Wireless positioning blood pressure transducer real-time monitoring and analyzing method based on Internet of Things

Abstract

The invention discloses a wireless positioning blood pressure transducer real-time monitoring and analyzing method based on the Internet of Things. The method comprises the following steps that S1, invasive pressure measuring signals are collected, wirelessly transmitted and preprocessed; s2, a Transform invasive pressure measurement signal analysis model is constructed; s3, performing local adaptive optimization on the hyper-parameters by adopting a bat algorithm; s4, performing global optimization on the bat algorithm hyper-parameters by using a sparrow search algorithm; s5, performing multi-sensor positioning data acquisition and Kalman filtering fusion to obtain real-time positioning; s6, the blood pressure anomaly detection result and the positioning data are fused, and a combined health risk assessment report is formed; and S7, the remote early warning module transmits the joint evaluation report and the real-time positioning information. The system has the advantages that real-time monitoring and accurate positioning of the blood pressure transducer are achieved by integrating the Internet of Things, multi-sensor positioning and intelligent data analysis, and efficient and accurate data support is provided for remote medical treatment and intelligent health management.

Description

Technical Field

[0001] The present invention relates to the technical field of Internet of Things, and in particular to a real-time monitoring and analysis method of a wireless positioning blood pressure transducer based on the Internet of Things. Background Art

[0002] Existing blood pressure monitoring technology mainly relies on traditional wired monitoring equipment or a single wireless transmission module, which has been widely used in clinical and family health management. However, with the rapid development of Internet of Things technology and wireless communication technology, traditional monitoring methods have gradually exposed many shortcomings and cannot meet the needs of modern telemedicine and intelligent health monitoring. In the existing technology, blood pressure monitoring equipment often uses wired connections or short-range wireless communications, and its data transmission distance and real-time performance are greatly limited, making it difficult to achieve truly cross-regional continuous monitoring. In addition, when locating the patient's position, traditional equipment usually only relies on a single technology such as GPS or Bluetooth, and the positioning accuracy and stability are difficult to guarantee, especially in indoor or weak signal environments. The positioning error is large and it is impossible to provide accurate location information for emergency rescue and health intervention.

[0003] In addition, existing blood pressure monitoring equipment relies on traditional statistical methods or shallow machine learning models for data processing and analysis, and fails to fully tap into the deep information and time series characteristics in physiological signals. This data processing method often has problems such as large signal noise interference, untimely abnormal detection, and delayed response of the early warning mechanism. It is difficult to detect abnormal blood pressure fluctuations in an emergency, thus affecting the telemedicine system's accurate assessment of the patient's status. At the same time, although traditional positioning technology can achieve location tracking to a certain extent, due to the lack of multi-sensor data fusion mechanism, it can often only provide information from a single data source, and cannot fully utilize the advantages of multiple wireless technologies such as GPS, Wi-Fi and Bluetooth to achieve high-precision, stable dynamic positioning.

[0004] In the context of the Internet of Things era, with the maturity of smart sensors, wireless communications and cloud computing technologies, medical health monitoring systems are gradually developing towards multi-dimensional data fusion, real-time monitoring and intelligent early warning. In recent years, some research and application attempts have been made to introduce Internet of Things technology into health monitoring systems, collect and transmit physiological data in real time through wireless communication devices, and use cloud platforms for data processing and intelligent analysis, and have made certain progress. However, the existing solutions still have shortcomings in data fusion and analysis methods, mainly manifested as follows: First, in terms of data collection, although the Internet of Things can realize real-time data transmission, due to the single type of sensors or simple data processing algorithms, the temporal and spatial correlation of different sensor data cannot be fully considered, resulting in poor data integration effect. Secondly, in terms of wireless positioning, due to the reliance on a single positioning method or the failure to introduce advanced data fusion technology, the accuracy and stability of positioning results in indoor environments and complex scenarios are still difficult to meet the requirements of actual applications. Thirdly, in terms of data analysis and early warning mechanisms, traditional methods cannot make full use of big data and deep learning technologies, resulting in insufficient sensitivity and timeliness in the detection of abnormal blood pressure fluctuations, thus failing to achieve effective remote intervention and emergency response.

[0005] In addition, existing technologies lack adaptive mechanisms in parameter optimization. Many devices and systems rely on fixed parameter settings, which are difficult to adapt to the differences in the environment and individual patients, further restricting the real-time and accuracy of the monitoring system. Especially in dynamic data fusion and positioning algorithms, traditional filtering methods such as Kalman filtering can achieve data smoothing and noise suppression to a certain extent, but when faced with multi-source heterogeneous data and complex environmental interference, their parameter settings often cannot be adjusted in real time, resulting in poor filtering effects, affecting the final positioning accuracy and reliability of the monitoring system.

[0006] Therefore, how to provide a real-time monitoring and analysis method for a wireless positioning blood pressure transducer based on the Internet of Things is an urgent problem that technicians in this field need to solve. Summary of the invention

[0007] One purpose of the present invention is to propose a real-time monitoring and analysis method for a wireless positioning blood pressure transducer based on the Internet of Things. The present invention makes full use of advanced technologies such as the Internet of Things, big data, wireless communication, multi-sensor fusion, Kalman filtering, and adaptive hyperparameter optimization algorithms, and describes in detail how to achieve accurate monitoring and efficient analysis of invasive pressure measurement signals and real-time positioning information through multi-sensor data acquisition, preprocessing, dynamic data fusion, and intelligent analysis. The method adopted by the present invention includes using a high-precision pressure sensor built into the blood pressure transducer to collect invasive pressure measurement signals in real time, and at the same time, by integrating multiple wireless positioning modules such as GPS, Wi-Fi, and Bluetooth, the location information of the device or patient is obtained in real time. The various types of data collected are transmitted to the cloud data processing platform through a wireless communication network. On this platform, the raw data is cleaned, time-synchronized, and noise-removed through a preprocessing algorithm to ensure data quality and consistency. Next, the present invention uses the Kalman filter to perform dynamic data fusion on the multi-sensor positioning data, realizes the correction and state estimation of multi-source data, and thus obtains accurate real-time positioning results; at the same time, in terms of blood pressure data monitoring, by introducing advanced signal processing technology and adaptive optimization algorithms, real-time monitoring and abnormality detection of invasive pressure measurement signals are performed to ensure that the monitoring process has high real-time and sensitivity.

[0008] According to an embodiment of the present invention, a real-time monitoring and analysis method of a wireless positioning blood pressure transducer based on the Internet of Things includes the following steps: S1. Collect invasive pressure measurement signals using a blood pressure transducer, transmit the invasive pressure measurement signals to a data processing platform in real time through a wireless communication module, and pre-process the transmitted invasive pressure measurement signals; S2. Construct an invasive pressure measurement signal analysis model based on the Transformer model, and use the multi-head self-attention mechanism to perform global feature extraction and time series modeling on the preprocessed invasive pressure measurement signal; S3, using the bat algorithm to perform local adaptive optimization of the hyperparameters of the invasive pressure measurement signal analysis model; S4, using a sparrow search algorithm to globally optimize the hyperparameters of the bat algorithm; S5, collecting multi-sensor positioning data including GPS, Wi-Fi and Bluetooth, and using Kalman filter to dynamically fuse the multi-sensor positioning data to obtain real-time positioning results; S6. Using the optimized invasive pressure measurement signal analysis model to perform data fusion on the abnormal detection results of the invasive pressure measurement signal and the real-time positioning results to form a joint health risk assessment report; S7. Transmit the joint health risk assessment report and real-time positioning information to the remote monitoring terminal through the remote early warning module.

[0009] Optionally, the S2 specifically includes: S21, the preprocessed invasive pressure measurement signal is represented as an input sequence ,in, is the sequence length, For the Signal characteristics at the time; S22, input sequence Perform position encoding and construct a position encoding matrix ; S23, input sequence With the position encoding matrix Add together to get the encoded input ; S24, code input Using a multi-head self-attention mechanism, for each attention head Compute the query matrix , key matrix And the numerical matrix ; S25. For each attention head Calculate the self-attention output and concatenate the outputs of each attention head to get the output matrix of the multi-head self-attention : ; in, is the number of attention heads, Indicates The output result obtained after the attention head calculates the self-attention mechanism, Represents a splicing operation; S26. Output matrix for multi-head self-attention Apply feedforward neural network to perform nonlinear transformation and high-level feature mapping: ; in, , are the weight matrices, , are the bias vectors, To find the maximum value, It is the output matrix after being processed by the feedforward neural network, and is used for global feature extraction and time series modeling of invasive pressure measurement signals.

[0010] Optionally, the S3 specifically includes: S31, the hyperparameter vector to be optimized for the Transformer model is ,in, Including the learning rate of each layer, the number of attention heads, the hidden layer dimension and the dropout ratio hyperparameters, is the number of hyperparameters; S32, use the adaptive initialization method to initialize the individual hyperparameter vector in the bat algorithm, record The initial vector of each bat is , ,in, and are the preset mean and standard deviation, is the population size, Indicates The initial hyperparameter vector of individual bats, represents normal distribution; S33. Define multi-objective fitness function : ; in, is the loss function, is the true value, is the model prediction value, To verify the number of samples, is a hyperparameter-based model complexity measure, is the weight coefficient; S34. For each individual bat , in In the iterations, the frequency parameter Updated to: ; in, and are the lower and upper bounds of the frequency, is a uniformly distributed random number, is a constant, is the current global optimal hyperparameter vector, Indicates The hyperparameter vector of each bat individual at the tth iteration; S35. Update the speed and hyperparameter vector of individual bats through dynamic weight and local search mechanism: ; ; in, For the Bat individuals in The speed at the iteration, is the updated speed, Indicates Bat individuals in The updated hyperparameter vector at iteration , Indicated in The local search disturbance term introduced in the iteration, the dynamic inertia weight for: ; Given, and are the maximum and minimum inertia weights, is the maximum number of iterations; S36. Update individual loudness and pulse emission rate: ; in, For the Bat individuals in The loudness at the iteration, For the Bat individuals in The pulse emission rate at the iteration, and are the adaptive update coefficients of loudness and pulse emission rate, is the disturbance scale factor, is the maximum number of iterations, and As the chaotic factors for updating loudness and pulse emission rate, For the Bat individuals in The loudness at the iteration, For the Bat individuals in The pulse emission rate at the iteration, and output the optimized hyperparameter vector when the termination condition is met .

[0011] Optionally, the S4 specifically includes: S41, the initial generated bat algorithm hyperparameter vector Initialize, where For the The hyperparameter vector of individual sparrows, is the sparrow population size; S42, using clustering algorithm to divide the sparrow population into subgroups, each subgroup is denoted by ; S43. In each subgroup Within, determine the local optimal hyperparameter vector : ; in, is the fitness function, Represents subgroup Any candidate hyperparameter vector in Indicates that in all subgroups In the candidate hyperparameter vector of The hyperparameter vector with the minimum value; S44, using the improved sparrow search algorithm structure, for each subgroup The hyperparameter vector of each sparrow individual is updated: ; in, Indicates The sparrow individual is in The hyperparameter vector at iteration , For the The adaptive attenuation coefficient of each sparrow, is the maximum number of iterations, is the random disturbance term, is the dynamic weight factor, is the reverse learning factor, is a chaotic perturbation sequence, is the disturbance scale factor, Indicates The sparrow individual is in The hyperparameter vector after the first stage update in the iteration, is an exponential function; S45. For the predetermined proportion of individuals with poor fitness in each subgroup, an adversarial learning mechanism is used to update the hyperparameter vector: ; in, and are the lower and upper bounds of each dimension of the hyperparameter vector respectively; S46. Introduce a memory archiving mechanism and record the historical optimal hyperparameter vector as , modify each candidate hyperparameter vector: ; in, Indicates The sparrow individual is in After iterations, the final updated hyperparameter vector meets the termination condition: ; in, is the preset convergence threshold, is the current iteration number, then the optimized bat algorithm hyperparameter vector is output .

[0012] Optionally, the S44 specifically includes: S441, for each individual sparrow in the sparrow population, adopt a local coordination strategy based on clustering to coordinately update the individuals in the same subgroup, and calculate the Euclidean distance between each individual sparrow and the local optimal solution of the subgroup : ; in, Indicates The hyperparameter vector of the sparrow individuals at the tth iteration, is the local optimal hyperparameter vector; S442: Based on the Euclidean distance, calculate the local diversity index of the individuals in the subgroup, and set the adaptive weight factor accordingly. : ; in, is an exponential function, For subgroups, is the fitness function, Represents subgroup Any candidate hyperparameter vector in ; S443, introduce the reverse learning mechanism and calculate the reverse learning factor for each sparrow individual : ; in, and are the upper and lower bounds of each dimension of the hyperparameter vector respectively; S444, Generate chaotic perturbation sequence using chaotic perturbation mechanism based on Logistic mapping : ; in, is the Logistic mapping parameter, and the initial value satisfy , Indicates the previous iteration The chaos disturbance value at the moment.

[0013] Optionally, the S5 specifically includes: S51, respectively using GPS, Wi-Fi and Bluetooth positioning modules to collect respective positioning data in real time, including signal strength, timestamp and related status information; S52, transmitting the collected positioning data to the Internet of Things gateway via the wireless communication network, and aggregating it to the central data processing platform; S53, pre-processing the received multi-source positioning data on the central data processing platform, performing data cleaning, time synchronization and abnormal data elimination; S54, inputting the pre-processed multi-source positioning data into the Kalman filter, initializing the Kalman filter state and error estimation; S55, using a Kalman filter to dynamically fuse the pre-processed multi-source positioning data, and automatically adjust the prediction and update process; S56: output the real-time positioning result after dynamic fusion processing by the Kalman filter, and transmit the real-time positioning result to the monitoring terminal.

[0014] The beneficial effects of the present invention are: The present invention realizes the real-time collection, dynamic data fusion and intelligent analysis of invasive pressure measurement signals and location information by organically combining advanced technologies such as the Internet of Things, big data, multi-sensor fusion, Kalman filtering and improved adaptive hyperparameter optimization algorithms. The present invention utilizes the built-in high-precision pressure sensor of the blood pressure transducer and wireless positioning modules such as GPS, Wi-Fi and Bluetooth to realize the efficient transmission and synchronous processing of multi-source data, and cleans the signal, synchronizes the time and removes abnormal data in the data preprocessing stage to ensure data quality. Subsequently, the Kalman filter is used to dynamically fuse the preprocessed multi-sensor positioning data, so that stable and accurate real-time positioning results can be output under various environmental conditions. At the same time, in terms of blood pressure data monitoring and analysis, the present invention introduces global feature extraction and time series modeling technology based on the Transformer model, and optimizes hyperparameters by jointly optimizing hyperparameters through the improved bat algorithm and sparrow search algorithm, so as to realize the adaptive adjustment of the data processing module in the face of individual differences and environmental changes, thereby improving the accuracy and response speed of abnormal detection.

[0015] The improved technology of the present invention not only overcomes the problems of insufficient real-time performance, low positioning accuracy, and difficulty in adaptive parameters of traditional blood pressure monitoring equipment under wired or single wireless communication, but also greatly improves the stability and robustness of the monitoring system by introducing multi-sensor data fusion and adaptive optimization mechanisms. Especially in indoor, complex environments, and scenes with large signal interference, through the dynamic data fusion of the Kalman filter, the system can effectively reduce noise interference and data errors, and ensure high-precision output of positioning information; at the same time, the improved hyperparameter optimization algorithm plays a key role in the blood pressure data analysis module, and can automatically adjust the model parameters to adapt to the ever-changing actual situation, realizing the timely detection and early warning of abnormal blood pressure conditions.

[0016] In addition, the present invention adopts a modular structure in the overall system design, and realizes seamless connection and efficient coordination of data between various functional modules, which not only meets the needs of real-time monitoring and precise positioning, but also provides a solid technical guarantee for telemedicine, intelligent health monitoring and emergency rescue. The system has achieved high efficiency and intelligence in various links such as wireless data transmission, data preprocessing, multi-sensor positioning and intelligent analysis, significantly improving the working efficiency and data utilization of medical monitoring equipment. By globally optimizing the hyperparameters of the bat algorithm through the improved sparrow search algorithm, the present invention further breaks through the limitations of traditional algorithms in local optimality and parameter adaptation, ensuring the stable operation and high-precision performance of the system in a changing environment, thereby providing reliable and real-time data support for health monitoring and medical intervention.

[0017] In summary, the present invention not only realizes the efficient collection, processing and intelligent analysis of invasive pressure measurement signals and real-time positioning data, but also has made significant progress in key technologies such as data fusion, parameter adaptation and multi-source positioning. The system as a whole has the advantages of high real-time performance, high positioning accuracy, high data processing accuracy and strong adaptability, which greatly improves the efficiency and safety of telemedicine, intelligent health monitoring and emergency rescue. Through these innovative technologies, the present invention provides a new technical path for building an efficient and intelligent remote health monitoring system, which not only meets the current needs of clinical and family health management, but also lays a solid foundation for the intelligent development of the future medical and health field. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings: Figure 1 This is a flow chart of the real-time monitoring and analysis method of a wireless positioning blood pressure transducer based on the Internet of Things proposed by the present invention; Figure 2 This is a schematic diagram of the modular design of the overall data processing and monitoring platform of the real-time monitoring and analysis method of the wireless positioning blood pressure transducer based on the Internet of Things proposed in the present invention. DETAILED DESCRIPTION

[0019] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, which only illustrate the basic structure of the present invention in a schematic manner, and therefore only show the components related to the present invention.

[0020] refer to Figure 1 and Figure 2 , a real-time monitoring and analysis method of a wireless positioning blood pressure transducer based on the Internet of Things, comprising the following steps: S1. Collect invasive pressure measurement signals using a blood pressure transducer, transmit the invasive pressure measurement signals to a data processing platform in real time through a wireless communication module, and pre-process the transmitted invasive pressure measurement signals; S2. Construct an invasive pressure measurement signal analysis model based on the Transformer model, and use the multi-head self-attention mechanism to perform global feature extraction and time series modeling on the preprocessed invasive pressure measurement signal; S3, using the bat algorithm to perform local adaptive optimization of the hyperparameters of the invasive pressure measurement signal analysis model; S4, using a sparrow search algorithm to globally optimize the hyperparameters of the bat algorithm; S5, collecting multi-sensor positioning data including GPS, Wi-Fi and Bluetooth, and using Kalman filter to dynamically fuse the multi-sensor positioning data to obtain real-time positioning results; S6. Using the optimized invasive pressure measurement signal analysis model to perform data fusion on the abnormal detection results of the invasive pressure measurement signal and the real-time positioning results to form a joint health risk assessment report; S7. Transmit the joint health risk assessment report and real-time positioning information to the remote monitoring terminal through the remote early warning module.

[0021] In this implementation, S2 specifically includes: S21, the preprocessed invasive pressure measurement signal is represented as an input sequence ,in, is the sequence length, For the Signal characteristics at the time; S22, input sequence Perform position encoding and construct a position encoding matrix ; S23, input sequence With the position encoding matrix Add together to get the encoded input ; S24, code input Using a multi-head self-attention mechanism, for each attention head Compute the query matrix , key matrix And the numerical matrix ; S25. For each attention head Calculate the self-attention output and concatenate the outputs of each attention head to get the output matrix of the multi-head self-attention : ; in, is the number of attention heads, Indicates The output result obtained after the attention head calculates the self-attention mechanism, Represents a splicing operation; S26. Output matrix for multi-head self-attention Apply feedforward neural network to perform nonlinear transformation and high-level feature mapping: ; in, , are the weight matrices, , are the bias vectors, To find the maximum value, It is the output matrix after being processed by the feedforward neural network, and is used for global feature extraction and time series modeling of invasive pressure measurement signals.

[0022] In this implementation, S3 specifically includes: S31, the hyperparameter vector to be optimized for the Transformer model is ,in, Including the learning rate of each layer, the number of attention heads, the hidden layer dimension and the dropout ratio hyperparameters, is the number of hyperparameters; S32, use the adaptive initialization method to initialize the individual hyperparameter vector in the bat algorithm, record The initial vector of each bat is , ,in, and are the preset mean and standard deviation, is the population size, Indicates The initial hyperparameter vector of individual bats, represents normal distribution; S33. Define multi-objective fitness function : ; in, is the loss function, is the true value, is the model prediction value, To verify the number of samples, is a hyperparameter-based model complexity measure, is the weight coefficient; S34. For each individual bat , in In the iterations, the frequency parameter Updated to: ; in, and are the lower and upper bounds of the frequency, is a uniformly distributed random number, is a constant, is the current global optimal hyperparameter vector, Indicates The hyperparameter vector of each bat individual at the tth iteration; S35. Update the speed and hyperparameter vector of individual bats through dynamic weight and local search mechanism: ; ; in, For the Bat individuals in The speed at the iteration, is the updated speed, Indicates Bat individuals in The updated hyperparameter vector at iteration , Indicated in The local search disturbance term introduced in the iteration, the dynamic inertia weight for: ; Given, and are the maximum and minimum inertia weights, is the maximum number of iterations; S36. Update individual loudness and pulse emission rate: ; in, For the Bat individuals in The loudness at the iteration, For the Bat individuals in The pulse emission rate at the iteration, and are the adaptive update coefficients of loudness and pulse emission rate, is the disturbance scale factor, is the maximum number of iterations, and As the chaotic factors for updating loudness and pulse emission rate, For the Bat individuals in The loudness at the iteration, For the Bat individuals in The pulse emission rate at the iteration, and output the optimized hyperparameter vector when the termination condition is met .

[0023] In this implementation manner, the S4 specifically includes: S41, the initial generated bat algorithm hyperparameter vector Initialize, where For the The hyperparameter vector of individual sparrows, is the sparrow population size; S42, using clustering algorithm to divide the sparrow population into subgroups, each subgroup is denoted by ; S43. In each subgroup Within, determine the local optimal hyperparameter vector : ; in, is the fitness function, Represents subgroup Any candidate hyperparameter vector in Indicates that in all subgroups In the candidate hyperparameter vector of The hyperparameter vector with the minimum value; S44, using the improved sparrow search algorithm structure, for each subgroup The hyperparameter vector of each sparrow individual is updated: ; in, Indicates The sparrow individual is in The hyperparameter vector at iteration , For the The adaptive attenuation coefficient of each sparrow, is the maximum number of iterations, is the random disturbance term, is the dynamic weight factor, is the reverse learning factor, is a chaotic perturbation sequence, is the disturbance scale factor, Indicates The sparrow individual is in The hyperparameter vector after the first stage update in the iteration, is an exponential function; S45. For the predetermined proportion of individuals with poor fitness in each subgroup, an adversarial learning mechanism is used to update the hyperparameter vector: ; in, and are the lower and upper bounds of each dimension of the hyperparameter vector respectively; S46. Introduce a memory archiving mechanism and record the historical optimal hyperparameter vector as , modify each candidate hyperparameter vector: ; in, Indicates The sparrow individual is in After iterations, the final updated hyperparameter vector meets the termination condition: ; in, is the preset convergence threshold, is the current iteration number, then the optimized bat algorithm hyperparameter vector is output .

[0024] In this implementation manner, the S44 specifically includes: S441, for each individual sparrow in the sparrow population, adopt a local coordination strategy based on clustering to coordinately update the individuals in the same subgroup, and calculate the Euclidean distance between each individual sparrow and the local optimal solution of the subgroup : ; in, Indicates The hyperparameter vector of the sparrow individuals at the tth iteration, is the local optimal hyperparameter vector; S442: Based on the Euclidean distance, calculate the local diversity index of the individuals in the subgroup, and set the adaptive weight factor accordingly. : ; in, is an exponential function, For subgroups, is the fitness function, Represents subgroup Any candidate hyperparameter vector in ; S443, introduce the reverse learning mechanism and calculate the reverse learning factor for each sparrow individual : ; in, and are the upper and lower bounds of each dimension of the hyperparameter vector respectively; S444, Generate chaotic perturbation sequence using chaotic perturbation mechanism based on Logistic mapping : ; in, is the Logistic mapping parameter, and the initial value satisfy , Indicates the previous iteration The chaos disturbance value at the moment.

[0025] In this implementation manner, S5 specifically includes: S51, respectively using GPS, Wi-Fi and Bluetooth positioning modules to collect respective positioning data in real time, including signal strength, timestamp and related status information; S52, transmitting the collected positioning data to the Internet of Things gateway via the wireless communication network, and aggregating it to the central data processing platform; S53, pre-processing the received multi-source positioning data on the central data processing platform, performing data cleaning, time synchronization and abnormal data elimination; S54, inputting the pre-processed multi-source positioning data into the Kalman filter, initializing the Kalman filter state and error estimation; S55, using a Kalman filter to dynamically fuse the pre-processed multi-source positioning data, and automatically adjust the prediction and update process; S56: output the real-time positioning result after dynamic fusion processing by the Kalman filter, and transmit the real-time positioning result to the monitoring terminal.

[0026] Embodiment 1: In order to verify the feasibility of the present invention in implementation, the present invention is applied to a large general hospital to perform remote health monitoring of patients in the emergency department and intensive care unit of the hospital. Traditional blood pressure monitoring and positioning technology has problems of data acquisition delay, low positioning accuracy and untimely abnormality detection in this scenario, and cannot meet the needs of critically ill patients for real-time monitoring and rapid intervention. The real-time monitoring and analysis method of wireless positioning blood pressure transducer based on the Internet of Things proposed in the present invention integrates high-precision blood pressure transducer, multi-sensor positioning technology, Kalman filter dynamic data fusion and improved adaptive hyperparameter optimization algorithm, so as to realize efficient collection, intelligent fusion and analysis of invasive pressure measurement signals and real-time positioning data, and provide more accurate, timely and intelligent technical support for clinical remote monitoring.

[0027] During the implementation process, the hospital equipped each critically ill patient with a blood pressure transducer with an integrated high-precision pressure sensor. The device can collect the patient's blood pressure data in real time, and obtain the patient's location information in the ward through the built-in GPS, Wi-Fi and Bluetooth modules. All data are transmitted to the cloud data processing platform through the hospital's internal wireless network. The platform pre-processes the data, including data cleaning, time synchronization and outlier removal to ensure data accuracy. The pre-processed multi-sensor positioning data is dynamically fused through the Kalman filter, which significantly improves the positioning accuracy, so that the average positioning error is controlled within 1.5 meters in the complex environment of the emergency department and intensive care unit; and the blood pressure data is monitored in real time through the global feature extraction and time series modeling technology based on the Transformer model. The hyperparameters are jointly optimized by the improved bat algorithm and the sparrow search algorithm, which effectively improves the detection accuracy of abnormal blood pressure fluctuations and controls the abnormal warning response time within 3 seconds.

[0028] To verify the beneficial effects of the present invention, the hospital continuously monitored 30 critically ill patients from November 2023 to January 2024. Statistical results show that after adopting the method of the present invention, the average delay in data acquisition is reduced to less than 0.8 seconds, while the average delay of the traditional method is 3.2 seconds; after the positioning data is fused by Kalman filtering, the average positioning error is 1.4 meters, which is 70% lower than the error of the traditional single positioning technology (about 4.8 meters); the average response time of the abnormal blood pressure monitoring module is 3 seconds, while the response time of the traditional monitoring equipment is about 6 seconds, and the accuracy of abnormal warning is increased to 95%, which is about 20 percentage points higher than that of traditional equipment. These data fully demonstrate the significant advantages of the present invention in real-time, positioning accuracy and abnormality detection.

[0029] In this embodiment, blood pressure data is seamlessly connected with multi-sensor positioning data through the Internet of Things platform, and Kalman filtering technology is used for dynamic fusion. Combined with the improved adaptive hyperparameter optimization algorithm, the system realizes efficient real-time processing and intelligent analysis of data, providing strong data support for remote monitoring, emergency rescue and health intervention. The entire system runs stably, and the modular design enables the various functional units to work together, realizing the automatic collection, processing and early warning of real-time monitoring data. After continuous operation testing, the stable operation rate of the system reached 99.5%, and the accuracy and timeliness of abnormal warnings were significantly improved, providing reliable protection for clinical emergency and patient safety management.

[0030] Table 1 Comparison of real-time positioning and blood pressure monitoring performance between traditional methods and the present invention ;

[0031] Table 1 systematically compares the performance of the traditional method and the present invention in terms of data acquisition delay, positioning error, abnormal warning response time, warning accuracy and system stable operation rate, and quantitatively demonstrates the technical advantages of the present invention. The data show that the present invention has achieved significant optimization in data transmission efficiency, positioning accuracy, warning response speed, etc., providing more accurate and efficient technical support for real-time monitoring and positioning analysis of blood pressure transducers.

[0032] In terms of data collection delay, the traditional method takes an average of 3.2 seconds, while the present invention reduces the delay to 0.8 seconds, a reduction of 75%, by optimizing the wireless communication protocol and improving the data compression efficiency. This improvement significantly improves the real-time nature of the data, allowing medical personnel to obtain patient blood pressure information faster, buying more time for emergency intervention.

[0033] In terms of positioning error, the average error of the traditional method is 4.8 meters, especially in indoor environments. The present invention uses GPS, Wi-Fi, and Bluetooth for multi-sensor data fusion, and uses Kalman filtering for dynamic correction, reducing the error to 1.4 meters and improving positioning accuracy by 70%. This improvement is extremely practical in complex environments such as hospital emergency rooms and ICUs, allowing medical staff to accurately locate patients and improve the reliability of remote monitoring systems.

[0034] In terms of abnormal warning response time, the traditional method takes an average of 6 seconds, while the present invention uses the Transformer model for feature extraction and combines it with an improved hyperparameter optimization strategy to shorten the response time to 3 seconds, a reduction of 50%. This means that the system can detect blood pressure abnormalities more quickly and notify medical staff as soon as possible, thereby improving the success rate of rescue.

[0035] In terms of early warning accuracy, the traditional method has many false positives and missed positives, with an accuracy rate of only 75%. The present invention uses an adaptive optimization algorithm to tune the deep learning model, which increases the accuracy rate to 95%, an increase of 20 percentage points, effectively reducing false positives and missed positives, and providing doctors with a more reliable basis for clinical decision-making assistance.

[0036] In terms of system stability, the traditional method has low long-term stability, with a system stability of 92%, and there is a risk of data loss or transmission interruption. The present invention optimizes the data processing and transmission mechanism, making the stable operation rate reach 99.5%, an increase of 7.5 percentage points, ensuring the continuity and efficiency of remote medical monitoring.

[0037] In summary, the present invention has demonstrated excellent performance in scenarios such as telemedicine, emergency monitoring, chronic disease management, and smart healthcare by optimizing data collection, wireless transmission, intelligent positioning, anomaly detection, and system stability. The system's low latency, high precision, and high reliability make it an efficient solution for medical health monitoring, providing strong technical support for the digital upgrade of the telemedicine industry.

[0038] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A real-time monitoring and analysis method for a wireless positioning blood pressure transducer based on the Internet of Things, characterized in that: The steps include: S1. Collect invasive pressure measurement signals using a blood pressure transducer, transmit the invasive pressure measurement signals to a data processing platform in real time through a wireless communication module, and pre-process the transmitted invasive pressure measurement signals; S2. Construct an invasive pressure measurement signal analysis model based on the Transformer model, and use the multi-head self-attention mechanism to perform global feature extraction and time series modeling on the preprocessed invasive pressure measurement signal; S3, using the bat algorithm to perform local adaptive optimization of the hyperparameters of the invasive pressure measurement signal analysis model; S4, using a sparrow search algorithm to globally optimize the hyperparameters of the bat algorithm; S5, collecting multi-sensor positioning data including GPS, Wi-Fi and Bluetooth, and using Kalman filter to dynamically fuse the multi-sensor positioning data to obtain real-time positioning results; S6. Using the optimized invasive pressure measurement signal analysis model to perform data fusion on the abnormal detection results of the invasive pressure measurement signal and the real-time positioning results to form a joint health risk assessment report; S7. Transmit the joint health risk assessment report and real-time positioning information to the remote monitoring terminal through the remote early warning module.

2. The method for real-time monitoring and analysis of a wireless positioning blood pressure transducer based on the Internet of Things according to claim 1 is characterized in that: The S2 specifically includes: S21, the preprocessed invasive pressure measurement signal is represented as an input sequence ,in, is the sequence length, For the Signal characteristics at the time; S22, input sequence Perform position encoding and construct a position encoding matrix ; S23, input sequence With the position encoding matrix Add together to get the encoded input ; S24, code input Using a multi-head self-attention mechanism, for each attention head Compute the query matrix , key matrix And the numerical matrix ; S25. For each attention head Calculate the self-attention output and concatenate the outputs of each attention head to get the output matrix of the multi-head self-attention ; S26. Output matrix for multi-head self-attention Apply feedforward neural network to perform nonlinear transformation and high-level feature mapping.

3. The method for real-time monitoring and analysis of a wireless positioning blood pressure transducer based on the Internet of Things according to claim 1 is characterized in that: The S3 specifically includes: S31, the hyperparameter vector to be optimized for the Transformer model is ,in, Including the learning rate of each layer, the number of attention heads, the hidden layer dimension and the Dropout ratio hyperparameters, is the number of hyperparameters; S32, use the adaptive initialization method to initialize the individual hyperparameter vector in the bat algorithm, record The initial vector of each bat is ; S33. Define multi-objective fitness function : ； in, is the loss function, is the true value, is the model prediction value, To verify the number of samples, is a hyperparameter-based model complexity measure, is the weight coefficient; S34. For each individual bat , in In the iterations, the frequency parameter Updated to: ； in, and are the lower and upper frequency bounds, respectively. is a uniformly distributed random number, is a constant, is the current global optimal hyperparameter vector, Indicates The hyperparameter vector of each bat individual at the tth iteration; S35. Update the speed and hyperparameter vector of individual bats through dynamic weight and local search mechanism: ； ； in, For the Bat individuals in The speed at the iteration, is the updated speed, Indicates Bat individuals in The updated hyperparameter vector at iteration , Indicated in The local search perturbation term introduced in the iteration, is the dynamic inertia weight; S36. Update individual loudness and pulse emission rate: ； in, For the Bat individuals in The loudness at the iteration, For the Bat individuals in The pulse emission rate at the iteration, and are the adaptive update coefficients of loudness and pulse emission rate, is the disturbance scale factor, is the maximum number of iterations, and As the chaotic factors for updating loudness and pulse emission rate, For the Bat individuals in The loudness at the iteration, For the Bat individuals in The pulse emission rate at the iteration, and output the optimized hyperparameter vector when the termination condition is met .

4. The method for real-time monitoring and analysis of a wireless positioning blood pressure transducer based on the Internet of Things according to claim 1 is characterized in that: The S4 specifically includes: S41, the initial generated bat algorithm hyperparameter vector Initialize, where For the The hyperparameter vector of individual sparrows, is the sparrow population size; S42, using clustering algorithm to divide the sparrow population into subgroups, each subgroup is denoted by ; S43. In each subgroup Within, determine the local optimal hyperparameter vector : ； in, is the fitness function, Represents subgroup Any candidate hyperparameter vector in Indicates that in all subgroups In the candidate hyperparameter vector of The hyperparameter vector with the minimum value; S44, using the improved sparrow search algorithm structure, for each subgroup The hyperparameter vector of each sparrow individual is updated: ； in, Indicates The sparrow individual is in The hyperparameter vector at iteration , For the The adaptive attenuation coefficient of each sparrow, is the maximum number of iterations, is the random disturbance term, is the dynamic weight factor, is the reverse learning factor, is a chaotic perturbation sequence, is the disturbance scale factor, Indicates The sparrow individual is in The hyperparameter vector after the first stage update in the iteration, is an exponential function; S45. For the predetermined proportion of individuals with poor fitness in each subgroup, an adversarial learning mechanism is used to update the hyperparameter vector: ； in, and are the lower and upper bounds of each dimension of the hyperparameter vector respectively; S46. Introduce a memory archiving mechanism and record the historical optimal hyperparameter vector as , modify each candidate hyperparameter vector: ； in, Indicates The sparrow individual is in After iterations, the final updated hyperparameter vector meets the termination condition: ； in, is the preset convergence threshold, is the current iteration number, then the optimized bat algorithm hyperparameter vector is output .

5. The method for real-time monitoring and analysis of a wireless positioning blood pressure transducer based on the Internet of Things according to claim 4 is characterized in that: The S44 specifically includes: S441, for each individual sparrow in the sparrow population, adopt a local coordination strategy based on clustering to coordinately update the individuals in the same subgroup, and calculate the Euclidean distance between each individual sparrow and the local optimal solution of the subgroup ; S442: Based on the Euclidean distance, calculate the local diversity index of the individuals in the subgroup, and set the adaptive weight factor accordingly. : ； in, is an exponential function, For subgroups, is the fitness function, Represents subgroup Any candidate hyperparameter vector in ; S443, introduce the reverse learning mechanism and calculate the reverse learning factor for each sparrow individual : ； in, and are the upper and lower bounds of each dimension of the hyperparameter vector respectively; S444, Generate chaotic perturbation sequence using chaotic perturbation mechanism based on Logistic mapping : ； in, is the Logistic mapping parameter, and the initial value satisfy , Indicates the previous iteration The chaos disturbance value at the moment.

6. The method for real-time monitoring and analysis of a wireless positioning blood pressure transducer based on the Internet of Things according to claim 1 is characterized in that: The S5 specifically includes: S51, respectively using GPS, Wi-Fi and Bluetooth positioning modules to collect respective positioning data in real time, including signal strength, timestamp and related status information; S52, transmitting the collected positioning data to the Internet of Things gateway via the wireless communication network, and aggregating it to the central data processing platform; S53, pre-processing the received multi-source positioning data on the central data processing platform, performing data cleaning, time synchronization and abnormal data elimination; S54, inputting the pre-processed multi-source positioning data into the Kalman filter, initializing the Kalman filter state and error estimation; S55, using a Kalman filter to dynamically fuse the pre-processed multi-source positioning data, and automatically adjust the prediction and update process; S56: output the real-time positioning result after dynamic fusion processing by the Kalman filter, and transmit the real-time positioning result to the monitoring terminal.

Images