Oxygen generator remote control system based on Internet of Things and method thereof

Abstract

The invention discloses an oxygenerator remote control system and method based on the Internet of Things, and belongs to the technical field of intelligent medical equipment and remote health management, and the method comprises the steps: S1, collecting multi-dimensional physiological data of a user in real time through a multi-source physiological signal collection module, a heart rate variability index and a body movement signal; s2, based on the multi-dimensional physiological data, processing is performed through a preset physiological state prediction model, and a unified physiological stress index is solved; s3, determining a predictive oxygen supply flow rate value in combination with the physiological stress index and a preset user basic oxygen supply flow rate; s4, according to the predictive oxygen supply flow velocity value, a control signal for a physical execution component of the oxygen generator is generated so as to dynamically adjust oxygen supply parameters, and conversion from passive compensation to active prediction is achieved.

Description

Technical Field

[0001] This invention relates to the field of intelligent medical devices and remote health management, specifically to a remote control system and method for an oxygen concentrator based on the Internet of Things. Background Technology

[0002] In current oxygen therapy technology, the control method of traditional oxygen concentrators mainly relies on a constant oxygen supply flow rate or manual adjustment by the user. This open-loop control method has inherent lag and non-personalization defects. These limitations prevent the equipment from providing optimal oxygen support according to the user's real-time physiological changes, affecting the effectiveness and safety of oxygen therapy.

[0003] The aforementioned situation and shortcomings are mainly due to the limitations of the control strategy.

[0004] The response is delayed and lacks foresight. Traditional oxygen supply adjustments usually occur after the user has already shown physiological discomfort such as a decrease in blood oxygen saturation. This is a passive compensatory behavior rather than an active preventive measure, missing the opportunity to intervene before the user shows obvious symptoms of hypoxia.

[0005] A fixed oxygen supply flow rate cannot adapt to the dynamic changes in oxygen demand of users in different activity states, such as rest, sleep, and light activity. This may result in insufficient oxygen supply when high flow is needed, or excessive oxygen supply when demand is low, leading to waste.

[0006] As a result, traditional oxygen therapy methods struggle to achieve automated, proactive, and personalized precise oxygen supply, and cannot form an intelligent health management system that can continuously learn and adapt to individual user differences.

[0007] The information disclosed in the background section above is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0008] The purpose of this invention is to provide a remote control system and method for an oxygen concentrator based on the Internet of Things, so as to solve the problems mentioned in the background art.

[0009] The technical solution of the present invention includes: S1 collects multi-dimensional physiological data of users in real time through a multi-source physiological signal acquisition module. The multi-dimensional physiological data includes blood oxygen saturation signal, respiratory rate signal, heart rate variability index and body movement signal. S2, based on multi-dimensional physiological data, is processed through a preset physiological state prediction model to calculate a unified physiological stress index; S3, combining the physiological stress index with the preset user baseline oxygen supply flow rate, determines the predictive oxygen supply flow rate value; S4 generates control signals for the physical actuators of the oxygen generator based on the predicted oxygen supply flow rate value, so as to dynamically adjust the oxygen supply parameters. S5 uploads a data package containing multi-dimensional physiological data and predictive oxygen flow rate values ​​to the cloud-based collaborative computing module for iterative optimization of the physiological state prediction model.

[0010] Preferably, the physiological state prediction model processing includes: Each physiological signal in the multidimensional physiological data is standardized based on its own preset reference rate of change to obtain a standardized change quantity; the standardized change quantity is then calculated using a nonlinear weighted fusion function to obtain a unified physiological stress index.

[0011] Preferably, determining the predictive oxygen supply flow rate includes: Based on the physiological stress index, the user's baseline oxygen supply flow rate and the preset maximum adjustment range, the initial oxygen supply flow rate is calculated; the initial oxygen supply flow rate is compared with the preset minimum safe oxygen supply flow rate, and the larger value is taken as the predictive oxygen supply flow rate value.

[0012] Preferably, the cloud-based collaborative computing module iteratively optimizes the physiological state prediction model, including: A loss function is constructed in the cloud-based collaborative computing module. The loss function is used to characterize the difference between the oxygen supply flow rate set by the system and the optimal flow rate required to maintain the user's physiological state. The loss function is optimized and solved using machine learning algorithms to calculate personalized weight combinations. The personalized weight combinations are then distributed to update the weight coefficients in the physiological state prediction model.

[0013] Preferably, the initial values ​​of the weight coefficients in the physiological state prediction model are derived from a general model formed by statistical analysis of large-scale clinical data.

[0014] Preferred options also include: The safety redundancy management module monitors the data quality of multi-dimensional physiological data and the user's core vital signs in real time. If the data quality is lower than the preset minimum effective threshold or the core vital signs are lower than the preset critical safety threshold, the control process based on the predictive oxygen supply flow rate value is interrupted, and the preset conservative oxygen supply mode is executed.

[0015] Preferably, the core vital sign is blood oxygen saturation; the conservative oxygen supply mode is to supply oxygen at a preset, constant and safe conservative oxygen supply flow rate.

[0016] A remote control system for an oxygen concentrator based on the Internet of Things (IoT) includes: The multi-source physiological signal acquisition module is used to collect multi-dimensional physiological data of users in real time, including blood oxygen saturation signal, respiratory rate signal, heart rate variability index and body movement signal; The edge computing module includes: The data processing unit is used to process multi-dimensional physiological data through a preset physiological state prediction model to calculate a unified physiological stress index. The decision unit is used to determine the predictive oxygen flow rate value by combining the physiological stress index with the preset user baseline oxygen flow rate. The adaptive oxygen supply control module is used to generate control signals for the physical actuators of the oxygen generator based on the predicted oxygen supply flow rate value. The cloud-based collaborative computing module receives data packets uploaded by the edge computing module, optimizes the physiological state prediction model, and sends the optimized model parameters back to the edge computing module.

[0017] Preferred options also include: The safety redundancy management module is used to monitor the data quality output by the multi-source physiological signal acquisition module and the user's core vital signs in real time. In response to the data quality being lower than the preset minimum effective threshold or the core vital signs being lower than the preset critical safety threshold, the adaptive oxygen supply control module is instructed to execute the preset conservative oxygen supply mode.

[0018] This invention provides an improved Internet of Things-based remote control system and method for oxygen concentrators, which has the following improvements and advantages compared with the prior art: 1. This invention realizes the transformation from passive compensation to active prediction; the core of the system is not merely responding to the current value of physiological indicators, but predicting future oxygen demand by analyzing their dynamic changing trends; the physiological state prediction model standardizes the rate of change of multi-dimensional physiological signals through the following calculations. 2. This invention establishes an oxygen supply control law that balances adaptive adjustment and absolute safety; based on the calculated physiological stress index, the system achieves precise control of the physical actuators of the oxygen generator through the following predictive oxygen supply flow rate calculation. 3. This invention introduces continuous learning and personalized adaptation evolutionary capabilities; through iterative optimization of the physiological state prediction model by the cloud collaborative computing module, the system surpasses the generic model at the factory; the cloud collaborative computing module analyzes long-term accumulated data packets, uses machine learning algorithms to optimize the loss function, and continuously adjusts the weight coefficients in the model; this mechanism enables the control strategy to gradually evolve from a universal initial state into a personalized model that highly matches the specific user's physiological response pattern; this is the essential difference between this and the unchanging control logic in existing technologies, achieving long-term, efficient, and precise oxygen therapy. 4. An independent safety redundancy guarantee has been established; the existence of the safety redundancy management module provides the system with fault safety guarantee independent of the main control logic; it continuously monitors data quality and core vital signs, and can forcibly interrupt predictive control and switch to the preset conservative oxygen supply mode when unreliable signals are detected or the user's physiological state deteriorates rapidly; this ensures that in extreme scenarios such as sensor failure or sudden emergency situations of users, the system's first reaction is to ensure safety rather than pursue the optimal solution, which significantly improves the reliability of the entire system and its credibility in practical applications. Attached Figure Description

[0019] The present invention will be further explained below with reference to the accompanying drawings and embodiments: Figure 1 This is a flowchart of a remote control method for an oxygen concentrator based on the Internet of Things according to the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0021] Example 1: Please see Figure 1 This invention provides a remote control method for an oxygen concentrator based on the Internet of Things, comprising: S1 collects multi-dimensional physiological data of users in real time through a multi-source physiological signal acquisition module. The multi-dimensional physiological data includes blood oxygen saturation signal, respiratory rate signal, heart rate variability index and body movement signal. S2, based on multi-dimensional physiological data, is processed through a preset physiological state prediction model to calculate a unified physiological stress index; S3, combining the physiological stress index with the preset user baseline oxygen supply flow rate, determines the predictive oxygen supply flow rate value; S4 generates control signals for the physical actuators of the oxygen generator based on the predicted oxygen supply flow rate value, so as to dynamically adjust the oxygen supply parameters. S5 uploads a data packet containing multi-dimensional physiological data and predictive oxygen flow rate values ​​to the cloud collaborative computing module for iterative optimization of the physiological state prediction model. In this embodiment, an IoT-based remote control method for an oxygen concentrator aims to achieve precise and forward-looking oxygen supply control to cope with the dynamic changes in the user's physiological state. The complete process of the method is designed as a continuously running closed-loop system, which transforms the user's real-time physiological state into precise oxygen supply adjustment commands through the synergistic effect of multiple functional steps. The multi-source physiological signal acquisition module collects multi-dimensional physiological data from the user in real time. The purpose of this module is to acquire, synchronously and in real time, various indicators that comprehensively reflect the user's physiological load. In this embodiment, the module is configured to continuously collect four key physiological data streams, including: blood oxygen saturation signal. It reflects the oxygenation level of the user's blood; respiratory rate signal It characterizes the user's respiratory system load; heart rate variability index This reveals the regulatory state of the autonomic nervous system and body movement signals. This quantifies the intensity of users' physical activities; these signals are collected synchronously to ensure the time alignment of subsequent data fusion analysis. Based on multi-dimensional physiological data, a unified physiological stress index is calculated through a pre-defined physiological state prediction model. This physiological state prediction model is a core algorithm module designed to transform the multi-source, heterogeneous raw physiological signals collected in previous steps into a standardized index that can uniformly represent the user's overall physiological stress level. The pre-defined model is not arbitrary; its initial mathematical structure and parameters are derived from statistical analysis and abstraction of the relationship between physiological signals and oxygen demand changes in numerous clinical cases. In this embodiment, the processing is executed within the edge computing module, using a series of mathematical operations to convert the real-time input signal stream... , , , This allows for the calculation of a single, dimensionless physiological stress index. ; By combining the physiological stress index with the preset user baseline oxygen supply flow rate, a predictive oxygen supply flow rate value is determined. The purpose of this step is to transform the abstract physiological stress index into a concrete, executable physical control quantity. The user baseline oxygen supply flow rate is defined as... This is a pre-set baseline oxygen supply value based on the user's static physiological information, such as age, weight, and medical history, designed to meet the user's basic oxygen demand at rest. This step receives the stress index output by the physiological state prediction model. And based on the exponential relationship of the base flow rate Dynamic adjustments are made to calculate a forward-looking predictive oxygen supply flow rate. ; Based on the predicted oxygen supply flow rate, control signals are generated for the physical actuators of the oxygen generator to dynamically adjust the oxygen supply parameters; the purpose of this step is to translate the decision results of the upper-level algorithm into direct control of the physical equipment; an adaptive oxygen supply control module receives the predicted oxygen supply flow rate from the edge computing module. The value serves as a high-level instruction; based on the instruction, the control module converts the numerical instruction into specific electronic control signals, such as adjusting the opening of the flow control valve or changing the operating speed of the compressor; in this way, the actual output oxygen flow of the oxygen generator is precisely and dynamically adjusted to match the predicted oxygen demand. Data packets containing multi-dimensional physiological data and predictive oxygen delivery flow rates are uploaded to the cloud-based collaborative computing module for iterative optimization of the physiological state prediction model. To achieve the system's personalized adaptive capabilities, the edge computing module periodically assembles data packets. These data packets contain not only the original input signals... , , , and the final control decision It can also include intermediate calculation results of the model and short-term physiological feedback after control is implemented; these data packets are uploaded to the cloud collaborative computing module, which uses its computing and storage resources to perform long-term, user-specific optimization of the physiological state prediction model. The method disclosed in this embodiment constructs a complete closed loop from real-time physiological data acquisition, edge-end intelligent prediction, device-end precise control to cloud-based personalized model optimization. It overcomes the limitations of traditional oxygen concentrators that rely on fixed oxygen supply or manual adjustment, and realizes the automation, foresight, and personalization of oxygen supply strategies. By predicting changes in the user's oxygen demand in real time, the system can proactively intervene before the user experiences obvious hypoxia symptoms, improving the response speed and matching degree of oxygen supply decisions, thereby optimizing the adaptability of the oxygen therapy process, and forming an intelligent health management system that can continuously learn and adapt to individual differences in users. The beneficial effects of this invention are rooted in the deep coupling between its overall architecture and internal algorithm, achieving a fundamental leap compared to existing technologies; Existing oxygen concentrators typically operate on a constant flow rate or rely on manual adjustment by the user. The inherent drawbacks of open-loop control are its lag and lack of personalization. Oxygen supply adjustments often occur after the user has already experienced physiological discomfort such as decreased blood oxygen saturation, making it a passive compensatory behavior rather than proactive prevention. Furthermore, a fixed flow rate cannot adapt to the dynamic changes in oxygen demand under different activity states, such as rest, sleep, and light activity, which may lead to insufficient oxygen supply or excessive waste.

[0022] Example 2 Physiological state prediction model processing includes: Each physiological signal in the multidimensional physiological data is standardized based on its own preset reference rate of change to obtain a standardized change; the standardized change is then calculated using a nonlinear weighted fusion function to obtain a unified physiological stress index. In this embodiment, the internal computational logic of the physiological state prediction model is limited; the core is to process the multi-dimensional raw signal stream into a unified physiological stress index through a two-step process. The physiological signals in the multidimensional physiological data are standardized based on their respective preset reference rates of change to obtain standardized changes. The purpose of this step is to eliminate computational barriers caused by differences in physical units and numerical ranges between different physiological signals, enabling them to be fused within a unified framework. Using blood oxygen saturation signals as an example... For example, its standardized variation The calculation method is designed as follows: , in, Through time series data The instantaneous rate of change obtained by differential calculation reflects the drastic degree of blood oxygenation change; This is a preset reference rate of change in blood oxygen saturation, whose value is set based on the statistical understanding of the significance of changes in blood oxygen in clinical medicine. For example, 1% / second serves as a benchmark to measure the relative magnitude of the current rate of change. Similarly, the respiratory rate signal... Heart rate variability index and body movement signals They also each used their own preset reference rate of change. Perform the same standardization process to obtain their respective standardized changes. ;in, This is the standardized variation of the respiratory rate signal; This represents the standardized change in the heart rate variability index. Standardized change in body movement signal; reference rate of change in respiratory rate. It can be set to 3 times / minute / second; The standardized changes are calculated using a nonlinear weighted fusion function to obtain a unified physiological stress index. The purpose of this step is to fuse multiple single-dimensional physiological change information into a comprehensive index that can represent the user's overall physiological load. The fusion function is designed as follows: , in, This is the instantaneous physiological stress index at time point t, and its value is expressed by the hyperbolic tangent function. It is naturally constrained within the interval (-1, 1), making it a dimensionless standardized exponent; t is the time point; Physiological stress index Weighting coefficients : Standardized change of each signal; i: Subscript used to iterate over each signal; By employing this two-step processing approach of standardization followed by fusion, this method can scientifically integrate physiological signals from different sources, units, and scales into a single stress index with clear physiological significance. Standardization ensures the comparability of different signals, while nonlinear weighted fusion can simulate complex physiological responses under the synergistic effect of various physiological indicators. This provides a high-quality, high-information-density decision-making basis for the subsequent accurate calculation of oxygen supply flow rate, improving the accuracy and robustness of the entire system's predictions. This represents a shift from passive compensation to active prediction. The core of the system is not merely responding to the current values ​​of physiological indicators, but rather predicting future oxygen demand by analyzing their dynamic trends. The physiological state prediction model standardizes the rates of change of multi-dimensional physiological signals through the following calculations: Standardized change quantity: instantaneous rate of change of any physiological signal; preset reference rate of change. ,in, Standardized variation Any physiological signal, Instantaneous rate of change Preset reference rate of change;

[0023] The physical meaning of the formula lies in unifying physiological signals of different physical units, such as the percentage change rate of blood oxygen saturation and the rate of change of respiratory rate, into a dimensionless relative degree of change; a preset reference rate of change is provided. The working principle is based on typical rates of change that are clinically significant, derived from large-scale clinical data statistics. It provides an objective benchmark for assessing the urgency or importance of current physiological signal changes. This rate-of-change analysis enables the system to capture early signals of physiological load before a user's core indicators, such as blood oxygen saturation, deteriorate significantly. This invention constructs a multi-dimensional, high-fidelity assessment of a user's physiological state. Compared to existing technologies that rely solely on a single blood oxygen signal, this solution integrates four multi-dimensional physiological data—blood oxygen saturation signal, respiratory rate signal, heart rate variability index, and body movement signal—through a non-linear weighted fusion function. Standardized variation of each signal in the weighting coefficients of the physiological stress index ,in, Physiological stress index Weighting coefficients : Standardized change of each signal; i: Subscript used to iterate over each signal; The logic of the formula can be derived from the instructions, receiving multiple standardized variations output by the preceding formula. As input; its core idea is that a single physiological indicator often cannot fully reflect the complex state of the human body, while multi-signal fusion can provide a more robust and comprehensive assessment; hyperbolic tangent function By applying this method, the weighted sum of multiple signals is mapped to a (-1, 1) interval, generating a unified physiological stress index. The index not only characterizes the degree of stress, but its positive or negative sign also reflects the increase or decrease of physiological load, providing a refined and rich input for subsequent oxygen supply decisions.

[0024] Example 3 Determining the predictive oxygen supply flow rate value includes: Based on the physiological stress index, the user's basic oxygen supply flow rate and the preset maximum adjustment range, the initial oxygen supply flow rate is calculated; the initial oxygen supply flow rate is compared with the preset minimum safe oxygen supply flow rate, and the larger value is taken as the predictive oxygen supply flow rate value. In this embodiment, the calculation process for determining the predictive oxygen supply flow rate is limited, and key safety boundary conditions are introduced. The underlying logic is that an initial oxygen supply flow rate is calculated based on the physiological stress index, and this initial flow rate is compared with a preset minimum safe oxygen supply flow rate; the larger value is taken as the final predictive oxygen supply flow rate value. This process ensures that the system output is always within an absolutely safe range; a complete predictive oxygen supply flow rate... The calculation formula is designed as follows: , in, It is a system prediction that should occur in the future. The target oxygen supply flow rate set within the time period, in liters per minute; Predicted time It is a configurable parameter preset according to the system's responsiveness and stability requirements. Its value is designed to effectively predict physiological changes while avoiding excessively frequent control that could lead to system oscillations. In practical applications, The typical value range can be set between 5 seconds and 30 seconds. Shorter... A certain value can make the system respond more quickly, while a longer value... The value helps improve the stability of control; It is the user's basic oxygen supply flow rate, which is a preset benchmark value based on the user's static information; This is the preset maximum adjustment range, representing the maximum positive or negative change allowed above the base flow rate. It is the maximum allowable adjustment range set according to the user's specific health condition and medical advice, such as the basic oxygen supply flow rate. 50%; The immediate physiological stress index is obtained from the previous step; the right side of the equation, and The dimensions of all are units of flow velocity, while Since it is dimensionless, The dimension of the quantity is flow velocity, and And the left side of the equals sign Maintaining consistent dimensions; the most crucial design element is the introduction of… functions and parameter; This refers to the preset minimum safe oxygen supply flow rate, which is a hard lower limit set in advance according to generally accepted medical safety standards, such as 0.5 liters / minute, and its function is to form a safety barrier. This calculation method achieves unity on two levels in terms of functionality; on the one hand, it unifies oxygen supply flow rate with physiological stress index. The correlation enables dynamic, real-time, and precise adjustment of oxygen supply; on the other hand, it achieves this by forcing a function to take a larger value and setting a minimum safe oxygen supply flow rate. The introduction of this feature sets an insurmountable safety lower limit for the system's output; this ensures that regardless of how stable the user's physiological state is, the output remains within a safe range. Even with a relatively large negative value, the oxygen supply output by the system will never be lower than the safety threshold required to maintain basic life support, thus logically eliminating the risk of oxygen interruption or insufficiency due to excessively stable physiological state, and greatly enhancing the safety of the system. An oxygen supply control law that balances adaptive adjustment and absolute safety was established. Based on the calculated physiological stress index, the system achieves precise control of the physical actuators of the oxygen generator through the following predictive oxygen supply flow rate calculation: Predictive oxygen supply flow rate, minimum safe oxygen supply flow rate, user baseline oxygen supply flow rate, maximum adjustment range, physiological stress index. ,in, Predictive oxygen supply flow rate Minimum safe oxygen supply flow rate User-based basic oxygen supply flow rate Maximum adjustment range Physiological stress index; The derivation of the formula is directly related to the aforementioned physiological stress index. Its progress lies in the establishment of a dual-guarantee mechanism; on the one hand, the oxygen supply flow rate... Able to supply oxygen around the user's basic oxygen flow rate According to the physiological stress index Linear floating adjustment enables real-time and adaptive oxygen supply; on the other hand, Function and preset minimum safe oxygen supply flow rate The combination of these elements forms an insurmountable safety baseline; preset parameters and The settings are based on medical standards and individual user circumstances, ensuring that no matter how the prediction model makes its decisions, the actual oxygen supply will never fall below the critical threshold for maintaining basic life needs, thus eliminating the risk of oxygen supply interruption caused by the algorithm.

[0025] Example 4 The cloud-based collaborative computing module iteratively optimizes the physiological state prediction model, including: A loss function is constructed in the cloud-based collaborative computing module. The loss function is used to characterize the difference between the oxygen supply flow rate set by the system and the optimal flow rate required to maintain the user's physiological state. The loss function is optimized and solved using machine learning algorithms to calculate a personalized weight combination. The personalized weight combination is then distributed to update the weight coefficients in the physiological state prediction model. The initial values ​​of the weight coefficients in the physiological state prediction model are derived from a general model formed by statistical analysis of large-scale clinical data; In this embodiment, the method of how the cloud-based collaborative computing module iteratively optimizes the physiological state prediction model and the source of the initial values ​​of the weight coefficients in the model are explained; these two features together constitute a mechanism for continuous learning and personalized adaptation. To ensure the model's effectiveness in the initial stage, the initial values ​​of the weight coefficients in the physiological state prediction model are not randomly set, but rather derived from a general model formed through statistical analysis of large-scale clinical data. Before the system is deployed to any specific user, a basic, universal physiological state prediction model has been constructed. This model is built by mining and analyzing massive amounts of anonymized clinical case data, abstracting a universal mathematical structure that can characterize the relationship between physiological load and changes in oxygen demand. The core adjustable parameter in the model is the weight coefficient. The initial value was obtained through this statistical analysis and represents the average physiological response pattern of the standard population. Once the system is running for a specific user, the cloud-based collaborative computing module is activated to iteratively optimize the physiological state prediction model; the technical logic of this process is as follows: A loss function is constructed in the cloud-based collaborative computing module. The loss function is a mathematical expression aimed at quantifying the predictive performance of the current model. In this embodiment, the function is designed to characterize the relationship between the system-set oxygen flow rate and the ability to maintain the user's blood oxygen saturation over a past period. The difference between the optimal flow rates required in the ideal range; The ideal range can be preset to the medically recognized healthy blood oxygen range, such as the lower limit of ideal blood oxygen saturation. The set value is 95%; while the optimal flow rate is... It is not an absolute prior value, but rather defined by analyzing historical data windows, indicating the ability to measure a user's blood oxygen saturation within that window. Maintain at the lower limit of the ideal range The minimum oxygen supply flow rate required above is determined by the cloud module through analysis of historical data windows, retrospectively identifying the window within which the user's blood oxygen saturation S(t) can be maintained within the lower limit of the ideal range. The theoretically optimal flow rate required above It was used as the optimization objective to evaluate the actual predicted oxygen supply flow rate at that time. Performance; loss function It is constructed as a composite function, aiming to simultaneously achieve both economic efficiency and safety in oxygen supply, and can be designed in the following form: , Where N: the total number of time points in the historical data window; It is the oxygen supply flow rate predicted by the model. It is the optimal flow velocity for the same period determined based on historical data; the first item The first term is the mean squared error term, used to penalize the deviation between the predicted flow rate and the optimal flow rate, thereby improving the accuracy and economy of oxygen supply; the second term... It is a penalty item if and only if the actual blood oxygen value Below the ideal lower limit When the value is positive, it is used to penalize predictive behaviors that lead to substandard blood oxygen levels, in order to ensure user safety. A coefficient with dimensions, in units of . , This is a hyperparameter determined during model training using methods such as cross-validation. It adjusts the strength of the low blood oxygen penalty term; a larger value indicates a more conservative model, prioritizing ensuring blood oxygen levels do not fall below a threshold. The loss function is adjusted using machine learning algorithms. By minimizing the solution, the optimized personalized weight combination can be obtained; The loss function is optimized using machine learning algorithms to calculate a personalized weight combination. The cloud module utilizes its computing resources to run machine learning algorithms such as backpropagation or reinforcement learning on the user's exclusive historical data, aiming to minimize the loss function and adjust the weight coefficients. The process involves iterative adjustments and optimizations; the final output is a personalized weight combination best suited to the specific physiological response pattern of a particular user. ; Personalized weight combinations are distributed to update the weight coefficients in the physiological state prediction model; this optimized parameter set is pushed from the cloud back to the user's corresponding edge computing module to replace the old weight coefficients in the model. By providing a universal initial value for the model weights derived from large-scale clinical statistics, the system is ensured to have reasonable and safe basic performance from the initial deployment stage. More importantly, the subsequent cloud optimization and parameter distribution mechanism constructs a closed loop of continuous learning. This mechanism enables the control model to gradually transform from a general model to a highly personalized adaptive model, with its decision logic becoming increasingly aligned with the user's unique physiological characteristics. This personalized adaptive capability is the key to achieving long-term, efficient, and precise oxygen supply management, and it is also the core advantage of this technical solution compared to static and non-personalized control strategies.

[0026] Example 5 Also includes: The safety redundancy management module monitors the data quality of multi-dimensional physiological data and the user's core vital signs in real time. If the data quality is lower than the preset minimum effective threshold or the core vital signs are lower than the preset critical safety threshold, the control process based on the predictive oxygen supply flow rate value is interrupted and the preset conservative oxygen supply mode is executed. The core vital sign is blood oxygen saturation; the conservative oxygen supply mode is to supply oxygen at a preset, constant and safe conservative oxygen supply flow rate. In this embodiment, a parallel security mechanism is further defined; the method also includes real-time monitoring of the data quality of multi-dimensional physiological data and the user's core vital signs through a security redundancy management module. The safety redundancy management module is a functionally independent monitoring unit. Its purpose is to continuously monitor the system's operational status and the user's safety baseline, serving as an independent verification mechanism. The module works in parallel, continuously monitoring two key states: one is the data quality of the signals output by the multi-source physiological signal acquisition module. ; Data quality It is a comprehensive quantitative indicator, with values ​​normalized to the range of [0, 1], where 1 represents the best quality; The calculation can take into account multiple sub-indicators, for example: , Q: A comprehensive quantitative data quality indicator; The quality of the photoplethysmography (PPG) signal itself can be evaluated based on signal-to-noise ratio (SNR) parameters such as the perfusion index. When the signal is weak or the noise is excessive... The value decreased; Represents body movement signals The degree of motion interference reflected can be expressed by the formula The calculation is performed, where max() is a function that takes the larger of the two values; M(t) is the body motion signal; and k is a preset proportional coefficient, which is adjusted when the user's activity is intense. Enlargement, leading to The value decreases, where k is based on the body movement signal. The preset normalization coefficient for the measurement range should be chosen to ensure that when the body motion reaches a level that affects signal quality, the value of k·M(t) is close to or greater than 1. For example, if The effective range is 0-50, and k=0.02 can be set. By taking the minimum value of each sub-index, it is ensured that the overall data quality is only considered when the quality of all key signals meets the standard. Only then is it considered valid; Secondly, the user's core vital signs; according to a further definition in claim 7, the core vital signs specifically refer to blood oxygen saturation. The absolute value; The core function of the module is to interrupt the control process based on the predictive oxygen supply flow rate and instruct the execution of the preset conservative oxygen supply mode when the data quality is lower than the preset minimum effective threshold or the core vital signs are lower than the preset critical safety threshold. The trigger condition is set as follows: Signal quality failure: Detected data quality indicators Below a preset minimum effective threshold Threshold The basis for this determination is the industry standard that ensures the signal has a sufficient signal-to-noise ratio to support effective computation, and prevents model input errors due to problems such as sensor detachment and signal interference. Critical vital signs: Blood oxygen saturation detected The absolute value is lower than a preset critical safety threshold. Threshold It is a critical value set according to medical safety standards, such as 88%. Once a user's blood oxygen level reaches the line, it indicates that the user may be in a dangerous state. Once any of the above conditions are triggered, the security redundancy management module is immediately activated, forcibly interrupting the predictive control process led by the edge computing module and ignoring its output. Command; simultaneously, the module directly sends a command to the adaptive oxygen supply control module to execute a preset conservative oxygen supply mode; as defined in claim 7, the conservative oxygen supply mode is to supply oxygen at a preset, constant, and safe conservative oxygen supply flow rate; the flow rate is defined as Its value is preset according to recognized medical emergency standards, aiming to provide users with a stable, sufficient and risk-free oxygen supply; for example, it can be preset to 3 liters / minute. The introduction of a safety redundancy management mechanism adds a crucial fault-safe layer to the entire intelligent control system. It can effectively address two core risks: unreliable input data due to hardware or environmental factors, and acute deterioration of the user's physiological state. In these situations, the system can decisively switch from a complex prediction mode that pursues the optimal outcome to a simple and conservative mode that ensures safety. This design greatly improves the system's reliability and application security in real and complex environments, ensuring that the user's life safety is always the highest priority in system operation.

[0027] Example 6 A remote control system for an oxygen concentrator based on the Internet of Things (IoT) includes: The multi-source physiological signal acquisition module is used to collect multi-dimensional physiological data of users in real time, including blood oxygen saturation signal, respiratory rate signal, heart rate variability index and body movement signal; The edge computing module includes: The data processing unit is used to process multi-dimensional physiological data through a preset physiological state prediction model to calculate a unified physiological stress index. The decision unit is used to determine the predictive oxygen flow rate value by combining the physiological stress index with the preset user baseline oxygen flow rate. The adaptive oxygen supply control module is used to generate control signals for the physical actuators of the oxygen generator based on the predicted oxygen supply flow rate value. The cloud-based collaborative computing module is used to receive data packets uploaded by the edge computing module, optimize the physiological state prediction model, and send the optimized model parameters back to the edge computing module. In this embodiment, an IoT-based remote control system for an oxygen concentrator is provided. The system is configured to execute any of the aforementioned methods, and its physical and logical structure includes the following interconnected modules: The multi-source physiological signal acquisition module is the data input terminal of the system. Its function is to collect multi-dimensional physiological data of users in real time. It integrates or connects sensors for measuring indicators such as blood oxygen saturation, respiratory rate, heart rate variability and body movement, and transmits the collected raw signal streams, namely blood oxygen saturation signal, respiratory rate signal, heart rate variability index and body movement signal, to the edge computing module in real time and synchronously. The edge computing module, which is the system's local processing and decision-making module, is deployed on the user side to ensure low-latency real-time response; it contains two key units: The data processing unit is used to process multi-dimensional physiological data through a preset physiological state prediction model to calculate a unified physiological stress index; the unit receives signals from the acquisition module and performs the standardization and nonlinear weighted fusion calculation detailed in claim 2 above. The decision unit is used to determine the predictive oxygen flow rate value by combining the physiological stress index with the preset user baseline oxygen flow rate; the unit receives the stress index calculated by the data processing unit and performs the oxygen flow rate calculation including the safety lower limit as detailed in claim 3 above. The adaptive oxygen supply control module is the physical execution end of the system. Its function is to generate control signals for the physical execution components of the oxygen generator based on the predicted oxygen supply flow rate value. It receives the predicted oxygen supply flow rate command from the decision unit of the edge computing module and converts it into electrical signals that can directly drive the internal components of the oxygen generator, such as flow valves and compressors, thereby realizing precise dynamic adjustment of the output oxygen flow rate. The cloud-based collaborative computing module is the system's remote computing and model optimization module. It establishes bidirectional communication with the edge computing module through the Internet of Things. Its function is to receive data packets uploaded by the edge computing module, optimize the physiological state prediction model, and send the optimized model parameters back to the edge computing module. As described in claim 4, it analyzes long-term accumulated user data to perform personalized training on the prediction model, thereby achieving continuous evolution of the control strategy. This system constructs a well-defined and collaborative whole by modularizing its functions. The edge computing module enables local real-time data processing and rapid decision-making, meeting the high immediacy requirements of medical scenarios. Meanwhile, the introduction of the cloud-based collaborative computing module utilizes the centralized computing and storage capabilities of the cloud to solve the problem of limited computing power at the edge, enabling complex, individual-specific model optimization and continuous learning. This edge-cloud collaborative system architecture balances real-time response and deep intelligence, making the entire remote control system both highly efficient and highly personalized and adaptive.

[0028] Example 7 Also includes: The safety redundancy management module is used to monitor the data quality output by the multi-source physiological signal acquisition module and the user's core vital signs in real time. In response to the data quality being lower than the preset minimum effective threshold or the core vital signs being lower than the preset critical safety threshold, the adaptive oxygen supply control module is instructed to execute the preset conservative oxygen supply mode. In this embodiment, a key security component is further integrated; the system also includes: The safety redundancy management module is a parallel monitoring unit with the highest priority in the system architecture. Its function is to monitor the data quality output by the multi-source physiological signal acquisition module and the user's core vital signs in real time. When the system is running, the module independently and continuously analyzes the validity of the input signals and the user's key physiological indicators, namely blood oxygen saturation. The module is configured to take immediate intervention measures in response to two preset abnormal situations: data quality falling below a preset minimum effective threshold or core vital signs falling below a preset critical safety threshold. It will exercise the highest control authority and instruct the adaptive oxygen supply control module to execute a preset conservative oxygen supply mode. This means that it will bypass and suspend the predictive control instructions of the edge computing module and instead enforce a preset, constant, and safe oxygen supply flow rate. By adding a dedicated security redundancy management module to the system architecture, this invention elevates security assurance from a methodological process to a clearly defined system component. This design provides functional independence, ensuring that even if the main control logic or edge computing module makes inappropriate judgments due to data errors, there is still an independent monitoring and arbitration mechanism to detect problems and execute contingency plans. This greatly enhances the system's robustness and fault tolerance, enabling the entire remote control system to exhibit higher reliability and security when facing real-world challenges such as sensor failures, signal interference, or sudden changes in user health. By solidifying the security monitoring function into an independent system module, it provides a robust systemic guarantee for user safety. It incorporates continuous learning and personalized adaptation capabilities; through iterative optimization of the physiological state prediction model via a cloud-based collaborative computing module, the system surpasses the factory-standard model; the cloud-based collaborative computing module analyzes long-accumulated data packets and utilizes machine learning algorithms to optimize the loss function, continuously adjusting the weight coefficients in the model. This mechanism allows control strategies to evolve from a general initial state to a personalized model that closely matches the physiological response patterns of specific users; this is the essential difference between this and the unchanging control logic in existing technologies, which enables long-term, efficient, and precise oxygen therapy. An independent safety redundancy guarantee has been established. The existence of the safety redundancy management module provides the system with fault safety guarantee independent of the main control logic. It continuously monitors data quality and core vital signs. When it detects unreliable signals or a sharp deterioration in the user's physiological state, it can forcibly interrupt predictive control and switch to the preset conservative oxygen supply mode. This ensures that in extreme scenarios such as sensor failure or sudden emergency situations of the user, the system's first reaction is to ensure safety rather than pursue the optimal solution, which significantly improves the reliability of the entire system and its credibility in practical applications.

[0029] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A remote control method for an oxygen concentrator based on the Internet of Things, characterized in that, include: S1 collects multi-dimensional physiological data of users in real time through a multi-source physiological signal acquisition module. The multi-dimensional physiological data includes blood oxygen saturation signal, respiratory rate signal, heart rate variability index and body movement signal. S2, based on multi-dimensional physiological data, is processed through a preset physiological state prediction model to calculate a unified physiological stress index; S3, combining the physiological stress index with the preset user baseline oxygen supply flow rate, determines the predictive oxygen supply flow rate value; S4 generates control signals for the physical actuators of the oxygen generator based on the predicted oxygen supply flow rate value, so as to dynamically adjust the oxygen supply parameters. S5 uploads a data package containing multi-dimensional physiological data and predictive oxygen flow rate values ​​to the cloud-based collaborative computing module for iterative optimization of the physiological state prediction model.

2. The method for remote control of an oxygen concentrator based on the Internet of Things according to claim 1, characterized in that, Physiological state prediction model processing includes: Each physiological signal in the multidimensional physiological data is standardized based on its own preset reference rate of change to obtain a standardized change quantity; the standardized change quantity is then calculated through a nonlinear weighted fusion function to obtain a unified physiological stress index.

3. The method for remote control of an oxygen concentrator based on the Internet of Things according to claim 1, characterized in that, Determining the predictive oxygen supply flow rate includes: Based on the physiological stress index, the user's baseline oxygen supply flow rate and the preset maximum adjustment range, the initial oxygen supply flow rate is calculated; the initial oxygen supply flow rate is compared with the preset minimum safe oxygen supply flow rate, and the larger value is taken as the predictive oxygen supply flow rate value.

4. The method for remote control of an oxygen concentrator based on the Internet of Things according to claim 1, characterized in that, The cloud-based collaborative computing module iteratively optimizes the physiological state prediction model, including: A loss function is constructed in the cloud-based collaborative computing module. The loss function is used to characterize the difference between the oxygen supply flow rate set by the system and the optimal flow rate required to maintain the user's physiological state. The loss function is optimized and solved using machine learning algorithms to calculate personalized weight combinations. The personalized weight combinations are then distributed to update the weight coefficients in the physiological state prediction model.

5. The method for remote control of an oxygen concentrator based on the Internet of Things according to claim 1, characterized in that, The initial values ​​of the weight coefficients in the physiological state prediction model are derived from a general model formed by statistical analysis of large-scale clinical data.

6. The method for remote control of an oxygen concentrator based on the Internet of Things according to claim 1, characterized in that, Also includes: The safety redundancy management module monitors the data quality of multi-dimensional physiological data and the user's core vital signs in real time. If the data quality is lower than the preset minimum effective threshold or the core vital signs are lower than the preset critical safety threshold, the control process based on the predictive oxygen supply flow rate value is interrupted, and the preset conservative oxygen supply mode is executed.

7. The method for remote control of an oxygen concentrator based on the Internet of Things according to claim 6, characterized in that, The core vital sign is blood oxygen saturation; the conservative oxygen supply mode is to supply oxygen at a preset, constant and safe conservative oxygen supply flow rate.

8. A remote control system for an oxygen concentrator based on the Internet of Things (IoT), comprising the remote control method for an oxygen concentrator based on the IoT as described in any one of claims 1-7, characterized in that, include: The multi-source physiological signal acquisition module is used to collect multi-dimensional physiological data of users in real time, including blood oxygen saturation signal, respiratory rate signal, heart rate variability index and body movement signal; The edge computing module includes: The data processing unit is used to process multi-dimensional physiological data through a preset physiological state prediction model to calculate a unified physiological stress index. The decision unit is used to determine the predictive oxygen flow rate value by combining the physiological stress index with the preset user baseline oxygen flow rate. The adaptive oxygen supply control module is used to generate control signals for the physical actuators of the oxygen generator based on the predicted oxygen supply flow rate value. The cloud-based collaborative computing module receives data packets uploaded by the edge computing module, optimizes the physiological state prediction model, and sends the optimized model parameters back to the edge computing module.

9. A remote control system for an oxygen concentrator based on the Internet of Things according to claim 8, characterized in that, Also includes: The safety redundancy management module is used to monitor the data quality output by the multi-source physiological signal acquisition module and the user's core vital signs in real time. In response to the data quality being lower than the preset minimum effective threshold or the core vital signs being lower than the preset critical safety threshold, the adaptive oxygen supply control module is instructed to execute the preset conservative oxygen supply mode.

Images