A preoperative waiting vital sign monitoring and early warning method based on rPPG

Abstract

This invention discloses a preoperative waiting vital sign monitoring and early warning method based on rPPG, belonging to the field of vital sign monitoring and early warning. Its purpose is to solve the technical problem of low accuracy in vital sign monitoring and alarms due to the lack of vital sign monitoring in waiting areas outside operating rooms and catheterization rooms, and the inability to dynamically set thresholds based on the individual conditions of patients in the waiting area. The monitoring and early warning method includes: acquiring facial video streams of patients in the waiting area; extracting facial features and vital sign time-series data from the facial video streams; comparing the facial features with patient information in the hospital information system to identify patient identity information and retrieve medical records; generating individual baselines for each indicator based on individualized physiological baselines in the medical records; generating dynamic thresholds for corresponding indicators based on individual baselines and key diagnostic information; and using the vital sign time-series data and the dynamic thresholds for corresponding indicators for monitoring and early warning of patient vital signs.

Description

Technical Field

[0001] This invention belongs to the field of medical technology and relates to the monitoring and early warning of vital signs, particularly to a method for monitoring and early warning of preoperative waiting vital signs based on rPPG. Background Technology

[0002] rPPG (Remote Photo-plethysmography) is a non-contact detection technology that uses a camera to capture subtle changes in skin color caused by heartbeats to estimate heart rate, thereby enabling the monitoring and early warning of patients' vital signs. It has been applied in fields such as health monitoring.

[0003] Currently, the mature application of rPPG technology in patient monitoring within medical environments primarily focuses on monitoring bedridden patients in inpatient wards. By installing cameras in the wards, the system captures subtle changes in the patient's skin color, measures vital signs, and sets fixed, universal thresholds for abnormal alarms. When the vital signs measured based on the patient's skin color changes exceed the set thresholds, the system generates an alarm, requiring medical staff to investigate and address the issue.

[0004] The invention patent application with application number 202510465425.8 discloses an online vital sign monitoring method based on rPPG emergency care, which is used to quickly determine the patient's vital signs and whether the patient needs pre-hospital emergency care before emergency care. It includes: acquiring a facial video of the face to be detected; determining the changes in subcutaneous blood volume in the face presented in the facial video, including images before and after the change; processing the changed images using rPPG to determine whether the vital signs corresponding to the face are preset emergency signs; if the vital signs are preset emergency signs, generating and outputting an emergency message; when determining whether the vital signs corresponding to the emergency indication information are preset emergency signs, it includes: determining the exponential function values ​​of the standard deviation of the first indication information, the second indication information, the third indication information, and the fourth indication information respectively, obtaining the first exponential function value, the second exponential function value, the third exponential function value, and the fourth exponential function value; determining the weights of the first exponential function value, the second exponential function value, the third exponential function value, and the fourth exponential function value in real time; using each weight, weighting the first exponential function value, the second exponential function value, the third exponential function value, and the fourth exponential function value to obtain the emergency indication information; determining the vital signs corresponding to the emergency indication information, and determining whether the vital signs are preset emergency signs.

[0005] Similar to the aforementioned patent application, existing technologies can utilize rPPG for monitoring in scenarios such as wards and pre-hospital emergency care. However, before surgery or catheterization, patients are typically accompanied by family members or transport workers to waiting areas outside the operating room or catheterization room. These areas are usually equipped with ordinary security cameras, limited to video monitoring and recording. Patients waiting for surgery or catheterization are often critically ill, lacking proper vital sign monitoring during the waiting period, and are at risk of continuous, unprofessional physiological monitoring. Furthermore, existing technologies for vital sign monitoring using rPPG do not integrate or correlate with hospital medical information systems (such as HIS and EMR). Patient images acquired using rPPG cannot be automatically matched or correlated with patient identity information, historical medical records, recent diagnoses, and personal physiological baseline data. Thresholds cannot be dynamically set based on the patient's individual condition, resulting in low accuracy of monitoring alarms. Summary of the Invention

[0006] The purpose of this invention is to address the technical problems in the prior art where there is a lack of vital sign monitoring in the waiting area outside the operating room and catheterization room, and the inability to dynamically set thresholds based on the individual conditions of patients in the waiting area, resulting in low accuracy of vital sign monitoring alarms. This invention provides a preoperative waiting vital sign monitoring and early warning method based on rPPG.

[0007] To achieve the above objectives, the present invention specifically adopts the following technical solution: A preoperative waiting vital signs monitoring and early warning method based on rPPG includes the following steps: Step 1: Data Collection; Acquire facial video streams of patients in the waiting area, extract facial features of patients from the facial video streams, and use rPPG technology to calculate the time series data of patients' vital signs from the facial video streams; Step 2, Identity Verification; The extracted facial features are compared with patient information in the hospital information system to identify the patient's identity and retrieve the patient's medical records; the retrieved medical records include individualized physiological baselines and key diagnostic information. Step 3: Generate individual baselines; Based on individualized physiological baselines, individual baselines for each indicator are generated; Step 4: Generate dynamic thresholds; Based on individual baselines and key diagnostic information, dynamic thresholds for corresponding indicators are generated. Step 5: Monitoring and early warning; The time-series data of vital signs calculated using rPPG technology and the dynamic thresholds of corresponding indicators are used for monitoring and early warning of patients' vital signs.

[0008] Furthermore, in step 1, the vital signs time series data include heart rate and respiratory rate calculated by rPPG technology, as well as blood oxygen saturation trend and blood pressure fluctuation index calculated based on the calculated blood oxygen saturation and blood pressure, respectively.

[0009] Furthermore, in step 3, when generating the individual baseline, the calculation formula for the individual baseline is: ; in, Indicates the number of resting measurements. This represents the i-th historical physiological indicator of the patient during the k-th time when the patient was at rest. This indicates the actual measured duration of rPPG. This indicates the i-th physiological indicator measured by rPPG at minute t within T minutes before the patient's emotions stabilize after entering the waiting area; The weighting coefficients representing historical physiological indicators. Represents the weighting coefficients of the measured rPPG data, and .

[0010] Furthermore, in step 4, when generating the dynamic threshold for the corresponding indicator, the formula for calculating the dynamic threshold is: ; in, This represents the individual baseline for the i-th physiological indicator. This represents the relative tolerance for fluctuations in the i-th physiological indicator based on the main diagnostic set D. This indicates the absolute red line offset.

[0011] Furthermore, in step 5, during the warning process, a dynamic threshold is used... Individual baseline and measured physiological indicators of rPPG Calculate real-time risk value If the real-time risk value If the risk value exceeds the preset risk intervention threshold, an early warning will be triggered; real-time risk value The calculation formula is: ; in, Represents a set of physiological indicators. This represents the risk weight of the i-th physiological indicator in the main diagnostic set D. This indicates the i-th physiological indicator measured by rPPG at minute t within T minutes before the patient's emotions stabilized after entering the waiting area. This represents the individual baseline for the i-th physiological indicator. This represents the dynamic threshold of the i-th physiological indicator; This represents the time penalty coefficient. Indicates the duration of continuous exceeding of the limit. This indicates a time-indexed penalty.

[0012] Furthermore, for patients with a primary diagnosis of atrial fibrillation and / or arrhythmia, the beat-by-beat interval is extracted using rPPG technology, and the temporal characteristic RMSSD of heart rate variability is calculated. If the temporal characteristic RMSSD is continuously monitored to be greater than the preset atrial fibrillation critical threshold, and is accompanied by a rapid increase in local ventricular rate, a "high-risk arrhythmia" warning is triggered. The formula for calculating the temporal characteristic RMSSD is: ; in, This indicates the total number of heartbeats displayed in the window. This indicates the period of the nth heartbeat; For patients with a primary diagnosis of myocardial infarction (D), heart rate was monitored using rPPG technology, and the shock severity index was calculated. When the heart rate derivative is positive and rises sharply, and the blood oxygen derivative is negative, if the shock deterioration index... A surge in the shock index triggers a "high-risk shock" warning; the shock deterioration index... The calculation formula is: ; in, This represents the macroscopic trend value of heart rate after "smoothing filtering". This represents the macroscopic trend value of blood oxygen saturation after "smoothing filtering". To represent a constant, let's set it to 10. -5 .

[0013] Furthermore, in step 5, after the warning is triggered, a two-level alarm is activated: Level 1 Alarm: Automatically generates alarm information including patient location, name, abnormal vital signs parameters, degree of deviation, and preliminary risk assessment, and pushes it to the responsible nurse's PDA in real time with the highest priority; Level 2 prompt: Announcements will be posted on the on-site information display screen in the waiting area.

[0014] The beneficial effects of this invention are as follows: 1. In this invention, facial features are extracted from the facial video stream of patients in the waiting area and compared with the hospital information system to obtain the patient's identity information and medical records. Based on these records, a personalized alarm threshold related to the patient's condition can be dynamically generated. Then, rPPG technology is used to calculate the patient's vital signs time-series data from the facial video stream in the waiting area and compare it with the dynamically generated alarm threshold. This enables seamless, continuous, and precise monitoring of high-risk patients waiting outside the operating room and catheterization room, solving the monitoring vacuum problem for these areas. The entire monitoring and early warning system is contactless, providing 24 / 7 continuous physiological parameter monitoring of patients in the waiting area without increasing the burden on medical staff or disturbing patients. This transforms passive patrols into proactive early warnings, significantly improving patient safety during the medical transition process. Furthermore, the dynamically generated personalized thresholds based on automatically acquired patient medical records are more tailored to the individual patient compared to fixed-threshold general alarms, reducing false alarms due to individual differences and significantly improving the accuracy of vital sign monitoring alarms.

[0015] 2. In this invention, real-time physiological data streams are deeply integrated with static medical record information, so that the monitoring data is no longer "anonymous" and becomes a decision-making basis with a complete clinical background.

[0016] 3. In addition to threshold comparison, this invention also extracts features directly from the original waveforms obtained by rPPG technology for some high-risk and critical illnesses, and obtains corresponding features / indices based on specific algorithms. This can more accurately identify clinically significant physiological deterioration events, significantly reduce false alarms caused by the characteristics of specific diseases, and make alarm information more valuable for clinical action.

[0017] 4. In this invention, after the early warning is triggered, a closed-loop alarm with multiple terminals and hierarchical collaboration is activated to shorten the intervention time. In addition, through precise push to PDA and synchronous prompts on the on-site screen, it is ensured that risk information can reach the responsible medical staff as soon as possible and assist them in quickly locating the patient, forming a rapid closed loop of "discovery-notification-location-treatment", which wins valuable time for rescue. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments.

[0020] Therefore, all other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0021] Example 1 This embodiment provides a preoperative waiting vital sign monitoring and early warning method based on rPPG, used for monitoring and warning of vital signs of patients waiting in the waiting area outside the operating room and catheterization room. Figure 1 As shown, this monitoring and early warning method includes five steps: data collection, identity verification, generation of individual baselines, generation of dynamic thresholds, and monitoring and early warning. The specific details of each step are as follows: Step 1: Data Collection; The system acquires facial video streams of patients in the waiting area, extracts facial features from the video streams, and uses rPPG technology to calculate the patients' vital signs time-series data from the facial video streams.

[0022] Visual perception terminals (existing terminals, such as high-definition cameras supporting visible light and near-infrared spectroscopy) are installed in the waiting areas of operating rooms and catheterization labs. These terminals have built-in or connected edge computing units responsible for video acquisition and face detection and tracking. The visual perception terminals capture facial video streams of patients in the waiting area and, using face detection and tracking algorithms (existing algorithms are acceptable), locate and crop standard facial images in real time, extracting facial features for subsequent identity verification. Using rPPG technology (remote photoplethysmography algorithm, existing algorithm), changes in the patient's skin color are identified from the facial video stream, and the patient's heart rate (HR), respiratory rate (RR), blood oxygen saturation (SpO2), blood pressure, and other vital signs are calculated in real time. Blood oxygen saturation and blood pressure are calculated (using existing methods) to obtain corresponding blood oxygen saturation trends and blood pressure fluctuation indices. The directly calculated heart rate and respiratory rate, along with the blood oxygen saturation trends and blood pressure fluctuation indices calculated from the calculated blood oxygen saturation and blood pressure, together constitute the vital signs time-series data.

[0023] The duration of each video stream processed by rPPG technology can be set to 5-10 seconds according to requirements.

[0024] Step 2, Identity Verification; The extracted facial features are compared with patient information in the hospital information system to identify the patient's identity and retrieve the patient's medical records.

[0025] The extracted facial features are compared with patient information in the hospital information system (HIS / electronic medical record EMR) to automatically and seamlessly identify the patient's identity and retrieve the patient's medical records from the hospital information system. The retrieved medical records include individualized physiological baselines and key diagnostic information.

[0026] Individualized physiological baseline refers to historical data such as heart rate, respiratory rate, blood oxygen saturation trend, and blood pressure variability index. These data are measured multiple times by the patient in a recent resting state (among which, blood oxygen saturation trend and blood pressure variability index are calculated from the corresponding measured blood oxygen saturation and blood pressure, rather than being directly measured).

[0027] Key diagnostic information refers to the primary diagnosis upon admission, which is the diagnosis given by the doctor based on the patient's self-report, examination results, etc., such as acute myocardial infarction, unstable angina, atrial fibrillation, heart failure, etc.

[0028] Step 3: Generate individual baselines; Individual baselines for each indicator are generated based on individualized physiological baselines in medical records.

[0029] This application does not use a fixed threshold, but dynamically generates an individual baseline based on the retrieved individualized physiological baseline. This dynamically generated individual baseline is the patient's exclusive safety detection threshold (which can be a specific value or a threshold range).

[0030] When generating an individual baseline, the formula for calculating the individual baseline is: ; in, The number of resting measurements is usually taken from the patient's most recent 3-5 stable measurements within the last 24 hours, i.e., N is usually 3, 4 or 5; This represents the i-th physiological indicator in the patient's history during the k-th resting state (i represents the trend of heart rate (HR), respiratory rate (RR), blood oxygen (SpO2) saturation, and blood pressure fluctuation index). This indicates the actual measured duration of rPPG. This refers to the i-th physiological indicator measured by rPPG at minute t within T minutes before the patient's emotions stabilize after entering the waiting area (usually set as the first 3-5 minutes after the patient sits down). The weighting coefficients representing historical physiological indicators, Represents the weighting coefficients of the measured rPPG data, and .

[0031] Step 4: Generate dynamic thresholds; Based on individual baselines and key diagnostic information, dynamic thresholds for corresponding indicators are generated.

[0032] After generating and obtaining individual baselines, based on the master diagnostic set D extracted from the system (which is a set of diagnostic information, each of which is usually in the form of a code, such as ICD=10), corresponding personalized upper and lower limit thresholds (i.e., dynamic thresholds) are calculated and generated for different indicators.

[0033] When generating the dynamic threshold for the corresponding indicator, the formula for calculating the dynamic threshold is: ; in, This represents the individual baseline for the i-th physiological indicator. This represents the relative tolerance for fluctuations of the i-th physiological indicator based on the main diagnostic set D (e.g., patients with hyperthyroidism or atrial fibrillation have a higher tolerance for heart rate fluctuations, which can be set to 20%-30%; for patients with severe heart failure or unstable angina, the tolerance can be set to 10%-15%; for patients with severe coronary heart disease, the tolerance is extremely low, which can be set to 5%). This indicates the absolute red line deviation, which can be a hard boundary difference specified in medical guidelines. It does not change with the baseline (intervention is necessary when blood oxygen SpO2 saturation drops by more than 5% or the absolute value is below 90%; upper limit of heart rate: 140 bpm; lower limit: 40 bpm, intervention is also necessary if it exceeds this range).

[0034] Step 5: Monitoring and early warning; The time-series data of vital signs calculated using rPPG technology and the dynamic thresholds of corresponding indicators are used for monitoring and early warning of patients' vital signs.

[0035] The system can directly compare the measured vital signs time-series data using rPPG technology with the corresponding dynamic thresholds. If the measured vital signs time-series data continuously exceeds the limits, an alert will be triggered. Alternatively, a risk value can be further generated based on the measured vital signs time-series data, and this risk value can be compared with a preset threshold. If the risk value continuously exceeds the limits, an alert will also be triggered.

[0036] Example 2 Patients waiting in the waiting area outside the operating room or catheterization room often experience a deterioration in their condition not due to a sudden change in a single indicator, but rather a sustained deviation in multiple indicators.

[0037] Therefore, in this embodiment, when performing monitoring and early warning, a risk scoring formula with a time penalty factor is used to calculate the real-time risk value based on a dynamic threshold. If the real-time risk value If the risk value exceeds the set risk intervention threshold, an early warning will be triggered. Real-time risk value. The calculation formula is: ; in, It represents a set of physiological indicators, including heart rate (HR), respiratory rate (RR), blood oxygen (SpO2) saturation trend, and blood pressure fluctuation index; This represents the risk weight of the i-th physiological indicator in the main diagnostic set D. This indicates the i-th physiological indicator measured by rPPG at minute t within T minutes before the patient's emotions stabilized after entering the waiting area. This represents the individual baseline for the i-th physiological indicator. This represents the dynamic threshold of the i-th physiological indicator; This represents the time penalty coefficient, which determines the "expansion (amplification) rate" of the risk score as the abnormal time increases. The larger the value, the steeper the curve, the lower the tolerance for "prolonged abnormality", and the more quickly an alarm is triggered (high sensitivity parameters (such as blood oxygen saturation, 0.3-0.5), low sensitivity parameters (such as heart rate, 0.05-0.1)). This indicates the duration of continuous deviation from the limit. It represents the length of time a specific physiological indicator i (such as blood oxygen or heart rate) remains outside the safe threshold. It does not require manual setting but is automatically calculated in real-time by the system program. Returning to the normal threshold range, Reset to zero immediately; This indicates a time-indexed penalty.

[0038] In this embodiment, for risk weights Risk weighting of heart rate in patients with atrial fibrillation Extremely high risk weights for respiratory rate (RR) and oxygen saturation (SpO2) in patients with asthma or COPD. The risk weight of the blood pressure variability index for hypertensive patients will be increased. It will be increased.

[0039] Penalty for time index This means that the longer the deviation time, the more exponentially the risk score amplifies, effectively filtering out instantaneous false alarms (such as data fluctuations of a few seconds caused by a patient suddenly turning around), and accurately capturing true physiological deterioration; among which... The larger the value, the lower the tolerance for long-term deviations, and the faster the warning is triggered.

[0040] Example 3 In addition to based on dynamic thresholds and / or real-time risk value In addition to comparing with the preset corresponding threshold, for some high-risk and severe cases, this embodiment also directly extracts features from the original waveforms obtained by rPPG technology, and obtains the corresponding features / indices based on specific algorithms, and performs early warning analysis based on the calculated features / indices.

[0041] For patients in the primary diagnostic set D with atrial fibrillation and / or arrhythmia, beat-by-beat intervals are extracted using rPPG technology, and the temporal characteristic RMSSD (root mean square of the difference between adjacent beat intervals) of heart rate variability (HRV) is calculated to detect rapid arrhythmias. If the temporal characteristic RMSSD is continuously monitored to be greater than the preset atrial fibrillation critical threshold, accompanied by a rapid increase in local ventricular rate, a "high-risk arrhythmia" warning is triggered. The formula for calculating the temporal characteristic RMSSD is: ; in, This indicates the total number of heartbeats displayed in the window. This indicates the nth heartbeat period.

[0042] For patients with myocardial infarction as the primary diagnosis set D, the "scissors difference" trend between compensatory increase in heart rate and progressive decrease in blood oxygen was detected using rPPG technology, and the shock deterioration index was calculated. When the heart rate derivative (slope) is positive and rising sharply, and the blood oxygen derivative is negative (i.e., declining), if the shock deterioration index... A surge in cases triggers a "high-risk shock" warning. Shock deterioration index. The calculation formula is: ; in, This represents the macroscopic trend value of heart rate after "smoothing filtering". The macroscopic trend value of blood oxygen saturation after "smoothing filtering" is obtained by using an exponentially weighted moving average algorithm to remove "short-term high-frequency fluctuations" caused by patient breathing, slight displacement, or changes in ambient light, while retaining "long-term low-frequency changes" caused by compensatory increases in heart rate and gradual decreases in blood oxygen due to cardiogenic shock. This represents a constant, usually set to 10. -5 This prevents the program from crashing when the blood oxygen derivative is 0.

[0043] Example 4 The continuously input real-time vital sign time-series data is compared and trended in real time with the dynamically generated personalized dynamic thresholds for the patient. Once data is found to continuously exceed limits (such as the dynamic threshold in Example 1 or the real-time risk value in Example 2), the system will take action. (Exceeding limits) or exhibiting a trend consistent with a specific disease deterioration pattern (such as the time-domain characteristic RMSSD or shock deterioration index in Example 3). This will immediately trigger an alert.

[0044] To avoid false alarms caused by sudden changes in patient position, an abnormality duration triggering time threshold can be set, with a recommended range of 10 to 30 seconds. An alert will only be triggered when the real-time vital signs data continuously exceed the personalized threshold for a period greater than the time threshold (i.e., continuous abnormal vital signs time - personalized threshold time > time threshold).

[0045] After the alert is triggered, a closed-loop alarm system with multi-terminal, hierarchical coordination is activated: Level 1 Alarm (Precise Push): Automatically generates alarm information including patient location, name, abnormal vital signs parameters, degree of deviation, and preliminary risk assessment, and pushes it to the responsible nurse's PDA in real time with the highest priority to ensure that key personnel are informed immediately; Level 2 prompts (on-site guidance): Simultaneously, on-site information display screens in the waiting area will display notices in a prominent but not panic-inducing manner (such as flashing the patient's bed / seat number and displaying the message "Please pay attention, medical staff") to assist medical staff in quickly locating the patient and reminding those nearby to provide initial assistance. Safety log recording: The entire alarm event, all relevant data, and subsequent handling and feedback from medical staff are fully recorded, forming an auditable medical safety closed-loop management log.

Claims

1. A preoperative waiting vital sign monitoring and early warning method based on rPPG, characterized in that, Includes the following steps: Step 1: Data Collection; Acquire facial video streams of patients in the waiting area, extract facial features of patients from the facial video streams, and use rPPG technology to calculate the time series data of patients' vital signs from the facial video streams; Step 2, Identity Verification; The extracted facial features are compared with patient information in the hospital information system to identify the patient's identity and retrieve the patient's medical records; the retrieved medical records include individualized physiological baselines and key diagnostic information. Step 3: Generate individual baselines; Based on individualized physiological baselines, generate individual baselines for each indicator; Step 4: Generate dynamic thresholds; Based on individual baselines and key diagnostic information, dynamic thresholds for corresponding indicators are generated. Step 5: Monitoring and early warning; The time-series data of vital signs calculated using rPPG technology and the dynamic thresholds of corresponding indicators are used for monitoring and early warning of patients' vital signs.

2. The method for preoperative waiting vital sign monitoring and early warning based on rPPG as described in claim 1, characterized in that, In step 1, the vital signs time series data include heart rate and respiratory rate calculated by rPPG technology, as well as blood oxygen saturation trend and blood pressure fluctuation index calculated based on the calculated blood oxygen saturation and blood pressure, respectively.

3. The method for preoperative waiting vital sign monitoring and early warning based on rPPG as described in claim 1, characterized in that, In step 3, when generating the individual baseline, the formula for calculating the individual baseline is: ； in, Indicates the number of resting measurements. This represents the i-th historical physiological indicator of the patient during the k-th time when the patient was at rest. This indicates the actual measured duration of rPPG. This indicates the i-th physiological indicator measured by rPPG at minute t within T minutes before the patient's emotions stabilize after entering the waiting area; The weighting coefficients representing historical physiological indicators, Represents the weighting coefficients of the measured rPPG data, and .

4. The method for preoperative waiting vital sign monitoring and early warning based on rPPG as described in claim 1, characterized in that, In step 4, when generating the dynamic threshold for the corresponding indicator, the formula for calculating the dynamic threshold is: ； in, This represents the individual baseline for the i-th physiological indicator. This represents the relative tolerance for fluctuations in the i-th physiological indicator based on the main diagnostic set D. This indicates the absolute red line offset.

5. The method for preoperative waiting vital sign monitoring and early warning based on rPPG as described in claim 1, characterized in that, In step 5, when an alert is issued, it is based on a dynamic threshold. Individual baseline and measured physiological indicators of rPPG Calculate real-time risk value If the real-time risk value If the risk value exceeds the preset risk intervention threshold, an early warning will be triggered; real-time risk value The calculation formula is: ； in, Represents a set of physiological indicators. This represents the risk weight of the i-th physiological indicator in the main diagnostic set D. This indicates the i-th physiological indicator measured by rPPG at minute t within T minutes before the patient's emotions stabilized after entering the waiting area. This represents the individual baseline for the i-th physiological indicator. This represents the dynamic threshold of the i-th physiological indicator; This represents the time penalty coefficient. Indicates the duration of continuous exceeding of the limit. This indicates a time-indexed penalty.

6. A preoperative waiting vital signs monitoring and early warning method based on rPPG as described in claim 1 or 5, characterized in that: For patients whose primary diagnosis is atrial fibrillation and / or arrhythmia, the beat-by-beat interval is extracted using rPPG technology, and the temporal characteristic RMSSD of heart rate variability is calculated. If the temporal characteristic RMSSD is continuously monitored to be greater than the preset atrial fibrillation critical threshold, and accompanied by a rapid increase in local ventricular rate, a "high risk of arrhythmia" warning is triggered. The formula for calculating the time-domain characteristic RMSSD is: ； in, This indicates the total number of heartbeats displayed in the window. This indicates the period of the nth heartbeat; For patients with a primary diagnosis of myocardial infarction (D), heart rate was monitored using rPPG technology, and the shock severity index was calculated. When the heart rate derivative is positive and rises sharply, and the blood oxygen derivative is negative, if the shock deterioration index... A surge in the shock index triggers a "high-risk shock" warning; the shock deterioration index... The calculation formula is: ； in, This represents the macroscopic trend value of heart rate after "smoothing filtering". This represents the macroscopic trend value of blood oxygen saturation after "smoothing filtering". To represent a constant, let's set it to 10. -5 .

7. The method for preoperative waiting vital sign monitoring and early warning based on rPPG as described in claim 1, characterized in that: In step 5, after the warning is triggered, a two-level alarm is activated: Level 1 Alarm: Automatically generates alarm information including patient location, name, abnormal vital signs parameters, degree of deviation, and preliminary risk assessment, and pushes it to the responsible nurse's PDA in real time with the highest priority; Level 2 prompt: Announcements will be posted on the on-site information display screen in the waiting area.

Images