Omnibearing patient monitoring and emergency treatment system for emergency department

Abstract

The invention relates to the technical field of emergency medicine, in particular to an emergency department patient all-dimensional monitoring and emergency treatment system which comprises a patient information management unit, a real-time detection unit, a behavior feature recognition unit, an intelligent early warning and decision support unit and a remote cooperation and communication unit. According to the invention, through a multi-mode monitoring architecture special for trauma and a dynamic anti-interference mechanism, accurate and continuous monitoring of hemodynamics and vital signs of an acute trauma patient is realized, hidden shock and other critical conditions are effectively identified, and non-contact multi-mode perception and intelligent behavior analysis technologies are utilized to realize accurate and continuous monitoring of the acute trauma patient. Behavior characteristics such as the body surface temperature, the body position micro-motion and the voiceprint of the patient are converted into clinical risk indexes, potential risks are warned in advance, multi-dimensional and high-precision monitoring and warning are provided for the emergency treatment trauma patient, and the timeliness and the accuracy of emergency diagnosis are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of emergency medicine, in particular to an all-round monitoring and first aid system for emergency department patients. Background Art

[0002] Emergency patients refer to people who are in critical condition or potentially life-threatening due to sudden illness, accidental injury, acute trauma, etc., and need to go to the hospital emergency department immediately for emergency medical treatment. The symptoms of such patients are often characterized by rapid onset, rapid progression, and complexity, requiring medical staff to quickly evaluate, diagnose, and intervene to save lives or prevent the condition from worsening. In the treatment of emergency patients, rapid and accurate monitoring and evaluation of the condition is the key to improving the success rate of rescue. With the development of medical technology, emergency treatment has put forward higher requirements on the timeliness and accuracy of the monitoring system. However, in the existing technology, the traditional monitoring system has problems such as delayed identification of hidden injuries and untimely emergency warnings due to the lag of physiological indicators and the limitation of single-dimensional evaluation.

[0003] Based on this, the present invention provides an all-round monitoring and first aid system for emergency department patients to solve the above-mentioned technical problems. Summary of the Invention

[0004] The purpose of the present invention is to provide an all-round monitoring and emergency system for patients in the emergency department. The trauma-specific multimodal monitoring architecture and dynamic anti-interference mechanism of the present invention can realize accurate and continuous monitoring of the hemodynamics and vital signs of patients with acute trauma, effectively identify critical conditions such as occult shock, and utilize non-contact multimodal perception and intelligent behavior analysis technology to convert the patient's behavioral characteristics such as surface temperature, body position micro-movements and voiceprints into clinical risk indicators, and provide early warning of potential risks, providing emergency trauma patients with multi-dimensional and high-precision monitoring and early warning, significantly improving the timeliness and accuracy of emergency diagnosis.

[0005] To achieve the above object, the present invention provides the following technical solutions: The present invention provides a comprehensive emergency monitoring and first aid system for emergency department patients, including a patient information management unit, a real-time detection unit, a behavior feature recognition unit, an intelligent early warning and decision support unit, and a remote collaboration and communication unit, wherein: The patient information management unit is used to quickly enter and prioritize key information of emergency patients through automatic recognition technology and dynamic data tagging; The real-time detection unit is used to perform accurate hemodynamic assessment and continuous vital sign monitoring of acute trauma patients through a trauma-specific multimodal monitoring architecture and a dynamic anti-interference mechanism; The behavioral feature recognition unit is used to dynamically identify and evaluate clinical risk indicators of patients in an emergency environment through non-contact multimodal perception and intelligent behavioral analysis; The intelligent early warning and decision support unit: based on the trauma progression prediction algorithm and the ISS score association model, provides risk warning and generates a graded intervention plan; The remote collaboration and communication unit is used to utilize AR annotation and secure data sharing to support interdisciplinary teams to participate in emergency and critical care in real time.

[0006] The patient information management unit includes an automatic identity recognition module, a data acquisition module, a priority marking module, and an information synchronization module, wherein: The automatic identity recognition module is used to quickly identify and bind the identity of emergency patients through barcode scanning, RFID or face recognition technology; The data acquisition module is used to automatically obtain key medical data such as the patient's admission information, basic medical history, and allergy history; The priority marking module is used to dynamically mark the patient's treatment priority according to the degree of injury and vital signs; The information synchronization module is used to synchronize patient information to relevant monitoring and diagnosis systems in real time.

[0007] The real-time detection unit includes a dual-mode monitoring module, a motion artifact suppression module, and a trauma-specific monitoring module, wherein: The dual-mode monitoring module is used to synchronously run impedance cardiography and laser Doppler to jointly evaluate bleeding volume and cardiac function; The motion artifact suppression module is used to dynamically filter signal noise caused by patient movement using IMU sensor data; The trauma-specific monitoring module is used to calculate the intra-abdominal pressure and perfusion index in real time and identify latent shock.

[0008] The trauma monitoring module calculates intra-abdominal pressure and perfusion index in real time to identify latent shock. The specific operations are as follows: A1: Intra-abdominal pressure measurement: Inject 50 ml of normal saline into the patient's bladder through the catheter, keep the patient in a supine position, and measure the intravesical pressure at the level of the pubic symphysis. ; Intra-abdominal pressure calculation formula: ,in, is the intra-abdominal pressure, is the current atmospheric pressure, and the measurement error after correction is ≤2mmHg; A2: Perfusion index calculation: ① Calculation of gastric mucosal pH: 1) Collect gastric mucosal carbon dioxide partial pressure through a nasogastric tube; 2) Simultaneously collect arterial carbon dioxide partial pressure; 3) Calculation formula: ,in, is the gastric mucosal pH, 0.03 is the carbon dioxide solubility coefficient; ② Central venous oxygen saturation monitoring: Blood samples are collected through a central venous catheter and the central venous oxygen saturation is measured using spectral analysis with an accuracy of ; A3: Latent shock identification logic: When any of the following conditions is met, the latent shock warning is triggered: 1) IAP ≥ 16 mmHg, and ; 2) IAP < 16 mmHg, but pHi < 7.30, ScvO2 < 65%, and lactate value Lac ≥ 2.5 mmol / L; A4: Warning level is based on the formula Calculate, when is judged as high risk.

[0009] The behavior feature recognition unit includes a non-contact sensing module and an intelligent behavior analysis module, wherein: The non-contact sensing module is used to capture abnormal body surface temperature, body position micro-movements, and pain-related voiceprint characteristics of the patient through infrared thermal imaging, millimeter-wave radar, and microphone array; The intelligent behavior analysis module converts body movements and facial expressions into clinical risk indicators based on a deep learning model.

[0010] The infrared thermal imaging is used to detect the distribution of surface temperature gradients and identify local ischemic areas with a temperature difference greater than 2°C. The millimeter-wave radar operates in the 60-64 GHz frequency band and captures trauma-specific body position changes of 0.1-5 Hz through the micro-Doppler effect. The microphone array uses beamforming technology to extract the characteristics of patient moans in the 80-300 Hz frequency band, with a signal-to-noise ratio of ≥15dB.

[0011] The intelligent behavior analysis module is based on a deep learning model and converts body movements and facial expressions into clinical risk indicators. The specific operations are as follows: B1: Behavioral feature data preprocessing: ①Segment the temperature field of infrared thermal imaging data and extract the temperature difference of the limb ends; ②Perform time-frequency transformation on millimeter-wave radar data to obtain respiratory rate curve and body acceleration characteristics; ③ The Openpose algorithm is used to extract the coordinates of 18 facial key points from the video image and calculate the expression parameters of frown amplitude and eyelid opening and closing; B2: Deep Learning Model Architecture: ①Adopt a multimodal fusion network, including: 1) Vision branch: 3D CNN network, which takes a sequence of facial key points as input and extracts dynamic features of facial expressions; 2) Radar branch: LSTM network, which takes respiratory rate and body acceleration time series data as input and extracts micro-motion pattern features; 3) Thermal imaging branch: 2D CNN network, which takes temperature field images as input and extracts features of abnormal surface temperature areas; ② The outputs of each branch are weighted and fused through the attention mechanism. The weight formula is: ,in, is the eigenvector of the i-th branch, 、 is a learnable parameter; B3: Clinical Risk Indicator Mapping: ① The model output layer uses a fully connected network and generates the risk probability value P through the Sigmoid activation function. The formula is: ,in is the fusion feature vector, is the Sigmoid function , W o is the weight matrix of the output layer, b o is the bias vector of the output layer; ②Convert the risk probability P into clinical indicators: Ⅰ. Pain level: , when Pain ≥ 7, it was judged as severe pain; II. Shock risk index: ,in, is the limb temperature difference, when When triggering an early warning; B4: Model training: ① The training dataset contains behavioral data of 1,000 emergency patients, annotated with clinical diagnosis results; ② Use the cross entropy loss function to optimize the model parameters. The formula is: , where y is the true label, and the Adam algorithm is iterated for 50 rounds.

[0012] The intelligent early warning and decision support unit includes a three-dimensional early warning matrix module, a time-sensitive intervention module, and a clinical pathway trigger module, wherein: The three-dimensional warning matrix module is used to generate risk levels by integrating physiological parameters, behavioral characteristics and ISS scores; The time-sensitive intervention module is used to push time-sensitive solutions based on the "golden hour" principle; The clinical pathway trigger module is used to automatically link to the hospital HIS system and retrieve the corresponding trauma treatment protocol.

[0013] The three-dimensional warning matrix module integrates physiological parameters, behavioral characteristics and ISS scores to generate risk levels. The specific operations are as follows: C1: Generates a composite risk index based on the following dimensions: ① Physiological dimension: systolic blood pressure <90 mmHg, heart rate >120 beats / min, each scored 2 points; ②Behavioral dimension: forced posture is scored as 3 points, pain groaning is scored as 1 point; ③ Trauma dimension: 1 point for every 5 points of ISS score; C2: When CRI ≥ 8 points, a red alert is activated and the corresponding trauma treatment protocol is pushed.

[0014] The remote collaboration and communication unit includes an AR space annotation module, a data sandbox module, and a multidisciplinary conversation module, wherein: The AR spatial annotation module is used by the expert to mark the puncture point or bleeding location on the patient's 3D body surface projection; The data sandbox module is used for encrypting and transmitting DICOM images and life trend data, and supports secure access by third-party devices; The multidisciplinary conversation module is used to establish a dedicated communication channel for the trauma team and supports real-time multi-party consultations using voice, text, and images.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention uses a trauma-specific multimodal monitoring architecture and a dynamic anti-interference mechanism to achieve accurate and continuous monitoring of the hemodynamics and vital signs of acute trauma patients, effectively identifying critical conditions such as occult shock. It also utilizes non-contact multimodal sensing and intelligent behavioral analysis technology to convert behavioral characteristics such as the patient's body surface temperature, body position micro-movements, and voiceprint into clinical risk indicators, providing early warning of potential risks. This provides emergency trauma patients with multi-dimensional, high-precision monitoring and early warning, significantly improving the timeliness and accuracy of emergency diagnosis. 2. The present invention uses a trauma progression prediction algorithm, an ISS score association model, and a three-dimensional warning matrix to integrate physiological parameters, behavioral characteristics, and trauma scores to generate risk levels. Combined with the "golden hour" principle, it pushes time-sensitive intervention plans and automatically retrieves clinical treatment protocols, providing emergency trauma patients with intelligent, graded risk warnings and decision-making support, effectively improving emergency treatment efficiency and standardization. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1This is a system diagram of a comprehensive monitoring and first aid system for emergency department patients according to the present invention; Figure 2 This is a system architecture diagram of a comprehensive emergency monitoring and first aid system for emergency department patients according to the present invention; Figure 3 This is a flowchart of the linkage between behavior recognition and early warning in the comprehensive monitoring and first aid system for emergency department patients of the present invention. DETAILED DESCRIPTION

[0017] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the embodiments described 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 any creative efforts shall fall within the scope of protection of the present invention.

[0018] like Figure 1-Figure 3 As shown, this embodiment provides an all-round monitoring and emergency system for emergency patients, including a patient information management unit, a real-time detection unit, a behavioral feature recognition unit, an intelligent early warning and decision support unit, and a remote collaboration and communication unit, wherein: the patient information management unit is used to quickly enter and prioritize key information of emergency patients through automatic recognition technology and dynamic data tagging; the real-time detection unit is used to perform accurate hemodynamic assessment and continuous vital sign monitoring of acute trauma patients through a trauma-specific multimodal monitoring architecture and a dynamic anti-interference mechanism; the behavioral feature recognition unit is used to dynamically identify and evaluate clinical risk indicators of patients in an emergency environment through contactless multimodal perception and intelligent behavioral analysis; the intelligent early warning and decision support unit is used to perform risk warnings based on the trauma progression prediction algorithm and the ISS score association model, and to generate a graded intervention plan; the remote collaboration and communication unit is used to support real-time participation of interdisciplinary teams in emergency and critical care by utilizing AR annotation and secure data sharing.

[0019] Among them, it should be noted that the patient information management unit quickly establishes patient files and marks priorities, the real-time detection unit and the behavioral feature recognition unit dynamically collect multimodal clinical data through physiological signal monitoring and non-contact behavioral analysis respectively, and the intelligent early warning and decision support unit integrates the above data with the ISS score to generate graded early warning and disposal plans, and finally the remote collaboration and communication unit realizes multidisciplinary AR collaborative treatment.

[0020] In this embodiment, it should also be noted that the patient information management unit includes an automatic identity recognition module, a data acquisition module, a priority marking module, and an information synchronization module, among which: the automatic identity recognition module is used to quickly identify and bind the identity of emergency patients through barcode scanning, RFID or face recognition technology; the data acquisition module is used to automatically obtain key medical data such as patient admission information, basic medical history, and allergy history; the priority marking module is used to dynamically mark the patient's treatment priority according to the degree of trauma and vital signs; the information synchronization module is used to synchronize patient information to relevant monitoring and treatment systems in real time.

[0021] Among them, it should be noted that the automatic identity recognition module quickly binds the patient's identity through biometric technology, the data collection module automatically obtains key medical data, the priority marking module dynamically evaluates the priority of treatment based on the degree of trauma and vital signs, and the information synchronization module pushes patient information to various diagnosis and treatment systems in real time.

[0022] Furthermore, it should be noted that the automatic identity recognition module uses a binocular camera for face recognition with an accuracy rate of 99.7% (FAR≤0.01%), and supports identity binding through eye area feature matching when the patient wears an oxygen mask; the priority marking module integrates the Emergency Severity Index (ESI) and Trauma Score (TS). When a patient meets both "ESI Level 1" and "TS≤10 points", he or she is automatically marked as red priority and is given priority in the allocation of emergency beds through the hospital's queuing system.

[0023] In this embodiment, it should also be noted that the real-time detection unit includes a dual-mode monitoring module, a motion artifact suppression module, and a trauma-specific monitoring module, wherein: the dual-mode monitoring module is used to synchronously run impedance cardiography and laser Doppler to jointly evaluate the amount of bleeding and cardiac function; the motion artifact suppression module is used to dynamically filter the signal noise caused by patient movement using IMU sensor data; the trauma-specific monitoring module is used to calculate the intra-abdominal pressure and perfusion index in real time to identify latent shock. The specific operations are as follows: A1: Intra-abdominal pressure measurement: 50 ml of normal saline is injected into the patient's bladder through a catheter, the patient is kept in a supine position, and the intra-bladder pressure value is measured at the level of the pubic symphysis. ; Intra-abdominal pressure calculation formula: ,in, is the intra-abdominal pressure, =The current atmospheric pressure, the measurement error after correction is ≤2mmHg; A2: Perfusion index calculation: ① Gastric mucosal pH calculation: 1) Collect gastric mucosal carbon dioxide partial pressure through nasogastric tube; 2) Simultaneously collect arterial carbon dioxide partial pressure; 3) Calculation formula: ,in, is the gastric mucosal pH, 0.03 is the carbon dioxide solubility coefficient; ② Central venous oxygen saturation monitoring: blood samples are collected through the central venous catheter, and the central venous oxygen saturation is measured using spectral analysis with an accuracy of A3: Latent shock recognition logic: When any of the following conditions is met, the latent shock warning is triggered: 1) IAP ≥ 16 mmHg, and 2) IAP < 16 mmHg, but pHi < 7.30, ScvO2 < 65% and lactate value Lac ≥ 2.5 mmol / L; A4: Warning level according to the formula Calculate, when is judged as high risk.

[0024] The above calculation process and related parameters do have a theoretical basis and have been verified in multiple clinical studies. The details are as follows: 1. Intra-abdominal pressure (IAP) calculation: formula Cystometry is a currently recognized standard method for assessing intra-abdominal pressure. This method calculates intra-abdominal pressure by measuring the pressure in the bladder and subtracting atmospheric pressure. The error is controlled within ±2 mmHg, ensuring the accuracy of the measurement.

[0025] 2. Calculation of gastric mucosal pH (pHi): pHi calculation formula Derived from the Henderson-Hasselbalch equation, it is used to assess the acid-base status of the gastric mucosa, where PaCO2 and PgCO2 represent the partial pressure of carbon dioxide in arterial blood and gastric mucosa, respectively, and 0.03 is the carbon dioxide solubility coefficient. This formula has been widely used in clinical practice to monitor tissue perfusion and oxygenation status.

[0026] 3. Central venous oxygen saturation (ScvO2) monitoring: The accuracy of measuring ScvO2 using spectral analysis can reach ±1%, which can accurately reflect the balance between systemic oxygen supply and oxygen consumption and is one of the important indicators for assessing shock.

[0027] 4. Latent shock identification logic: The established warning conditions are based on extensive clinical data and research results and can effectively identify early-stage shock. For example, when IAP ≥ 16 mmHg and pHi < 7.35, or IAP < 16 mmHg but pHi < 7.30, ScvO2 < 65%, and Lac ≥ 2.5 mmol / L, these conditions indicate the risk of occult shock.

[0028] 5. Warning level calculation formula: The system comprehensively considers three key indicators: intra-abdominal pressure, gastric mucosal pH and central venous oxygen saturation, with reasonable weight distribution, and can comprehensively assess the patient's shock risk. When RI ≥ 0.7, it is judged as high risk, which helps to take timely intervention measures.

[0029] Among them, it should be noted that the dual-mode monitoring module simultaneously evaluates the amount of bleeding and cardiac function, the motion artifact suppression module ensures the signal accuracy in a dynamic environment, and the trauma-specific monitoring module accurately identifies latent shock.

[0030] Furthermore, it should be noted that in the dual-mode monitoring module, the impedance cardiogram sampling frequency is set to 128Hz. Based on the impedance cardiogram signal, the Kubicek formula is used to calculate cardiac output (CO). Cardiac output refers to the amount of blood pumped by the heart per unit time, with an error of ≤5%. Laser Doppler uses a 670nm wavelength light source to measure mesenteric microcirculatory blood flow velocity. When the blood flow velocity is less than 15cm / s and the CO decreases by 20%, it is determined to be in the early stage of hemorrhagic shock. For intraperitoneal pressure measurement in the trauma monitoring module, the catheter model is F16-F18, and the saline solution is maintained at 37°C to avoid bladder irritation. During measurement, the patient must remain in the supine position for at least 2 minutes to ensure data stability.

[0031] In this embodiment, it should also be noted that the behavioral feature recognition unit includes a non-contact sensing module and an intelligent behavioral analysis module. The non-contact sensing module uses infrared thermal imaging, millimeter-wave radar, and a microphone array to capture abnormal body surface temperature, micro-posture movements, and pain-related voiceprint characteristics. Infrared thermal imaging detects surface temperature gradients and identifies ischemic areas with temperature differences greater than 2°C. The millimeter-wave radar operates in the 60-64 GHz frequency band and uses the micro-Doppler effect to capture trauma-specific body position changes of 0.1-5 Hz. The microphone array uses beamforming technology to extract patient moaning characteristics in the 80-300 Hz frequency band, achieving a signal-to-noise ratio of ≥15 dB. The intelligent behavioral analysis module, based on a deep learning model, converts body movements and facial expressions into clinical risk indicators. The specific operations are as follows: B1: Behavioral feature data preprocessing: ① Segment the temperature field of infrared thermal imaging data and extract the temperature difference of the limb ends; ② Perform time-frequency transformation on millimeter-wave radar data to obtain the respiratory rate curve and body acceleration features; ③ Use the Openpose algorithm to extract the coordinates of 18 facial key points from the video image and calculate the expression parameters of frown amplitude and eyelid opening and closing; B2: Deep learning model architecture: ① Use a multimodal fusion network, including: 1) Visual branch: 3D CNN network, input is the sequence of facial key points, extract the dynamic features of expression; 2) Radar branch: LSTM network, input is the time series data of respiratory rate and body acceleration, extract the micro-motion pattern features; 3) Thermal imaging branch: 2D CNN network, input is the temperature field image, extract the features of abnormal surface temperature area; ② The output of each branch is weighted fusion through the attention mechanism, and the weight formula is: ,in, is the eigenvector of the i-th branch, 、 is a learnable parameter; B3: Clinical risk indicator mapping: ① The model output layer uses a fully connected network to generate the risk probability value P through the Sigmoid activation function, and the formula is: ,in is the fusion feature vector, is the Sigmoid function , W o is the weight matrix of the output layer, b o is the bias vector of the output layer; ② Convert the risk probability P into clinical indicators: I. Pain level: When Pain ≥ 7, it is judged as severe pain; II. Shock risk index: ,in, is the limb temperature difference, when B4: Model training: ① The training dataset contains behavioral data of 1,000 emergency patients, annotated with clinical diagnosis results; ② The cross entropy loss function is used to optimize the model parameters, the formula is: , where y is the true label, and the Adam algorithm is iterated for 50 rounds.

[0032] Among them, it should be noted that the non-contact perception module collects the patient's surface temperature, body posture dynamics and voiceprint characteristics in a multi-modal manner, and the intelligent behavior analysis module transforms the multi-source behavior data into clinical risk indicators based on the deep learning model.

[0033] Furthermore, it should be noted that the infrared thermal imager uses a vanadium oxide (VOx) detector with a resolution of 384×288 and a temperature sensitivity of ≤0.05°C, capable of identifying ischemic areas with a temperature difference of >2°C at the extremities (e.g., fingers). The millimeter-wave radar utilizes FMCW technology, transmitting linear frequency modulation signals in the 60-64 GHz frequency band. It uses the micro-Doppler effect to extract 0.1-5 Hz respiratory and body motion signals, achieving a motion amplitude detection accuracy of 1 mm. The multimodal fusion network of the intelligent behavior analysis module uses a ResNet-3D architecture for the visual branch, consisting of 16 residual blocks. The input is a 16-frame sequence of facial keypoints (18 2D coordinates per frame). The system uses spatiotemporal feature extraction to identify facial fasciculations that precede epileptic seizures.

[0034] In this embodiment, it should also be noted that the intelligent early warning and decision support unit includes a three-dimensional early warning matrix module, a time-sensitive intervention module, and a clinical pathway trigger module. The three-dimensional early warning matrix module is used to generate a risk level by integrating physiological parameters, behavioral characteristics, and the ISS score. The specific operations are as follows: C1: Generates a composite risk index based on the following dimensions: ① Physiological dimension: Systolic blood pressure <90 mmHg and heart rate >120 bpm are each scored 2 points; ② Behavioral dimension: Forced posture is scored 3 points, and groaning in pain is scored 1 point; ③ Trauma dimension: Every 5 points in the ISS score is scored 1 point; C2: When the CRI is ≥8, a red alert is triggered and the corresponding trauma treatment protocol is pushed. The time-sensitive intervention module is used to push time-sensitive plans based on the "golden hour" principle; and the clinical pathway trigger module is used to automatically connect to the hospital's HIS system and retrieve the corresponding trauma treatment protocol.

[0035] Among them, it should be noted that the three-dimensional early warning matrix module evaluates the patient's risk level in multiple dimensions, the time-sensitive intervention module generates a time-sensitive treatment plan based on the "golden 1 hour" principle, and the clinical pathway trigger module automatically calls the matching trauma treatment protocol.

[0036] Furthermore, it should be noted that the three-dimensional early warning matrix module's composite risk index calculation includes physiological dimensions, including lactate levels (Lac ≥ 4mmol / L, 3 points) and urine output (<0.5ml / kg / h, 2 points). The behavioral dimension also includes a score for "confused behavior" (e.g., inability to follow instructions, 4 points). The trauma dimension combines the ISS score and injury location, adding an additional 2 points to the weighting for patients with craniocerebral injury. The time-sensitive intervention module, based on blockchain timestamp technology, records the patient's injury time (obtained through the behavioral feature recognition unit's instantaneous motion detection), hospital admission time, and the execution time of each intervention measure. A level 3 audible and visual alarm is triggered when the remaining golden time is ≤15 minutes.

[0037] In this embodiment, it should also be noted that the remote collaboration and communication unit includes an AR spatial annotation module, a data sandbox module, and a multidisciplinary conversation module, among which: the AR spatial annotation module: used by the expert side to mark the puncture point or bleeding location on the patient's 3D body surface projection; the data sandbox module: used for encrypted transmission of DICOM images and vital trend data, supporting secure access by third-party devices; the multidisciplinary conversation module: used to establish an exclusive communication channel for the trauma team, supporting real-time multi-party consultation using voice, text, and images.

[0038] Among them, it should be noted that the AR spatial annotation module realizes anatomical positioning visualization, the data sandbox module ensures the safe interaction of multimodal medical data, and the multidisciplinary conversation module establishes an efficient collaborative channel.

[0039] Furthermore, it should be noted that the AR spatial annotation module supports mixed reality devices (such as Microsoft HoloLens). Experts can project CT images on the patient's actual body surface and mark the liver rupture area through gestures, with a marking error of ≤3mm; the data sandbox module uses the AES-256 encryption algorithm to transmit DICOM images, complies with HIPAA privacy protection standards, and supports real-time transmission of 4K video under 5G networks (delay ≤50ms).

[0040] Throughout this specification, references to terms such as "one embodiment," "example," or "specific example" indicate that the specific features, structures, materials, or characteristics described in conjunction with that embodiment or example are included in at least one embodiment or example of the present invention. In this specification, schematic representations of these terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0041] The preferred embodiments of the present invention disclosed above are intended only to help illustrate the present invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the present invention to the specific embodiments described. Obviously, many modifications and variations are possible based on the content of this specification. These embodiments are selected and described in detail in this specification to better explain the principles and practical applications of the present invention, thereby enabling those skilled in the art to better understand and utilize the present invention. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. A comprehensive emergency monitoring system for patients in the emergency department, characterized by: It includes patient information management unit, real-time detection unit, behavior feature recognition unit, intelligent early warning and decision support unit, and remote collaboration and communication unit, among which: The patient information management unit is used to quickly enter and prioritize key information of emergency patients through automatic recognition technology and dynamic data tagging; The real-time detection unit is used to perform accurate hemodynamic assessment and continuous vital sign monitoring of acute trauma patients through a trauma-specific multimodal monitoring architecture and a dynamic anti-interference mechanism; The behavioral feature recognition unit is used to dynamically identify and evaluate clinical risk indicators of patients in an emergency environment through non-contact multimodal perception and intelligent behavioral analysis; The intelligent early warning and decision support unit: based on the trauma progression prediction algorithm and the ISS score association model, provides risk warning and generates a graded intervention plan; The remote collaboration and communication unit is used to utilize AR annotation and secure data sharing to support interdisciplinary teams to participate in emergency and critical care in real time.

2. The all-round monitoring and emergency system for emergency patients according to claim 1, characterized in that: The patient information management unit includes an automatic identity recognition module, a data acquisition module, a priority marking module, and an information synchronization module, wherein: The automatic identity recognition module is used to quickly identify and bind the identity of emergency patients through barcode scanning, RFID or face recognition technology; The data acquisition module is used to automatically obtain key medical data such as the patient's admission information, basic medical history, and allergy history; The priority marking module is used to dynamically mark the patient's treatment priority according to the degree of injury and vital signs; The information synchronization module is used to synchronize patient information to relevant monitoring and diagnosis systems in real time.

3. The all-round monitoring and emergency system for emergency patients according to claim 1, characterized in that: The real-time detection unit includes a dual-mode monitoring module, a motion artifact suppression module, and a trauma-specific monitoring module, wherein: The dual-mode monitoring module is used to synchronously run impedance cardiography and laser Doppler to jointly evaluate bleeding volume and cardiac function; The motion artifact suppression module is used to dynamically filter signal noise caused by patient movement using IMU sensor data; The trauma-specific monitoring module is used to calculate the intra-abdominal pressure and perfusion index in real time and identify latent shock.

4. The all-round monitoring and emergency system for emergency patients according to claim 3, characterized in that: The trauma monitoring module calculates intra-abdominal pressure and perfusion index in real time to identify latent shock. The specific operations are as follows: A1: Intra-abdominal pressure measurement: Inject 50 ml of normal saline into the patient's bladder through the catheter, keep the patient in a supine position, and measure the intravesical pressure at the level of the pubic symphysis. ; Intra-abdominal pressure calculation formula: ,in, is the intra-abdominal pressure, is the current atmospheric pressure, and the measurement error after correction is ≤2mmHg; A2: Perfusion index calculation: ① Calculation of gastric mucosal pH: 1) Collect gastric mucosal carbon dioxide partial pressure through a nasogastric tube; 2) Simultaneously collect arterial carbon dioxide partial pressure; 3) Calculation formula: ,in, is the gastric mucosal pH, 0.03 is the carbon dioxide solubility coefficient; ② Central venous oxygen saturation monitoring: Blood samples are collected through a central venous catheter and the central venous oxygen saturation is measured using spectral analysis with an accuracy of ; A3: Latent shock identification logic: When any of the following conditions is met, the latent shock warning is triggered: 1) IAP ≥ 16 mmHg, and ; 2) IAP < 16 mmHg, but pHi < 7.30, ScvO2 < 65%, and lactate value Lac ≥ 2.5 mmol / L; A4: Warning level is based on the formula Calculate, when is judged as high risk.

5. The all-round monitoring and emergency system for emergency patients according to claim 1, characterized in that: The behavior feature recognition unit includes a non-contact sensing module and an intelligent behavior analysis module, wherein: The non-contact sensing module is used to capture abnormal body surface temperature, body position micro-movements, and pain-related voiceprint characteristics of the patient through infrared thermal imaging, millimeter-wave radar, and microphone array; The intelligent behavior analysis module converts body movements and facial expressions into clinical risk indicators based on a deep learning model.

6. The all-round monitoring and emergency system for emergency patients according to claim 5, characterized in that: The infrared thermal imaging is used to detect the distribution of surface temperature gradients and identify local ischemic areas with a temperature difference greater than 2°C. The millimeter-wave radar operates in the 60-64 GHz frequency band and captures trauma-specific body position changes of 0.1-5 Hz through the micro-Doppler effect. The microphone array uses beamforming technology to extract the characteristics of patient moans in the 80-300 Hz frequency band, with a signal-to-noise ratio of ≥15dB.

7. The all-round monitoring and emergency system for emergency patients according to claim 5, characterized in that: The intelligent behavior analysis module is based on a deep learning model and converts body movements and facial expressions into clinical risk indicators. The specific operations are as follows: B1: Behavioral feature data preprocessing: ①Segment the temperature field of infrared thermal imaging data and extract the temperature difference of the limb ends; ②Perform time-frequency transformation on millimeter-wave radar data to obtain respiratory rate curve and body acceleration characteristics; ③ The Openpose algorithm is used to extract the coordinates of 18 facial key points from the video image and calculate the expression parameters of frown amplitude and eyelid opening and closing; B2: Deep Learning Model Architecture: ①Adopt a multimodal fusion network, including: 1) Vision branch: 3D CNN network, which takes a sequence of facial key points as input and extracts dynamic features of facial expressions; 2) Radar branch: LSTM network, which takes respiratory rate and body acceleration time series data as input and extracts micro-motion pattern features; 3) Thermal imaging branch: 2D CNN network, which takes temperature field images as input and extracts features of abnormal surface temperature areas; ② The outputs of each branch are weighted and fused through the attention mechanism. The weight formula is: ,in, is the eigenvector of the i-th branch, 、 is a learnable parameter; B3: Clinical Risk Indicator Mapping: ① The model output layer uses a fully connected network and generates the risk probability value P through the Sigmoid activation function. The formula is: ,in is the fusion feature vector, is the Sigmoid function , W o is the weight matrix of the output layer, b o is the bias vector of the output layer; ②Convert the risk probability P into clinical indicators: Ⅰ. Pain level: , when Pain ≥ 7, it was judged as severe pain; II. Shock risk index: ,in, is the limb temperature difference, when When triggering an early warning; B4: Model training: ① The training dataset contains behavioral data of 1,000 emergency patients, annotated with clinical diagnosis results; ② Use the cross entropy loss function to optimize the model parameters. The formula is: , where y is the true label, and the Adam algorithm is iterated for 50 rounds.

8. The all-round monitoring and emergency system for emergency patients according to claim 1, characterized in that: The intelligent early warning and decision support unit includes a three-dimensional early warning matrix module, a time-sensitive intervention module, and a clinical pathway trigger module, wherein: The three-dimensional warning matrix module is used to generate risk levels by integrating physiological parameters, behavioral characteristics and ISS scores; The time-sensitive intervention module is used to push time-sensitive solutions based on the "golden hour" principle; The clinical pathway trigger module is used to automatically link to the hospital HIS system and retrieve the corresponding trauma treatment protocol.

9. The all-round monitoring and emergency system for emergency patients according to claim 8, characterized in that: The three-dimensional warning matrix module integrates physiological parameters, behavioral characteristics and ISS scores to generate risk levels. The specific operations are as follows: C1: Generates a composite risk index based on the following dimensions: ① Physiological dimension: systolic blood pressure <90 mmHg, heart rate >120 beats / min, each scored 2 points; ②Behavioral dimension: forced posture is scored as 3 points, pain groaning is scored as 1 point; ③ Trauma dimension: 1 point for every 5 points of ISS score; C2: When CRI ≥ 8 points, a red alert is activated and the corresponding trauma treatment protocol is pushed.

10. The all-round monitoring and emergency system for emergency patients according to claim 1, characterized in that: The remote collaboration and communication unit includes an AR space annotation module, a data sandbox module, and a multidisciplinary conversation module, wherein: The AR spatial annotation module is used by the expert to mark the puncture point or bleeding location on the patient's 3D body surface projection; The data sandbox module is used for encrypting and transmitting DICOM images and life trend data, and supports secure access by third-party devices; The multidisciplinary conversation module is used to establish a dedicated communication channel for the trauma team and supports real-time multi-party consultations using voice, text, and images.

Images