Sickbed with intelligent monitoring system

Abstract

The invention provides a sickbed with an intelligent monitoring system, and relates to the technical field of medical instruments. The sickbed with the intelligent monitoring system comprises a sickbed body, an environment field analysis module, a medical equipment signal multiplexing module, a multi-modal data fusion engine and a clinical decision module, and the environment field analysis module comprises an electromagnetic field monitoring unit, a sound field imaging unit and a thermal field entropy change analysis unit. The non-contact type multi-physical-field monitoring technology is adopted, the skin injury risk caused by a contact type sensor is thoroughly avoided, comprehensive and accurate monitoring of vital signs such as respiration, the heart rate and the body temperature is achieved by deeply excavating the multi-source heterogeneous data value of medical equipment, and a more efficient monitoring mode is provided for clinical monitoring.

Description

Technical Field

[0001] This invention relates to the field of medical device technology, specifically to a hospital bed equipped with an intelligent monitoring system. Background Art

[0002] Bedside patient monitoring refers to the continuous and comprehensive collection, analysis, and evaluation of physiological and pathological indicators of bedridden patients using various equipment and technologies. It encompasses monitoring basic vital signs such as heart rate, blood pressure, respiratory rate, and body temperature, which directly reflect the patient's basic physical function. It also includes monitoring blood oxygen saturation to understand the patient's oxygenation status and determine if respiratory function is normal; electrocardiogram (ECG) monitoring captures information about the heart's electrical activity, helping to detect cardiac problems such as arrhythmias. Furthermore, it may involve assessing the patient's level of consciousness, mental state, pain intensity, and other subjective feelings and physical functions. Through the comprehensive monitoring and analysis of this multi-dimensional data, medical staff can promptly detect changes in the patient's condition and potential risks, thereby quickly adjusting treatment plans, providing timely and effective medical care, and ensuring the patient's life, health, and safety.

[0003] Existing bedside monitoring technologies have revealed numerous shortcomings in practical use: traditional contact sensors, such as ECG electrodes and pulse oximeter clips, require prolonged contact with the patient's skin, easily causing pressure sores and allergic reactions, and severely restricting the patient's freedom of movement, potentially causing secondary injury, especially to burn or critically ill patients. Regarding monitoring effectiveness, single monitoring methods have significant limitations; for example, respiratory monitoring belts cannot detect pleural effusion, and infrared thermometers struggle to distinguish between infection-induced fever and changes in ambient temperature, leading to a high false alarm rate. Even more troubling is the fragmented nature of various medical devices; valuable information such as the vibration signals from infusion pumps and glare from monitor screens is not fully utilized, forcing medical staff to switch between multiple terminals and manually compare data, severely impacting treatment efficiency. All of these factors pose significant challenges to clinical monitoring. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a hospital bed with an intelligent monitoring system, which solves the problem that existing monitoring methods mostly rely on contact sensors and have relatively limited monitoring efficiency.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solution: a hospital bed with an intelligent monitoring system, comprising a hospital bed body, an environmental field analysis module, a medical equipment signal multiplexing module, a multimodal data fusion engine, and a clinical decision module, wherein the environmental field analysis module comprises an electromagnetic field monitoring unit, a sound field imaging unit, and a thermal field entropy change analysis unit;

[0006] The electromagnetic field monitoring unit uses the MIMO antenna array of the existing Wi-Fi router in the ward to capture electromagnetic waves reflected from the patient's chest cavity in real time in the 2.4GHz / 5GHz band. Based on the channel state information and phase difference, it analyzes the respiratory rate and heart rate using the following formula:

[0007] Δφ(t)=arg(H k (t))-arg(H k (t0))

[0008] Where H k (t) represents the channel response of the k-th subcarrier at time t;

[0009] The sound field imaging unit collects the diffraction field characteristics of ambient background noise through a wideband microphone array distributed at the four corners of the bed, and reconstructs a three-dimensional thoracic cavity motion model using a time-reversal mirror algorithm.

[0010] The thermal field entropy change analysis unit uses infrared cameras in the ward to acquire the distribution of thermal radiation on the body surface and calculates the rate of change of entropy in a local area to identify signs of infection.

[0011] The medical device signal multiplexing module includes an infusion tubing vibration sensor and a monitor optical resolution unit.

[0012] The infusion tubing vibration sensor is connected to the built-in pressure sensor of the infusion pump. The vibration signal in the 0.01-10Hz frequency band is extracted by wavelet packet decomposition and cross-correlation analysis is performed with the ECG waveform.

[0013] The optical analysis unit of the monitor captures the jitter characteristics of iris pixels in the reflective area of ​​the monitor screen through a camera, and calculates the intracranial pressure fluctuation based on the optical flow method.

[0014] The multimodal data fusion engine uses a federated Kalman filter to align the spatiotemporal characteristics of electromagnetic fields, sound fields, and medical device signals, and outputs fused physiological parameters.

[0015] The clinical decision module triggers tiered warnings based on fused data and writes the warning information into the electronic medical record system via the HL7FHIR protocol.

[0016] Preferably, the electromagnetic field monitoring unit includes:

[0017] a) The mathematical model for an adaptive beamforming controller that dynamically adjusts the Wi-Fi signal beam towards the patient's chest cavity and suppresses multipath interference is as follows:

[0018]

[0019] Where H is the channel matrix and d is the desired beam direction;

[0020] b) An LSTM-based respiratory signal denoising network, the input layer receives the CSI phase sequence, the output layer generates the denoised respiratory waveform, and the network structure contains 3 hidden layers;

[0021] c) Sleep apnea detection algorithm: when the amplitude of sleep apnea decreases by 90% for 5 consecutive respiratory cycles and the duration is >15 seconds, a Level I warning is triggered.

[0022] Preferably, the sound field imaging unit performs:

[0023] 1) Solving for the sound source distribution by inversely applying the Helmholtz equation:

[0024]

[0025] Where p(r) is the reconstructed sound pressure field with a spatial resolution of 0.5 mm;

[0026] 2) Method for detecting lung consolidation areas: When the acoustic impedance value is >2000 Rayl and lasts for 30 minutes, a red warning area is marked in the three-dimensional model;

[0027] 3) Cough event localization algorithm: Extracts pulse signals in the 1-3kHz frequency band through time-frequency analysis, with a localization accuracy error of <2cm. 3 .

[0028] Preferably, the infusion tubing vibration sensor includes:

[0029] A. Cross-modal analysis module for vibration signals and ECG, calculating the normalized mutual information entropy of both.

[0030]

[0031] When I norm An atrial fibrillation event is defined as a value >0.85.

[0032] B. Adaptive compensation mechanism for thrill intensity, dynamically adjusting the detection threshold T(v) = 0.02e based on the infusion flow rate v. －0.1v +0.01.

[0033] Preferably, the multimodal data fusion engine implements:

[0034] a. Device-level time synchronization based on the IEEE 1588v2 protocol, clock skew compensation amount

[0035] b. Method for establishing the spatial coordinate system of the hospital bed: with the geometric center of the bed as the origin, the Z-axis is vertically upward and the X-axis is along the long axis of the bed.

[0036] cD-S evidence theory decision fusion: When electromagnetic field and acoustic field data conflict, the trust function is calculated.

[0037]

[0038] Select monitoring results with a trust level > 0.9 as the final output.

[0039] Preferably, the clinical decision module includes:

[0040] ①. The biomechanical analysis unit for nursing operations calculates the stability index of the force applied by the nurses by using the current ripple characteristic ΔI(t) of the bed drive motor:

[0041]

[0042] When S < 0.7, it indicates that the patient has muscle weakness.

[0043] ②. The mobile DR device linkage interface reconstructs the micro-motion trajectory of the bones using the scattered radiation field when the X-ray machine passes the bed, generating a fracture healing rate report.

[0044] Preferably, the system security control includes:

[0045] (1). Biometric data desensitization processing: Conditional generative adversarial network is used to transform the original signal.

[0046]

[0047] Ensure that the output data cannot be traced back to a specific patient;

[0048] (2) Electromagnetic radiation safety control: Wi-Fi transmission power is limited to 10mW, and the SAR value meets the following requirements:

[0049]

[0050] Where σ is the tissue conductivity and ρ is the density.

[0051] Preferably, the system includes a self-maintenance subsystem:

[0052] The sound field array phase calibration is performed automatically every day by sending a 5kHz calibration audio source to adjust the delay compensation of each microphone. Where d i Microphone spacing;

[0053] Based on the federated learning model update mechanism, each hospital locally trains the LSTM denoising network parameters θ. i The central server aggregates and generates a global model:

[0054]

[0055] Where n i This represents the amount of data at each node.

[0056] This invention provides a hospital bed equipped with an intelligent monitoring system. It has the following beneficial effects:

[0057] This invention provides a hospital bed with an intelligent monitoring system. The technology utilizes a non-contact multi-physics collaborative sensing mechanism, employing the MIMO antenna array of an existing Wi-Fi router in the ward to capture electromagnetic wave phase disturbances. Combined with wideband microphone array sound field imaging and infrared thermal radiation entropy change analysis, it achieves non-invasive and accurate monitoring of respiration, heart rate, and signs of infection. This completely avoids the skin damage risks associated with traditional contact sensors, deeply explores the value of idle signals from medical devices, and establishes a cross-modal physiological parameter correlation model through wavelet packet decomposition of infusion tubing vibration signals and optical flow analysis of monitor screen reflections. This overcomes the blind spots of monitoring from a single data source. Based on federated Kalman filtering and DS evidence theory, a multi-source heterogeneous data fusion engine is constructed to achieve spatiotemporal alignment and conflict resolution of electromagnetic fields, sound fields, and device signals, significantly improving the robustness of monitoring in complex clinical environments.

[0058] This invention provides a hospital bed with an intelligent monitoring system. Utilizing a closed-loop decision-making mechanism, it organically combines tiered early warning, electronic medical record interaction, and biomechanical analysis. Through the HL7 FHIR protocol, it achieves systematic output of clinical decisions, integrates conditional generative adversarial network anonymization processing and dynamic SAR value adjustment technology, constructs a privacy and security protection system throughout the entire data lifecycle, and is equipped with automatic sound field calibration and federated learning model update functions. By compensating for device drift with a 5kHz calibration signal, it continuously optimizes system performance, resulting in an intelligent hospital bed with higher environmental adaptability, data intelligence, and clinical practicality. Attached Figure Description

[0059] Figure 1 This is a schematic diagram of the system framework of the present invention. Detailed Implementation

[0060] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only 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 ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0061] like Figure 1As shown, this embodiment of the invention provides a hospital bed with an intelligent monitoring system, including a hospital bed body, an environmental field analysis module, a medical device signal multiplexing module, a multimodal data fusion engine, and a clinical decision-making module. The environmental field analysis module includes an electromagnetic field monitoring unit, a sound field imaging unit, and a thermal field entropy change analysis unit.

[0062] The electromagnetic field monitoring unit uses the MIMO antenna array of the existing Wi-Fi router in the ward to capture electromagnetic waves reflected from the patient's chest cavity in real time in the 2.4GHz / 5GHz band. Based on the channel state information (CSI) phase difference, it analyzes the respiratory rate and heart rate using the following formula:

[0063] Δφ(t)=arg(H k (t))-arg(H k (t0))

[0064] Where H k (t) represents the channel response of the k-th subcarrier at time t;

[0065] The sound field imaging unit collects the diffraction field characteristics of the ambient background noise through a wideband microphone array (frequency response 20Hz-20kHz) distributed at the four corners of the bed, and reconstructs a three-dimensional thoracic cavity motion model using a time-reversal mirror algorithm.

[0066] The thermal field entropy change analysis unit uses infrared cameras in the ward to acquire the distribution of thermal radiation on the body surface and calculates the rate of change of entropy in local areas to identify signs of infection.

[0067] The medical device signal multiplexing module includes an infusion tubing vibration sensor and a monitor optical resolution unit;

[0068] The infusion tubing vibration sensor is connected to the built-in pressure sensor of the infusion pump. The vibration signal in the 0.01-10Hz frequency band is extracted by wavelet packet decomposition and cross-correlation analysis is performed with the ECG waveform.

[0069] The optical analysis unit of the monitor captures the jitter characteristics of iris pixels in the reflective area of ​​the monitor screen through a camera, and calculates the intracranial pressure fluctuation based on the optical flow method.

[0070] The multimodal data fusion engine uses a federated Kalman filter to align the spatiotemporal characteristics of electromagnetic fields, sound fields, and medical device signals, and outputs fused physiological parameters.

[0071] The clinical decision module triggers tiered warnings based on fused data and writes the warning information into the electronic medical record system via the HL7FHIR protocol.

[0072] The electromagnetic field monitoring unit includes:

[0073] a) The mathematical model for an adaptive beamforming controller that dynamically adjusts the Wi-Fi signal beam towards the patient's chest cavity and suppresses multipath interference is as follows:

[0074]

[0075] Where H is the channel matrix and d is the desired beam direction;

[0076] b) An LSTM-based respiratory signal denoising network, the input layer receives the CSI phase sequence, and the output layer generates the denoised respiratory waveform. The network structure contains 3 hidden layers (128-64-32 neurons).

[0077] c) Sleep apnea detection algorithm: when the amplitude of sleep apnea decreases by 90% for 5 consecutive respiratory cycles and the duration is >15 seconds, a Level I warning is triggered.

[0078] The sound field imaging unit performs the following:

[0079] 1) Solving for the sound source distribution by inversely applying the Helmholtz equation:

[0080]

[0081] Where p(r) is the reconstructed sound pressure field with a spatial resolution of 0.5 mm;

[0082] 2) Method for detecting lung consolidation areas: When the acoustic impedance value is >2000 Rayl and lasts for 30 minutes, a red warning area is marked in the three-dimensional model;

[0083] 3) Cough event localization algorithm: Extracts pulse signals in the 1-3kHz frequency band through time-frequency analysis, with a localization accuracy error of <2cm. 3 .

[0084] Infusion line vibration sensors include:

[0085] A. Cross-modal analysis module for vibration signals and ECG, calculating the normalized mutual information entropy of both.

[0086]

[0087] When I norm An atrial fibrillation event is defined as a value >0.85.

[0088] B. Adaptive compensation mechanism for thrill intensity, dynamically adjusting the detection threshold T(v) = 0.02e based on the infusion flow rate v. －0.1v +0.01.

[0089] Multimodal data fusion engine implementation:

[0090] a. Device-level time synchronization based on the IEEE 1588v2 protocol, clock skew compensation amount

[0091] b. Method for establishing the spatial coordinate system of the hospital bed: with the geometric center of the bed as the origin, the Z-axis is vertically upward and the X-axis is along the long axis of the bed.

[0092] cD-S evidence theory decision fusion: When electromagnetic field and acoustic field data conflict, the trust function is calculated.

[0093]

[0094] Select monitoring results with a trust level > 0.9 as the final output.

[0095] The clinical decision module includes:

[0096] ①. The biomechanical analysis unit for nursing operations calculates the stability index of the force applied by the nurses by using the current ripple characteristic ΔI(t) of the bed drive motor:

[0097]

[0098] When S < 0.7, it indicates that the patient has muscle weakness.

[0099] ②. The mobile DR device linkage interface reconstructs the micro-motion trajectory of the bones using the scattered radiation field when the X-ray machine passes the bed, generating a fracture healing rate report.

[0100] System security controls include:

[0101] (1) Desensitization of biometric data: Conditional Generative Adversarial Network (cGAN) is used to transform the original signal.

[0102]

[0103] Ensure that the output data cannot be traced back to a specific patient;

[0104] (2) Electromagnetic radiation safety control: Wi-Fi transmission power is limited to 10mW, and the SAR value meets the following requirements:

[0105]

[0106] Where σ is the tissue conductivity and ρ is the density.

[0107] The system includes a self-maintenance subsystem:

[0108] The sound field array phase calibration is performed automatically every day by sending a 5kHz calibration audio source to adjust the delay compensation of each microphone. Where d iMicrophone spacing;

[0109] Based on the federated learning model update mechanism, each hospital locally trains the LSTM denoising network parameters θ. i The central server aggregates and generates a global model:

[0110]

[0111] Where n i This represents the amount of data at each node.

[0112] Specifically, this invention integrates multi-dimensional sensing mechanisms of electromagnetic, acoustic, and thermal fields to construct a basic architecture for non-invasive physiological monitoring. Utilizing the MIMO antenna array of existing Wi-Fi routers in hospital wards, the communication infrastructure is transformed into a biomechanical sensor. By analyzing the phase perturbation characteristics of electromagnetic waves in the 2.4GHz / 5GHz band, millimeter-level dynamic capture of chest cavity movements is achieved. The acoustic field imaging unit employs a wideband microphone array and a Helmholtz equation inverse solution algorithm, overcoming the bottleneck of environmental noise interference in acoustic monitoring. It can accurately reconstruct a three-dimensional vibration model of lung tissue in high background noise environments. Thermal field entropy change analysis, through infrared thermal imaging and dynamic entropy value calculation, effectively distinguishes between abnormal thermal radiation caused by local tissue infection and environmental temperature fluctuations, solving the misjudgment problem of traditional single temperature monitoring. This technology avoids physical pressure on the patient's skin from contact sensors, eliminating clinical risks such as electrode detachment and allergic reactions, making it particularly suitable for special scenarios such as burns and intensive care.

[0113] This invention transforms traditionally untapped medical device output information, such as tremor signals from infusion tubing and optical features of monitor screens, into high-value monitoring data sources. The infusion pump's built-in pressure sensor signal is analyzed using wavelet packet decomposition to extract micro-tremor features in the 0.01-10Hz frequency band. Cross-correlation analysis with ECG signals enables early warning of cardiovascular events. Optical analysis of the monitor screen's reflective area utilizes iris pixel jitter features and an improved optical flow algorithm to derive intracranial pressure fluctuation parameters, pioneering a new method for non-contact intracranial pressure monitoring. This signal multiplexing mechanism not only reduces reliance on dedicated monitoring probes but, more importantly, breaks down data barriers between medical devices, enabling a closed-loop feedback loop between the infusion process, vital sign monitoring, and intelligent analysis of the bedside system.

[0114] This invention employs a hybrid architecture combining federated Kalman filters and DS evidence theory to overcome the core technical challenge of spatiotemporal alignment of multimodal data. By utilizing the IEEE 1588v2 protocol to achieve device-level microsecond-level time synchronization, and combining this with dynamic calibration of the bed's spatial coordinate system, it ensures accurate matching of electromagnetic field data, acoustic imaging results, and medical device signals under a unified spatiotemporal reference. When outputs from different sensing modalities conflict, the confidence calculation model based on DS evidence theory automatically identifies the data source with the highest credibility, significantly improving monitoring robustness in complex scenarios such as electromagnetic interference from multiple devices in the ICU and disturbances caused by nursing operations. This fusion mechanism upgrades the detection of critical events such as apnea and atrial fibrillation from single-parameter threshold judgment to multi-dimensional evidence chain verification, greatly reducing the risk of false alarms and missed alarms.

[0115] The clinical decision-making module of this invention constructs a complete closed loop from data acquisition to medical intervention. The tiered early warning system automatically triggers a differentiated response mechanism based on multimodal fusion results: Level I warnings, such as abnormal respiratory rate, directly activate bedside alarms, while Level II warnings, such as muscle weakness, generate electronic medical record prompts. Deep integration with the hospital information system via the HL7 FHIR protocol enables advanced functions such as structured writing of early warning information and automatic verification of medical orders. The biomechanical analysis unit quantifies the force of nursing staff's operations and the patient's physical response by analyzing the current ripple characteristics of the bed drive motor, providing objective quantitative indicators for rehabilitation assessment. The system's self-maintenance mechanism automatically compensates for the microphone array's latency deviation using a 5kHz acoustic calibration signal, and continuously optimizes the model under a federated learning framework, ensuring that monitoring accuracy does not decrease with equipment aging or environmental changes, thus solving the pain point of traditional systems requiring frequent manual calibration.

[0116] This invention employs Conditional Generative Adversarial Networks (cGANs) at the data acquisition end to irreversibly transform raw biometric signals. By generating anonymized data with statistical characteristics consistent with real physiological signals but without traceability to the individual, it fundamentally eliminates the risk of privacy leaks. During signal transmission, dynamic SAR value control is implemented, calculating electromagnetic radiation absorption rate in real time based on the patient's position and automatically limiting Wi-Fi transmission power within safe thresholds. Homomorphic encryption technology is used in the data storage stage, ensuring that the fused data in the electronic medical record system remains encrypted yet still allows for early warning analysis. This meets the requirements of regulations such as HIPAA while avoiding the data utilization degradation problem caused by traditional encryption. This end-to-end protection system of "acquisition-transmission-storage" provides a system-level solution for the secure deployment of medical IoT devices.

[0117] This technology redefines the clinical functional boundaries of intelligent hospital beds through the synergistic effect of three dimensions: innovation in physical layer perception, breakthroughs in signal analysis algorithms, and optimization of decision-making mechanisms. It integrates traditionally independent aspects such as vital sign monitoring, medical equipment management, and nursing quality assessment into a unified intelligent platform. This not only improves the accuracy of individual monitoring parameters but, more importantly, establishes the ability to dynamically correlate multiple parameters. For example, cross-modal analysis of infusion tremor signals and ECG data can predict cardiovascular events 15 minutes in advance, and correlation analysis of thermal entropy change trends with respiratory patterns can identify early lung infections. This system-level innovation transforms hospital beds from passive monitoring devices into active diagnostic and treatment nodes, providing a new technological carrier for precision medicine.

[0118] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A hospital bed with an intelligent monitoring system, comprising a bed body, an environmental field analysis module, a medical equipment signal multiplexing module, a multimodal data fusion engine, and a clinical decision-making module, characterized in that: The environmental field analysis module includes an electromagnetic field monitoring unit, an acoustic field imaging unit, and a thermal field entropy change analysis unit. The electromagnetic field monitoring unit uses the MIMO antenna array of the existing Wi-Fi router in the ward to capture electromagnetic waves reflected from the patient's chest cavity in real time in the 2.4GHz / 5GHz band. Based on the channel state information and phase difference, it analyzes the respiratory rate and heart rate using the following formula: Δφ(t)=arg(H k (t))-arg(H k (t0)) Where H k (t) represents the channel response of the k-th subcarrier at time t; The sound field imaging unit collects the diffraction field characteristics of ambient background noise through a wideband microphone array distributed at the four corners of the bed, and reconstructs a three-dimensional thoracic cavity motion model using a time-reversal mirror algorithm. The thermal field entropy change analysis unit uses infrared cameras in the ward to acquire the distribution of thermal radiation on the body surface and calculates the rate of change of entropy in a local area to identify signs of infection. The medical device signal multiplexing module includes an infusion tubing vibration sensor and a monitor optical resolution unit. The infusion tubing vibration sensor is connected to the built-in pressure sensor of the infusion pump. The vibration signal in the 0.01-10Hz frequency band is extracted by wavelet packet decomposition and cross-correlation analysis is performed with the ECG waveform. The optical analysis unit of the monitor captures the jitter characteristics of iris pixels in the reflective area of ​​the monitor screen through a camera, and calculates the intracranial pressure fluctuation based on the optical flow method. The multimodal data fusion engine uses a federated Kalman filter to align the spatiotemporal characteristics of electromagnetic fields, sound fields, and medical device signals, and outputs fused physiological parameters. The clinical decision module triggers tiered warnings based on fused data and writes the warning information into the electronic medical record system via the HL7FHIR protocol.

2. A hospital bed with an intelligent monitoring system according to claim 1, characterized in that: The electromagnetic field monitoring unit includes: a) The mathematical model for an adaptive beamforming controller that dynamically adjusts the Wi-Fi signal beam towards the patient's chest cavity and suppresses multipath interference is as follows: Where H is the channel matrix and d is the desired beam direction; b) An LSTM-based respiratory signal denoising network, the input layer receives the CSI phase sequence, the output layer generates the denoised respiratory waveform, and the network structure contains 3 hidden layers; c) Sleep apnea detection algorithm: when the amplitude of sleep apnea decreases by 90% for 5 consecutive respiratory cycles and the duration is >15 seconds, a Level I warning is triggered.

3. A hospital bed with an intelligent monitoring system according to claim 1, characterized in that: The acoustic field imaging unit performs: 1) Solving for the sound source distribution by inversely applying the Helmholtz equation: Where p(r) is the reconstructed sound pressure field with a spatial resolution of 0.5 mm; 2) Method for detecting lung consolidation areas: When the acoustic impedance value is >2000 Rayl and lasts for 30 minutes, a red warning area is marked in the three-dimensional model; 3) Cough event localization algorithm: Extracts pulse signals in the 1-3kHz frequency band through time-frequency analysis, with a localization accuracy error of <2cm. 3 .

4. A hospital bed with an intelligent monitoring system according to claim 1, characterized in that: The infusion tubing vibration sensor includes: A. Cross-modal analysis module for vibration signals and ECG, calculating the normalized mutual information entropy of both. When I norm An atrial fibrillation event is defined as a value >0.

85. B. Adaptive compensation mechanism for thrill intensity, dynamically adjusting the detection threshold T(v) = 0.02e based on the infusion flow rate v. －0.1v +0.

01.

5. A hospital bed with an intelligent monitoring system according to claim 1, characterized in that: The multimodal data fusion engine achieves the following: a. Device-level time synchronization based on the IEEE 1588v2 protocol, clock skew compensation amount b. Method for establishing the spatial coordinate system of the hospital bed: with the geometric center of the bed as the origin, the Z-axis is vertically upward and the X-axis is along the long axis of the bed. cD-S evidence theory decision fusion: When electromagnetic field and acoustic field data conflict, the trust function is calculated. Select monitoring results with a trust level > 0.9 as the final output.

6. A hospital bed with an intelligent monitoring system according to claim 1, characterized in that: The clinical decision module includes: ①. The biomechanical analysis unit for nursing operations calculates the stability index of the force applied by the nurses by using the current ripple characteristic ΔI(t) of the bed drive motor: When S < 0.7, it indicates that the patient has muscle weakness. ②. The mobile DR device linkage interface reconstructs the micro-motion trajectory of the bones using the scattered radiation field when the X-ray machine passes the bed, generating a fracture healing rate report.

7. A hospital bed with an intelligent monitoring system according to claim 1, characterized in that: The system security control includes: (1). Biometric data desensitization processing: Conditional generative adversarial network is used to transform the original signal. Ensure that the output data cannot be traced back to a specific patient; (2) Electromagnetic radiation safety control: Wi-Fi transmission power is limited to 10mW, and the SAR value meets the following requirements: Where σ is the tissue conductivity and ρ is the density.

8. A hospital bed with an intelligent monitoring system according to claim 1, characterized in that: The system includes a self-maintaining subsystem: The sound field array phase calibration is performed automatically every day by sending a 5kHz calibration audio source to adjust the delay compensation of each microphone. Where d i Microphone spacing; Based on the federated learning model update mechanism, each hospital locally trains the LSTM denoising network parameters θ. i The central server aggregates and generates a global model: Where n i This represents the amount of data at each node.

Images