Oxygen supply pressure dynamic adjusting method and system based on medical cabin

Abstract

The invention discloses an oxygen supply pressure dynamic adjusting method and system based on a medical cabin, and relates to the technical field of medical equipment and medical information.The method comprises the steps that the real-time oxygen supply pressure of the medical cabin and physiological data of a patient are collected; predicting pressure disturbance of an oxygen supply network through a graph neural network, and identifying a risk medical cabin; monitoring oxygen pressure mutation at an inlet of an oxygen supply pipeline, and identifying an interfered medical cabin; based on the real-time physiological data of the patient, generating a dynamic physiological baseline through parameter empirical Bayesian calibration; calculating a comprehensive deviation index of the physiological data and the baseline through sliding window integration, and determining a compensation control vector; generating a compensation instruction set by combining the abrupt change of oxygen pressure and the compensation control vector; the executed physiological parameters and oxygen supply pressure data are fed back to the medical health big data platform, and closed-loop adjustment is achieved. The problems that an existing medical cabin oxygen supply system is insufficient in accuracy, weak in anti-interference capacity, difficult in personalized adjustment and the like due to control lag, passive response and static reference setting are effectively solved.

Description

Technical Field

[0001] This invention relates to the field of medical equipment and medical information technology, specifically to a method and system for dynamic adjustment of oxygen supply pressure based on a medical cabin. Background Technology

[0002] In hyperbaric oxygen therapy or intensive care, the stability of oxygen supply pressure in the medical chamber is crucial to patient treatment outcomes and safety. Currently, most medical chamber oxygen supply systems employ a feedback control mechanism based on fixed thresholds. This means that adjustment is only triggered after the chamber pressure or a patient's physiological parameter deviates from a set range. This approach cannot address the propagation of pressure disturbances caused by sudden changes in oxygen consumption in other chambers, leaving patients in affected chambers at risk of oxygen pressure fluctuations. Furthermore, existing technologies lack the ability to perceive the overall topology of the oxygen supply network, making it impossible to predict the source and propagation path of disturbances in advance. This results in the adjustment behavior remaining in a passive "follow-up" state, hindering the achievement of proactive and precise control.

[0003] On the other hand, existing systems, when setting baseline physiological parameters for patients, typically rely on population averages or patients' own static historical data, failing to fully consider the dynamic changes in physiological parameters caused by different states such as sleep, activity, and emotions. They cannot accurately distinguish between changes in a patient's true physiological needs and abnormalities caused by external oxygen supply interference, potentially leading to miscompensation or undercompensation. Furthermore, traditional control strategies often rely on the instantaneous deviation of a single parameter, lacking an assessment of the comprehensive trends of multiple physiological parameters over time. The resulting compensation instructions are not precise or robust enough, hindering the realization of personalized, adaptive regulation. Summary of the Invention

[0004] (a) Technical problems to be solved To address the shortcomings of existing technologies, this invention provides a method and system for dynamic adjustment of oxygen supply pressure in medical cabins. This invention uses a graph neural network algorithm to predict pressure disturbances in the oxygen supply network, proactively identifying high-risk medical cabins. By combining a dynamic physiological baseline calibrated based on parameter experience Bayesian and a comprehensive deviation index obtained through sliding window integration, a precise compensation control vector is generated. This vector provides rapid and coordinated dynamic compensation for the oxygen supply pressure of disturbed medical cabins. This effectively overcomes the insufficient adjustment precision caused by control lag and rigid benchmarks in existing technologies, achieving a shift from passive response to active intervention and from group benchmarks to personalized adaptation. This ensures the stability of oxygen supply pressure and patient treatment safety, solving the technical problems described in the background section.

[0005] (II) Technical Solution To achieve the above objectives, the present invention provides a method for dynamically adjusting oxygen supply pressure based on a medical cabin, comprising the following steps: S1: Obtain real-time oxygen supply pressure of the medical cabin, physiological parameter benchmarks of patients of the same age and with the same disease, and real-time physiological data of patients in the cabin from the medical and health big data sharing platform; S2: Based on the graph neural network algorithm, the real-time oxygen supply pressure of the entire medical cabin oxygen supply network is analyzed by data stream, and the pressure disturbance that is about to occur in the network is predicted hundreds of milliseconds to several seconds in advance. The medical cabins where the pressure disturbance is about to occur are identified as high-risk medical cabins. S3: Monitor the oxygen pressure at the inlet of the oxygen supply pipe of other medical cabins that are in the same oxygen supply network as the risk medical cabin. If the oxygen pressure at the inlet of the oxygen supply pipe of a certain medical cabin suddenly changes beyond the preset inlet disturbance threshold, the medical cabin is identified as the disturbed medical cabin. S4: Based on the real-time physiological data of the patient, perform parameter empirical Bayes calibration on the physiological parameter benchmark of the patients of the same age and the same disease to obtain a dynamic physiological baseline that adaptively adjusts with the patient's sleep, activity, mood and other states; S5: Perform sliding window integration based on the patient's physiological data and the dynamic physiological baseline to obtain the comprehensive deviation index between the two, and determine the patient's compensation control vector based on the comprehensive deviation index; S6: Based on the sudden change in oxygen pressure at the inlet of the oxygen supply pipeline and the compensation control vector, adjust the oxygen supply pressure compensation command of the disturbed medical oxygen chamber to generate a compensation command set for the disturbed medical oxygen chamber. S7: Feed back the patient's physiological parameters and medical cabin oxygen supply pressure response data after the execution of the compensation instruction set to the medical and health big data sharing platform.

[0006] Furthermore, real-time oxygen supply pressure in the medical cabin, baseline physiological parameters of patients of the same age and with the same condition, and real-time physiological data of patients inside the cabin are obtained from the medical and health big data sharing platform, including: The real-time oxygen supply pressure of the medical cabin is obtained from the oxygen supply pressure sensor of the medical cabin, which is uploaded from the medical and health big data sharing platform. Extract the age and medical record information of the patients in the cabin from the electronic medical record database of the medical and health big data sharing platform; Based on the patient's age and medical record information, physiological parameter benchmarks for patients with similar diseases and similar disease severity that are within 5 years of each other in the historical case database of the medical and health big data sharing platform are retrieved. If no similar patients matching the criteria are found, the patient's physiological parameter benchmarks, constructed based on their own historical physiological data, are obtained from the medical and health big data sharing platform. The patient's real-time physiological data is obtained by acquiring physiological signals from a multi-parameter vital signs monitor deployed in the target medical cabin through the medical and health big data sharing platform.

[0007] Furthermore, the graph neural network algorithm is used to perform data stream analysis on the real-time oxygen supply pressure of the entire medical cabin oxygen supply network, predicting pressure disturbances that will occur in the network hundreds of milliseconds to several seconds in advance, and identifying medical cabins with impending pressure disturbances as high-risk medical cabins, including: The topology of the oxygen supply network is analyzed to obtain the node connection relationship of the entire oxygen supply network; Based on the node connection relationship, each medical cabin is abstracted as a graph node, and the oxygen supply pipes connecting the medical cabins are abstracted as edges, thus constructing a directed graph of the oxygen supply network with medical cabins as nodes and oxygen supply pipes as edges. The real-time oxygen supply pressure is input as a node feature into the directed graph of the oxygen supply network. Based on the directed graph of the oxygen supply network, data flow analysis is performed to predict the pressure prediction value of the downstream medical cabin node in the next hundreds of milliseconds to several seconds. Based on the flow direction of oxygen in the oxygen supply network and the propagation process of pressure disturbance in the oxygen supply network, forward reasoning is performed on the pressure of the downstream medical cabin node to obtain the predicted pressure value of the downstream medical cabin node in the next hundreds of milliseconds to several seconds. The predicted pressure value is compared with a preset pressure threshold. If the value exceeds the threshold, the downstream medical cabin is determined to be a high-risk medical cabin.

[0008] Furthermore, the step of performing topology analysis on the entire oxygen supply network to obtain the node connection relationships of the entire oxygen supply network includes: The location of the medical cabin nodes in the oxygen supply network is obtained based on the medical and health big data sharing platform, and the distribution of medical cabin nodes in the entire medical cabin oxygen supply network is obtained. The node connection relationship of the oxygen supply network is constructed based on the node distribution of the medical cabin and the connection relationship of the oxygen supply pipeline in the oxygen supply network.

[0009] Furthermore, the monitoring of the oxygen pressure at the inlet of the oxygen supply pipes of other medical cabins located in the same oxygen supply network as the risk medical cabin, and the identification of the medical cabin as an affected medical cabin if the sudden change in the inlet oxygen pressure of a certain medical cabin exceeds a preset inlet disturbance threshold, includes: The oxygen pressure range within a unit time window is calculated for the inlet oxygen pressure to obtain the transient fluctuation amplitude of the inlet oxygen pressure of the medical cabin. The transient fluctuation amplitude is compared with a preset inlet disturbance threshold to identify nodes where the oxygen pressure transient fluctuation amplitude exceeds the inlet disturbance threshold; The identified nodes are marked as oxygen pressure mutation events, and the nodes are identified as the interfered medical cabin.

[0010] Furthermore, the step of performing empirical Bayesian calibration on the physiological parameter benchmarks of patients of the same age and with the same condition based on the patient's real-time physiological data to obtain a dynamic physiological baseline that adaptively adjusts with the patient's sleep, activity, mood, and other states includes: The patient's activity status is detected using pressure sensors and millimeter-wave radar in the medical cabin. Based on the historical physiological parameter data of each physiological state under the qualified physiological parameter benchmark, Gaussian distribution fitting is performed to obtain the prior probability distribution of each physiological state of patients of the same age and the same disease. Based on the physiological parameters of the patient under the current physiological state, a Bayesian update is performed on the prior probability distribution that matches the current physiological state to obtain the posterior distribution of the patient's physiological parameters under the current physiological state. The mean of the posterior distribution of the physiological parameters is calculated to obtain the personalized mean parameter of the posterior distribution; The standard deviation of the posterior distribution of the physiological parameters is calculated to obtain the personalized standard deviation parameter of the posterior distribution; Using the personalized mean parameter as the central benchmark, and the personalized standard deviation parameter and a preset confidence coefficient as a measure of fluctuation range, the statistical upper and lower limits of the personalized confidence interval are calculated to obtain the personalized confidence interval of the patient in the current physiological activity state. The personalized confidence interval is defined as the dynamic physiological baseline.

[0011] Furthermore, the step of performing sliding window integration based on the patient's physiological data and the dynamic physiological baseline to obtain a comprehensive deviation index, and determining the patient's compensation control vector based on the comprehensive deviation index, includes: Motion artifact removal was performed on the time-series data of prefrontal cortex blood oxygen saturation in the real-time physiological data to obtain a standard prefrontal cortex blood oxygen saturation value sequence. Using the sampling time of the standard prefrontal cortex blood oxygen saturation value sequence as a reference, the baseline data in the dynamic physiological baseline under the same physiological state as the current acquisition time are time-aligned and resampled to obtain the prefrontal cortex blood oxygen saturation baseline sequence corresponding to the standard value sequence at the time point. The absolute deviation of the blood oxygen saturation value sequence in the standard prefrontal cortex is calculated at each time point from the baseline blood oxygen saturation value sequence in the prefrontal cortex to generate the absolute deviation sequence of blood oxygen saturation. The same operation was performed on the remaining data in the physiological baseline to obtain the absolute deviation sequence of all data in the physiological baseline. The cumulative absolute deviation is obtained by performing an exponential decay time integral operation within a fixed-duration sliding window on the absolute deviation sequence of a single data point, and the average of the cumulative absolute deviations of all data points is used as the comprehensive deviation index. The comprehensive deviation index is input into the oxygen supply compensation model in the medical and health big data sharing platform, and the oxygen supply compensation control vector of the patient is output.

[0012] Furthermore, the adjustment of the oxygen supply pressure compensation command for the disturbed medical oxygen chamber based on the sudden change in oxygen pressure at the inlet of the oxygen supply pipeline and the compensation control vector, generating a compensation command set for the disturbed medical oxygen chamber, includes: Based on the transient fluctuation amplitude of the sudden change in inlet oxygen pressure, the compensation control vector is dynamically gain-corrected to obtain the corrected compensation control vector of the disturbed medical oxygen chamber. Based on the modified compensation control vector, the oxygen supply pressure command of the interfered medical cabin is adjusted to generate a compensation command set for the interfered medical cabin.

[0013] Furthermore, the step of feeding back the patient's physiological parameters and the medical cabin oxygen supply pressure response data after the execution of the compensation instruction set to the medical and health big data sharing platform includes: Collect real-time physiological parameters of the patient and oxygen supply pressure response data of the medical cabin after the compensation command is executed; The patient's real-time physiological parameters and the oxygen supply pressure response data of the medical cabin are integrated into a data structure to generate a feedback data package containing information on the correlation between the patient's physiological state and oxygen supply pressure. The feedback data packet is sent to the medical and health big data sharing platform in a standardized format by calling the real-time data transmission interface provided by the medical and health big data sharing platform.

[0014] A dynamic oxygen supply pressure regulation system based on a medical cabin includes: Data acquisition module: used to acquire real-time oxygen supply pressure of the medical cabin, physiological parameter benchmarks of patients of the same age and with the same disease, and real-time physiological data of patients in the cabin from the medical and health big data sharing platform; Risk Disturbance Module: Used to perform data stream analysis on the real-time oxygen supply pressure of the entire medical cabin oxygen supply network based on graph neural network algorithm, predict the pressure disturbance that will occur in the network hundreds of milliseconds to several seconds in advance, and identify the medical cabins where the pressure disturbance is about to occur as risk medical cabins; Interference identification module: used to monitor the oxygen pressure at the inlet of the oxygen supply pipe of other medical cabins that are in the same oxygen supply network as the risk medical cabin. If the oxygen pressure at the inlet of the oxygen supply pipe of a certain medical cabin suddenly changes beyond the preset inlet disturbance threshold, the medical cabin is identified as the interfered medical cabin. Baseline calibration module: used to perform empirical Bayesian calibration of physiological parameters of patients of the same age and with the same disease based on the real-time physiological data of the patient, so as to obtain a dynamic physiological baseline that adaptively adjusts with the patient's sleep, activity, mood and other states; Compensation control module: used to perform sliding window integration based on the patient's physiological data and the dynamic physiological baseline to obtain a comprehensive deviation index between the two, and to determine the patient's compensation control vector based on the comprehensive deviation index; Command adjustment module: used to adjust the oxygen supply pressure compensation command of the disturbed medical oxygen chamber based on the sudden change in oxygen pressure at the inlet of the oxygen supply pipeline and the compensation control vector, and generate a compensation command set for the disturbed medical oxygen chamber; Data feedback module: used to feed back the patient's physiological parameters and medical cabin oxygen supply pressure response data after the execution of the compensation instruction set to the medical and health big data sharing platform.

[0015] (III) Beneficial Effects This invention provides a method and system for dynamic adjustment of oxygen supply pressure based on a medical cabin, which has the following beneficial effects: 1. By using graph neural network algorithms to perform topology modeling and data flow analysis on the oxygen supply network of medical wards, it is possible to predict pressure disturbances that will occur in the network hundreds of milliseconds to several seconds in advance, identify high-risk and disturbed medical wards, and achieve proactive early warning and intervention for oxygen supply pressure fluctuations. Combined with dynamic gain correction of the compensation control vector based on the transient fluctuation amplitude of inlet oxygen pressure mutations, a precise compensation instruction set is generated, which effectively overcomes the lag problem of traditional feedback control, significantly improves the response speed and control accuracy of the oxygen supply system, and ensures the stability and safety of oxygen supply for patients during treatment.

[0016] 2. Employing a parametric empirical Bayesian calibration method, and combining real-time physiological data of patients with physiological parameter benchmarks of patients of the same age and with the same condition, a dynamic physiological baseline is constructed that adaptively adjusts according to the patient's sleep, activity, and emotional state, realizing the transformation of physiological benchmarks from group-based to individualized. Furthermore, by calculating the comprehensive deviation index between multiple physiological parameters and the dynamic physiological baseline through sliding window integration, a compensating control vector is generated. The system can more accurately distinguish between changes in the patient's actual physiological needs and external oxygen supply interference, thereby making more reasonable and robust oxygen supply adjustment decisions and improving the adaptability and safety of treatment. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the process of a dynamic adjustment method for oxygen supply pressure based on a medical cabin according to the present invention. Figure 2 This is a schematic diagram of the structure of a dynamic oxygen supply pressure regulation system based on a medical cabin according to the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Please see Figure 1 This invention provides a method for dynamically adjusting oxygen supply pressure based on a medical cabin, comprising the following steps: S1: Obtain real-time oxygen supply pressure of the medical cabin, physiological parameter benchmarks of patients of the same age and with the same disease, and real-time physiological data of patients in the cabin from the medical and health big data sharing platform.

[0020] In this embodiment, obtaining real-time oxygen supply pressure of the medical cabin, physiological parameter benchmarks of patients of the same age and with the same condition, and real-time physiological data of patients inside the cabin from the medical and health big data sharing platform includes: The real-time oxygen supply pressure of the medical cabin is obtained from the oxygen supply pressure sensor of the medical cabin, which is uploaded from the medical and health big data sharing platform. Extract the age and medical record information of the patients in the cabin from the electronic medical record database of the medical and health big data sharing platform; Based on the patient's age and medical record information, physiological parameter benchmarks for patients with similar diseases and similar disease severity that are within 5 years of each other in the historical case database of the medical and health big data sharing platform are retrieved. If no similar patients matching the criteria are found, the patient's physiological parameter benchmarks, constructed based on their own historical physiological data, are obtained from the medical and health big data sharing platform. The patient's real-time physiological data is obtained by acquiring physiological signals from a multi-parameter vital signs monitor deployed in the target medical cabin through the medical and health big data sharing platform.

[0021] Preferably, this step involves systematically acquiring three types of key data from a medical and health big data sharing platform: real-time oxygen supply pressure in the medical cabin, physiological parameter benchmarks for patients of the same age and with the same condition, and real-time physiological data of patients inside the cabin. Real-time oxygen supply pressure refers to the oxygen pressure value continuously measured and uploaded by pressure sensors installed on the oxygen supply pipeline of the medical cabin; it directly reflects the immediate working status of the oxygen supply system inside the medical cabin.

[0022] The baseline physiological parameters for patients of the same age and with the same condition are reference standards derived from statistical analysis of historical case data, used to assess the normal range of the current patient's physiological state. Real-time physiological data of patients within the medical cabin are continuously collected through monitoring equipment integrated within the cabin, reflecting changes in the patient's current physiological state.

[0023] Specifically, the real-time oxygen supply pressure of the medical cabin is obtained from the oxygen supply pressure sensor of the medical cabin, which is uploaded by the sensor. The oxygen supply pressure sensor is a precision measuring device specifically installed in the oxygen supply system of the medical cabin, capable of continuously monitoring pressure changes in the oxygen delivery pipeline at a high sampling frequency.

[0024] Real-time oxygen supply pressure is a sequence of pressure readings collected by sensors and transmitted to the platform in real time via a data interface. This data includes precise timestamps, accurately reflecting pressure fluctuations within the oxygen supply system. The acquisition process involves calling the platform's device data interface, subscribing to and receiving the corresponding pressure data stream according to the medical pod's unique identifier, ensuring timely detection of any abnormal pressure changes at any point in the oxygen supply network.

[0025] More specifically, the age and medical record information of patients in the cabin are extracted from the electronic medical record database of the aforementioned medical and health big data sharing platform. The electronic medical record database is a centralized database that stores complete medical records of patients, including structured data such as basic personal information, diagnostic history, and treatment records. Age refers to the patient's actual physiological age, accurately recorded in one-year increments.

[0026] The medical record information is a comprehensive dataset, including disease diagnosis codes, disease names, medical history summaries, descriptions of clinical symptoms, and physician-assessed disease severity levels. The extraction operation involves performing a precise matching query in the electronic medical record database using the patient's identification code to retrieve and return the patient's latest age information and complete medical record, providing accurate filtering criteria for subsequent benchmark data retrieval.

[0027] Based on the patient's age and medical record information, physiological parameter benchmarks are retrieved from the historical case database of the medical and health big data sharing platform for patients with the same type of disease and similar disease severity whose age difference is no more than 5 years.

[0028] The historical case database is a data repository that has accumulated a large number of past treatment cases, storing complete treatment records and physiological monitoring data of discharged patients. An age difference of no more than 5 years means that the search scope is limited to the current patient's age plus or minus 5 years, ensuring that the reference groups are comparable in basic physiological function. For the same type of disease, matching with the same or highly similar disease diagnostic classification is required; for similar disease severity, matching is performed using standardized clinical assessment scales.

[0029] Physiological parameter benchmarks are representative ranges of physiological indicators derived from statistical analysis of successfully matched case populations, including key parameters such as baseline values ​​for blood oxygen saturation, heart rate, and respiratory rate. The retrieval process involves performing a multi-condition joint query in the historical case database. First, the database is filtered by disease type and severity. Then, age ranges are precisely matched. Finally, statistical analysis is performed on the physiological data of all cases that meet the criteria to generate statistically significant physiological parameter reference benchmarks.

[0030] Meanwhile, if no similar patients matching the criteria are found, the patient's physiological parameter benchmark, constructed based on their own historical physiological data, is obtained from the medical and health big data sharing platform. Their own historical physiological data refers to the time series data of physiological parameters recorded by monitoring devices during the patient's previous treatment cycles.

[0031] The constructed physiological parameter benchmarks are derived from the cleaning, filtering, and statistical analysis of these individual historical data, representing personalized normal ranges for physiological parameters. This operation serves as a backup plan when a population benchmark is unavailable. The system automatically switches to the patient's personal historical database, extracts their physiological parameter records under stable conditions, and generates a unique physiological parameter benchmark through statistical calculations, ensuring a reliable reference standard is available under any circumstances.

[0032] Finally, the patient's real-time physiological data is obtained by acquiring physiological signals from the multi-parameter vital signs monitor deployed within the target medical cabin through the aforementioned medical and health big data sharing platform. The multi-parameter vital signs monitor is a high-precision medical monitoring device integrated within the medical cabin, continuously collecting the patient's physiological signals through multiple sensors.

[0033] Physiological signals include raw bioelectrical signals such as photoplethysmography (PPG), electrocardiogram (ECG), and respiratory waveforms. Real-time physiological data are clinically significant parameter values ​​obtained after these raw signals have been processed and converted by the device's built-in algorithms, such as blood oxygen saturation, heart rate, and respiratory rate. The acquisition process involves receiving timestamped physiological parameter data packets uploaded in real time from the designated medical pod monitor via the platform's data acquisition interface, forming a continuous monitoring curve of the patient's physiological state.

[0034] Overall, this step establishes a comprehensive data acquisition pipeline to ensure the acquisition of all-round data, from equipment operating status to patient physiological information. This multi-level data collection mechanism provides complete data support for subsequent pressure disturbance prediction and personalized oxygen supply adjustment, taking into account both the real-time status of the equipment system and individual patient differences and group medical experience, thus laying a solid data foundation for achieving precision medical cabin oxygen supply control.

[0035] S2: Based on the graph neural network algorithm, the real-time oxygen supply pressure of the entire medical cabin oxygen supply network is analyzed by data stream. The pressure disturbances that are about to occur in the network are predicted hundreds of milliseconds to several seconds in advance, and the medical cabins where pressure disturbances are about to occur are identified as high-risk medical cabins.

[0036] In this embodiment, the data stream analysis of the real-time oxygen supply pressure of the entire medical cabin oxygen supply network based on the graph neural network algorithm, predicting the pressure disturbances that will occur in the network hundreds of milliseconds to several seconds in advance, and identifying the medical cabins where pressure disturbances are about to occur as high-risk medical cabins, includes: The topology of the oxygen supply network is analyzed to obtain the node connection relationship of the entire oxygen supply network; Based on the node connection relationship, each medical cabin is abstracted as a graph node, and the oxygen supply pipes connecting the medical cabins are abstracted as edges, thus constructing a directed graph of the oxygen supply network with medical cabins as nodes and oxygen supply pipes as edges. The real-time oxygen supply pressure is input as a node feature into the directed graph of the oxygen supply network. Based on the directed graph of the oxygen supply network, data flow analysis is performed to predict the pressure prediction value of the downstream medical cabin node in the next hundreds of milliseconds to several seconds. Based on the flow direction of oxygen in the oxygen supply network and the propagation process of pressure disturbance in the oxygen supply network, forward reasoning is performed on the pressure of the downstream medical cabin node to obtain the predicted pressure value of the downstream medical cabin node in the next hundreds of milliseconds to several seconds. The predicted pressure value is compared with a preset pressure threshold. If the value exceeds the threshold, the downstream medical cabin is determined to be a high-risk medical cabin.

[0037] Preferably, this step involves performing topology analysis on the oxygen supply network to obtain the node connection relationships of the entire oxygen supply network. The oxygen supply network refers to a complex oxygen supply system composed of a central oxygen source, various levels of oxygen delivery pipelines, and multiple medical cabins. Topology analysis is a process of systematically analyzing the physical connections and spatial layout relationships between all components of the network to fully understand its system architecture.

[0038] The node connectivity precisely describes the exact location coordinates of each medical module within the entire network, as well as the connection paths and gas flow directions established between them via oxygen supply pipelines of various specifications. This crucial operation requires retrieving comprehensive network configuration information from the infrastructure management database of the healthcare big data sharing platform. This includes a table of the physical location coordinates of all medical modules and a matrix of connection relationships for each segment of the oxygen supply pipeline. This information is then processed by a specialized network analysis engine to accurately reconstruct the complete topology of the entire oxygen supply network in digital space.

[0039] In detail, based on the node connection relationships, each medical pod is abstracted as a graph node, and the oxygen supply pipes connecting the medical pods are abstracted as edges, constructing a directed graph of the oxygen supply network with medical pods as nodes and oxygen supply pipes as edges. Graph nodes are the basic building blocks in graph theory, specifically referring to a single medical pod unit after abstraction. Each node stores a unique identifier and state attribute of the corresponding medical pod.

[0040] Oxygen supply pipelines are the physical channels that actually connect the various medical cabins for oxygen delivery. They are abstracted as edges connecting two nodes in a graph, and the weight of the edge reflects the physical characteristics of the pipeline, such as its length and diameter. A directed graph is a graph structure where the edges have a clear direction. Here, the direction strictly follows the actual flow direction of oxygen in the pipeline, from the oxygen source to each medical cabin, forming a unidirectional oxygen delivery path.

[0041] The construction operation uses a specialized graph construction engine to transform the parsed physical connections into a standard graph data structure. This process includes assigning a unique node identifier to each medical pod, creating edge objects with directional attributes for each pipe segment, and establishing connections between nodes and edges based on the actual network topology. Ultimately, it generates a complete digital twin model that represents the topological characteristics of the oxygen supply network.

[0042] More specifically, the real-time oxygen supply pressure is input as a node feature into the directed graph of the oxygen supply network. Based on the directed graph, data flow analysis is performed to predict the pressure values ​​of downstream medical cabin nodes within the next few hundred milliseconds to several seconds. The real-time oxygen supply pressure is the instantaneous oxygen pressure reading monitored and uploaded in real time by each medical cabin node through its oxygen supply pressure sensor, reflecting the current pressure status of the node.

[0043] Node characteristics are attribute information used in the graph to describe the dynamic state of nodes; here, they specifically refer to the real-time oxygen supply pressure value sequence that changes continuously over time for each medical module node. Data flow analysis refers to the continuous processing of the pressure data sequence that is generated over time, observing and analyzing the propagation patterns and dynamic trends of this data in a complex directed graph structure.

[0044] Downstream medical module nodes are adjacent or indirectly connected medical module units located behind a specific node in the direction of oxygen flow. Pressure prediction values ​​are inferred based on the real-time state and topology of the entire network. By analyzing the propagation characteristics of pressure fluctuations, the precise pressure values ​​that downstream nodes may experience within the next few hundred milliseconds to several seconds are predicted.

[0045] This crucial operation involves assigning the real-time acquired network pressure data stream to each node in the graph according to their corresponding relationships. Then, by utilizing the complex connection relationships and directional information inherent in the graph structure, the transmission process and attenuation law of pressure signal fluctuations in the network are simulated, thereby enabling accurate and quantitative predictions of the pressure change trends of downstream nodes.

[0046] Based on the flow direction of oxygen in the oxygen supply network and the propagation process of pressure disturbances in the network, forward inference is performed on the pressure of downstream medical module nodes to obtain predicted pressure values ​​for those nodes in the next few hundred milliseconds to several seconds. The oxygen flow direction is determined by the engineering design of the oxygen supply system and is a fixed path for unidirectional delivery of oxygen from the central gas source through various pipelines to the terminal medical module.

[0047] Pressure disturbance refers to localized abnormal pressure fluctuations caused by factors such as the start-up or shutdown of oxygen equipment in a medical ward within a network, or sudden changes in oxygen consumption. The propagation process precisely describes how these pressure fluctuations, following fluid dynamics principles, gradually propagate along the oxygen supply network and affect other connected nodes. Forward reasoning strictly follows the fixed direction of oxygen flow, recursively inferring the intensity and time delay of the pressure disturbance event's impact on subsequent downstream nodes according to the network topology.

[0048] This fine-tuning operation follows the graph structure analysis in the previous step. Specifically, based on the determined oxygen flow path, it simulates the complete process of how a pressure disturbance event, generated from an upstream node, propagates through a complex pipeline network and gradually affects each downstream node. This requires considering multiple factors such as pipeline impedance, gas velocity, and node capacity, and then quantitatively calculates the precise pressure value of a specific downstream node in the very short timescale of the future.

[0049] Finally, the predicted pressure value is compared with a preset pressure threshold. If the predicted value is exceeded, the downstream medical cabin is determined to be a high-risk medical cabin. The predicted pressure value is the pressure value at a specific future time point obtained through complex reasoning calculations in the preceding steps. The preset pressure threshold is a pressure safety limit range pre-set according to the clinical treatment safety specifications and equipment operation requirements of the medical cabin. This threshold is usually determined based on the minimum oxygen supply pressure required to ensure patient treatment safety and comfort, combined with the technical specifications of the equipment manufacturer.

[0050] A high-risk medical cabin refers to a medical cabin unit whose predicted pressure value will exceed the safe operating range, facing the risk of insufficient oxygen supply or excessive pressure fluctuations. The comparison operation involves comparing the predicted pressure value of each downstream node with the node's personalized pressure threshold, performing a precise logical judgment on whether the limit has been exceeded. Once the predicted pressure of a medical cabin node is determined to exceed the threshold range, it is immediately marked as a high-risk medical cabin, and the severity of the exceedance and the expected time of occurrence are recorded, thus completing the critical transition from pressure prediction to risk identification.

[0051] Overall, this series of steps achieves a highly forward-looking ability to sense and warn of pressure disturbances by precisely abstracting the physical oxygen supply network into a computational graph model and continuously injecting real-time operational data. Its significant effect lies in revolutionizing the monitoring mode of the oxygen supply system from the traditional passive reception of current status information to proactively predicting system risks in the near future.

[0052] This allows the operations and maintenance management unit to accurately pinpoint potentially affected medical pods before pressure imbalances actually occur and have a substantial impact on patient treatment, identify risk areas where oxygen supply fluctuations are about to occur, and gain valuable preparation time for subsequent precise compensation and adjustment. This fundamentally enhances the operational reliability, treatment safety, and patient comfort of the medical pod oxygen supply system.

[0053] In this embodiment, the step of performing topology analysis on the entire oxygen supply network to obtain the node connection relationships of the entire oxygen supply network includes: The location of the medical cabin nodes in the oxygen supply network is obtained based on the medical and health big data sharing platform, and the distribution of medical cabin nodes in the entire medical cabin oxygen supply network is obtained. The node connection relationship of the oxygen supply network is constructed based on the node distribution of the medical cabin and the connection relationship of the oxygen supply pipeline in the oxygen supply network.

[0054] Preferably, this step involves obtaining the node locations of medical cabins in the oxygen supply network based on a medical and health big data sharing platform, thus obtaining the node distribution of the entire medical cabin oxygen supply network. The medical and health big data sharing platform is a centralized data management platform integrating medical equipment information and patient data. The oxygen supply network refers to a complete oxygen supply system consisting of a central oxygen source, oxygen pipelines, and multiple interconnected medical cabins. Node location refers to the specific installation coordinates of each medical cabin in physical space, including floor location, area number, and spatial coordinate information.

[0055] The distribution of medical pod nodes is a collection of all medical pod location information in digital space, forming a spatial layout map of the oxygen supply network. This acquisition operation calls the device location query interface provided by the medical and health big data sharing platform, sends a request command containing network identifiers to the platform, receives and parses the medical pod location data packets returned by the platform, extracts the unique device code of each medical pod and its corresponding physical location coordinates, and finally generates a node distribution map containing the spatial information of all medical pods.

[0056] In detail, the node connection relationship of the oxygen supply network is constructed based on the distribution of medical cabin nodes and the connection relationship of oxygen supply pipelines in the oxygen supply network. The distribution of medical cabin nodes is the set of spatial locations of medical cabins obtained in the aforementioned steps.

[0057] The oxygen supply pipeline connection relationship refers to the actual physical connection information of the oxygen delivery pipelines connecting various medical modules, including connection attributes such as pipeline start point, end point, pipeline length, and pipe diameter. The node connection relationship is the final result that needs to be constructed, which fully describes the connection path and gas flow direction established between all medical modules in the oxygen supply network through oxygen supply pipelines.

[0058] This construction operation utilizes a specialized network topology building engine to fuse medical cabin node distribution data with oxygen supply pipeline connection data. First, each medical cabin is mapped to a node object in the topology graph based on its location. Then, connecting edges are established between corresponding nodes according to the pipeline connections. The direction of the edges strictly follows the actual flow direction of oxygen in the pipeline, and the edge attributes include the physical parameters of the corresponding pipeline. This process requires precise matching of device codes in the pipeline connections with medical cabin identifiers in the node distribution, ensuring that each pipeline connection is correctly associated with the medical cabin nodes at both ends.

[0059] More specifically, the process of constructing node connections involves several refined processing steps. First, the distribution data of medical cabin nodes is standardized to ensure that all location information uses a unified coordinate system and unit of measurement. Then, the integrity of the oxygen supply pipeline connections is verified to confirm that the start and end points of each pipeline record can be found in the corresponding medical cabin in the node distribution.

[0060] Next, an association mapping between nodes and edges is established, creating connection ports for each medical pod node and establishing directed connections between nodes based on pipeline connections. Finally, a standardized node connection description file is generated, which fully records the topology of the entire oxygen supply network, including the attribute information of all nodes and the connection attributes of all edges, providing an accurate structural foundation for the subsequent construction of the directed graph of the oxygen supply network.

[0061] Overall, this series of steps systematically processed medical module location information and pipeline connection data, achieving a precise conversion from a physical oxygen supply network to a digital topology model. Its significant effect lies in establishing a complete digital twin of the oxygen supply network, integrating scattered equipment location information and pipeline connection relationships into a unified topology description.

[0062] This node connection not only accurately reflects the physical connectivity of the oxygen supply network but also clarifies the direction and path of oxygen flow, providing a crucial network structure foundation for subsequent data flow analysis and pressure disturbance prediction based on graph neural networks. This refined topology analysis ensures that the dynamic oxygen supply pressure regulation method's understanding of the network structure is completely consistent with the actual situation, thereby guaranteeing the accuracy of prediction results and the reliability of regulation decisions.

[0063] S3: Monitor the oxygen pressure at the inlet of the oxygen supply pipe of other medical cabins that are in the same oxygen supply network as the risk medical cabin. If the oxygen pressure at the inlet of the oxygen supply pipe of a certain medical cabin suddenly changes beyond the preset inlet disturbance threshold, the medical cabin is identified as the disturbed medical cabin.

[0064] In this embodiment, monitoring the oxygen pressure at the inlet of the oxygen supply pipes of other medical cabins located in the same oxygen supply network as the risk medical cabin, and identifying the medical cabin as an affected medical cabin if the sudden change in the inlet oxygen pressure of a certain medical cabin exceeds a preset inlet disturbance threshold, includes: The oxygen pressure range within a unit time window is calculated for the inlet oxygen pressure to obtain the transient fluctuation amplitude of the inlet oxygen pressure of the medical cabin. The transient fluctuation amplitude is compared with a preset inlet disturbance threshold to identify nodes where the oxygen pressure transient fluctuation amplitude exceeds the inlet disturbance threshold; The identified nodes are marked as oxygen pressure mutation events, and the nodes are identified as the interfered medical cabin.

[0065] Preferably, this step involves calculating the oxygen pressure range within a unit time window to obtain the transient fluctuation amplitude of the inlet oxygen pressure of the medical chamber. Inlet oxygen pressure specifically refers to the real-time oxygen pressure measurement value collected at the inlet of the oxygen supply pipeline in the medical chamber by precision pressure sensors. These sensors continuously monitor the oxygen pressure status entering the medical chamber at millisecond-level frequencies. The unit time window is a pre-set fixed monitoring period, typically three to five seconds in length. This period is sufficient to capture clinically significant pressure changes without masking transient abrupt changes due to excessive time.

[0066] The calculation of oxygen pressure range involves statistically analyzing all continuously collected inlet oxygen pressure readings within a specific time window to identify the maximum and minimum pressure values ​​during that period, and then accurately calculating the numerical difference between them.

[0067] The transient fluctuation amplitude is a quantitative result obtained through this range calculation. It accurately reflects the maximum range of change in inlet oxygen pressure within the monitoring period in specific pressure units, directly characterizing the severity and potential hazards of pressure surges. This calculation operation is continuously and cyclically executed by a dedicated data processing unit. Data is re-acquired and a complete range calculation is performed at each new time window, thus forming a continuous dynamic monitoring of the inlet oxygen pressure fluctuation characteristics.

[0068] In detail, the transient fluctuation amplitude is compared with a preset inlet disturbance threshold to identify nodes where the oxygen pressure transient fluctuation amplitude exceeds the inlet disturbance threshold. The transient fluctuation amplitude is a pressure fluctuation quantification index obtained through rigorous calculation, which specifically represents the range of pressure change in kilopascals or megapascals.

[0069] The preset inlet disturbance threshold is an important safety parameter that is pre-set according to the clinical treatment safety standards and equipment technical specifications of the medical cabin. This threshold clearly defines the maximum safe boundary that the inlet oxygen pressure fluctuation is allowed to be. It is usually determined based on the minimum pressure stability required to maintain normal respiratory treatment for patients and the equipment tolerance characteristics of the medical cabin oxygen supply system.

[0070] In this context, "node" specifically refers to each individual medical cabin unit monitored within the oxygen supply network topology. The comparison operation involves a dedicated threshold comparator that compares the real-time calculated transient fluctuation amplitude of each medical cabin with a preset inlet disturbance threshold, performing precise numerical judgment. The identification process is an automated screening mechanism that quickly and accurately identifies medical cabin nodes whose transient fluctuation amplitude exceeds the preset inlet disturbance threshold. The identification results of these nodes clearly indicate that their inlet oxygen pressure has experienced abnormal fluctuations exceeding the safe allowable range, potentially affecting the treatment of patients within the cabin.

[0071] More specifically, the identified nodes are marked as oxygen pressure surge events to determine if they are the affected medical cabins. Identified nodes refer to medical cabin units whose inlet oxygen pressure transient fluctuation amplitude exceeds the safe range, as confirmed through strict threshold comparison. Oxygen pressure surge event marking is a detailed status identification operation that adds specific logical labels and event flags to these confirmed abnormal nodes, comprehensively recording key information about oxygen supply interference events.

[0072] This tagging operation typically includes accurately recording the specific timestamp of the event, the precise magnitude of the fluctuation, the duration, and the node's unique device identifier. It also records the event's location relative to the at-risk medical bay. A disrupted medical bay is a medical bay unit that is ultimately and formally identified as experiencing or about to experience oxygen supply instability caused by the propagation of pressure disturbances in the oxygen supply network.

[0073] The confirmed operation involves formally classifying these nodes as disrupted medical wards after completing comprehensive event tagging, updating their detailed status information to the active monitoring list, and establishing a complete event file. This provides clear intervention targets and necessary decision-making basis for subsequent targeted oxygen supply pressure compensation adjustments.

[0074] Overall, this series of steps establishes a highly efficient, sensitive, and reliable mechanism for detecting and identifying oxygen supply disturbances. Its core effectiveness lies in capturing abnormal fluctuations in oxygen pressure at the inlet of each medical chamber in the oxygen supply network in real time through a systematic monitoring process. Furthermore, by employing precise quantitative analysis and rigorous threshold comparisons, it quickly and accurately locates medical chamber units that are actually affected by the propagation of pressure disturbances.

[0075] This identification method based on transient fluctuation amplitude characteristics, combined with a carefully designed unit time window monitoring strategy, can effectively distinguish between normal physiological pressure fluctuations and harmful pathological pressure disturbances, ensuring that subsequent compensatory regulation is only triggered when clinically significant oxygen supply interference occurs.

[0076] This precise identification capability avoids unnecessary and frequent adjustments that would place an additional burden on the oxygen supply system of the medical pod, while also enabling timely and accurate responses to real oxygen supply risks. Thus, in a complex multi-pod oxygen supply network environment, it provides reliable technical assurance for the treatment safety and oxygen supply stability of each patient in the medical pod.

[0077] S4: Based on the patient's real-time physiological data, perform parameter empirical Bayes calibration on the physiological parameter benchmarks of patients of the same age and with the same disease to obtain a dynamic physiological baseline that adaptively adjusts with the patient's sleep, activity, mood and other states.

[0078] In this embodiment, the step of performing empirical Bayesian calibration on the physiological parameter benchmarks of patients of the same age and with the same disease based on the patient's real-time physiological data to obtain a dynamic physiological baseline that adaptively adjusts with the patient's sleep, activity, mood, and other states includes: The patient's activity status is detected using pressure sensors and millimeter-wave radar in the medical cabin. Based on the historical physiological parameter data of each physiological state under the qualified physiological parameter benchmark, Gaussian distribution fitting is performed to obtain the prior probability distribution of each physiological state of patients of the same age and the same disease. Based on the prior probability distribution of the patient's physiological parameters in the current physiological state, a Bayesian update is performed to obtain the posterior distribution of the patient's physiological parameters in the current physiological state. The mean of the posterior distribution of the physiological parameters is calculated to obtain the personalized mean parameter of the posterior distribution; The standard deviation of the posterior distribution of the physiological parameters is calculated to obtain the personalized standard deviation parameter of the posterior distribution; Using the personalized mean parameter as the central benchmark, and the personalized standard deviation parameter and a preset confidence coefficient as a measure of fluctuation range, the statistical upper and lower limits of the personalized confidence interval are calculated to obtain the personalized confidence interval of the patient in the current physiological activity state. The personalized confidence interval is defined as the dynamic physiological baseline.

[0079] Preferably, this step involves detecting the patient's activity status using pressure sensors and millimeter-wave radar within the medical cabin. The pressure sensors in the medical cabin are distributed arrays of flexible micro-pressure-sensitive elements installed on the mattress, seat, and footrests within the cabin. These sensors continuously monitor minute changes in pressure distribution between the patient and the contact surfaces at a sampling frequency of 30-50Hz. When the patient changes posture, turns over, raises their legs, or experiences heart rate fluctuations within the cabin, it triggers shifts in pressure peaks and changes in distribution patterns in the contact area. These changes are captured by the pressure sensors and converted into digital pressure matrix signals.

[0080] Simultaneously, a 24GHz medical-grade millimeter-wave radar module integrated into the center of the medical cabin's roof and side walls emits frequency-modulated continuous waves. By analyzing the phase shift, time delay difference, and Doppler frequency shift characteristics of the received echo signals, a three-dimensional motion map of the patient's whole body and local limbs is constructed. The millimeter-wave radar has strong penetrating power, capable of penetrating clothing and bedding, accurately identifying the amplitude, speed, and trajectory of the patient's limb movements, while also capturing minute physiological movements such as chest rise and fall.

[0081] In detail, the pressure distribution data collected by the pressure sensor and the motion characteristic data acquired by the millimeter-wave radar are integrated and processed by the in-cabin embedded data fusion processor: First, the two types of data are time-stamped and subjected to anti-interference preprocessing (the pressure signal is filtered by moving average to remove vibration interference, and the radar signal is filtered by CFAR constant false alarm rate detection to remove environmental reflection noise). Subsequently, the centroid coordinates, pressure distribution variance, number of peak regions, and fluctuation frequency of the pressure signal are extracted. At the same time, the human body height, movement speed, trajectory curvature, chest cavity displacement, and Doppler frequency shift peak of the radar signal are extracted to form a 12-dimensional fused feature vector. Then, the feature vector is uploaded to the medical and health big data sharing platform via 5G / wired network. The platform calls a lightweight MobileNet classification model trained on a medical scene labeled dataset, and performs matching calculations between the feature vector and the parameter matrix of the pre-trained model to output the confidence probability of various activity states. Finally, using a confidence threshold of ≥0.8 as the criterion, the fused features are mapped to predefined activity state categories—including resting oxygen inhalation state, mild activity state (turning over / adjusting posture, feature vector matching the "small range of center of gravity shift + limb micro-movement" template), moderate activity state (sitting up / walking slowly, feature vector matching the "height increase + continuous trajectory" template), vigorous activity state (restlessness / struggling, feature vector matching the "high frequency fluctuation + trajectory disorder" template), and abnormal activity state (falling / leaving the cabin, feature vector matching the "sudden pressure drop + height change / target disappearance" template). If the confidence of all categories is <0.8, it is marked as "pending confirmation state," triggering the platform to push a review reminder to the medical staff terminal, combined with real-time sensor raw data to assist manual judgment.

[0082] In detail, this step involves performing Gaussian distribution fitting on historical physiological parameter data for each physiological state based on a qualifying physiological parameter benchmark, to obtain the prior probability distribution of each physiological state for patients of the same age and with the same condition. The qualifying physiological parameter benchmark refers to historical physiological data of a strictly screened group of patients of the same age and with the same condition, obtained from a medical and health big data sharing platform.

[0083] Specifically, historical physiological parameter data for each physiological state refers to time-series data of key physiological indicators such as blood oxygen saturation, heart rate, and respiratory rate recorded in similar patients under the same activity state. Gaussian distribution fitting is a statistical method used to model the probability distribution of these historical data to find their mathematical expectation and variance parameters. The prior probability distribution is a statistical model obtained through this fitting, describing the normal fluctuation range of each parameter in a specific physiological state for patients of the same type, reflecting the general physiological characteristics of this patient group.

[0084] More specifically, based on the physiological parameters of the patient in their current physiological state, a Bayesian update is performed on the prior probability distribution matching the current physiological state to obtain the posterior distribution of the patient's physiological parameters in the current physiological state. The physiological parameters of the patient in their current physiological state are physiological measurements of the patient in their current activity state, collected in real time by monitoring equipment within the medical cabin.

[0085] The prior probability distribution matching the current physiological state refers to the prior distribution model corresponding to the detected patient activity state. Bayesian update is a calculation process that combines the prior probability distribution with the currently observed actual physiological parameters of the patient using Bayesian statistical principles. The posterior distribution, obtained after Bayesian update, is a probability distribution that incorporates both group prior knowledge and individual real-time data, and it more accurately reflects the patient's physiological characteristics in the current state.

[0086] The mean of the posterior distribution of the physiological parameters is calculated to obtain the personalized mean parameter of the posterior distribution. The posterior distribution of the physiological parameters is a probability distribution model obtained after Bayesian updating. The mean calculation is a mathematical method to calculate the expected value of this probability distribution. The personalized mean parameter is this calculated expected value, which represents the central tendency of the physiological parameters adjusted based on the individual characteristics of the patient, and is the most likely baseline value of the physiological parameters for the patient in the current physiological state.

[0087] Simultaneously, the standard deviation of the posterior distribution of the physiological parameters is calculated to obtain the personalized standard deviation parameter of the posterior distribution. The standard deviation is calculated using statistical methods to determine the standard deviation of the posterior probability distribution. The personalized standard deviation parameter is this calculated standard deviation value, which quantifies the normal fluctuation range of the patient's physiological parameters around the personalized mean, reflecting the individual's unique physiological stability characteristics.

[0088] Next, using the personalized mean parameter as the central benchmark, and the personalized standard deviation parameter and a preset confidence coefficient as a measure of fluctuation range, the statistical upper and lower limits of the personalized confidence interval are calculated to obtain the personalized confidence interval of the patient under the current physiological activity state.

[0089] The personalized mean parameter is the individualized central baseline calculated in the preceding steps. The personalized standard deviation parameter is a statistic characterizing individual fluctuations. The preset confidence coefficient is a probability guarantee set according to clinical needs, typically taken as a specific quantile of the corresponding standard normal distribution. The statistical upper and lower limits are boundary values ​​calculated by adding and subtracting the product of the personalized standard deviation parameter and the confidence coefficient from the personalized mean parameter. The personalized confidence interval is the numerical range determined by these two boundary values, defining the normal fluctuation range of the patient's physiological parameters under the current physiological state.

[0090] Finally, the personalized confidence intervals are defined as the dynamic physiological baseline. The personalized confidence intervals are individualized normal ranges obtained through complete calculation. The dynamic physiological baseline is the final, adaptively adjusted reference standard for physiological parameters that changes with the patient's physiological state. It retains the rationality of population statistics while incorporating individual patient characteristics, accurately reflecting the normal physiological fluctuation range of patients under different conditions.

[0091] Overall, this series of steps establishes a complete and sophisticated mechanism for constructing personalized physiological baselines. Its core effect lies in accurately identifying the patient's physiological state through multi-sensor fusion technology and organically combining group medical knowledge with individual real-time data using parametric empirical Bayesian methods to generate a dynamic physiological baseline that adaptively adjusts according to the patient's state.

[0092] This dynamic baseline not only takes into account the normal changes in physiological parameters caused by different states of patients such as sleep, activity, and emotions, but also quantifies the individual-specific physiological fluctuation characteristics through confidence intervals, providing a precise and personalized reference standard for accurately distinguishing between physiological fluctuations and pathological abnormalities.

[0093] This refined benchmark calibration significantly improves the accuracy and adaptability of oxygen supply pressure regulation, ensuring that the medical cabin oxygen supply system can respond promptly to changes in actual oxygen demand while avoiding unnecessary regulatory interventions due to normal individual physiological fluctuations. This, in turn, safeguards the patient's treatment safety and comfort at the level of personalized medicine.

[0094] S5: Perform sliding window integration based on the patient's physiological data and the dynamic physiological baseline to obtain a comprehensive deviation index between the two, and determine the patient's compensation control vector based on the comprehensive deviation index.

[0095] In this embodiment, the step of performing sliding window integration based on the patient's physiological data and the dynamic physiological baseline to obtain a comprehensive deviation index, and determining the patient's compensation control vector based on the comprehensive deviation index, includes: Motion artifact removal was performed on the time-series data of prefrontal cortex blood oxygen saturation in the real-time physiological data to obtain a standard prefrontal cortex blood oxygen saturation value sequence. Using the sampling time of the standard prefrontal cortex blood oxygen saturation value sequence as a reference, the baseline data in the dynamic physiological baseline under the same physiological state as the current acquisition time are time-aligned and resampled to obtain the prefrontal cortex blood oxygen saturation baseline sequence corresponding to the standard value sequence at the time point. The absolute deviation of the blood oxygen saturation value sequence in the standard prefrontal cortex is calculated at each time point from the baseline blood oxygen saturation value sequence in the prefrontal cortex to generate the absolute deviation sequence of blood oxygen saturation. The same operation was performed on the remaining data in the physiological baseline to obtain the absolute deviation sequence of all data in the physiological baseline. The cumulative absolute deviation is obtained by performing an exponential decay time integral operation within a fixed-duration sliding window on the absolute deviation sequence of a single data point, and the average of the cumulative absolute deviations of all data points is used as the comprehensive deviation index. The comprehensive deviation index is input into the oxygen supply compensation model in the medical and health big data sharing platform, and the oxygen supply compensation control vector of the patient is output.

[0096] Preferably, this step involves performing a sliding window integration based on the patient's physiological data and a dynamic physiological baseline to obtain a comprehensive deviation index, and then determining the compensation control vector accordingly. This process aims to quantify the dynamic deviation between the patient's real-time physiological state and a personalized baseline through detailed time-series analysis, providing accurate data for subsequent oxygen supply pressure compensation.

[0097] Sliding window integration involves setting a fixed-duration analysis window over a continuous time series. This window slides over time, and a weighted integration is performed on the data within the window. The comprehensive deviation index is a scalar value obtained using this method, reflecting the overall deviation level of multiple physiological parameters from the dynamic physiological baseline over a specific time period. The compensation control vector is a set of parameterized instructions generated based on this index, used to guide the pressure regulation operation of the medical cabin's oxygen supply system.

[0098] In detail, motion artifact removal was first performed on the time-series data of prefrontal cortex blood oxygen saturation in the real-time physiological data. This prefrontal cortex blood oxygen saturation time-series data is continuously acquired by a near-infrared spectral sensor installed in the medical cabin, reflecting changes in the concentration of oxyhemoglobin in the patient's prefrontal cortex. Motion artifact removal is a data preprocessing technique specifically designed to address non-physiological noise in physiological signals, aiming to identify and eliminate signal distortion and transient abrupt changes caused by patient limb movements, changes in body position, or slight sensor displacement.

[0099] This operation employs wavelet transform-based or adaptive filtering signal processing algorithms to perform multi-scale decomposition and reconstruction of the original time-series data, effectively separating physiological signals from motion interference components, and ultimately outputting a pure and stable sequence of standard prefrontal cortex blood oxygen saturation values. This crucial preprocessing step significantly improves the signal quality for subsequent analysis, laying a reliable foundation for accurately assessing blood oxygenation status.

[0100] More specifically, using the sampling time point of the preprocessed standard prefrontal cortex oxygen saturation value sequence as a benchmark, the benchmark data in the dynamic physiological baseline that are in the same physiological state as the current acquisition time are time-aligned and resampled. The dynamic physiological baseline is a personalized physiological parameter reference range obtained through parametric empirical Bayesian calibration methods, which can adaptively adjust according to different states of the patient, such as sleep, activity, and mood.

[0101] Time alignment refers to precisely matching the time axis of dynamic physiological baseline data with the timestamps of the real-time acquired standard value sequence, ensuring that the two are directly comparable at exactly the same time points. Resampling processing uses linear interpolation or cubic spline interpolation algorithms to adjust the resolution of the time series of dynamic physiological baseline data, making its sampling interval completely consistent with the real-time data sequence.

[0102] After the above processing, a baseline sequence of prefrontal cortex blood oxygen saturation that strictly corresponds to the standard value sequence at a given time point is obtained. This sequence accurately represents the ideal reference trajectory of the patient's blood oxygen saturation under the current specific physiological state.

[0103] The absolute deviation of standard prefrontal cortex oxygen saturation values ​​is calculated point-by-point from the time-aligned prefrontal cortex oxygen saturation baseline sequence. Point-by-time absolute deviation calculation refers to taking the absolute difference between the real-time oxygen saturation measurement and the baseline reference value at each same time stamp, thereby generating a continuous oxygen saturation absolute deviation sequence.

[0104] This sequence fully records the instantaneous deviation between the patient's actual blood oxygen saturation value and the expected baseline value at each sampling moment, accurately quantifying the real-time fluctuations of blood oxygen parameters. This refined point-by-point comparison method can effectively capture subtle and transient physiological changes, providing high-precision input data for subsequent time-integration analysis.

[0105] Meanwhile, other physiological parameters included in the dynamic physiological baseline are also processed using the same procedures described above. The dynamic physiological baseline typically includes several key physiological parameters, such as heart rate variability, respiratory rate, and pulse pressure. Each parameter requires independent motion artifact removal, time alignment, resampling, and calculation of absolute deviation at each time point.

[0106] Through this series of standardized operations, the absolute deviation sequence corresponding to each data point in the physiological baseline is finally obtained. This step ensures the consistency and comprehensiveness of the multi-physiological parameter deviation assessment, and constructs a complete multidimensional deviation dataset for comprehensively assessing the patient's overall physiological status.

[0107] Then, for each individual physiological parameter's absolute deviation sequence, an exponentially decaying time integral is performed within a fixed-length sliding window. The fixed-length sliding window is a pre-defined time interval of fixed length that slides continuously forward along the time axis, covering a continuous segment of deviation data points each time. The exponentially decaying time integral is a special weighted integration method that assigns weights to deviation data points at different times within the window that decrease exponentially with time, with higher weights for recent data points and weights for older data points approaching zero, thereby calculating the cumulative absolute deviation within each sliding window.

[0108] This computational mechanism emphasizes the contribution of recent bias to the overall assessment, making the calculation results more sensitive to the latest physiological trends. Subsequently, the cumulative absolute biases calculated for all physiological parameters within the same time window are arithmetically averaged to obtain the comprehensive bias index. This index, as an integrated indicator, effectively characterizes the overall deviation of multiple physiological parameters from the personalized dynamic baseline within the recent time window.

[0109] Finally, the calculated comprehensive deviation index is input into the oxygen supply compensation model deployed in the medical and health big data sharing platform. The oxygen supply compensation model is a data-driven model trained on a large amount of historical clinical data and oxygen supply regulation cases, and it encapsulates a complex mapping relationship from physiological deviations to oxygen supply regulation strategies.

[0110] The model first determines the basic pressure adjustment amount based on the numerical range of the comprehensive deviation index using a piecewise linear function. For example, when the index is in the middle range, the adjustment amount is scaled proportionally. At the same time, the adjustment duration is dynamically calculated based on the rate of change of the index to ensure rapid response to sudden deviations. Finally, the above parameters are weighted, fused, and limited based on the transient fluctuation amplitude of the oxygen pressure at the current oxygen supply pipeline inlet, generating a final compensation control vector that includes the target pressure adjustment amount, adjustment duration, and rate of change curve. This process achieves precise and automatic conversion from physiological deviations to compensation commands through defined mathematical relationships.

[0111] The compensation control vector is a set of structured output parameters, including key control commands such as the target oxygen supply pressure adjustment amount, the duration of pressure regulation, and the shape of the pressure change rate curve. These commands will directly guide the oxygen supply actuator of the medical module to perform precise pressure compensation operations.

[0112] Overall, this series of steps utilizes a systematic and standardized data processing and analysis method to achieve a complete transformation from real-time monitoring of multi-channel physiological signals to the generation of personalized oxygen supply control commands. Its significant effect lies in its ability to dynamically, comprehensively, and accurately assess the overall deviation of a patient's physiological state from their personalized baseline, and based on this, generate scientific and precise oxygen supply compensation strategies.

[0113] This not only greatly improves the accuracy and timeliness of dynamic adjustment of oxygen supply pressure in the medical cabin, but also effectively distinguishes between normal physiological fluctuations and deviations caused by abnormal oxygen supply, thereby optimizing the overall control performance of the medical cabin oxygen supply system while ensuring the safety and comfort of patient treatment.

[0114] S6: Based on the sudden change in oxygen pressure at the inlet of the oxygen supply pipeline and the compensation control vector, adjust the oxygen supply pressure compensation command of the disturbed medical oxygen chamber to generate a compensation command set for the disturbed medical oxygen chamber.

[0115] In this embodiment, the step of adjusting the oxygen supply pressure compensation command of the disturbed medical oxygen chamber based on the sudden change in oxygen pressure at the oxygen supply pipeline inlet and the compensation control vector, and generating a compensation command set for the disturbed medical oxygen chamber, includes: Based on the transient fluctuation amplitude of the sudden change in inlet oxygen pressure, the compensation control vector is dynamically gain-corrected to obtain the corrected compensation control vector of the disturbed medical oxygen chamber. Based on the modified compensation control vector, the oxygen supply pressure command of the interfered medical cabin is adjusted to generate a compensation command set for the interfered medical cabin.

[0116] Preferably, this step involves adjusting the oxygen supply pressure compensation command for the disturbed medical oxygen chamber based on sudden changes in oxygen pressure at the oxygen supply pipeline inlet and the compensation control vector, to generate a complete operation process specific to the disturbed medical oxygen chamber. This process aims to deeply integrate the detected physical oxygen supply disturbance characteristics with the compensation needs derived from the analysis of the patient's physiological state, and through a precise command adjustment mechanism, form a control strategy that can quickly offset the impact of oxygen supply pressure fluctuations.

[0117] Oxygen pressure surge at the inlet of the oxygen supply pipeline specifically refers to a significant and drastic change in oxygen pressure detected at the inlet of the oxygen supply pipeline in the medical cabin within a very short period of time. This pressure surge is usually caused by the sudden start-up or shutdown of oxygen-using equipment in other medical cabins within the same oxygen supply network.

[0118] The compensation control vector is a set of parameterized adjustment instructions obtained by analyzing the comprehensive deviation index between multiple physiological parameters of the patient and the dynamic physiological baseline. It includes key control information such as the suggested oxygen supply pressure adjustment amount, adjustment direction, and the degree of adjustment. The compensation instruction set is the final generated collection of specific control commands with temporal and logical relationships, which are directly issued to the oxygen supply pressure regulation actuator of the medical module.

[0119] In detail, the compensation control vector is first dynamically adjusted based on the transient fluctuation amplitude of the inlet oxygen pressure mutation, thereby obtaining a corrected compensation control vector specifically for the disturbed medical oxygen chamber. The transient fluctuation amplitude of the inlet oxygen pressure mutation is a quantitative indicator obtained by calculating the difference between the maximum and minimum inlet oxygen pressure within a unit time window, which accurately characterizes the severity and energy intensity of this oxygen pressure mutation.

[0120] Dynamic gain correction is a real-time, adaptive online optimization technique for control parameters. Its core idea is to dynamically adjust the gain coefficients of each control component in the compensation control vector based on the magnitude of the monitored transient fluctuations, thereby achieving precise scaling of the compensation intensity. The specific operation process is as follows: First, the calculated transient fluctuation amplitude is multiplied by a gain adjustment coefficient pre-set according to the characteristics of the oxygen supply network and clinical safety requirements, thus generating a real-time dynamic gain factor.

[0121] Subsequently, this dynamic gain factor is multiplied one by one by each pressure adjustment component contained in the compensation control vector; the final output is a new corrected compensation control vector with adaptive amplitude adjustment. This key step ensures that the compensation response strength to sudden changes in oxygen supply pressure is proportional to the actual severity of the disturbance, avoiding pressure oscillations that may be caused by overcompensation under slight disturbances, and preventing the inability to quickly restore pressure stability due to insufficient compensation under severe disturbances.

[0122] More specifically, based on the obtained modified compensation control vector, the current oxygen supply pressure command of the disturbed medical cabin is further refined according to this vector, ultimately generating a complete and immediately executable set of compensation commands. The modified compensation control vector at this point embodies an optimized control strategy that comprehensively considers the intensity of external oxygen supply disturbances and the individual physiological compensation needs of the patient.

[0123] The oxygen supply pressure command is the sequence of target pressure setpoints that the medical cabin's oxygen supply control system is currently executing or preparing to execute. The core of the adjustment operation is to map the control strategy contained in the correction compensation control vector to the specific oxygen supply pressure command. This process is achieved through specific command fusion logic, such as performing vector addition between the pressure adjustment component in the vector and the current baseline pressure command, or replanning the trajectory of the pressure command change based on the time parameter in the vector.

[0124] The compensation instruction set, as the final output, is a structured data set that includes not only the adjusted target oxygen supply pressure value but also details the pressure change rate required to reach the target pressure, the stabilization duration after reaching the target, and possible phased adjustment parameters. The generation of this instruction set is accomplished by a dedicated instruction compilation module. This module converts the adjusted pressure parameters into a sequence of control codes with clear timing and logical relationships that the underlying oxygen supply actuators can directly recognize and execute, ensuring that each instruction is executed accurately.

[0125] Overall, this series of steps, by introducing a dynamic gain correction mechanism and a refined instruction adjustment process, constructs an intelligent compensation control closed loop capable of rapidly responding to pressure disturbances in the oxygen supply network and closely integrating with the patient's real-time physiological state. Its significant effect lies in achieving a leap from passive pressure anomaly monitoring to proactive, forward-looking, and precise compensation tailored to the patient's physiological needs.

[0126] This regulatory mechanism, which deeply integrates the disturbance characteristics of the physical system with the biological response characteristics of the patient, greatly enhances the robustness and adaptability of the medical cabin oxygen supply system in the face of internal or external disturbances. It fundamentally ensures the continuous stability of the oxygen pressure inhaled by the patient during hyperbaric oxygen therapy or intensive care and the safety of the treatment environment, thus providing a solid technical guarantee for the patient's treatment safety and rehabilitation effect.

[0127] S7: Feed back the patient's physiological parameters and medical cabin oxygen supply pressure response data after the execution of the compensation instruction set to the medical and health big data sharing platform.

[0128] In this embodiment, feeding back the patient's physiological parameters and the oxygen supply pressure response data of the medical cabin after the execution of the compensation instruction set to the medical and health big data sharing platform includes: Collect real-time physiological parameters of the patient and oxygen supply pressure response data of the medical cabin after the compensation command is executed; The patient's real-time physiological parameters and the oxygen supply pressure response data of the medical cabin are integrated into a data structure to generate a feedback data package containing information on the correlation between the patient's physiological state and oxygen supply pressure. The feedback data packet is sent to the medical and health big data sharing platform in a standardized format by calling the real-time data transmission interface provided by the medical and health big data sharing platform.

[0129] Preferably, this step involves a complete operational process of feeding back the patient's physiological parameters and the oxygen supply pressure response data of the medical cabin after the execution of the compensation instruction set to the medical and health big data sharing platform. This process constitutes a closed loop in the entire dynamic adjustment process of oxygen supply pressure. By systematically collecting data on the actual effects of the compensation measures, it provides crucial empirical evidence for the continuous optimization and verification of the entire adjustment method.

[0130] The real-time physiological parameters of the patient after the execution of the compensation command set specifically refer to the continuous data sequence reflecting changes in the patient's physiological state, newly collected by various high-precision monitoring devices integrated within the medical cabin after oxygen supply pressure compensation adjustment is implemented on the disturbed medical cabin. The oxygen supply pressure response data of the medical cabin refers to the actual pressure change trajectory generated by the actuators of the oxygen supply system and the final stable state parameters reached during the complete execution of the compensation command.

[0131] In detail, the first step is to execute the operation of collecting real-time physiological parameters of the patient and oxygen supply pressure response data of the medical cabin after the compensation command is executed. The specific process of collecting real-time physiological parameters of the patient is to continuously monitor and record the changes of various physiological indicators of the patient after experiencing oxygen supply pressure compensation adjustment by a multi-parameter vital signs monitor deployed in the medical cabin. These key physiological parameters include, but are not limited to, indicators reflecting oxygenation status and cardiovascular function, such as prefrontal cortex blood oxygen saturation, instantaneous heart rate variability, and respiratory rhythm waveform.

[0132] Meanwhile, the process of collecting oxygen supply pressure response data in the medical cabin is achieved by using high dynamic response pressure sensors precisely installed on key nodes of the oxygen supply pipeline in the medical cabin. These sensors record the actual dynamic response curve of the oxygen supply pressure under the compensation command with a time resolution of milliseconds. These response data specifically include important parameters characterizing the dynamic characteristics of the system, such as the initial pressure response delay time, the instantaneous rate of pressure rise or fall, the stabilization time to reach the target pressure, and the possible pressure overshoot.

[0133] The acquisition process is highly automated. The data recording mechanism is automatically triggered the moment the compensation command is executed and continues to run until the oxygen supply pressure is restored and maintained in a stable state for a preset observation period, thereby ensuring that all key dynamic response information during the entire adjustment process is captured completely and without omission.

[0134] More specifically, the process involves integrating the patient's real-time physiological parameters with the oxygen supply pressure response data from the medical cabin, generating a complete feedback data package containing information on the correlation between the patient's physiological state and oxygen supply pressure. Data structure integration is a sophisticated data processing procedure; its core task is to precisely align and logically reorganize heterogeneous data from different acquisition sources, with different formats and sampling characteristics, according to a unified spatiotemporal reference standard.

[0135] The patient's real-time physiological parameters are essentially a stream of equally spaced sampled data showing the continuous variation of multiple physiological parameters over time, while the oxygen supply pressure response data from the medical cabin is a high-density sequence of sampled values ​​showing the continuous variation of oxygen supply pressure parameters over time. The integration process first requires precise millisecond-level alignment of the timestamps of these two data streams to eliminate potential time synchronization errors caused by different acquisition devices. Then, following a predefined data structure template, the measured values ​​of various physiological parameters collected at the same time are combined with the corresponding measured values ​​of oxygen supply pressure to form a structured record with complete spatiotemporal correlation.

[0136] The feedback data packet generated through this process is a structured data set with strict format specifications. It not only contains all the original time-series physiological parameters and stress dynamic response data, but also systematically includes important metadata such as the unique identifier of this compensation regulation event, the precise timestamps of the start and end of the compensation instruction, the medical cabin equipment number involved, the anonymized identifier of the patient receiving treatment, and the version of the compensation control vector applied. This constitutes a complete and traceable contextual data unit.

[0137] This involves executing a call to the real-time data transmission interface provided by the medical and health big data sharing platform, sending feedback data packets to the platform in a standardized format. The real-time data transmission interface is a dedicated data receiving channel specifically designed and provided by the medical and health big data sharing platform for accessing data from various medical devices. This interface typically features high-concurrency data processing capabilities, low-latency transmission characteristics, and stringent security mechanisms.

[0138] The specific operation of calling this interface is achieved through a dedicated data transmission module deployed in the local control system of the medical pod. This module establishes a secure network connection in accordance with the interface technical specifications published by the platform, completes a strict identity authentication process, and pre-verifies the format and integrity of the data to be transmitted.

[0139] Standardized format here specifically refers to adherence to the unified data exchange protocol established by the medical and health big data sharing platform. This includes specific data encoding methods, transport layer protocols, data encapsulation specifications, and encryption standards. The entire transmission process involves serializing and encapsulating the integrated and verified feedback data packets according to the platform's standardized format, pushing them in real-time through an encrypted secure transmission channel to the data receiving endpoint designated by the medical and health big data sharing platform, and awaiting confirmation of successful reception from the platform.

[0140] Overall, this series of steps successfully constructed a complete and reliable data feedback loop by establishing standardized data acquisition processes, implementing refined data structure integration, and adopting standardized data transmission mechanisms. A significant benefit of this loop is that it enables, for the first time, quantifiable traceability and objective evaluation of the entire process of oxygen supply pressure compensation and adjustment in a single medical cabin, providing rich and high-quality real-world operational data for the medical and health big data sharing platform.

[0141] The continuously accumulated feedback data can not only be used to monitor the operation status and adjustment effect of each medical ward in real time, but more importantly, it provides valuable empirical materials for long-term data mining and in-depth analysis. This provides a solid data foundation for continuously optimizing the parameter settings of the oxygen supply compensation model, verifying the effectiveness of the control strategy, and improving the performance of the dynamic adjustment algorithm. As a result, the dynamic adjustment method of oxygen supply pressure in medical wards is constantly evolving towards a higher level of accuracy, reliability and adaptability, and ultimately provides continuous technical support for ensuring the safety and comfort of patients during treatment.

[0142] Please see Figure 2 This invention provides a dynamic oxygen supply pressure regulation system based on a medical cabin, comprising: Data acquisition module: used to acquire real-time oxygen supply pressure of the medical cabin, physiological parameter benchmarks of patients of the same age and with the same disease, and real-time physiological data of patients in the cabin from the medical and health big data sharing platform; Risk Disturbance Module: Used to perform data stream analysis on the real-time oxygen supply pressure of the entire medical cabin oxygen supply network based on graph neural network algorithm, predict the pressure disturbance that will occur in the network hundreds of milliseconds to several seconds in advance, and identify the medical cabins where the pressure disturbance is about to occur as risk medical cabins; Interference identification module: used to monitor the oxygen pressure at the inlet of the oxygen supply pipe of other medical cabins that are in the same oxygen supply network as the risk medical cabin. If the oxygen pressure at the inlet of the oxygen supply pipe of a certain medical cabin suddenly changes beyond the preset inlet disturbance threshold, the medical cabin is identified as the interfered medical cabin. Baseline calibration module: used to perform empirical Bayesian calibration of physiological parameters of patients of the same age and with the same disease based on the real-time physiological data of the patient, so as to obtain a dynamic physiological baseline that adaptively adjusts with the patient's sleep, activity, mood and other states; Compensation control module: used to perform sliding window integration based on the patient's physiological data and the dynamic physiological baseline to obtain a comprehensive deviation index between the two, and to determine the patient's compensation control vector based on the comprehensive deviation index; Command adjustment module: used to adjust the oxygen supply pressure compensation command of the disturbed medical oxygen chamber based on the sudden change in oxygen pressure at the inlet of the oxygen supply pipeline and the compensation control vector, and generate a compensation command set for the disturbed medical oxygen chamber; Data feedback module: used to feed back the patient's physiological parameters and medical cabin oxygen supply pressure response data after the execution of the compensation instruction set to the medical and health big data sharing platform.

[0143] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0144] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0145] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for dynamically adjusting oxygen supply pressure based on a medical cabin, characterized in that, The method includes: S1: Obtain real-time oxygen supply pressure of the medical cabin, physiological parameter benchmarks of patients of the same age and with the same disease, and real-time physiological data of patients in the cabin from the medical and health big data sharing platform; S2: Based on the graph neural network algorithm, the real-time oxygen supply pressure of the entire medical cabin oxygen supply network is analyzed by data stream, and the pressure disturbance that is about to occur in the network is predicted hundreds of milliseconds to several seconds in advance. The medical cabins where the pressure disturbance is about to occur are identified as high-risk medical cabins. S3: Monitor the oxygen pressure at the inlet of the oxygen supply pipe of other medical cabins that are in the same oxygen supply network as the risk medical cabin. If the oxygen pressure at the inlet of the oxygen supply pipe of a certain medical cabin suddenly changes beyond the preset inlet disturbance threshold, the medical cabin is identified as the disturbed medical cabin. S4: Based on the real-time physiological data of the patient, perform parameter empirical Bayes calibration on the physiological parameter benchmark of the patients of the same age and the same disease to obtain a dynamic physiological baseline that adaptively adjusts with the patient's sleep, activity, mood and other states; S5: Perform sliding window integration based on the patient's physiological data and the dynamic physiological baseline to obtain the comprehensive deviation index between the two, and determine the patient's compensation control vector based on the comprehensive deviation index; S6: Based on the sudden change in oxygen pressure at the inlet of the oxygen supply pipeline and the compensation control vector, adjust the oxygen supply pressure compensation command of the disturbed medical oxygen chamber to generate a compensation command set for the disturbed medical oxygen chamber. S7: Feed back the patient's physiological parameters and medical cabin oxygen supply pressure response data after the execution of the compensation instruction set to the medical and health big data sharing platform.

2. The method for dynamically adjusting oxygen supply pressure based on a medical cabin as described in claim 1, characterized in that, The acquisition of real-time oxygen supply pressure, physiological parameter benchmarks for patients of the same age and with the same condition, and real-time physiological data of patients within the medical cabin from the medical and health big data sharing platform includes: The real-time oxygen supply pressure of the medical cabin is obtained from the oxygen supply pressure sensor of the medical cabin, which is uploaded from the medical and health big data sharing platform. Extract the age and medical record information of the patients in the cabin from the electronic medical record database of the medical and health big data sharing platform; Based on the patient's age and medical record information, physiological parameter benchmarks for patients with the same type of disease and similar disease severity that are within 5 years of age from the historical case database of the medical and health big data sharing platform are retrieved. If no similar patients matching the criteria are found, the patient's physiological parameter benchmarks, constructed based on their own historical physiological data, are obtained from the medical and health big data sharing platform. The patient's real-time physiological data is obtained by acquiring physiological signals from a multi-parameter vital signs monitor deployed in the target medical cabin through the medical and health big data sharing platform.

3. The method for dynamically adjusting oxygen supply pressure based on a medical cabin as described in claim 1, characterized in that, The graph neural network algorithm performs real-time oxygen supply pressure data stream analysis on the entire medical cabin oxygen supply network, predicting impending pressure disturbances in the network hundreds of milliseconds to several seconds in advance, and identifying medical cabins with impending pressure disturbances as high-risk medical cabins, including: The topology of the oxygen supply network is analyzed to obtain the node connection relationship of the entire oxygen supply network; Based on the node connection relationship, each medical cabin is abstracted as a graph node, and the oxygen supply pipes connecting the medical cabins are abstracted as edges, thus constructing a directed graph of the oxygen supply network with medical cabins as nodes and oxygen supply pipes as edges. The real-time oxygen supply pressure is input as a node feature into the directed graph of the oxygen supply network. Based on the directed graph of the oxygen supply network, data flow analysis is performed to predict the pressure prediction value of the downstream medical cabin node in the next hundreds of milliseconds to several seconds. Based on the flow direction of oxygen in the oxygen supply network and the propagation process of pressure disturbance in the oxygen supply network, forward reasoning is performed on the pressure of the downstream medical cabin node to obtain the predicted pressure value of the downstream medical cabin node in the next hundreds of milliseconds to several seconds. The predicted pressure value is compared with a preset pressure threshold. If the value exceeds the threshold, the downstream medical cabin is determined to be a high-risk medical cabin.

4. The method for dynamically adjusting oxygen supply pressure based on a medical cabin as described in claim 3, characterized in that, The process of performing topology analysis on the entire oxygen supply network to obtain the node connection relationships of the entire oxygen supply network includes: The location of the medical cabin nodes in the oxygen supply network is obtained based on the medical and health big data sharing platform, and the distribution of medical cabin nodes in the entire medical cabin oxygen supply network is obtained. The node connection relationship of the oxygen supply network is constructed based on the node distribution of the medical cabin and the connection relationship of the oxygen supply pipeline in the oxygen supply network.

5. The method for dynamically adjusting oxygen supply pressure based on a medical cabin as described in claim 1, characterized in that, The monitoring of oxygen pressure at the inlet of the oxygen supply pipes of other medical cabins located in the same oxygen supply network as the risky medical cabin, if the sudden change in the inlet oxygen pressure of a certain medical cabin exceeds a preset inlet disturbance threshold, then that medical cabin is identified as a disturbed medical cabin, including: The oxygen pressure range within a unit time window is calculated for the inlet oxygen pressure to obtain the transient fluctuation amplitude of the inlet oxygen pressure of the medical cabin. The transient fluctuation amplitude is compared with a preset inlet disturbance threshold to identify nodes where the oxygen pressure transient fluctuation amplitude exceeds the inlet disturbance threshold; The identified nodes are marked as oxygen pressure mutation events, and the nodes are identified as the interfered medical cabin.

6. The method for dynamically adjusting oxygen supply pressure based on a medical cabin as described in claim 1, characterized in that, The process involves performing empirical Bayesian calibration on the physiological parameter benchmarks of patients of the same age and with the same condition based on the patient's real-time physiological data to obtain a dynamic physiological baseline that adaptively adjusts with the patient's sleep, activity, mood, and other states, including: The patient's activity status is detected using pressure sensors and millimeter-wave radar in the medical cabin. Based on the historical physiological parameter data of each physiological state under the qualified physiological parameter benchmark, Gaussian distribution fitting is performed to obtain the prior probability distribution of each physiological state of patients of the same age and the same disease. Based on the physiological parameters of the patient under the current physiological state, a Bayesian update is performed on the prior probability distribution that matches the current physiological state to obtain the posterior distribution of the patient's physiological parameters under the current physiological state. The mean of the posterior distribution of the physiological parameters is calculated to obtain the personalized mean parameter of the posterior distribution; The standard deviation of the posterior distribution of the physiological parameters is calculated to obtain the personalized standard deviation parameter of the posterior distribution; Using the personalized mean parameter as the central benchmark, and the personalized standard deviation parameter and a preset confidence coefficient as a measure of fluctuation range, the statistical upper and lower limits of the personalized confidence interval are calculated to obtain the personalized confidence interval of the patient in the current physiological activity state. The personalized confidence interval is defined as the dynamic physiological baseline.

7. The method for dynamically adjusting oxygen supply pressure based on a medical cabin as described in claim 1, characterized in that, The process of performing sliding window integration based on the patient's physiological data and the dynamic physiological baseline to obtain a comprehensive deviation index, and determining the patient's compensation control vector based on the comprehensive deviation index, includes: Motion artifact removal was performed on the time-series data of prefrontal cortex blood oxygen saturation in the real-time physiological data to obtain a standard prefrontal cortex blood oxygen saturation value sequence. Using the sampling time of the standard prefrontal cortex blood oxygen saturation value sequence as a reference, the baseline data in the dynamic physiological baseline under the same physiological state as the current acquisition time are time-aligned and resampled to obtain the prefrontal cortex blood oxygen saturation baseline sequence corresponding to the standard value sequence at the time point. The absolute deviation of the blood oxygen saturation value sequence in the standard prefrontal cortex is calculated at each time point from the baseline blood oxygen saturation value sequence in the prefrontal cortex to generate the absolute deviation sequence of blood oxygen saturation. The same operation was performed on the remaining data in the physiological baseline to obtain the absolute deviation sequence of all data in the physiological baseline. The cumulative absolute deviation is obtained by performing an exponential decay time integral operation within a fixed-duration sliding window on the absolute deviation sequence of a single data point, and the average of the cumulative absolute deviations of all data points is used as the comprehensive deviation index. The comprehensive deviation index is input into the oxygen supply compensation model in the medical and health big data sharing platform, and the oxygen supply compensation control vector of the patient is output.

8. The method for dynamic adjustment of oxygen supply pressure based on a medical cabin as described in claim 1, characterized in that, The method of adjusting the oxygen supply pressure compensation command of the disturbed medical oxygen chamber based on the sudden change in oxygen pressure at the inlet of the oxygen supply pipeline and the compensation control vector, and generating a compensation command set for the disturbed medical oxygen chamber, includes: Based on the transient fluctuation amplitude of the sudden change in inlet oxygen pressure, the compensation control vector is dynamically gain-corrected to obtain the corrected compensation control vector of the disturbed medical oxygen chamber. Based on the modified compensation control vector, the oxygen supply pressure command of the interfered medical cabin is adjusted to generate a compensation command set for the interfered medical cabin.

9. The method for dynamically adjusting oxygen supply pressure based on a medical cabin as described in claim 1, characterized in that, The step of feeding back the patient's physiological parameters and the oxygen supply pressure response data of the medical cabin after the execution of the compensation instruction set to the medical and health big data sharing platform includes: Collect real-time physiological parameters of the patient and oxygen supply pressure response data of the medical cabin after the compensation command is executed; The patient's real-time physiological parameters and the oxygen supply pressure response data of the medical cabin are integrated into a data structure to generate a feedback data package containing information on the correlation between the patient's physiological state and oxygen supply pressure. The feedback data packet is sent to the medical and health big data sharing platform in a standardized format by calling the real-time data transmission interface provided by the medical and health big data sharing platform.

10. A dynamic oxygen supply pressure regulation system based on a medical cabin, used to implement the dynamic oxygen supply pressure regulation method based on a medical cabin as described in any one of claims 1-9, characterized in that, The system includes: Data acquisition module: used to acquire real-time oxygen supply pressure of the medical cabin, physiological parameter benchmarks of patients of the same age and with the same disease, and real-time physiological data of patients in the cabin from the medical and health big data sharing platform; Risk Disturbance Module: Used to perform data stream analysis on the real-time oxygen supply pressure of the entire medical cabin oxygen supply network based on graph neural network algorithm, predict the pressure disturbance that will occur in the network hundreds of milliseconds to several seconds in advance, and identify the medical cabins where the pressure disturbance is about to occur as risk medical cabins; Interference identification module: used to monitor the oxygen pressure at the inlet of the oxygen supply pipe of other medical cabins that are in the same oxygen supply network as the risk medical cabin. If the oxygen pressure at the inlet of the oxygen supply pipe of a certain medical cabin suddenly changes beyond the preset inlet disturbance threshold, the medical cabin is identified as the interfered medical cabin. Baseline calibration module: used to perform empirical Bayesian calibration of physiological parameters of patients of the same age and with the same disease based on the real-time physiological data of the patient, so as to obtain a dynamic physiological baseline that adaptively adjusts with the patient's sleep, activity, mood and other states; Compensation control module: used to perform sliding window integration based on the patient's physiological data and the dynamic physiological baseline to obtain a comprehensive deviation index between the two, and to determine the patient's compensation control vector based on the comprehensive deviation index; Command adjustment module: used to adjust the oxygen supply pressure compensation command of the disturbed medical oxygen chamber based on the sudden change in oxygen pressure at the inlet of the oxygen supply pipeline and the compensation control vector, and generate a compensation command set for the disturbed medical oxygen chamber; Data feedback module: used to feed back the patient's physiological parameters and medical cabin oxygen supply pressure response data after the execution of the compensation instruction set to the medical and health big data sharing platform.

Images