Construction method and system for medical health dynamic prediction model

Abstract

The invention relates to the technical field of medical monitoring, in particular to a construction method and system for a medical health dynamic prediction model. The method comprises the following steps: acquiring hospital structure data and constructing an intra-hospital millimeter wave radar network to obtain the intra-hospital millimeter wave radar network; obtaining RFID identification data and carrying out personnel distribution marking to obtain a hospital personnel distribution structure model; carrying out posture sign integration based on a millimeter wave radar network in the hospital to obtain posture sign data of personnel in the hospital; constructing a personnel abnormal sign dynamic prediction model according to the posture sign data of the personnel in the hospital; performing dynamic prediction on the posture sign data of the personnel in the hospital to obtain abnormal sign prediction personnel data, and performing position identification on the abnormal sign prediction personnel data to obtain abnormal sign prediction personnel position data; and information integration is carried out according to the abnormal sign prediction personnel position data, so that RFID early warning information is obtained. According to the invention, the early warning capability of abnormal signs can be improved.

Description

Technical Field

[0001] The present invention relates to the field of medical monitoring technology, and in particular to a method and system for constructing a medical health dynamic prediction model. Background Art

[0002] In modern medical and health management, the construction of dynamic prediction models has received increasing attention. These models are designed to monitor and analyze the health status of patients in real time, so as to detect potential health risks in a timely manner and take corresponding intervention measures. Early health monitoring mainly relied on traditional examination methods, such as physical examinations, medical records, and laboratory tests. Although these methods can provide patients with health information to a certain extent, they often have defects such as poor real-time performance, difficult data integration, and delayed feedback. For example, traditional vital sign monitoring often relies on manual recording and analysis by medical staff, which is not only time-consuming and labor-intensive, but also easily affected by human factors, resulting in insufficient data accuracy and timeliness. In addition, traditional methods usually lack effective real-time performance in data collection and processing, and often fail to reflect changes in patients' health status in a timely manner. This is especially important for patients with acute illnesses or those in intensive care, because their physical condition may change dramatically in a short period of time, and any delay may lead to serious consequences. Summary of the invention

[0003] Based on this, it is necessary for the present invention to provide a method and system for constructing a medical health dynamic prediction model to solve at least one of the above technical problems.

[0004] To achieve the above purpose, a method for constructing a dynamic prediction model for medical health includes the following steps: Step S1: Acquire hospital structure data, and perform three-dimensional structural modeling of the hospital building according to the hospital structure data, thereby obtaining a three-dimensional structural model of the hospital building; construct an in-hospital millimeter-wave radar network based on the three-dimensional structural model of the hospital building, thereby obtaining an in-hospital millimeter-wave radar network; Step S2: Acquire RFID identification data, and perform personnel distribution marking on the three-dimensional structure model of the hospital building according to the RFID identification data, thereby obtaining a hospital personnel distribution structure model; Step S3: collecting millimeter-wave reflection signals in the hospital based on the millimeter-wave radar network in the hospital, thereby obtaining millimeter-wave reflection signal data in the hospital, and integrating posture and vital signs of personnel in the hospital according to the millimeter-wave reflection signal data in the hospital, thereby obtaining posture and vital signs data of personnel in the hospital; Step S4: abnormal physical sign monitoring is performed on the posture and physical sign data of the personnel in the hospital, thereby obtaining abnormal physical sign detection data of the personnel, and a dynamic prediction model for abnormal physical signs of the personnel is constructed according to the abnormal physical sign detection data of the personnel; Step S5: using the personnel abnormal physical sign dynamic prediction model to dynamically predict the posture physical sign data of the in-hospital personnel, thereby obtaining abnormal physical sign predicted personnel data, and using the hospital personnel distribution structure model to identify the abnormal physical sign predicted personnel positions on the abnormal physical sign predicted personnel data, thereby obtaining abnormal physical sign predicted personnel position data; Step S6: Integrate the predicted personnel RFID warning information based on the personnel location data predicted by abnormal physical signs, thereby obtaining RFID warning information and transmitting it to the RFID management platform to execute the personnel warning task.

[0005] The three-dimensional structural model in the present invention can clearly display the internal spatial layout of the hospital, which is helpful to optimize resource allocation and emergency response. By constructing a millimeter wave radar network, real-time monitoring of the internal environment of the hospital is achieved, and the perception of the dynamic state of the patient is enhanced. According to the RFID identification data, the distribution of various types of personnel in the hospital can be accurately marked, which is convenient for quickly locating patients and medical staff. The personnel distribution information is combined with the hospital building model to provide comprehensive spatial data support to help optimize personnel flow and resource allocation. Through the millimeter wave reflection signal, the patient's posture changes can be captured in real time, and potential health risks can be identified in time. Millimeter wave technology is not affected by lighting conditions and can work effectively in a variety of environments, improving the accuracy and reliability of data. Real-time monitoring of abnormal signs can help medical staff take intervention measures before the problem becomes serious, thereby improving patient safety. By constructing a dynamic prediction model, the system can analyze historical data, identify potential health risk trends, and thus formulate personalized health management plans. By combining abnormal sign prediction with the personnel distribution model, the location of patients who need attention can be accurately identified, providing guidance for medical staff. The response efficiency to patients with abnormal signs is improved, the time for medical staff to find patients is reduced, and the overall medical efficiency is improved. Integrate and transmit the RFID warning information of abnormal signs predictors to the management platform to ensure rapid response. Through systematic warning information processing, the hospital's response speed and decision-making ability in dealing with emergencies can be improved. In summary, these steps can significantly improve the real-time and accuracy of health monitoring in hospitals through the integration and innovation of technology, and provide patients with safer and more timely medical services. At the same time, it also reduces the workload of medical staff and improves overall medical efficiency and patient satisfaction.

[0006] Optionally, step S1 specifically includes: Step S11: Acquire hospital structure data, and divide the hospital structure data into regional structures, thereby obtaining hospital ward regional structure data and hospital outpatient regional structure data; Step S12: integrating the regional spatial layout of the hospital ward regional structure data and the hospital outpatient regional structure data respectively, thereby obtaining the hospital ward regional spatial layout data and the hospital outpatient regional spatial layout data; Step S13: performing a regional spatial channel intersection operation on the hospital ward regional spatial layout data and the hospital outpatient regional spatial layout data, thereby obtaining regional connection channel data; Step S14: performing regional structural spatial association on the hospital ward regional structural data and the hospital outpatient regional structural data according to the regional connection channel data, thereby obtaining hospital regional structural connection data; Step S15: performing three-dimensional structural modeling of the hospital building according to the hospital area structural connection data, thereby obtaining a three-dimensional structural model of the hospital building; Step S16: Acquire the hospital millimeter-wave radar deployment data, and construct an in-hospital millimeter-wave radar network for the three-dimensional structure model of the hospital building according to the hospital millimeter-wave radar deployment data, thereby obtaining the in-hospital millimeter-wave radar network.

[0007] The present invention can understand the functional layout and space utilization of the hospital in a more systematic way by acquiring comprehensive hospital structure data. The division of regional structure can help identify the functional requirements of different regions, thereby optimizing the resource allocation of wards and outpatient areas and improving the efficiency of medical services. By integrating the spatial layout of wards and outpatient clinics, the flow routes of patients and medical staff can be better designed, congestion can be reduced, and the fluency of medical services can be improved. Reasonable spatial layout can improve the patient's medical experience and make it easier for patients to find the required service area. Through intersection operations, key connection channels in the hospital can be identified to ensure efficient connections between important areas. Clarifying the structure of the channel is helpful for rapid evacuation and rescue in emergency situations (such as fire or other disasters). By analyzing the regional connection channels, the connection mode between different areas can be optimized, the coordination of the overall space can be improved, the functionality of the hospital can be improved, the linkage between different departments can be smoother, and work efficiency can be improved. Three-dimensional modeling makes the structure of the hospital more intuitive, which is convenient for designers and managers to carry out space planning and design evaluation. The three-dimensional model can provide decision makers with accurate information to help make decisions such as resource allocation and building renovation. The millimeter-wave radar network can accurately locate and track people and equipment within the hospital, improving safety and management efficiency. Through millimeter-wave radar data, hospitals can achieve intelligent environmental monitoring and management, and improve overall service levels.

[0008] Optionally, step S16 is specifically: Step S161: Acquire hospital millimeter wave radar deployment data, and perform feature extraction on the hospital millimeter wave radar deployment data, thereby obtaining radar deployment location data; Step S162: constructing a millimeter wave radar deployment network for the hospital millimeter wave radar deployment data and the three-dimensional structure model of the hospital building according to the radar deployment location data, thereby obtaining a millimeter wave radar deployment structure network; Step S163: performing millimeter wave propagation simulation based on the millimeter wave radar deployment structure model to obtain millimeter wave propagation simulation data; Step S164: estimating the propagation coverage range according to the millimeter wave propagation simulation data, thereby obtaining propagation coverage range data; Step S165: Optimize the radar deployment of the millimeter-wave radar deployment structure network according to the propagation coverage data to obtain the in-hospital millimeter-wave radar network, and upload the in-hospital millimeter-wave radar network to the hospital millimeter-wave radar management platform to perform the radar deployment optimization task.

[0009] The present invention ensures the reliability of basic data in subsequent steps by obtaining accurate deployment data. Extracting features helps identify different types of radar data, making subsequent analysis more effective. Providing a basis for the deployment location of the radar helps to improve the coverage effect and data quality of the radar. By combining the radar deployment location with the three-dimensional model of the hospital building, a clear structured network is formed, which is convenient for subsequent analysis and optimization. The construction of the three-dimensional structural model visualizes the radar deployment situation, helps relevant personnel understand and manage the deployment effect, and can reasonably arrange the installation location of the radar to ensure the best coverage and monitoring effect. By simulating the millimeter wave propagation in different environments, the system performance is predicted in advance to avoid problems that may arise after actual deployment. Considering the complex environmental factors in the hospital (such as walls, equipment, etc.), improving the adaptability and accuracy of the propagation model, it can provide a scientific basis for the management to formulate a more effective millimeter wave radar deployment strategy. Accurately estimating the coverage range can better determine whether the existing deployment meets the needs and avoid blind spots. By analyzing the coverage range, radar resources can be more reasonably configured to avoid unnecessary investment. According to the data of the propagation coverage range, the deployment plan is continuously optimized to improve the overall performance of the radar system. Uploading the radar network information to the management platform makes real-time monitoring and management more efficient. Provide basic data for subsequent technology updates and iterations, and support the hospital's continuous development in millimeter-wave radar technology.

[0010] Optionally, step S2 specifically includes: Step S21: Acquire RFID identification data, and extract RFID identification features from the RFID identification data, thereby obtaining RFID identification tag data and RFID identification location data; Step S22: classifying the identification tags according to the RFID identification tag data, thereby obtaining the RFID doctor tag data and the RFID patient tag data; Step S23: performing tag position association on the RFID doctor tag data and the RFID patient tag data respectively based on the RFID identification position data, thereby obtaining the RFID doctor position data and the RFID patient position data; Step S24: integrating hospital personnel distribution according to the RFID doctor location data and the RFID patient location data, thereby obtaining hospital personnel distribution data; Step S25: marking the personnel distribution of the three-dimensional structure model of the hospital building according to the hospital personnel distribution data, thereby obtaining a hospital personnel distribution structure model.

[0011] The RFID technology in the present invention can realize real-time tracking of personnel and objects, and improve the efficiency of hospital management. By extracting features, the accuracy of identification can be improved and the possibility of misidentification can be reduced. Classifying doctor and patient labels facilitates effective management during the medical process and improves communication efficiency. By clarifying identity classification, it helps to improve the security within the hospital and prevent identity fraud. By associating location data, doctors and patients can be accurately located, which facilitates the rational allocation of resources. The accuracy of positioning information helps to improve patients' medical experience and reduce waiting time. Analyzing the distribution data of personnel in the hospital can optimize human resource allocation and improve the efficiency of medical services. Dynamic monitoring of the distribution of hospital personnel helps to respond quickly to emergencies. Displaying the distribution of personnel through a three-dimensional model improves managers' understanding of the hospital layout and helps optimize space utilization.

[0012] Optionally, step S3 specifically includes: Step S31: collecting millimeter-wave reflection signals within the hospital based on the millimeter-wave radar network within the hospital, thereby obtaining millimeter-wave reflection signal data within the hospital; Step S32: performing reflection signal pattern recognition based on the millimeter wave reflection signal data within the hospital, thereby obtaining millimeter wave reflection signal pattern data; Step S33: performing posture recognition of the in-hospital personnel based on the millimeter wave reflection signal pattern data, thereby obtaining posture data of the in-hospital personnel; Step S34: extracting periodic features of the reflected signal according to the millimeter wave reflected signal pattern data, thereby obtaining periodic reflected signal data, and integrating the vital signs of the in-hospital personnel based on the periodic reflected signal data, thereby obtaining the vital signs data of the in-hospital personnel; Step S35: integrating the in-hospital personnel posture data and the in-hospital personnel vital sign data, thereby obtaining the in-hospital personnel posture and vital sign data.

[0013] The millimeter wave radar in the present invention can penetrate clothing and obstacles, provide high-precision reflection signal data, and is suitable for various environments. It can realize real-time monitoring of personnel dynamics, helping hospitals to understand personnel positions and activities in a timely manner. By identifying the reflection signal pattern, valuable information can be extracted to provide a basis for subsequent data processing. Enhance the intelligence level of hospital management, improve data analysis capabilities, and support decision-making. Accurately identifying the posture of personnel helps to analyze the behavior patterns of personnel and assess their health status. Timely detection of abnormal postures (such as falls, etc.) can improve the safety of patients and enhance the responsiveness of medical services. By extracting the features of periodic reflection signals, data related to physiological status can be obtained to support health monitoring. Integrating vital sign data into a comprehensive health record helps doctors make more accurate medical decisions. Integrating posture data with vital sign data can provide a more comprehensive health assessment and enhance the personalization of medical services. Provide data support for hospitals to help make more scientific and accurate decisions in treatment and care.

[0014] Optionally, step S33 is specifically: Step S331: extracting signal reflection amplitude pattern features and signal Doppler frequency shift pattern features from the millimeter wave reflection signal pattern data, thereby obtaining signal reflection amplitude pattern data and signal Doppler frequency shift pattern data; Step S332: performing signal reflection amplitude statistics based on the signal reflection amplitude pattern data, thereby obtaining high-value reflection amplitude signal pattern data and low-value reflection amplitude signal pattern data; Step S333: performing signal Doppler frequency shift statistics according to the signal Doppler frequency shift pattern data, thereby obtaining significant Doppler frequency shift signal pattern data and shallow Doppler frequency shift signal pattern data; Step S334: performing a signal pattern intersection operation on the high-value reflection amplitude signal pattern data and the shallow Doppler frequency shift amount signal pattern data, thereby obtaining the stationary signal pattern data of the personnel; performing a signal pattern intersection operation on the low-value reflection amplitude signal pattern data and the significant Doppler frequency shift amount signal pattern data, thereby obtaining the moving signal pattern data of the personnel; Step S335: construct a personnel posture signal pattern recognition model for the personnel stationary signal pattern data and the personnel moving signal pattern data, and perform in-hospital personnel posture recognition on the millimeter wave reflection signal pattern data through the personnel posture signal pattern recognition model, so as to obtain in-hospital personnel posture data.

[0015] The present invention can effectively identify the existence and motion state of different objects (such as personnel) by extracting reflection amplitude and Doppler shift features. This step provides a basis for subsequent data analysis, ensuring that the extracted information has sufficient discrimination and can reflect the static and moving state of personnel. Dividing the signal into high-value and low-value reflection amplitude pattern data helps to screen out more valuable information. For example, high reflection amplitude usually points to relatively static objects, while low reflection amplitude indicates relatively moving objects. This classification can improve the accuracy of subsequent analysis. By analyzing the Doppler shift, the motion state and speed of the object can be determined, and the signal patterns of significant movement and less movement can be distinguished. This is particularly important for dynamic monitoring, which can capture the activity state of personnel in real time and provide support for real-time monitoring systems. Through intersection operations, static and moving personnel can be identified more accurately. This logical operation combines data with different characteristics, helps to further reduce false alarms, ensures that the system can more reliably identify static or moving personnel, and provide more accurate data support for security monitoring systems. By building a posture recognition model using static and motion signal pattern data, intelligent recognition of people's postures can be achieved. This enables analysis of the behavior of people within the hospital, and can monitor and identify the status of people in real time, providing important support for security, rescue and other applications.

[0016] Optionally, step S34 is specifically: Step S341: extracting periodic features of the reflection signal according to the millimeter wave reflection signal pattern data, thereby obtaining periodic reflection signal data; Step S342: performing periodic signal division based on the periodic reflection signal data, thereby obtaining short periodic reflection signal data and long periodic reflection signal data; Step S343: dividing the periodic reflection signal data into signal waveforms, thereby obtaining peak waveform periodic reflection signal data and flat waveform periodic reflection signal data; Step S344: performing a periodic reflection signal intersection operation based on the short periodic reflection signal data and the peak waveform periodic reflection signal data, thereby obtaining the personnel's periodic heart rate signal data; performing a periodic reflection signal intersection operation based on the long periodic reflection signal data and the smooth waveform periodic reflection signal data, thereby obtaining the personnel's periodic breathing frequency signal data; Step S345: Integrate the vital signs of the in-hospital personnel based on the periodic heart rate signal data and the periodic respiratory rate signal data of the personnel, so as to obtain the vital sign data of the in-hospital personnel.

[0017] By extracting the periodic characteristics of the reflection signal, the present invention can effectively identify and capture the signal pattern related to human activities, which lays the foundation for subsequent signal analysis, can filter out non-periodic noise, improve the availability of the signal, and ensure that the subsequent analysis is more accurate. Dividing the periodic reflection signal data into short-period and long-period signals is helpful for independent analysis of different physiological characteristics (such as heart rate and respiratory rate). This division can help identify signal changes related to different physiological states, and facilitate the targeted extraction of required data. By classifying the signal waveform, waveform features with different physiological significance can be separated, such as spike waveforms are usually related to rapid physiological states, while flat waveforms may be related to stable states. This process makes the interpretation of physiological signals more intuitive and facilitates the identification of abnormal signals or states. For the intersection operation of heart rate and respiratory rate signals, more accurate physiological indicators can be effectively integrated to reflect the health status of personnel. By processing the intersection of short-period and long-period signals, clinically significant physical sign data can be extracted to provide data support for health monitoring and clinical diagnosis. The integrated heart rate and respiratory rate signals form complete vital signs data of hospital personnel, which can provide important basis for medical monitoring, personnel health assessment and emergency response. The integration of this data can provide medical institutions with real-time monitoring capabilities and improve the understanding of patients' health status and management efficiency.

[0018] Optionally, step S4 is specifically: Step S41: dividing the posture and vital sign data of the in-hospital personnel, thereby obtaining the posture data and vital sign data of the in-hospital personnel; Step S42: performing posture duration threshold statistics on the posture data of the in-hospital personnel, thereby obtaining the posture duration threshold data of the personnel, and performing abnormal posture duration detection on the posture data of the in-hospital personnel according to the posture duration threshold data of the personnel, thereby obtaining the abnormal posture duration data of the personnel and the normal posture duration data of the personnel; Step S43: Perform statistics on the range of changes in physical signs according to the physical sign data of the in-hospital personnel, so as to obtain the time data of the range of changes in the high-value physical signs of the personnel and the time data of the range of changes in the low-value physical signs of the personnel; Step S44: performing a time intersection operation on the personnel's high-level vital sign change amplitude time data and the personnel's abnormal posture duration data, so as to obtain abnormal personnel vital sign data; performing a time intersection operation on the personnel's low-level vital sign change amplitude time data and the personnel's normal posture duration data, so as to obtain normal personnel vital sign data; Step S45: merging the abnormal person's vital sign data and the normal person's vital sign data to obtain abnormal person's vital sign detection data; Step S46: constructing a dynamic prediction model for abnormal vital signs of personnel based on the abnormal vital sign detection data of personnel.

[0019] The present invention can effectively classify the data by dividing the posture and vital sign data of the hospital personnel, making the subsequent analysis more efficient, laying the foundation for the subsequent abnormal detection and model building, and ensuring that each data set has a clear purpose. By counting the posture duration, it is possible to identify which postures are normal and which may be abnormal, and this process helps to reduce the false alarm rate. Through the threshold of the duration, flexible standards can be set for different scenarios to meet actual needs. The statistics of the amplitude of changes in vital signs can help identify changes in the health status of personnel and timely discover potential health problems. By dividing the amplitude of changes in high and low amounts, personnel with different risk levels can be classified and managed, and targeted intervention can be provided. Through time intersection, personnel with abnormal postures and vital sign changes in a specific time period can be accurately identified, and the accuracy of abnormal detection can be improved. By combining posture and vital sign data, the health status of personnel and its changes can be understood more comprehensively. By merging the vital sign data of abnormal and normal personnel, a comprehensive data set can be established, which is convenient for in-depth analysis and model training. The merged data can be more conveniently visualized to help decision makers quickly understand the situation. By building a dynamic prediction model, potential abnormal situations can be predicted in advance to provide support for preventive measures. The model can learn patterns in historical data and provide intelligent suggestions for hospital management, improving overall safety and efficiency.

[0020] Optionally, step S6 specifically includes: Step S61: performing personnel RFID tag identification on the abnormal physical sign prediction personnel location data according to the hospital personnel distribution data, thereby obtaining the abnormal physical sign prediction personnel RFID tag data; Step S62: performing RFID tag spatial flow prediction on the RFID tag data of the abnormal physical sign prediction personnel through the hospital personnel distribution structure model, thereby obtaining the abnormal physical sign prediction personnel spatial flow data; Step S63: acquiring RFID identifier deployment data, and selecting an early warning RFID identifier for the abnormal physical sign prediction personnel spatial flow data and the RFID identifier deployment data, thereby obtaining the RFID identifier data to be warned; Step S64: Integrate the warning information based on the RFID tag data of the person predicted to have abnormal physical signs and the RFID identifier data to be warned, so as to obtain the RFID warning information, and transmit it to the RFID management platform to execute the personnel warning task.

[0021] The present invention can monitor the location of personnel in real time and help locate abnormal signs prediction personnel through the identification of RFID tags. The collected RFID tag data provides a basis for subsequent analysis to ensure the accuracy and real-time nature of the data. Using the spatial flow data of RFID tags, the flow of abnormal signs prediction personnel can be monitored in real time to discover potential risk areas. By predicting the flow, the hospital can reasonably allocate human resources and medical equipment to reduce unnecessary waiting time. By comprehensively analyzing the spatial flow data of abnormal signs prediction personnel and the deployment data of RFID identifiers, the most effective identifier can be accurately selected to enhance the accuracy of early warning. Selecting a suitable identifier helps to respond quickly, take timely intervention measures for potential risks, and enhance the emergency response capabilities of the hospital. By integrating various types of data (RFID tag data and early warning identifier data), all relevant information is ensured to be centralized, which is convenient for rapid analysis and decision-making. After the RFID early warning information is transmitted to the management platform, hospital managers can quickly obtain information, conduct emergency treatment, and improve management efficiency.

[0022] Optionally, the present specification also provides a system for constructing a medical health dynamic prediction model, which is used to execute the method for constructing a medical health dynamic prediction model as described above, and the system for constructing a medical health dynamic prediction model includes: The three-dimensional structure modeling module is used to obtain hospital structure data, and perform three-dimensional structure modeling of the hospital building based on the hospital structure data, thereby obtaining a three-dimensional structure model of the hospital building; based on the three-dimensional structure model of the hospital building, the in-hospital millimeter wave radar network is constructed, thereby obtaining an in-hospital millimeter wave radar network; The personnel distribution analysis module is used to obtain RFID identification data and mark the personnel distribution of the three-dimensional structure model of the hospital building according to the RFID identification data, so as to obtain the hospital personnel distribution structure model; A personnel posture and vital signs integration module is used to collect millimeter-wave reflection signals in the hospital based on the millimeter-wave radar network in the hospital, so as to obtain millimeter-wave reflection signal data in the hospital, and integrate the posture and vital signs of personnel in the hospital based on the millimeter-wave reflection signal data in the hospital, so as to obtain the posture and vital signs data of personnel in the hospital; The abnormal physical sign monitoring module is used to monitor the abnormal physical sign data of the hospital personnel's posture and physical sign data, thereby obtaining the personnel's abnormal physical sign detection data, and constructing the personnel's abnormal physical sign dynamic prediction model based on the personnel's abnormal physical sign detection data; The abnormal physical sign prediction module is used to dynamically predict the abnormal physical signs of personnel in the hospital by using the abnormal physical sign dynamic prediction model, so as to obtain abnormal physical sign predicted personnel data, and to identify the abnormal physical sign predicted personnel positions by using the hospital personnel distribution structure model, so as to obtain abnormal physical sign predicted personnel position data; The early warning information integration module is used to predict personnel location data based on abnormal physical signs and integrate personnel RFID early warning information, thereby obtaining RFID early warning information and transmitting it to the RFID management platform to execute personnel early warning tasks.

[0023] The system for constructing a dynamic prediction model for medical health of the present invention can implement any one of the methods for constructing a dynamic prediction model for medical health of the present invention, and is used to combine the operation and signal transmission medium between various modules to complete the method for constructing a dynamic prediction model for medical health. The modules within the system cooperate with each other, thereby improving the real-time and accuracy of data collection, and also enhancing the early warning capability of abnormal physical signs through the dynamic prediction model. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments thereof made with reference to the following drawings: Figure 1 A schematic diagram of the steps of the method for constructing a dynamic prediction model for medical health according to the present invention; Figure 2 Detailed step flow diagram of step S1 in the present invention; Figure 3 Detailed step flow diagram of step S2 in the present invention; The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0025] The technical method of the present invention is described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0026] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0027] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0028] To achieve this, please refer to Figures 1 to 3 The present invention provides a method for constructing a dynamic prediction model for medical health, the method comprising the following steps: Step S1: Acquire hospital structure data, and perform three-dimensional structural modeling of the hospital building according to the hospital structure data, thereby obtaining a three-dimensional structural model of the hospital building; construct an in-hospital millimeter-wave radar network based on the three-dimensional structural model of the hospital building, thereby obtaining an in-hospital millimeter-wave radar network; In this embodiment, detailed structural data of the hospital, including floor plans, walls, doors and windows, etc., are obtained through laser scanning or building information modeling (BIM) technology. Using software such as Revit or SketchUp, these data are input for 3D modeling to generate a 3D structural model of the hospital building. Subsequently, based on the model, the location of the millimeter-wave radar equipment is configured, and the millimeter-wave radar network in the hospital is designed to ensure coverage of all important areas (such as wards, operating rooms, reception areas), forming a complete millimeter-wave radar network.

[0029] Step S2: Acquire RFID identification data, and perform personnel distribution marking on the three-dimensional structure model of the hospital building according to the RFID identification data, thereby obtaining a hospital personnel distribution structure model; In this embodiment, when obtaining RFID identification data, RFID readers are installed in various key areas of the hospital, combined with the RFID tags worn by each staff member and patient, to collect personnel location data in real time. This data is integrated into the three-dimensional structural model of the hospital building to mark the distribution of personnel in the building. In specific implementation, the GIS system can be used to visualize the RFID data to help hospital managers understand the real-time flow and density of personnel.

[0030] Step S3: collecting millimeter-wave reflection signals in the hospital based on the millimeter-wave radar network in the hospital, thereby obtaining millimeter-wave reflection signal data in the hospital, and integrating posture and vital signs of personnel in the hospital according to the millimeter-wave reflection signal data in the hospital, thereby obtaining posture and vital signs data of personnel in the hospital; In this embodiment, multiple millimeter wave sensors are deployed based on the millimeter wave radar network in the hospital, and signals are collected through time synchronization technology. In the signal processing process, the high-frequency signal analysis algorithm is used to extract the reflected signal and generate the millimeter wave reflected signal data in the hospital. Combined with these signal data, machine learning algorithms (such as convolutional neural networks) are used to analyze the posture and vital signs of personnel, extract key information such as walking, sitting, standing, etc., and generate posture and vital sign data of personnel in the hospital.

[0031] Step S4: abnormal physical sign monitoring is performed on the posture and physical sign data of the personnel in the hospital, thereby obtaining abnormal physical sign detection data of the personnel, and a dynamic prediction model for abnormal physical signs of the personnel is constructed according to the abnormal physical sign detection data of the personnel; In this embodiment, an abnormal detection algorithm, such as an isolation forest or a support vector machine (SVM), is applied to the collected posture and vital sign data to monitor abnormal vital signs. The normal and abnormal postures (such as falls and abnormal gait) are identified by training the model, and personnel vital sign abnormality detection data is generated. Based on these data, a dynamic prediction model is constructed to use historical data and real-time data to predict possible abnormal situations in the future.

[0032] Step S5: using the personnel abnormal physical sign dynamic prediction model to dynamically predict the posture physical sign data of the in-hospital personnel, thereby obtaining abnormal physical sign predicted personnel data, and using the hospital personnel distribution structure model to identify the abnormal physical sign predicted personnel positions on the abnormal physical sign predicted personnel data, thereby obtaining abnormal physical sign predicted personnel position data; In this embodiment, the established dynamic prediction model is used to analyze the real-time posture and vital signs data to identify possible abnormal vital signs predicted personnel. Combined with the hospital personnel distribution structure model, the specific positions of these personnel are located through spatial location algorithms (such as KNN or distance matrix analysis), and the abnormal vital signs predicted personnel location data are generated to provide information support for subsequent warnings.

[0033] Step S6: Integrate the predicted personnel RFID warning information based on the personnel location data predicted by abnormal physical signs, thereby obtaining RFID warning information and transmitting it to the RFID management platform to execute the personnel warning task.

[0034] In this embodiment, the location data of the personnel with abnormal physical signs is predicted and RFID warning information is integrated. Through the integrated RFID management platform, the location information of the personnel with abnormal physical signs is combined with the real-time monitoring data to achieve dynamic warning. For example, if abnormal physical signs frequently appear in a certain area, the system automatically generates warning information and notifies relevant medical staff to ensure timely response and processing.

[0035] Optionally, step S1 specifically includes: Step S11: Acquire hospital structure data, and divide the hospital structure data into regional structures, thereby obtaining hospital ward regional structure data and hospital outpatient regional structure data; In this embodiment, the basic structural data of the hospital is obtained through the hospital management system, including the floor plan of the building and the functional division of each area (such as wards, outpatient clinics, operating rooms, etc.). The building floor plan of the hospital is analyzed using geographic information system (GIS) technology, and divided according to the functional areas and the physical structure of the building. The specific implementation method is to use GIS software to import the building floor plan into the system, and by setting different attributes (such as the number of wards, outpatient clinic types), mark the data and divide it into ward areas and outpatient areas, so as to obtain the corresponding regional structure data.

[0036] Step S12: integrating the regional spatial layout of the hospital ward regional structure data and the hospital outpatient regional structure data respectively, thereby obtaining the hospital ward regional spatial layout data and the hospital outpatient regional spatial layout data; In this embodiment, after obtaining the structural data of the ward area and the outpatient area, the layout of these two types of areas is integrated using a spatial analysis tool. Graphics processing software (such as AutoCAD or SketchUp) can be used for spatial analysis. The structural data of the ward and outpatient department can be imported into the graphics processing software, and the layout visualization can be performed based on the spatial utilization efficiency of the area and the equipment deployment location of the area, and finally the spatial layout data of the hospital ward area and the spatial layout data of the hospital outpatient area are output.

[0037] Step S13: performing a regional spatial channel intersection operation on the hospital ward regional spatial layout data and the hospital outpatient regional spatial layout data, thereby obtaining regional connection channel data; In this embodiment, spatial analysis software (such as QGIS or ArcGIS) is used to perform an intersection operation on the spatial layout data of the ward area and the spatial layout data of the outpatient area. This operation will identify the public passages, entrances and exits, and intersection areas between different areas. For example, a map layer is generated to identify the connection points between the main passage of the hospital and each area, thereby forming regional connection channel data. This can provide the necessary information basis for subsequent layout analysis and network construction.

[0038] Step S14: performing regional structural spatial association on the hospital ward regional structural data and the hospital outpatient regional structural data according to the regional connection channel data, thereby obtaining hospital regional structural connection data; In this embodiment, according to the obtained regional connection channel data, a graph theory algorithm (such as Dijkstra algorithm) is used to analyze the regional structural connection of the hospital. By constructing an adjacency matrix, the ward area and the outpatient area are associated to form regional structural connection data. The graph theory method can also be used to regard each area as a node and each channel as an edge to construct a regional connection network. In this way, the connection between the ward area and the outpatient area can be clearly displayed, such as direct channels and indirect channels.

[0039] Step S15: performing three-dimensional structural modeling of the hospital building according to the hospital area structural connection data, thereby obtaining a three-dimensional structural model of the hospital building; In this embodiment, 3D modeling is performed based on the hospital regional structural connection data. Modeling software (such as Revit or SketchUp) is used to input the 3D structure and structural materials of the ward and outpatient areas into the software to construct a 3D structural model of the hospital building. Combined with the regional connection data, it is ensured that the connection channels between the various areas in the model are accurately represented.

[0040] Step S16: Acquire the hospital millimeter-wave radar deployment data, and construct an in-hospital millimeter-wave radar network for the three-dimensional structure model of the hospital building according to the hospital millimeter-wave radar deployment data, thereby obtaining the in-hospital millimeter-wave radar network.

[0041] In this embodiment, the deployment data of the millimeter-wave radar including the radar installation location and signal strength is obtained. Based on this data, the three-dimensional structure model of the hospital is networked. Using professional software (such as MATLAB or Simulink), the coverage area of ​​the radar and the connection between them are marked in the model, so as to obtain the millimeter-wave radar network diagram in the hospital, ensure the effective coverage of the radar network and meet the real-time monitoring needs of the hospital.

[0042] Optionally, step S16 is specifically: Step S161: Acquire hospital millimeter wave radar deployment data, and perform feature extraction on the hospital millimeter wave radar deployment data, thereby obtaining radar deployment location data; In this embodiment, the deployment data of the existing millimeter-wave radar equipment in the hospital is collected, including the specific location, model, operating frequency and parameter settings of each radar. In addition, it is necessary to collect environmental factors related to the radar, such as wall materials, furniture distribution, etc., which will affect the propagation characteristics of the millimeter-wave signal. Through data cleaning and sorting, structured radar deployment data is formed, laying the foundation for subsequent feature extraction. Subsequently, the specific location, orientation and other position parameters of each radar are feature extracted to obtain radar deployment location data.

[0043] Step S162: constructing a millimeter wave radar deployment network for the hospital millimeter wave radar deployment data and the three-dimensional structure model of the hospital building according to the radar deployment location data, thereby obtaining a millimeter wave radar deployment structure network; In this embodiment, based on the radar deployment location data, the deployment of the hospital millimeter-wave radar is combined with the three-dimensional structural model of the hospital building to construct a deployment network of the millimeter-wave radar. Computer-aided design (CAD) software is used to superimpose the radar position and the three-dimensional model of the hospital to analyze the relationship between the radar coverage area and the building structure. Through graph algorithms and network optimization algorithms (such as the shortest path algorithm), a millimeter-wave radar deployment structure network is generated to ensure that the signal coverage of each radar is maximized and complementary to improve the overall monitoring capability. The installation location of the millimeter-wave radar can also be accurately marked on the building model. Then, using network construction algorithms including graph algorithms and network optimization algorithms, the network topology of the millimeter-wave radar is designed, taking into account the mutual coverage and signal interference between radars. Different deployment schemes are simulated by simulation software to optimize network connections and signal coverage, and finally a complete millimeter-wave radar deployment structure network is formed, which will help improve the overall monitoring and positioning accuracy.

[0044] Step S163: performing millimeter wave propagation simulation based on the millimeter wave radar deployment structure model to obtain millimeter wave propagation simulation data; In this embodiment, computer simulation software (such as CST Microwave Studio or HFSS) is used to simulate millimeter wave propagation. First, the operating frequency, transmission power and environmental parameters (such as temperature, humidity, etc.) of the millimeter wave radar need to be input. By setting the three-dimensional model of the hospital building in detail, the propagation characteristics of the signal in different materials (such as concrete, glass, etc.) are simulated. The simulation will generate a detailed signal intensity distribution map to help analyze the propagation characteristics of millimeter waves inside the hospital and provide basic data for subsequent coverage estimation.

[0045] Step S164: estimating the propagation coverage range according to the millimeter wave propagation simulation data, thereby obtaining propagation coverage range data; In this embodiment, after obtaining the millimeter wave propagation simulation data, the propagation coverage is estimated. This process requires analyzing the signal strength and coverage of different deployment locations, and using statistical methods to evaluate the signal strength distribution of each area. Interpolation methods (such as Kriging interpolation) can be used to infer the coverage of unmeasured areas, and a coverage map can be generated through visualization tools. The goal is to identify the signal strength of each area in the hospital and its effective coverage to form propagation coverage data. Algorithms (such as Monte Carlo simulation or signal strength attenuation model) can also be used to analyze the simulation data to determine the effective coverage area under a specific signal strength threshold.

[0046] Step S165: Optimize the radar deployment of the millimeter-wave radar deployment structure network according to the propagation coverage data to obtain the in-hospital millimeter-wave radar network, and upload the in-hospital millimeter-wave radar network to the hospital millimeter-wave radar management platform to perform the radar deployment optimization task.

[0047] In this embodiment, the millimeter-wave radar deployment structure network is optimized based on the propagation coverage data. Use optimization algorithms (such as genetic algorithms, particle swarm optimization, etc.) to evaluate the impact of different radar configurations on coverage and find the best deployment solution. The optimization process involves adjusting the placement of existing radars, adding new radars, or changing signal transmission strategies. After completion, the optimized millimeter-wave radar network data is uploaded to the hospital millimeter-wave radar management platform to optimize the actual deployment of the radar, facilitate real-time monitoring and subsequent maintenance, and ensure efficient operation and continuous optimization of the system.

[0048] Optionally, step S2 specifically includes: Step S21: Acquire RFID identification data, and extract RFID identification features from the RFID identification data, thereby obtaining RFID identification tag data and RFID identification location data; In this embodiment, RFID tag signals are captured in real time by deploying RFID readers at different locations in the hospital. Each RFID tag has a built-in unique identification code (UID) that can identify objects or people. The data extraction module parses the signal and extracts feature information, including the tag's UID, timestamp, and signal strength. Through data cleaning, it is ensured that accurate RFID identification tag data is obtained, such as the UID of doctors and patients, and their corresponding identification location data, such as "Emergency Department".

[0049] Step S22: classifying the identification tags according to the RFID identification tag data, thereby obtaining the RFID doctor tag data and the RFID patient tag data; In this embodiment, the extracted RFID tag data is classified according to preset classification standards (such as UID prefix, tag type, etc.). Through database query, the tags are divided into two categories: RFID doctor tags and RFID patient tags. For example, tags with UIDs starting with "DOC" are classified as doctor tags, while tags with UIDs starting with "PAT" are classified as patient tags. The system automatically generates a list of doctor and patient tags, and records the number and basic information of each type of tag.

[0050] Step S23: performing tag position association on the RFID doctor tag data and the RFID patient tag data respectively based on the RFID identification position data, thereby obtaining the RFID doctor position data and the RFID patient position data; In this embodiment, based on the obtained RFID identification location data, the system associates each doctor tag with the patient tag. Using the coordinate matching algorithm, the location information of the nearest RFID reader / writer for each tag is identified. Assuming that a doctor is in the operating room, the doctor's UID and its corresponding operating room coordinates will be recorded to form the RFID doctor location data. For example, the doctor's UID "DOC001" is associated with the nearest location coordinates (X:10, Y:20) of his UID.

[0051] Step S24: integrating hospital personnel distribution according to the RFID doctor location data and the RFID patient location data, thereby obtaining hospital personnel distribution data; In this embodiment, after obtaining the RFID doctor and patient location data, the data analysis tool is used to integrate all personnel location data to generate hospital personnel distribution data. The heat map algorithm is used to analyze the personnel concentration areas in different time periods and identify high-mobility areas, such as emergency rooms and wards. The system generates a visual report showing the distribution of personnel. For example, in a certain time period, there are 10 doctors and 30 patients in the emergency room, which is convenient for management and resource allocation.

[0052] Step S25: marking the personnel distribution of the three-dimensional structure model of the hospital building according to the hospital personnel distribution data, thereby obtaining a hospital personnel distribution structure model.

[0053] In this embodiment, the hospital's building information model (BIM) technology is used to overlay the hospital's personnel distribution data onto the three-dimensional structural model. Each department and floor is marked with the corresponding number of doctors and patients, and different colors are used to represent the density of personnel. This model can be used to optimize hospital layout and resource allocation and improve management efficiency.

[0054] Optionally, step S3 specifically includes: Step S31: collecting millimeter-wave reflection signals within the hospital based on the millimeter-wave radar network within the hospital, thereby obtaining millimeter-wave reflection signal data within the hospital; In this embodiment, the millimeter wave radar sensor in the constructed in-hospital millimeter wave radar network transmits millimeter wave signals at a specific frequency (such as 76-81GHz) and receives reflected signals. The sampling rate of each sensor is set to 100Hz to obtain high-precision reflection data. Data collection lasts for 10 minutes, and reflection signals in different time periods are collected for subsequent analysis.

[0055] Step S32: performing reflection signal pattern recognition based on the millimeter wave reflection signal data within the hospital, thereby obtaining millimeter wave reflection signal pattern data; In this embodiment, a deep learning algorithm is used to perform pattern recognition on the collected reflection signal. First, the frequency domain features are extracted using feature extraction technology (such as short-time Fourier transform). Then, a convolutional neural network (CNN) is used to train the frequency domain features of the reflection signal data to identify different signal patterns. In addition, a threshold judgment algorithm is set to screen out effective reflection signal patterns, thereby obtaining millimeter wave reflection signal pattern data to ensure that each posture and activity state can be accurately identified.

[0056] Step S33: performing posture recognition of the in-hospital personnel based on the millimeter wave reflection signal pattern data, thereby obtaining posture data of the in-hospital personnel; In this embodiment, a convolutional neural network (CNN) is used for posture recognition based on the obtained signal pattern data. The network input is the feature map extracted from the signal, and through multi-layer convolution and pooling operations, the probabilities of different posture categories are finally output. During the recognition process, the model is trained using the labeled data set to improve the recognition accuracy, and finally the posture data of the hospital personnel is generated.

[0057] Step S34: extracting periodic features of the reflected signal according to the millimeter wave reflected signal pattern data, thereby obtaining periodic reflected signal data, and integrating the vital signs of the in-hospital personnel based on the periodic reflected signal data, thereby obtaining the vital signs data of the in-hospital personnel; In this embodiment, the obtained reflection signal pattern data is analyzed in time series, and the periodic characteristics of the signal are extracted using wavelet transform. Physiological characteristics such as heart rate and respiratory rate are identified, and characteristic thresholds are set to filter noise. The extracted periodic characteristic data are integrated to form the vital sign data of the hospital personnel, and compared and analyzed with the pattern data to further understand the health status of the personnel.

[0058] Step S35: integrating the in-hospital personnel posture data and the in-hospital personnel vital sign data, thereby obtaining the in-hospital personnel posture and vital sign data.

[0059] In this embodiment, the posture data and the vital sign data are integrated, and a weighted algorithm is used to evaluate the correlation between each posture and vital sign. For example, the correlation coefficient obtained by calculation reflects that a certain posture (such as exercise) is highly correlated with a specific physiological state (such as increased heart rate). Then, the data fusion technology is used to fuse several groups of postures and vital signs with high correlation, and the comprehensive posture and vital sign data of the hospital personnel is output to provide more comprehensive information support for subsequent applications.

[0060] Optionally, step S33 is specifically: Step S331: extracting signal reflection amplitude pattern features and signal Doppler frequency shift pattern features from the millimeter wave reflection signal pattern data, thereby obtaining signal reflection amplitude pattern data and signal Doppler frequency shift pattern data; In this embodiment, the reflected signal data is collected by the millimeter wave radar device. First, the signal is analyzed in the frequency domain using the fast Fourier transform (FFT) to extract the signal reflection amplitude pattern characteristics. This involves converting the original signal into a frequency domain representation and identifying the amplitude of each frequency component. Then, the Doppler effect analysis is used to extract the signal Doppler frequency shift pattern characteristics, calculate the frequency change of the signal, and then obtain information reflecting the relative motion of the target. This process can filter noise by setting a threshold to improve the validity and accuracy of the data.

[0061] Step S332: performing signal reflection amplitude statistics based on the signal reflection amplitude pattern data, thereby obtaining high-value reflection amplitude signal pattern data and low-value reflection amplitude signal pattern data; In this embodiment, based on the obtained signal reflection amplitude pattern data, the reflection amplitude is classified using statistical analysis methods (such as mean, variance, etc.). A certain threshold is set, and signals with reflection amplitudes higher than this value are defined as "high reflection amplitude signal pattern data", while those lower than this value are defined as "low reflection amplitude signal pattern data". The distribution of reflection amplitudes can be visualized by using a histogram, thereby effectively identifying and separating different categories of signal patterns.

[0062] Step S333: performing signal Doppler frequency shift statistics according to the signal Doppler frequency shift pattern data, thereby obtaining significant Doppler frequency shift signal pattern data and shallow Doppler frequency shift signal pattern data; In this embodiment, the frequency change value extracted from the signal Doppler frequency shift pattern data. By calculating the mean and standard deviation of the Doppler frequency shift, the signal can be divided into "significant Doppler frequency shift amount signal pattern data" and "simple Doppler frequency shift amount signal pattern data". Specifically, the threshold discrimination method can be used to set different frequency shift thresholds for different application scenarios (such as crowd movement, stationary state, etc.) to enhance the accuracy of recognition.

[0063] Step S334: performing a signal pattern intersection operation on the high-value reflection amplitude signal pattern data and the shallow Doppler frequency shift amount signal pattern data, thereby obtaining the stationary signal pattern data of the personnel; performing a signal pattern intersection operation on the low-value reflection amplitude signal pattern data and the significant Doppler frequency shift amount signal pattern data, thereby obtaining the moving signal pattern data of the personnel; In this embodiment, an intersection operation is performed on the "high reflection amplitude signal pattern data" and the "shallow Doppler frequency shift amount signal pattern data" to obtain the "personnel stationary signal pattern data". The intersection operation is performed on the "low reflection amplitude signal pattern data" and the "significant Doppler frequency shift amount signal pattern data" to obtain the "personnel movement signal pattern data". In implementation, the intersection of the pattern data can be achieved through logical operations (such as AND operations) to ensure that the screened signals accurately reflect the state of the target.

[0064] Step S335: construct a personnel posture signal pattern recognition model for the personnel stationary signal pattern data and the personnel moving signal pattern data, and perform in-hospital personnel posture recognition on the millimeter wave reflection signal pattern data through the personnel posture signal pattern recognition model, so as to obtain in-hospital personnel posture data.

[0065] In this embodiment, a personnel posture signal pattern recognition model is constructed based on the personnel stationary signal pattern data and the personnel motion signal pattern data. A machine learning algorithm (such as a support vector machine, a neural network, etc.) is used for training, with the reflection signal as input and the posture recognition as output, so that the model can recognize and classify different personnel postures. By inputting the millimeter wave reflection signal pattern data into the model, real-time recognition of the posture of personnel in the hospital is achieved. In specific implementation, a feedback mechanism can be set up to continuously optimize the model parameters using the recognition results, thereby improving the accuracy and robustness of recognition.

[0066] Optionally, step S34 is specifically: Step S341: extracting periodic features of the reflection signal according to the millimeter wave reflection signal pattern data, thereby obtaining periodic reflection signal data; In this embodiment, the periodic features in the millimeter wave reflection signal pattern data are extracted by time domain analysis. The time domain signal is converted into the frequency domain using Fast Fourier Transform (FFT) to identify the main frequency components. The noise is filtered by setting a threshold, and only the periodic signals above the threshold are retained to obtain valid periodic reflection signal data.

[0067] Step S342: performing periodic signal division based on the periodic reflection signal data, thereby obtaining short periodic reflection signal data and long periodic reflection signal data; In this embodiment, based on the extracted periodic reflection signal data, the sliding window method (the window length can be set to 500ms and the step length is 100ms) is used to divide the signal into multiple short time periods. The spectrum characteristics in each window are further analyzed, and the signal is divided into short-periodic reflection signals (for example, the period is less than 1s) and long-periodic reflection signals (for example, the period is greater than 1s) by calculating the period and frequency of the signal.

[0068] Step S343: dividing the periodic reflection signal data into signal waveforms, thereby obtaining peak waveform periodic reflection signal data and flat waveform periodic reflection signal data; In this embodiment, waveform analysis is performed on the periodic reflection signal data, and a peak detection algorithm is used to identify the peak waveform and the gentle waveform in the signal. The peak waveform is defined as a signal whose instantaneous amplitude exceeds a certain threshold (such as a maximum amplitude of 0.8), and the gentle waveform is a signal with a small amplitude change (for example, a rate of change of less than 0.1). Through these characteristics, peak waveform periodic reflection signal data and gentle waveform periodic reflection signal data are obtained respectively.

[0069] Step S344: performing a periodic reflection signal intersection operation based on the short periodic reflection signal data and the peak waveform periodic reflection signal data, thereby obtaining the personnel's periodic heart rate signal data; performing a periodic reflection signal intersection operation based on the long periodic reflection signal data and the smooth waveform periodic reflection signal data, thereby obtaining the personnel's periodic breathing frequency signal data; In this embodiment, the short-periodic reflection signal data and the peak waveform periodic reflection signal data are intersected, and the heart rate-related features are extracted using an intersection algorithm (such as an AND operation) to generate personnel periodic heart rate signal data. Similarly, the long-periodic reflection signal data and the smooth waveform periodic reflection signal data are intersected to extract respiratory frequency-related features to generate personnel periodic respiratory frequency signal data.

[0070] Step S345: Integrate the vital signs of the in-hospital personnel based on the periodic heart rate signal data and the periodic respiratory rate signal data of the personnel, so as to obtain the vital sign data of the in-hospital personnel.

[0071] In this embodiment, the heart rate and respiratory rate of the personnel are estimated based on the periodic heart rate signal data and the periodic respiratory rate signal data of the personnel, respectively, and then the weighted average method or Kalman filtering algorithm is used to integrate the data to obtain the vital signs data of the hospital personnel. It is stored in the database system and can be displayed in real time on the monitoring terminal to provide effective health monitoring basis for medical staff. By analyzing the waveform of the heart rate signal, the peak of the heartbeat is identified, which can be achieved by finding the local maximum value in the signal. The heart rate is calculated according to the time interval between the peaks, and the formula is usually used: \text{heart rate (bpm)}=\frac{60}{\text{RR interval (seconds)}}, where the RR interval is the time difference between two adjacent heartbeat peaks. In order to improve the stability of the heart rate estimation, the calculated heart rate value can be subjected to sliding average processing. Estimation of respiratory rate can be: waveform analysis of respiratory signals to identify the cycles of inhalation and exhalation; by analyzing the changes in the signal, determine the switching point between inhalation and exhalation to calculate the time of each cycle; use a similar formula to calculate the respiratory rate: \text{Respiratory rate (times / minute)}=\frac{60}{\text{Cycle time (seconds)}}. Similarly, a sliding average can be performed on the respiratory rate to reduce instantaneous fluctuations. Then different weights are assigned to heart rate and respiratory rate, and weighted averages are performed based on clinical experience or statistical analysis. This approach ensures that more important data has a greater impact on the final result. Using the Kalman filter algorithm, heart rate and respiratory rate are used as state variables, and the estimated values ​​are dynamically updated by establishing a system model and a measurement model to improve the accuracy and stability of the estimate. The integrated data can be updated in real time to ensure that medical staff can obtain the latest health monitoring data in a timely manner.

[0072] Optionally, step S4 is specifically: Step S41: dividing the posture and vital sign data of the in-hospital personnel, thereby obtaining the posture data and vital sign data of the in-hospital personnel; In this embodiment, feature engineering technology is used to divide the posture and vital sign data of hospital personnel into posture data (such as movement, stillness, etc.) and vital sign data (such as heart rate, respiratory rate).

[0073] Step S42: performing posture duration threshold statistics on the posture data of the in-hospital personnel, thereby obtaining the posture duration threshold data of the personnel, and performing abnormal posture duration detection on the posture data of the in-hospital personnel according to the posture duration threshold data of the personnel, thereby obtaining the abnormal posture duration data of the personnel and the normal posture duration data of the personnel; In this embodiment, the time series analysis method is used to calculate the duration of each posture for the posture data of the hospital personnel. For example, the sliding window technology can be used to count the time distribution of each posture, and a threshold value (such as a duration of more than 5 minutes) can be set. Based on these data, an abnormal detection algorithm (such as a Z-score-based method) is applied to identify postures with abnormal duration (such as long-term immobility greater than the set threshold of 4 hours or a nighttime immobility threshold of 10 hours), and finally two sets of data are generated: abnormal posture duration data and normal duration data.

[0074] Step S43: Perform statistics on the range of changes in physical signs according to the physical sign data of the in-hospital personnel, so as to obtain the time data of the range of changes in the high-value physical signs of the personnel and the time data of the range of changes in the low-value physical signs of the personnel; In this embodiment, the variation range of the vital sign data is obtained by calculating the standard deviation of the data. For example, the fluctuation of the heart rate is recorded, and the thresholds of the high and low variation ranges are set (such as a high variation when it is higher than 2 times the standard deviation). Through these calculations, the high and low vital sign variation range time data will be obtained, which can be displayed using a chart visualization tool to help analysis.

[0075] Step S44: performing a time intersection operation on the personnel's high-level vital sign change amplitude time data and the personnel's abnormal posture duration data, so as to obtain abnormal personnel vital sign data; performing a time intersection operation on the personnel's low-level vital sign change amplitude time data and the personnel's normal posture duration data, so as to obtain normal personnel vital sign data; In this embodiment, the data intersection method is used to compare the high-value vital sign change amplitude time data with the abnormal posture duration data, identify the time period that occurs under these two conditions at the same time, and obtain the abnormal personnel vital sign data. Relatively speaking, the same intersection operation is performed on the low-value vital sign change amplitude time data and the normal duration data to obtain the normal personnel vital sign data. Data processing can be performed efficiently using Python's Pandas library.

[0076] Step S45: merging the abnormal person's vital sign data and the normal person's vital sign data to obtain abnormal person's vital sign detection data; In this embodiment, the abnormal person's vital sign data is merged with the normal person's vital sign data to construct a comprehensive data set. This data set will include timestamp, person ID, vital sign value and its status (abnormal or normal). A database management system (such as MySQL) can be used to store and manage these data to ensure data integrity and traceability.

[0077] Step S46: constructing a dynamic prediction model for abnormal vital signs of personnel based on the abnormal vital sign detection data of personnel.

[0078] In this embodiment, a dynamic prediction model is constructed based on the abnormal vital sign detection data of the personnel. For example, a machine learning algorithm (such as random forest or LSTM) is used to train the abnormal vital sign detection data of the personnel to establish an abnormal prediction model. The accuracy of the model can then be evaluated by a cross-validation method, and the model can be updated and optimized online using real-time data streams.

[0079] Optionally, step S6 specifically includes: Step S61: performing personnel RFID tag identification on the abnormal physical sign prediction personnel location data according to the hospital personnel distribution data, thereby obtaining the abnormal physical sign prediction personnel RFID tag data; In this embodiment, the RFID tags of the marked personnel in the hospital personnel distribution data and their corresponding location information are compared with the location data of the abnormal physical signs predicted personnel to identify the RFID tag information and corresponding location information of the abnormal physical signs predicted personnel.

[0080] Step S62: performing RFID tag spatial flow prediction on the RFID tag data of the abnormal physical sign prediction personnel through the hospital personnel distribution structure model, thereby obtaining the abnormal physical sign prediction personnel spatial flow data; In this embodiment, based on the personnel distribution structure model of the hospital, the RFID tag data of the abnormal signs predicted personnel are analyzed to establish a flow prediction model. This model takes into account the daily activity paths, peak hours for medical consultations, and flow patterns of personnel. By performing regression analysis on historical data, the possible position changes of these personnel within a specific time period are predicted to generate spatial flow data. For example, if a patient is often in a department between 9 and 11 a.m. in previous records, the model will predict his future flow at the same time. The daily activity paths, peak hours for medical consultations, and flow patterns can be obtained by analyzing the hospital's RFID records.

[0081] Step S63: acquiring RFID identifier deployment data, and selecting an early warning RFID identifier for the abnormal physical sign prediction personnel spatial flow data and the RFID identifier deployment data, thereby obtaining the RFID identifier data to be warned; In this embodiment, the deployment data of the RFID identifier is obtained, including the location information, effective identification range and working status of the identifier. These deployment data are cross-analyzed with the spatial flow data of abnormal physical sign prediction personnel to determine which RFID identifiers can effectively capture abnormal physical sign personnel. Through algorithm evaluation, the identifier that can monitor and send out warning signals in real time on its expected path is selected, and finally the RFID identifier data to be warned is obtained. For example, if a certain identifier has a high reading efficiency in a specific area, it will be included in the list to be warned.

[0082] Step S64: Integrate the warning information based on the RFID tag data of the person predicted to have abnormal physical signs and the RFID identifier data to be warned, so as to obtain the RFID warning information, and transmit it to the RFID management platform to execute the personnel warning task.

[0083] In this embodiment, the warning information is integrated based on the collected abnormal signs prediction personnel RFID tag data and the RFID identifier data to be warned. The identified abnormal signs, the location information of the prediction personnel, and the status of the warning identifier are integrated into an operational information format. Subsequently, this information is transmitted to relevant medical staff in real time through the hospital's RFID management platform. An automated report will be generated to prompt medical staff to monitor and subsequently handle potential abnormal signs, thereby improving the hospital's emergency response capabilities. For example, if a certain identifier captures abnormal signs within a predetermined time, it will automatically notify the corresponding department to handle it.

[0084] Optionally, the present specification also provides a system for constructing a medical health dynamic prediction model, which is used to execute the method for constructing a medical health dynamic prediction model as described above, and the system for constructing a medical health dynamic prediction model includes: The three-dimensional structure modeling module is used to obtain hospital structure data, and perform three-dimensional structure modeling of the hospital building based on the hospital structure data, thereby obtaining a three-dimensional structure model of the hospital building; based on the three-dimensional structure model of the hospital building, the in-hospital millimeter wave radar network is constructed, thereby obtaining an in-hospital millimeter wave radar network; The personnel distribution analysis module is used to obtain RFID identification data and mark the personnel distribution of the three-dimensional structure model of the hospital building according to the RFID identification data, so as to obtain the hospital personnel distribution structure model; A personnel posture and vital signs integration module is used to collect millimeter-wave reflection signals in the hospital based on the millimeter-wave radar network in the hospital, so as to obtain millimeter-wave reflection signal data in the hospital, and integrate the posture and vital signs of personnel in the hospital based on the millimeter-wave reflection signal data in the hospital, so as to obtain the posture and vital signs data of personnel in the hospital; The abnormal physical sign monitoring module is used to monitor the abnormal physical sign data of the hospital personnel's posture and physical sign data, thereby obtaining the personnel's abnormal physical sign detection data, and constructing the personnel's abnormal physical sign dynamic prediction model based on the personnel's abnormal physical sign detection data; The abnormal physical sign prediction module is used to dynamically predict the abnormal physical signs of personnel in the hospital by using the abnormal physical sign dynamic prediction model, so as to obtain abnormal physical sign predicted personnel data, and to identify the abnormal physical sign predicted personnel positions by using the hospital personnel distribution structure model, so as to obtain abnormal physical sign predicted personnel position data; The early warning information integration module is used to predict personnel location data based on abnormal physical signs and integrate personnel RFID early warning information, thereby obtaining RFID early warning information and transmitting it to the RFID management platform to execute personnel early warning tasks.

[0085] The system for constructing a dynamic prediction model for medical health of the present invention can implement any one of the methods for constructing a dynamic prediction model for medical health of the present invention, and is used to combine the operation and signal transmission medium between various modules to complete the method for constructing a dynamic prediction model for medical health. The modules within the system cooperate with each other, thereby improving the real-time and accuracy of data collection, and also enhancing the early warning capability of abnormal physical signs through the dynamic prediction model.

[0086] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0087] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A method for constructing a dynamic prediction model for medical health, characterized in that: The following steps are involved: Step S1: Acquire hospital structure data, and perform three-dimensional structural modeling of the hospital building according to the hospital structure data, thereby obtaining a three-dimensional structural model of the hospital building; Based on the three-dimensional structural model of the hospital building, the millimeter-wave radar network in the hospital is constructed to obtain the millimeter-wave radar network in the hospital; Step S2: Acquire RFID identification data, and perform personnel distribution marking on the three-dimensional structure model of the hospital building according to the RFID identification data, thereby obtaining a hospital personnel distribution structure model; Step S3: collecting millimeter-wave reflection signals in the hospital based on the millimeter-wave radar network in the hospital, thereby obtaining millimeter-wave reflection signal data in the hospital, and integrating posture and vital signs of personnel in the hospital according to the millimeter-wave reflection signal data in the hospital, thereby obtaining posture and vital signs data of personnel in the hospital; Step S4: abnormal physical sign monitoring is performed on the posture and physical sign data of the personnel in the hospital, thereby obtaining abnormal physical sign detection data of the personnel, and a dynamic prediction model for abnormal physical signs of the personnel is constructed according to the abnormal physical sign detection data of the personnel; Step S5: using the personnel abnormal physical sign dynamic prediction model to dynamically predict the posture physical sign data of the in-hospital personnel, thereby obtaining abnormal physical sign predicted personnel data, and using the hospital personnel distribution structure model to identify the abnormal physical sign predicted personnel positions on the abnormal physical sign predicted personnel data, thereby obtaining abnormal physical sign predicted personnel position data; Step S6: Integrate the predicted personnel RFID warning information based on the personnel location data predicted by abnormal physical signs, thereby obtaining RFID warning information and transmitting it to the RFID management platform to execute the personnel warning task.

2. The method for constructing a dynamic prediction model for medical health according to claim 1, characterized in that: Step S1 is specifically as follows: Step S11: Acquire hospital structure data, and divide the hospital structure data into regional structures, thereby obtaining hospital ward regional structure data and hospital outpatient regional structure data; Step S12: integrating the regional spatial layout of the hospital ward regional structure data and the hospital outpatient regional structure data respectively, thereby obtaining the hospital ward regional spatial layout data and the hospital outpatient regional spatial layout data; Step S13: performing a regional spatial channel intersection operation on the hospital ward regional spatial layout data and the hospital outpatient regional spatial layout data, thereby obtaining regional connection channel data; Step S14: performing regional structural spatial association on the hospital ward regional structural data and the hospital outpatient regional structural data according to the regional connection channel data, thereby obtaining hospital regional structural connection data; Step S15: performing three-dimensional structural modeling of the hospital building according to the hospital area structural connection data, thereby obtaining a three-dimensional structural model of the hospital building; Step S16: Acquire the hospital millimeter-wave radar deployment data, and construct an in-hospital millimeter-wave radar network for the three-dimensional structure model of the hospital building according to the hospital millimeter-wave radar deployment data, thereby obtaining the in-hospital millimeter-wave radar network.

3. The method for constructing a dynamic prediction model for medical health according to claim 2, characterized in that: Step S16 is specifically as follows: Step S161: Acquire hospital millimeter wave radar deployment data, and perform feature extraction on the hospital millimeter wave radar deployment data, thereby obtaining radar deployment location data; Step S162: constructing a millimeter wave radar deployment network for the hospital millimeter wave radar deployment data and the three-dimensional structure model of the hospital building according to the radar deployment location data, thereby obtaining a millimeter wave radar deployment structure network; Step S163: performing millimeter wave propagation simulation based on the millimeter wave radar deployment structure model to obtain millimeter wave propagation simulation data; Step S164: estimating the propagation coverage range according to the millimeter wave propagation simulation data, thereby obtaining propagation coverage range data; Step S165: Optimize the radar deployment of the millimeter-wave radar deployment structure network according to the propagation coverage data to obtain the in-hospital millimeter-wave radar network, and upload the in-hospital millimeter-wave radar network to the hospital millimeter-wave radar management platform to perform the radar deployment optimization task.

4. The method for constructing a dynamic prediction model for medical health according to claim 1, characterized in that: Step S2 is specifically as follows: Step S21: Acquire RFID identification data, and extract RFID identification features from the RFID identification data, thereby obtaining RFID identification tag data and RFID identification location data; Step S22: classifying the identification tags according to the RFID identification tag data, thereby obtaining the RFID doctor tag data and the RFID patient tag data; Step S23: performing tag position association on the RFID doctor tag data and the RFID patient tag data respectively based on the RFID identification position data, thereby obtaining the RFID doctor position data and the RFID patient position data; Step S24: integrating hospital personnel distribution according to the RFID doctor location data and the RFID patient location data, thereby obtaining hospital personnel distribution data; Step S25: marking the personnel distribution of the three-dimensional structure model of the hospital building according to the hospital personnel distribution data, thereby obtaining a hospital personnel distribution structure model.

5. The method for constructing a dynamic prediction model for medical health according to claim 1, characterized in that: Step S3 is specifically as follows: Step S31: collecting millimeter-wave reflection signals within the hospital based on the millimeter-wave radar network within the hospital, thereby obtaining millimeter-wave reflection signal data within the hospital; Step S32: performing reflection signal pattern recognition based on the millimeter wave reflection signal data within the hospital, thereby obtaining millimeter wave reflection signal pattern data; Step S33: performing posture recognition of the in-hospital personnel based on the millimeter wave reflection signal pattern data, thereby obtaining posture data of the in-hospital personnel; Step S34: extracting periodic features of the reflected signal according to the millimeter wave reflected signal pattern data, thereby obtaining periodic reflected signal data, and integrating the vital signs of the in-hospital personnel based on the periodic reflected signal data, thereby obtaining the vital signs data of the in-hospital personnel; Step S35: integrating the in-hospital personnel posture data and the in-hospital personnel vital sign data, thereby obtaining the in-hospital personnel posture and vital sign data.

6. The method for constructing a dynamic prediction model for medical health according to claim 5, characterized in that: Step S33 is specifically as follows: Step S331: extracting signal reflection amplitude pattern features and signal Doppler frequency shift pattern features from the millimeter wave reflection signal pattern data, thereby obtaining signal reflection amplitude pattern data and signal Doppler frequency shift pattern data; Step S332: performing signal reflection amplitude statistics based on the signal reflection amplitude pattern data, thereby obtaining high-value reflection amplitude signal pattern data and low-value reflection amplitude signal pattern data; Step S333: performing signal Doppler frequency shift statistics according to the signal Doppler frequency shift pattern data, thereby obtaining significant Doppler frequency shift signal pattern data and shallow Doppler frequency shift signal pattern data; Step S334: performing a signal pattern intersection operation on the high-value reflection amplitude signal pattern data and the shallow Doppler frequency shift amount signal pattern data, thereby obtaining the stationary signal pattern data of the personnel; performing a signal pattern intersection operation on the low-value reflection amplitude signal pattern data and the significant Doppler frequency shift amount signal pattern data, thereby obtaining the moving signal pattern data of the personnel; Step S335: construct a personnel posture signal pattern recognition model for the personnel stationary signal pattern data and the personnel moving signal pattern data, and perform in-hospital personnel posture recognition on the millimeter wave reflection signal pattern data through the personnel posture signal pattern recognition model, so as to obtain in-hospital personnel posture data.

7. The method for constructing a dynamic prediction model for medical health according to claim 5, characterized in that: Step S34 is specifically as follows: Step S341: extracting periodic features of the reflection signal according to the millimeter wave reflection signal pattern data, thereby obtaining periodic reflection signal data; Step S342: performing periodic signal division based on the periodic reflection signal data, thereby obtaining short periodic reflection signal data and long periodic reflection signal data; Step S343: dividing the periodic reflection signal data into signal waveforms, thereby obtaining peak waveform periodic reflection signal data and flat waveform periodic reflection signal data; Step S344: performing a periodic reflection signal intersection operation based on the short periodic reflection signal data and the peak waveform periodic reflection signal data, thereby obtaining the personnel's periodic heart rate signal data; performing a periodic reflection signal intersection operation based on the long periodic reflection signal data and the smooth waveform periodic reflection signal data, thereby obtaining the personnel's periodic breathing frequency signal data; Step S345: Integrate the vital signs of the in-hospital personnel based on the periodic heart rate signal data and the periodic respiratory rate signal data of the personnel, so as to obtain the vital sign data of the in-hospital personnel.

8. The method for constructing a dynamic prediction model for medical health according to claim 1, characterized in that: Step S4 is specifically as follows: Step S41: dividing the posture and vital sign data of the in-hospital personnel, thereby obtaining the posture data and vital sign data of the in-hospital personnel; Step S42: performing posture duration threshold statistics on the posture data of the in-hospital personnel, thereby obtaining the posture duration threshold data of the personnel, and performing abnormal posture duration detection on the posture data of the in-hospital personnel according to the posture duration threshold data of the personnel, thereby obtaining the abnormal posture duration data of the personnel and the normal posture duration data of the personnel; Step S43: Perform statistics on the range of changes in physical signs according to the physical sign data of the in-hospital personnel, so as to obtain the time data of the range of changes in the high-value physical signs of the personnel and the time data of the range of changes in the low-value physical signs of the personnel; Step S44: performing a time intersection operation on the personnel's high-level vital sign change amplitude time data and the personnel's abnormal posture duration data, thereby obtaining abnormal personnel vital sign data; Perform time intersection operation on the time data of the change amplitude of the low-level vital signs of personnel and the normal duration data of the posture of personnel, so as to obtain the normal vital signs data of personnel; Step S45: merging the abnormal person's vital sign data and the normal person's vital sign data to obtain abnormal person's vital sign detection data; Step S46: constructing a dynamic prediction model for abnormal vital signs of personnel based on the abnormal vital sign detection data of personnel.

9. The method for constructing a dynamic prediction model for medical health according to claim 1, characterized in that: Step S6 is specifically as follows: Step S61: performing personnel RFID tag identification on the abnormal physical sign prediction personnel location data according to the hospital personnel distribution data, thereby obtaining the abnormal physical sign prediction personnel RFID tag data; Step S62: performing RFID tag spatial flow prediction on the RFID tag data of the abnormal physical sign prediction personnel through the hospital personnel distribution structure model, thereby obtaining the abnormal physical sign prediction personnel spatial flow data; Step S63: acquiring RFID identifier deployment data, and selecting an early warning RFID identifier for the abnormal physical sign prediction personnel spatial flow data and the RFID identifier deployment data, thereby obtaining the RFID identifier data to be warned; Step S64: Integrate the warning information based on the RFID tag data of the person predicted to have abnormal physical signs and the RFID identifier data to be warned, so as to obtain the RFID warning information, and transmit it to the RFID management platform to execute the personnel warning task.

10. A system for constructing a medical health dynamic prediction model, characterized in that: Used to execute the method for constructing a medical health dynamic prediction model as claimed in claim 1, the system for constructing a medical health dynamic prediction model comprises: The three-dimensional structure modeling module is used to obtain hospital structure data, and perform three-dimensional structure modeling of the hospital building based on the hospital structure data, thereby obtaining a three-dimensional structure model of the hospital building; based on the three-dimensional structure model of the hospital building, the in-hospital millimeter wave radar network is constructed, thereby obtaining an in-hospital millimeter wave radar network; The personnel distribution analysis module is used to obtain RFID identification data and mark the personnel distribution of the three-dimensional structure model of the hospital building according to the RFID identification data, so as to obtain the hospital personnel distribution structure model; A personnel posture and vital signs integration module is used to collect millimeter-wave reflection signals in the hospital based on the millimeter-wave radar network in the hospital, so as to obtain millimeter-wave reflection signal data in the hospital, and integrate the posture and vital signs of personnel in the hospital based on the millimeter-wave reflection signal data in the hospital, so as to obtain the posture and vital signs data of personnel in the hospital; The abnormal physical sign monitoring module is used to monitor the abnormal physical sign data of the hospital personnel's posture and physical sign data, thereby obtaining the personnel's abnormal physical sign detection data, and constructing the personnel's abnormal physical sign dynamic prediction model based on the personnel's abnormal physical sign detection data; The abnormal physical sign prediction module is used to dynamically predict the abnormal physical signs of personnel in the hospital by using the abnormal physical sign dynamic prediction model, so as to obtain abnormal physical sign predicted personnel data, and to identify the abnormal physical sign predicted personnel positions by using the hospital personnel distribution structure model, so as to obtain abnormal physical sign predicted personnel position data; The early warning information integration module is used to predict personnel location data based on abnormal physical signs and integrate personnel RFID early warning information, thereby obtaining RFID early warning information and transmitting it to the RFID management platform to execute personnel early warning tasks.

Images