gradient deviation value is obtained through the backpropagation algorithm along the staged hierarchical mapping chain, and a staged temperature damage compensation matrix is ​​constructed. Based on the real-time phased instantaneous oxygen supply flow rate and the corresponding oxygen supply phase duration, a piecewise integral kernel function is constructed using a trapezoidal integral algorithm to calculate the cumulative oxygen supply amount for each oxygen supply phase. The second standard loss is obtained by combining the cumulative oxygen supply of each oxygen supply stage with the standard cumulative oxygen supply of the corresponding stage.

4. A medical oxygen flow metering and control system as described in claim 3, characterized in that, The construction and training process of the segmented standardized econometric model also includes: The staged nonlinear correction algorithm is fine-tuned by constructing a weighted combined loss function based on the first oxygen supply metering loss, the staged temperature damage compensation matrix and the second standard loss; when the weighted combined loss function meets the corresponding comprehensive loss threshold and the first oxygen supply metering loss is less than the first loss threshold under the preset training period, the trained segmented standardized metering model is obtained. When either the weighted combined loss function or the first oxygen supply metering loss fails to meet the corresponding loss threshold under the preset training period, an abnormal oxygen supply valve opening warning is issued, and the oxygen supply valve opening is readjusted N times. The temperature change gradient value and the corresponding compensation power value after the adjustment are collected to perform iterative adjustment and training on the phased nonlinear correction algorithm until the weighted combined loss function meets the corresponding comprehensive loss threshold and the first oxygen supply metering loss is less than the first loss threshold. At the same time, the compensation value of the oxygen supply valve opening for the corresponding oxygen supply stage is output. The compensation value of the oxygen supply valve opening degree of the corresponding oxygen supply stage is mapped to the hierarchical mapping chain corresponding to the stage-by-stage hierarchical mapping chain to obtain the stage-by-stage hierarchical compensation mapping chain and perform opening degree early warning compensation.

5. A medical oxygen flow metering and control system as described in claim 4, characterized in that, The stepped oxygen supply flow rate mapping rule base is constructed by combining a hash algorithm and a tree database, based on the disease type, the severity score corresponding to each disease type, the number of oxygen supply stages corresponding to the corresponding severity score, the interval duration and average oxygen supply flow rate corresponding to each oxygen supply stage, and is used to match the stepped oxygen supply flow rate mapping rules corresponding to the severity score of the disease type; the segmented billing rule is constructed by the average oxygen supply flow rate of the corresponding stage and the billing unit price corresponding to the average oxygen supply flow rate.

6. A medical oxygen flow metering and control system as described in claim 5, characterized in that, The detection results are fed back to the oxygen supply billing node to adjust the billing status and trigger an early warning, including: The symptoms of the current user are extracted from the electronic medical record subsystem, and the severity score of the symptoms is obtained by combining the fuzzy comprehensive evaluation algorithm. Simultaneously, based on the current user's symptom characteristics and extracted basic user information, combined with information gain and mutual information algorithms, the corresponding user's access sensitivity is obtained; Based on the user's access sensitivity, combined with hierarchical permission mapping and access role access level, the current user's symptom characteristics and symptom severity score are transmitted to the oxygen supply end through the transmission synchronization node.

7. A medical oxygen flow metering and control system as described in claim 6, characterized in that, The process of feeding back the test results to the oxygen supply billing node to adjust the billing status and trigger an early warning also includes: Based on the current user's symptom characteristics and symptom severity score, the step-by-step oxygen supply flow rate mapping rules are obtained from the step-by-step oxygen supply flow rate mapping rule library using a depth-first search algorithm. The system responds to the stepped oxygen supply flow rate mapping rule and simultaneously obtains the cumulative oxygen supply corresponding to each stage through a segmented standardized metering model. Based on the cumulative oxygen supply for each stage and the segmented billing rules, the cumulative billing result and total billing result for each stage are obtained.

8. A medical oxygen flow metering and control system as described in claim 7, characterized in that, The process of feeding back the test results to the oxygen supply billing node to adjust the billing status and trigger an early warning also includes: While responding to the step-type oxygen supply flow rate mapping rule, the oxygen concentration value is monitored in real time by the configured oxygen concentration measurement sensor, and the oxygen concentration change prediction function is fitted by combining the radial basis kernel function. Based on real-time monitoring of oxygen concentration values, combined with an oxygen concentration change prediction function and a preset oxygen concentration anomaly threshold, the time points corresponding to the oxygen concentration anomalies during the oxygen supply process are obtained. The time point of the abnormal oxygen concentration is fed back to the oxygen supply billing node, generating a pre-oxygen supply metering stop command. When the oxygen concentration value is less than the abnormal oxygen concentration threshold, the cumulative oxygen flow metering and cumulative billing metering are stopped through the pre-oxygen supply metering stop command, and the abnormal result is fed back to the oxygen metering display interface.

9. A medical oxygen flow metering and control system as described in claim 8, characterized in that, The process of feeding back the test results to the oxygen supply billing node to adjust the billing status and trigger an early warning also includes: Based on the current user’s disease type and disease severity score, combined with the disease type-reimbursement ratio rules preset by the reimbursement node, the cumulative reimbursement ratio for the current user is obtained. Based on the cumulative reimbursement ratio corresponding to the current user and the cumulative oxygen supply measured by the segmented standardized measurement model, the corresponding cumulative billing is reimbursed and allocated. The allocation result is transmitted to the information storage node corresponding to the current user in the electronic medical record subsystem through the transmission synchronization node and displayed in real time.

Images