Monitoring and early warning method, system and device for inoculation observation period

Abstract

The disclosure provides a method, system and device for monitoring and early warning of inoculation observation period, relating to the technical field of medical Internet of Things. The method comprises: continuously acquiring a plurality of source original observation vectors of the inoculated person in the observation period; synchronously extracting external confounding factors, calculating confounding offset and performing causal intervention correction on the plurality of source original observation vectors to obtain a deconfounded signal; performing effectiveness determination based on the coupling logic of the deconfounded signal and the external confounding factors to obtain an effectiveness mark; for the deconfounded signal with the effectiveness mark being effective, performing multi-modal weighted risk scoring, and combining a first path of extracting skin mucosa abnormalities accompanied by cardiovascular, respiratory or gastrointestinal abnormal characteristic signals, and a second path of not having skin mucosa abnormalities but accompanied by cardiovascular or respiratory acute abnormal characteristic signals, to output a graded early warning instruction. The technical scheme of the disclosure effectively excludes normal activities and emotional interference, and realizes early screening of suspected allergic events.

Description

Technical Field

[0001] This disclosure relates to the field of medical Internet of Things and artificial intelligence, and in particular to a monitoring and early warning method, device, equipment and storage medium for the observation period after vaccination. Background Technology

[0002] The 30-minute observation period after vaccination is the golden window for preventing rapid-type severe allergic reactions such as anaphylactic shock. Currently, clinical observation management mainly relies on manual rounds, but this method has the problems of long rounds and high subjectivity, which can easily lead to missed diagnoses of faint signs in the prodromal stage of occult allergies, thus delaying the best time for treatment.

[0003] With the widespread adoption of smart wearable devices, some existing technologies attempt to use conventional vital sign monitoring equipment (such as pulse oximeters and heart rate bracelets) to assist in observation. However, the observation area environment is complex, and normal physical activity and emotional stress (such as panic attacks or emotional stress) can easily cause drastic fluctuations in heart rate and blood pressure. Traditional monitoring solutions lack interference elimination mechanisms and multimodal cross-validation logic specific to the observation scenario, resulting in an extremely high false positive rate. This not only leads to the ineffective consumption of medical resources but also fails to achieve accurate triage and automated closed-loop intervention for truly fatal circulatory failure or acute cardiopulmonary abnormalities. Summary of the Invention

[0004] This disclosure provides a method, apparatus, device, and storage medium for monitoring and early warning during the observation period after vaccination, in order to solve or alleviate one or more technical problems in the prior art.

[0005] In a first aspect, this disclosure provides a monitoring and early warning method for the observation period after vaccination, characterized by comprising the following steps: Continuously acquire multi-source raw observation vectors of the recipients during the observation period; among which, the multi-source raw observation vectors include physiological observation vectors output by the vital signs acquisition device, visual observation vectors output by the visual acquisition device, and spatial posture observation vectors output by the spatial positioning device; Simultaneously extract multiple preset external confounding factors, calculate confounding offsets, and perform causal intervention correction on the multi-source original observation vectors to obtain deconfounded signals; The validity is determined by the coupled logic of the decontamination signal and the external contamination factor, and a validity label is obtained. For decontamination signals marked as valid, perform multimodal weighted risk scoring and, in conjunction with preset dual-path judgment rules, output graded early warning instructions; The dual-path determination rule includes: a first path triggered when the visual observation vector extracts abnormal skin and mucous membrane features and the physiological observation vector contains abnormal feature signals in the cardiovascular, respiratory, or gastrointestinal dimensions; and a second path triggered when there are no abnormal skin and mucous membrane features and the physiological observation vector contains acute abnormal feature signals in the cardiovascular or respiratory dimensions.

[0006] Secondly, this disclosure provides a monitoring and early warning system for the observation period after vaccination, characterized in that it includes: The data acquisition module is used to continuously acquire multi-source raw observation vectors of the recipients during the observation period. The multi-source raw observation vectors include physiological observation vectors output by the vital signs acquisition device, visual observation vectors output by the visual acquisition device, and spatial posture observation vectors output by the spatial positioning device. The edge computing module includes a promiscuous correction unit and a validity determination gateway. It is used to simultaneously extract multiple preset external promiscuous factors, calculate the promiscuous offset, and perform causal intervention correction on the multi-source original observation vector to obtain the depromiscuous signal. Based on the coupling logic of the depromiscuous signal and the external promiscuous factors, it performs validity determination and outputs validity labels. The comprehensive analysis module is used to perform multimodal weighted risk scoring on decontamination signals marked as valid, and output graded early warning instructions in combination with preset dual-path judgment rules; The dual-path determination rule includes: a first path triggered when the visual observation vector extracts abnormal skin and mucous membrane features and the physiological observation vector contains abnormal feature signals in the cardiovascular, respiratory, or gastrointestinal dimensions; and a second path triggered when there are no abnormal skin and mucous membrane features and the physiological observation vector contains acute abnormal feature signals in the cardiovascular or respiratory dimensions.

[0007] Thirdly, this disclosure provides a monitoring and early warning device for the observation period after vaccination, comprising: The acquisition module is used to continuously acquire multi-source raw observation vectors of the recipient during the observation period; among which, the multi-source raw observation vectors include physiological observation vectors output by the vital signs acquisition device, visual observation vectors output by the visual acquisition device, and spatial posture observation vectors output by the spatial positioning device. The decontamination module is used to simultaneously extract multiple preset external contamination factors, calculate the contamination offset, and perform causal intervention correction on the multi-source original observation vectors to obtain the decontamination signal. The determination module is used to determine the validity based on the coupled logic of the de-proliferated signal and the external proliferating factor, and obtain the validity label. The output module is used to perform multimodal weighted risk scoring on decontamination signals marked as valid, and output graded early warning instructions in combination with preset dual-path judgment rules; The dual-path determination rule includes: a first path triggered when the visual observation vector extracts abnormal skin and mucous membrane features and the physiological observation vector contains abnormal feature signals in the cardiovascular, respiratory, or gastrointestinal dimensions; and a second path triggered when there are no abnormal skin and mucous membrane features and the physiological observation vector contains acute abnormal feature signals in the cardiovascular or respiratory dimensions.

[0008] Fourthly, an electronic device is provided, comprising: At least one processor; and The memory is communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform any of the methods described in the present disclosure.

[0009] Fifthly, a non-transitory computer-readable storage medium is provided storing computer instructions, wherein the computer instructions are used to cause the computer to perform any of the methods according to embodiments of the present disclosure.

[0010] In a sixth aspect, a computer program product is provided, including a computer program that, when executed by a processor, implements any of the methods according to embodiments of the present disclosure.

[0011] The beneficial effects of the technical solution provided in this disclosure include at least the following: By simultaneously extracting external confounding factors and performing causal intervention correction on multi-source observation vectors, false fluctuations in physiological indicators caused by recipients' daily activities and stress were effectively filtered out, thereby achieving early screening and accurate graded warning of real risks and significantly improving the level of intelligent safety management during the vaccination observation period.

[0012] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this disclosure, nor is it intended to limit the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0013] In the accompanying drawings, unless otherwise specified, the same reference numerals throughout the various drawings denote the same or similar parts or elements. These drawings are not necessarily drawn to scale. It should be understood that these drawings depict only some embodiments provided according to this disclosure and should not be construed as limiting the scope of this disclosure.

[0014] Figure 1This is a flowchart illustrating the monitoring and early warning method for the observation period after vaccination provided in accordance with embodiments of this disclosure; Figure 2 This is a module architecture diagram of a monitoring and early warning system for the observation period after vaccination, provided according to an embodiment of this disclosure; Figure 3 This is a schematic diagram of a scenario provided according to an embodiment of this disclosure; Figure 4 This is a schematic diagram of the structure of a monitoring and early warning device for the observation period after vaccination, provided according to an embodiment of this disclosure; Figure 5 This is a block diagram of an electronic device used to implement embodiments of the present disclosure. Detailed Implementation

[0015] The present disclosure will now be described in further detail with reference to the accompanying drawings. The same reference numerals in the drawings denote elements that have the same or similar functions. Although various aspects of embodiments are shown in the drawings, they are not necessarily drawn to scale unless specifically indicated otherwise.

[0016] Furthermore, to better illustrate this disclosure, numerous specific details are set forth in the following detailed description. Those skilled in the art will understand that this disclosure can be practiced without certain specific details. In some instances, methods, means, components, and circuits well known to those skilled in the art have not been described in detail in order to highlight the main points of this disclosure.

[0017] Before introducing the technical solutions of the embodiments of this disclosure, the technical terms that may be used in this disclosure will be further explained: Multi-source raw observation vectors refer to the set of unprocessed objective physical quantities synchronously collected by heterogeneous sensor devices (such as wearable pulse oximeters, camera arrays, positioning beacons, etc.) distributed in different spatial locations. Causal Intervention Correction refers to a signal processing mechanism based on a causal inference mathematical model. It introduces interference quantifiers (such as Do-Calculus) to mathematically remove the superficial spurious correlation effects of external environmental or behavioral variables on target features, thereby extracting the true internal evolutionary trend of things.

[0018] Current vaccination observation and monitoring protocols often rely on manual timing or simple external smart bracelets for vital sign monitoring. This traditional method has significant technical limitations: First, traditional devices often only trigger alarms when there are late-stage signs of severe illness, such as a drop in blood pressure or blood oxygen levels, failing to capture subtle, early signs of allergic reactions (such as early skin redness combined with slight heart rate changes), leading to severe delays in warnings. Second, the observation environment is complex; normal movement of the recipient, mechanical falls due to children playing, and excessive panic about adverse vaccine reactions (psychogenic emotional fluctuations) can all cause drastic fluctuations in heart rate and blood pressure. Traditional systems lack multimodal cross-validation and decongestant algorithms, easily misjudging these physiological data fluctuations caused by daily behavior or psychological factors as critical events, resulting in a very high false positive rate and ineffective consumption of medical resources. Furthermore, the data flow in traditional warning systems is often fragmented, failing to achieve automatic closed-loop linkage from physical location of personnel to deployment of emergency medical equipment after a danger assessment.

[0019] To at least partially address one or more of the aforementioned problems and other potential issues, this disclosure proposes a monitoring and early warning method for the observation period after vaccination. By constructing a multi-source perception architecture integrating physiological, visual, and spatial posture data, and innovatively introducing a causal intervention decontamination gateway based on interference thresholds, adaptive amplification of genuine, subtle potential hazards and forced interception of false abnormal signals are achieved under a unified computational model. This method significantly improves the accuracy of early detection of abnormal states during the observation period and, through trimodal cross-validation and dual-path decision rules, effectively ensures robustness in identifying various phenotypic states. Ultimately, it achieves a closed loop from hazard feature extraction to fully automated, second-level response in physical emergency resource allocation.

[0020] This disclosure provides a monitoring and early warning method for the observation period after vaccination. This method can be applied to a monitoring and early warning device for the observation period after vaccination. The device is located in an electronic device, which includes, but is not limited to, fixed devices and / or mobile devices. For example, fixed devices include, but are not limited to, edge computing nodes, local servers, or cloud servers deployed in medical institutions; mobile devices include, but are not limited to, smart wearable terminals worn by vaccine recipients (such as smart bracelets, pulse oximeter rings), panoramic visual monitoring devices deployed in the observation area, and smart mobile dispatch terminals held by medical staff. In some possible implementations, the monitoring and early warning method can also be implemented by a processor calling computer-readable instructions stored in memory.

[0021] Figure 1 This is a flowchart illustrating the monitoring and early warning method for the inoculation observation period provided in the embodiments of this disclosure. Figure 1 As shown, the method includes the following steps: S110. Continuously acquire multi-source raw observation vectors of the recipient during the observation period. Among them, the multi-source raw observation vectors include physiological observation vectors output by the vital sign acquisition device, visual observation vectors output by the visual acquisition device, and spatial posture observation vectors output by the spatial positioning device.

[0022] In this embodiment, the multi-source raw observation vector can be understood as the initial state data set acquired by the system from various sensing terminals without interference removal. Specifically, the physiological observation vector can be obtained by continuously collecting vital signs such as heart rate, pulse oxygen saturation, blood pressure, and skin impedance using a non-invasive smart wearable device. The visual observation vector can be obtained by combining a multi-camera array deployed within the area with artificial intelligence image recognition algorithms, used to quantify abnormal states of the skin surface. The spatial posture observation vector can be obtained by combining high-precision indoor positioning technologies such as ultra-wideband (UWB) with a triaxial accelerometer, used to acquire position coordinates and basic posture.

[0023] For example, in a 30-minute observation period after vaccination, the vital signs acquisition device outputs the recipient's heart rate as 100 beats / min and systolic blood pressure as 110 mmHg in real time; this is the physiological observation vector. The visual acquisition device captures that the recipient's facial erythema pixels account for 10% of the total; this is the visual observation vector. The spatial positioning device outputs the recipient's current three-dimensional coordinates and their current walking state; this is the spatial posture observation vector.

[0024] S120. Simultaneously extract multiple preset external confounding factors, calculate the confounding offset, and perform causal intervention correction on the multi-source original observation vector to obtain the deconfounded signal.

[0025] In this embodiment, external confounding factors can be understood as interfering variables that cause pseudo-fluctuations in physiological indicators of the recipient under non-pathological states (such as daily activities, emotional changes, or external environmental stimuli). Causal intervention correction is a computational mechanism based on a causal inference framework (such as introducing the Do-Calculus operator) aimed at stripping away superficial correlations and restoring the true pathological changes. In specific implementation, the system senses and quantifies the current activity or emotional state in real time, calculates the normal data fluctuations that these states should cause (i.e., confounding offsets), and then subtracts the offset from the multi-source original observation vector.

[0026] For example, a recipient's heart rate increased to 110 beats per minute due to brisk walking in the observation area. The system extracted the external confounding factor of "walking," and calculated that this level of walking would cause the heart rate to deviate from the normal range by approximately 20 beats per minute. The system corrected this through causal intervention, subtracting the confounding deviation from the original heart rate to obtain a deconfounded signal after removing motion interference (at which point the corrected core heart rate appears stable).

[0027] S130. Validity determination is performed based on the coupled logic of the de-proliferated signal and the external proliferating factor to obtain a validity label.

[0028] In this embodiment, the coupling logic can be understood as the mutual corroboration relationship between different data dimensions in terms of medical pathology or physical laws. The validity marker is used to identify whether the current abnormal signal is a genuine risk indicator or invalid noise. Specifically, the authenticity of the signal is confirmed by determining whether the direction of change of the decontamination signal and the state of the contamination factor conform to a conventional combination of false positive characteristics.

[0029] For example, when an abnormally high systolic blood pressure is detected in a recipient, the system checks the recipient's current emotional confounding factor. If the recipient is in a state of extreme tension, the coupling logic determines that the blood pressure increase is caused by tension, thus invalidating the validity flag of the abnormal blood pressure signal and avoiding false alarms.

[0030] S140. For signals marked as valid after confounding, perform multimodal weighted risk scoring and, in conjunction with preset dual-path judgment rules, output graded early warning instructions. The dual-path judgment rules include a first path that extracts abnormal skin / mucous membrane features accompanied by abnormal cardiovascular, respiratory, or gastrointestinal signals, and a second path that does not possess abnormal skin / mucous membrane features but is accompanied by acute abnormal cardiovascular or respiratory signals.

[0031] In this embodiment, the multimodal weighted risk score refers to the process of assigning differentiated weights to effective signals from multiple dimensions such as physiological, visual, and spatial aspects and then summing them. A higher score indicates a greater risk. The dual-path judgment rule is a parallel verification logic customized to cover different pathological manifestations. In specific implementation, the system compares the calculated comprehensive score with a preset risk threshold. If obvious skin erythema or edema is visually identified, accompanied by fluctuations in any internal organ's vital signs, an instruction is triggered along the first path. If the visual appearance is completely normal, but blood pressure or blood oxygen shows a sharp drop, an instruction is triggered along the second path.

[0032] For example, in a severe allergy case, the recipient may not exhibit severe allergic reactions such as rashes on the skin and mucous membranes, but because it can rapidly progress to acute respiratory / circulatory failure, the system uses physiological equipment to determine that their cardiovascular and respiratory indicators are abnormally high. At this point, the system issues a red alert along the second pathway, reminding medical staff that acute cardiopulmonary failure may be occurring.

[0033] According to the scheme of this disclosure, by simultaneously extracting external confounding factors and performing causal intervention correction, and combining multimodal weighting and dual-path judgment rules, the problem of high false positive rate caused by the susceptibility of traditional single sensors to daily activities and emotional interference, as well as high underreporting rate of cardiovascular, respiratory and digestive tract abnormalities, is solved, thereby improving the detection rate of very early real risk events.

[0034] It should be specifically noted that the technical solution provided in this disclosure belongs to the field of computer data processing and IoT device automation control. During operation, this solution does not perform any form of disease diagnosis, nor does it include any medical interventions or treatments directly affecting the human body. Essentially, this system performs mathematical modeling, filtering, noise reduction, and logical comparison of electrical signal waveforms, image pixel features, and spatial coordinate data collected by heterogeneous sensors. The final output of tiered early warning commands serves only as an intermediate computer processing result, used to drive relevant electronic hardware devices to perform specific physical actions, such as triggering on-site audible and visual alarms, displaying pop-up prompts on medical personnel's dispatch terminals, pushing three-dimensional spatial coordinates to specific terminals, and opening the electronic locks of emergency supply cabinets. Any subsequent clinical decisions made by medical personnel are independent of this system.

[0035] In one possible implementation, before obtaining the original observation vectors from multiple sources, a step of establishing multimodal baseline data is also included: S101. Obtain the recipient's resting physiological baseline and visual baseline before vaccination.

[0036] In this embodiment of the disclosure, the resting physiological baseline and visual baseline can be understood as the basic physical characteristics and appearance standards of an individual in a calm state without external interference. In specific implementation, the average value of stable physiological data and normal facial features within one minute can be collected and stored by a device during the queuing, registration and health inquiry process before vaccination.

[0037] For example, the system collects data at the vaccination station and finds that an adult's resting baseline heart rate is 75 beats per minute and their complexion is not abnormally rosy. This data is recorded as the resting physiological baseline and visual baseline.

[0038] S102. After the recipient enters the observation area and is determined to be seated and stationary for a preset time threshold, the initial spatial coordinates are automatically locked and a positioning baseline is established.

[0039] In this embodiment of the disclosure, the positioning baseline is used to establish the initial reference anchor point of the recipient within the safe observation area. Specifically, the system uses an indoor positioning antenna to continuously track the rate of change of coordinates. When it is found that the coordinates fluctuate within a few seconds (e.g., 5 seconds) within a very small tolerance range, it is determined that the person has taken their seat and their current coordinates are automatically written into the database.

[0040] For example, the recipient walks to a seat in the observation area D and sits down. The system detects that the coordinates of this position remain stable for more than 5 seconds, and then locks the coordinates (X1, Y1) as the individual's positioning baseline.

[0041] S103. During the observation period, if the spatial body posture observation vector indicates that the recipient has moved without triggering abnormal posture characteristics, it is determined to be a physiological activity confounding factor, the positioning baseline is updated and no abnormal alarm signal is triggered.

[0042] In this embodiment, the system allows normal free movement. Specifically, if positioning data indicates a person has moved, but the system cannot visually detect a dangerous posture such as a fall or collapse, it interprets the movement as a physiological activity such as going to the restroom or walking. In this case, the positioning reference is only refreshed in the background, and an alarm is blocked.

[0043] For example, the recipient stands up normally from their seat and walks towards the water dispenser. Because their posture is stable, the system identifies this as a physiological activity confounding factor, automatically updates the positioning baseline to the side of the water dispenser, and remains silent throughout the process, without triggering any abnormal alarms.

[0044] According to the scheme of this disclosure, by introducing a dynamic updating positioning baseline and resting physiological visual baseline comparison mechanism, the problem that fixed baselines cannot adapt to the free movement of people in the observation area is solved, thereby achieving the effect of reducing false alarm rate and improving baseline comparison accuracy without restricting the normal activities of the recipient.

[0045] In one possible implementation, the multiple external confounding factors include at least stress and anxiety factors. Simultaneous extraction of the pre-defined multiple external confounding factors includes the following steps: S121. Based on physiological observation vectors, extract skin impedance, systolic blood pressure and heart rate features in real time.

[0046] In this embodiment, skin impedance refers to the change in the conductivity of the human epidermis, which is usually directly related to sympathetic nerve excitation and sweat gland secretion. Specifically, microcurrent feedback values ​​are acquired in real time through electrode sensors on the wearable device, and heart rate values ​​from the blood pressure monitor and photoplethysmography (PPG) sensor are simultaneously captured.

[0047] For example, the system collects data in real time and extracts the current skin impedance as 100 kΩ, systolic blood pressure as 135 mmHg, and heart rate as 105 beats / min.

[0048] S122. When a decreasing trend in skin impedance and an increasing trend in heart rate are detected, a stress response is determined, and the trend characteristics of systolic blood pressure are further extracted for reverse verification.

[0049] In this embodiment of the disclosure, the stress response is the autonomic nervous system response of the human body when faced with stimuli. Specifically, the system uses a time sliding window to compare the rate of change of features. After detecting a decrease in impedance and an increase in heart rate, it triggers reverse verification logic, that is, it retrieves the indicator of vasoconstriction (systolic blood pressure) as the basis for decision-making.

[0050] For example, during the observation period, if a recipient experiences excessive anxiety about adverse vaccine reactions (such as panic due to soreness at the injection site), the system detects a sudden 20% drop in skin resistance within one minute (due to cold sweats from anxiety), and a simultaneous increase in heart rate of 25 beats per minute. The system determines that a stress response has been triggered and immediately retrieves the systolic blood pressure data from the next second for verification.

[0051] S123. If systolic blood pressure is higher than the resting physiological baseline, it is determined to be emotional stress dominated by sympathetic excitation, and the quantitative value of the tension and anxiety factor is increased. If systolic blood pressure is lower than the baseline, it is determined to be a non-simple sympathetic hyperactivity state, and the quantitative value of the tension and anxiety factor is restricted.

[0052] In this embodiment of the disclosure, the actual risk is separated by the contradictory physiological manifestations. Anxiety during vaccination can trigger sympathetic nerve excitation, resulting in increased heart rate, rapid breathing, and elevated systolic blood pressure; if the anxiety stimulus continues, it can further induce the vasovagal reflex, leading to a decrease in blood pressure and heart rate, or even syncope.

[0053] For example, the system detects that the systolic blood pressure is 15 mmHg higher than the baseline upon entering the room. The system determines this to be typical psychogenic emotional stress caused by the initial stage of panic (sympathetic hyperactivity), rather than vasodilatory shock caused by allergies. It then increases the stress and anxiety factor value to 0.8 to avoid false alarms. Conversely, if blood pressure shows a significant downward trend, since hypotension (whether due to tissue hypoperfusion caused by shock or vagal syncope caused by persistent panic) is an abnormal sign requiring immediate medical attention, the system determines this to be a non-simple sympathetic hyperactivity state (including vasodilation due to shock, or syncope due to decreased blood pressure and heart rate secondary to strong vagal nerve excitation after sustained sympathetic tension). The stress and anxiety factor value is limited to a very low level of 0.1 to prevent genuine danger signals from being intercepted as "emotional noise," thus allowing the system to smoothly initiate allergy or syncope alarm commands.

[0054] According to the scheme of this disclosure embodiment, by verifying the inverse characteristics of skin impedance, heart rate and systolic blood pressure, the problem that panic attacks (psychogenic reactions) during the clinical observation period are easily confused with syncope and collapse symptoms and early shock symptoms is solved. This achieves the objective effect of accurately quantifying abstract emotional factors and greatly enhances the system's ability to resist interference from psychological factors.

[0055] In one possible implementation, validity is determined based on the coupled logic of the decontamination signal and the external contamination factor to obtain a validity label, including the following steps: S131. When there are abnormal feature signals in the physiological observation vector or visual observation vector, call the associated external confounding factor for verification.

[0056] In this embodiment, the associated call is a mapping verification mechanism. Specifically, the system pre-establishes a mapping table to clearly identify which abnormalities in a vital sign are easily affected by which external behaviors. When a vital sign goes out of bounds, the corresponding external factor state value is immediately read from memory.

[0057] In one example, when the system detects the abnormal signal of "heart rate abnormally rising to 120 beats / min", the system immediately calls the current status of the associated "exercise factor" and "stress and anxiety factor" for investigation according to the mapping rules.

[0058] S132. If and only if the quantization assignment of the associated external confounding factor does not reach the preset interference threshold, the validity flag of the abnormal feature signal is set to valid to obtain the validity flag.

[0059] In this embodiment of the disclosure, the interference threshold refers to the maximum upper limit of non-pathological noise that the system can tolerate. In specific implementation, the abnormal vital sign signal is only allowed after a comparison operation confirms that the external interference is extremely small.

[0060] For example, the system detects that the recipient's exercise factor is 0.1 and anxiety factor is 0.2, both far below the set interference threshold of 0.6. This means that the abnormal heart rate is not caused by running or stress, and the system immediately marks the abnormal heart rate signal as "valid".

[0061] S133. If the quantitative assignment of the tension and anxiety factor reaches the interference threshold, the validity marker of the abnormal characteristic signal triggered by the simultaneous increase in heart rate, respiratory rate and systolic blood pressure will be forcibly set to invalid, and it will be prohibited to input it into the calculation step of the multimodal weighted risk score.

[0062] In this embodiment, a forced interception mechanism is employed. Specifically, once the system determines that the emotional factor has exceeded the limit, it directly cuts off these emotion-driven false positive signs at the underlying data bus, preventing them from entering the complex scoring calculations at the upper level.

[0063] For example, the stress and anxiety factor was assigned a value as high as 0.9 (exceeding the threshold). Even though the recipient's heart rate was soaring and breathing was rapid, the system still forcibly marked these characteristics as "invalid" and recorded their scores as zero in the subsequent total risk score calculation.

[0064] According to the solution of this disclosure embodiment, by establishing a strong verification and forced interception mechanism based on interference threshold, the problem of massive vital sign data becoming disordered and thus overwhelming the real analysis model is solved. This achieves the objective effect of saving ineffective computing power and completely blocking the transmission of false positive signals.

[0065] In one possible implementation, when acquiring physical fall features from spatial posture observation vectors, a three-mode cross-fusion determination mechanism is employed, including the following steps: S111. Simultaneously acquire the triaxial acceleration weightlessness mutation features in the physiological observation vector, the vertical axis skeletal point drop features based on key point detection in the visual observation vector, and the coordinate mutation features output by the spatial positioning device.

[0066] In this embodiment of the disclosure, tri-mode cross-fusion refers to the simultaneous use of three sensors based on different physical principles to confirm the same event. Specifically, the rapid flip curve of the Z-axis acceleration is captured from the wrist-worn device, while the rapid falling rate of human skeletal feature points (such as shoulder and head nodes) in the vertical direction of the Y-axis is retrieved from the camera image processing module, and the high-frequency jump information of the three-dimensional coordinates emitted by the indoor positioning tag is extracted simultaneously.

[0067] For example, if the recipient suddenly collapses, the wristband transmits a waveform of weightlessness below 1G for a short period of time. The camera identifies that the skeletal structure of the recipient's head falls 1.2 meters in one second, and the positioning system measures that the recipient's Z-axis coordinates quickly reach the bottom.

[0068] S112. Output a confirmed physical fall marker if and only if the weightlessness mutation feature, the vertical axis skeletal point drop feature, and the coordinate mutation feature simultaneously meet the triggering conditions within the same time window.

[0069] In this embodiment, the time window refers to an extremely short time-aligned interval on the order of microseconds or milliseconds. Specifically, the system uses timestamp matching, requiring three distinct sensor signals to trigger an alarm simultaneously within the same second before outputting a fall conclusion.

[0070] For example, if the recipient's bracelet simply falls to the ground, triggering a sudden change in weightlessness and coordinates, but the camera detects that the human skeleton remains upright, the system rejects the fall assessment and does not output a physical fall marker because none of the three conditions are met simultaneously.

[0071] According to the solution of the present disclosure, by aligning the three-mode cross-axis of three-axis acceleration, visual key points and spatial coordinates, the problem that traditional single anti-fall devices (such as those relying solely on accelerometers) are prone to false alarms due to device detachment or bending over to pick up objects is solved, thereby achieving the objective effect of greatly improving the robustness and accuracy of physical fall event judgment.

[0072] In one possible implementation, multimodal weighted risk scoring is performed by calculating a weighted sum of preset base weights and corresponding scores for each dimension of features. The method also includes an adaptive weight enhancement step for weak precursor signals. S141. Set up a precursor clue trigger library that includes early warning features.

[0073] In this embodiment, the precursor cue trigger library is a set of extremely subtle signs that indicate impending organ dysfunction. Specifically, during system initialization, specific continuous small-scale change trends are written into a memory cache.

[0074] For example, the system's pre-set trigger library contains rules: if the pulse oxygen level shows a step-like decline of 0.5% each time within three minutes, although it does not fall below the traditional alarm line of 95%, it has been included as a dangerous precursor clue.

[0075] S142. When any early warning feature in the precursor cue trigger library is marked as valid as a valid decontamination signal, dynamic multiplication logic is triggered to dynamically increase the weight coefficient of the dimension containing the early warning feature above the preset base weight. The early warning feature includes at least: pulse oxygen saturation showing a continuous downward trend below the critical alarm threshold, or the confidence level of skin erythema diffusion in the visual observation vector reaching the preset early warning threshold.

[0076] In this embodiment of the disclosure, the dynamic multiplication logic refers to the system breaking the fixed linear weight allocation and increasing the calculation proportion of key risk features in real time. In specific implementation, once the pattern matching is successful, the system directly multiplies the preset basic weight of that dimension by a multiplication factor (such as 1.5 times or 2 times) and then performs a weighted sum operation.

[0077] In one example, an adult's blood oxygen saturation level gradually decreased from 99% to 95%. Although 95% is insufficient to trigger an alarm on a standard hospital pulse oximeter, the system detected a "continuous downward trend" and immediately increased the weight of the pulse oximetry dimension from 0.15 to 0.30. This significantly amplified the slight hypoxia trend in the final overall score.

[0078] According to the solution of this disclosure embodiment, by constructing a precursor clue trigger library and introducing an adaptive weight enhancement mechanism, the problem of serious lag in early warning when facing slowly deteriorating hidden diseases by traditional static scoring models is solved, thereby achieving the objective effect of locking in risks and amplifying subtle hidden dangers several minutes before the occurrence of substantial organ failure.

[0079] In one possible implementation, during the execution of multimodal weighted risk scoring, severe schizophrenia cross-validation and circuit breaker mechanisms are executed in parallel, including the following steps: S143. When a physical fall marker or postural imbalance feature is detected, the physiological observation vector or visual observation vector is forcibly invoked to perform cross-validation of tissue hypoperfusion features.

[0080] In this embodiment of the disclosure, collapse refers to pathological paralysis caused by circulatory failure due to severe allergies. Specifically, after the system detects a fall in terms of mechanical force, it immediately launches a parallel, highest-priority thread to specifically check the blood supply parameters of the cardiovascular system.

[0081] For example, the system has just confirmed a fall using tri-mode fusion. To distinguish between a trip and shock, the system forces the module to read current blood pressure and facial image data within 50 milliseconds.

[0082] S144. If the cross-validation conditions are met, such as a sudden drop in systolic blood pressure, a drop in pulse oxygen to the critical threshold, or visual characteristics of a pale, clammy, and cold face, then the condition is confirmed as an effective collapse state, triggering the circuit breaker mechanism.

[0083] In this embodiment, the circuit breaker mechanism refers to an extreme protection procedure that bypasses the standard procedure, similar to a fuse in a circuit. In specific implementation, the system compares cardiovascular or blood supply appearance indicators using a logical OR gate. If any one of them is critical, the circuit breaker state is directly activated.

[0084] For example, after detecting a fall, the system checks the blood pressure and finds that the systolic pressure has dropped to 80 mmHg (a sudden drop in systolic pressure). The system determines that the brain has lost perfusion, confirms that this is a valid collapse caused by shock, and immediately activates the circuit breaker.

[0085] S145. After the circuit breaker mechanism is triggered, the linear weighted calculation step of the multimodal weighted risk score is bypassed, the extreme value of the multimodal weighted risk score is assigned, and the highest level of graded early warning instruction is directly output.

[0086] In this embodiment of the disclosure, assigning an extreme value means ignoring the scores of all other normal dimensions and directly giving the system's highest full score. In specific implementation, the underlying scheduler directly skips matrix multiplication and accumulation operations, directly fills the comprehensive risk total score register (for example, writes it to 100 points), and broadcasts the most urgent red signal to the outside world.

[0087] For example, once the circuit breaker is activated, the system no longer slowly calculates whether the recipient has erythema or subjective perception score, but directly assigns a full score of 100 and issues a red warning command to call the rescue team within seconds.

[0088] According to the scheme of this disclosure embodiment, by combining cross-validation of physical falls and tissue hypoperfusion characteristics with a circuit breaker mechanism that skips levels, the problem that traditional weighted algorithms are easily diluted by a large number of normal indicators, resulting in low alarm levels for critical falls is solved. This achieves the objective effect of eliminating false alarms for general falls while providing a top-level response in seconds for lethal circulatory failure.

[0089] In one possible implementation, the extraction of visual observation vectors combines features from multiple systems and incorporates actively reported observation vectors for verification, specifically including the following steps: S113. The erythema diffusion features of the skin and mucous membrane dimension, the nausea and vomiting features of the digestive dimension, and the respiratory dyspnea postural features of the respiratory dimension are extracted in parallel through a neural network model.

[0090] In this embodiment, the neural network model can be a lightweight convolutional neural network, specifically trained to capture specific pathological image distributions. Specifically, image frames are simultaneously fed into multiple different classifier branches to identify skin color gradients, laryngospasm during swallowing, and rapid chest rise and fall postures, respectively.

[0091] For example, the images captured by the camera were analyzed by a neural network, which not only yielded the skin dimension conclusion that "the area of ​​erythema spread accounts for 15%", but also simultaneously analyzed the postural characteristics of breathing difficulties, such as "the torso is leaning forward and the hands are covering the chest and neck".

[0092] S114. When visual perception captures the itching and rubbing motion, and a sudden drop in skin impedance in the physiological observation vector is detected simultaneously, a high-risk confidence level is output.

[0093] In this embodiment of the disclosure, the joint verification of subjective and objective factors involves binding external visual behavior with internal physiological electrical signals for verification. Specifically, if the model determines that the recipient is continuously scratching their arms, neck, or other parts of their torso, and at the same time the microcurrent sensor reports a sharp decrease in impedance (due to tissue fluid leakage or extreme sympathetic hyperactivity), then a confidence probability close to full is directly output.

[0094] For example, if the camera detects that the recipient is repeatedly scratching their neck, the system will compare and find that the skin resistance at their wrist drops sharply from the normal level, and immediately output a high risk of severe skin system involvement with a confidence level of 0.95.

[0095] S115. Receive subjective symptom codes from the active reporting module. If the codes include symptoms such as a foreign body sensation in the throat or hoarseness, assign them an independent high weight in the multimodal weighted risk score.

[0096] In this embodiment, the active reporting module can be an interactive screen or a personal button device configured in the observation seat, used to acquire proprioceptive sensations that cannot be detected by cameras and sensors. Specifically, it parses the symptom codes input by the recipient via button presses; once a high-risk subjective term indicating airway obstruction is identified, its calculated weight matrix is ​​forcibly rewritten.

[0097] For example, the recipient clicked on the subjective symptom code "throat tightness or blockage" on the seat screen. The system recognized this as a precursor to acute laryngeal edema and applied a very high independent coefficient directly to the scoring item for this symptom, ensuring it was sufficient to affect the overall score.

[0098] According to the scheme of Embodiment 0 of this disclosure, by extracting pathological information from multiple visual dimensions and cross-verifying it with objective impedance and subjective reporting codes, the problem of easily missing critical airway symptoms in visual blind spots by a single perception dimension is solved, thereby achieving the effect of comprehensively covering diseases of the body surface and internal organs and significantly improving the overall detection sensitivity.

[0099] In one possible implementation, the specific logic of the dual-path determination rule includes: S146, First Path: When the confidence level of abnormal skin and mucous membrane features output by the visual observation vector is greater than or equal to the preset warning threshold, and there is any valid abnormal feature in the corresponding dimension of the respiratory, digestive or circulatory system, the first type of high-level warning instruction is triggered.

[0100] In this embodiment, the first path simulates the typical disease progression of most severe allergic reactions in medicine, which initially manifest as skin reactions. Specifically, the system employs a "serial gate" logic, requiring the detection of a definitive skin abnormality (e.g., exceeding the confidence threshold) before iterating through the other three major internal organ signs. If any one of these indicators triggers an alarm, the pathway is declared open.

[0101] For example, the system determines that the recipient not only has a trunk erythema spread confidence level of 0.8 (exceeding the standard), but also has severe abdominal pain signals in the digestive dimension. At this time, the system considers both internal organs and the body surface to be affected, and naturally triggers a high-level warning along the first path.

[0102] S147, Second Path: When the confidence level of abnormal skin and mucous membrane features is less than the preset warning threshold, and the physiological observation vector detects a surge in respiratory rate, and at the same time, an acute drop in systolic blood pressure or an acute drop in pulse oxygen reaches the corresponding critical threshold, the second type of high-level warning instruction is triggered.

[0103] In this embodiment, the second path is a fallback option specifically designed for a very small number of atypical cases that show no skin abnormalities but directly progress to organ failure. In practice, even if the skin assessment module determines that there are no abnormalities in appearance (with extremely low confidence), as long as the underlying core vital signs show signs of rapid deterioration, the system can bypass the skin review and file a separate case.

[0104] For example, the recipient's face and torso may be clean with no signs of rashes, but the system detects in the background that their respiratory rate suddenly spikes to 35 breaths per minute and their oxygen saturation drops rapidly below 95%. At this point, the system completely ignores the peaceful appearance of the skin and decisively triggers a cardiopulmonary emergency alarm along the second pathway.

[0105] According to the solution of this disclosure embodiment, by customizing a dual-mode parallel judgment and distribution path with and without skin lesions, the technical blind spot of serious missed diagnosis of a few atypical fulminant cardiopulmonary shocks due to over-reliance on skin morphology is solved, thereby achieving the objective effect of full phenotypic and no blind spots in the coverage of various sudden critical events during the observation period.

[0106] In one possible implementation, calculating the promiscuous offset includes the following steps: S124. The confounding offset is obtained by summing the products of the pre-calibrated weight coefficients of each category of confounding factors and the corresponding confounding factor characteristic offset functions.

[0107] In this embodiment, a mathematical modeling method for quantifying the superposition effect of multiple disturbances is disclosed. Each category of confounding factors has its own weight allocation, and the vital sign offset function refers to a nonlinear curve fitted based on a large amount of healthy sample data, representing the magnitude of numerical fluctuation caused by a certain disturbance. Specifically, the overall offset magnitude is calculated through vector dot product or algebraic summation.

[0108] For example, the system determines that both "shouting loudly" (weight 0.3, corresponding to a heart rate offset of 10) and "walking briskly" (weight 0.7, corresponding to a heart rate offset of 15) are present simultaneously. The system multiplies the weights by the function outputs and then adds them together: 0.3 × 10 + 0.7 × 15 = 13.5. The system concludes that the total heart rate spurious offset caused by the current environment is 13.5 beats / min.

[0109] According to the solution of this disclosure, by constructing a cumulative model of the weights of multiple mixed factors and the nonlinear offset function, the computing power problem of the inability to accurately estimate the degree of interference caused by the intertwining of multiple interferences in a complex observation environment is solved, thereby realizing the effect of transforming abstract interference actions into deterministic values ​​that can be directly deducted by the computer.

[0110] In one possible implementation, outputting tiered warning instructions includes the following steps: S148. Compare the overall risk score obtained from the multimodal weighted risk scoring with the preset segmented critical threshold, and output a yellow alert, orange warning, or red warning instruction.

[0111] In this embodiment, the segmented critical threshold is a stepped alarm scale calibrated using historical large sample data through ROC cutoff point testing. Specifically, the calculated continuous score is fed into a series of interval comparators; different score segments trigger different color-level electronic signals.

[0112] For example, the system presets a risk range of 40 to 70 points as medium risk. If a recipient's final calculated total risk score is 55 points, the system will compare the scores within the range and send an orange alert to the backend, indicating that immediate intervention is required.

[0113] S149. When a red warning command is output or the circuit breaker mechanism is triggered, the system automatically locks and pushes the recipient's spatial location coordinates to the monitoring terminal, and simultaneously generates a preset emergency resource allocation prompt.

[0114] In this embodiment, the data flow is a closed loop at the end, designed to drive physical medical response as quickly as possible. Specifically, when a red signal indicating extreme danger is generated, the system's underlying layer directly calls the hardware interface to package and send the recipient's latest UWB coordinates to a medical staff handheld tablet, remotely unlocking the ambulance's electronic lock or displaying a detailed list of emergency supplies on the screen.

[0115] For example, if the recipient triggers the circuit breaker mechanism, the system will not only sound an alarm, but also display a pop-up window on the nurse station screen: "Critical Alert! The recipient is in row 3 of section C. Please bring 1:1000 adrenaline and an oxygen cylinder immediately."

[0116] According to the solution of this disclosure embodiment, a graded instruction is issued by comparing the total risk score in a tiered manner, and the coordinate positioning is automatically linked with emergency resources at the most critical level. This solves the problem of the early warning system being disconnected from clinical treatment, which leads to excessive time spent searching for patients. Thus, a fully automatic second-level response closed loop effect is achieved from risk discovery to precise positioning and then to the allocation of intervention medical supplies and equipment.

[0117] For simple syncope events determined to be non-allergic, the system will also push an abnormality alert to the medical staff terminal to achieve comprehensive observation and safety protection.

[0118] Figure 2 This is a module architecture diagram of a monitoring and early warning system for the vaccination observation period provided according to embodiments of this disclosure. For example... Figure 2 As shown, the system includes: The data acquisition module 201 is used to continuously acquire multi-source raw observation vectors of the recipient during the observation period. The multi-source raw observation vectors include physiological observation vectors output by the vital signs acquisition device, visual observation vectors output by the visual acquisition device, and spatial posture observation vectors output by the spatial positioning device.

[0119] The edge computing module 202 includes a promiscuous correction unit and a validity determination gateway. It is used to simultaneously extract multiple preset external promiscuous factors, calculate the promiscuous offset, and perform causal intervention correction on the multi-source original observation vector to obtain a depromiscuous signal. Based on the coupling logic between the depromiscuous signal and the external promiscuous factors, it performs validity determination and outputs a validity label.

[0120] The comprehensive analysis module 203 is used to perform multimodal weighted risk scoring on decontamination signals marked as valid, and output graded early warning instructions in combination with preset dual-path judgment rules; The dual-path determination rule includes: a first path triggered when the visual observation vector extracts abnormal features of the skin and mucous membranes, and the physiological observation vector contains abnormal feature signals in the cardiovascular, respiratory, or gastrointestinal dimensions; and a second path triggered when there are no abnormal features of the skin and mucous membranes, but the physiological observation vector contains acute abnormal feature signals in the cardiovascular or respiratory dimensions.

[0121] In one possible implementation, the various logical modules (data acquisition module, edge computing module, and comprehensive analysis module) in the aforementioned monitoring and early warning system for the vaccination observation period can be implemented based on a specific physical hardware architecture. Combined with... Figure 3 The schematic diagram shown illustrates the application scenario of the vaccination observation period, and provides a detailed description of the system physical architecture provided in this embodiment.

[0122] like Figure 3 As shown, the monitoring and early warning system for the observation period after vaccination is mainly deployed in the physical space as follows: wearable multi-vital sign sensor terminal 30, panoramic visual monitoring unit 20, UWB positioning base station 40, active reporting module 50, edge computing engine 60, and backend medical terminal display screen 70, early warning prompt module 80 and blockchain network 90.

[0123] Specifically, the mapping relationship and workflow between each logical module of the system and the physical hardware are as follows: Physical implementation of the data acquisition module: The function of this module is jointly completed by the front-end sensing hardware distributed in the observation area.

[0124] Wearable multi-signal sensing terminal 30: worn on the wrist or other parts of the recipient 10, for continuous vital sign monitoring, real-time collection of physiological electrical and mechanical signals such as heart rate (HR), skin conductance (GSR) and triaxial acceleration (ACC) of the recipient 10, and transmitted to edge computing engine 60 through low-power wireless transmission.

[0125] Panoramic visual monitoring unit 20: Deployed at the top or around the observation area, it is used to continuously collect visual image streams (including facial, neck, upper limb and whole body features) of the recipient 10 and transmit them to the edge computing engine 60.

[0126] UWB positioning base station 40: Deployed in the observation area, it transmits high-precision positioning and tracking signals to obtain centimeter-level location coordinates and movement trajectory data of the recipient 10 in real time.

[0127] Active reporting module 50: This can be a mobile tablet or interactive panel configured next to the recipient 10, used to receive the recipient's own subjective symptom code trigger 51 (such as clicking the "throat tightness" button) and synchronously integrate the subjective signal into the data stream.

[0128] Physical implementation of the edge computing module and the comprehensive analysis module: The functions of these two core computing modules are mainly undertaken by the edge computing engine 60 (such as a local AI edge server or smart gateway) deployed locally in the observation area.

[0129] Edge computing engine 60 aggregates raw data from multiple sources from the aforementioned sensing devices and performs multimodal weighted fusion analysis and causal intervention decontamination algorithms locally.

[0130] By performing promiscuous correction, validity determination, and dual-path risk scoring on the local edge, the system does not need to upload a large amount of video streams and high-frequency physiological data to the cloud, greatly reducing network latency and ensuring a second-level response to critical shock events.

[0131] Tiered warning command output and physical linkage: When the edge computing engine 60 outputs a tiered warning command, it will directly drive the backend physical response devices, forming a complete emergency response closed loop. Medical terminal display screen 70: Receives and displays the graded early warning results and the real-time location trajectory map of the vaccine recipient to medical staff 100, so that medical staff can accurately locate the location of the vaccine recipient who has an abnormality.

[0132] Early warning module 80: Receives audible and visual alarm signals from edge computing engine 60 and triggers audible and visual alerts (such as red strobe lights and buzzers) at the corresponding alarm level on site.

[0133] Blockchain Network 90: Edge computing engine 60 will generate monitoring data, early warning logs and handling records, and use the SHA-256 algorithm to generate hash values ​​for data storage on the blockchain, ensuring that medical monitoring data is tamper-proof and fully traceable.

[0134] Figure 4 This is a schematic diagram of the structure of a monitoring and early warning device 400 for the inoculation observation period provided according to an embodiment of this disclosure. Figure 4 As shown, the device includes: The acquisition module 401 is used to continuously acquire multi-source raw observation vectors of the recipient during the observation period; wherein, the multi-source raw observation vectors include physiological observation vectors output by the vital signs acquisition device, visual observation vectors output by the visual acquisition device, and spatial posture observation vectors output by the spatial positioning device. The decontamination module 402 is used to simultaneously extract multiple preset external contamination factors, calculate the contamination offset, and perform causal intervention correction on the multi-source original observation vector to obtain the decontamination signal. The determination module 403 is used to determine the validity based on the coupled logic of the decontamination signal and the external contamination factor, and obtain the validity mark.

[0135] The output module 404 is used to perform multimodal weighted risk scoring on the decontamination signals marked as valid, and output graded warning instructions in combination with preset dual-path judgment rules. The dual-path judgment rules include: a first path triggered when the visual observation vector extracts abnormal skin and mucous membrane features, and the physiological observation vector contains abnormal feature signals in the cardiovascular, respiratory, or gastrointestinal dimensions; and a second path triggered when there are no abnormal skin and mucous membrane features, but the physiological observation vector contains acute abnormal feature signals in the cardiovascular or respiratory dimensions.

[0136] It should be noted that the specific execution principle of each module in the above system and device embodiments is completely consistent with the corresponding steps in the above method embodiments. Their conceptual analysis, specific examples and technical effects are the same as the aforementioned method content, and will not be repeated here.

[0137] Figure 5 This is a structural block diagram of an electronic device according to an embodiment of the present disclosure. Figure 5 As shown, the electronic device includes a memory 510 and a processor 520. The memory 510 stores a computer program that can run on the processor 520. The number of memories 510 and processors 520 can be one or more. The memory 510 can store one or more computer programs, which, when executed by the electronic device, cause the electronic device to perform the methods provided in the above-described method embodiments. The electronic device may also include a communication interface 530 for communicating with external devices and performing data exchange and transmission.

[0138] If the memory 510, processor 520, and communication interface 530 are implemented independently, they can be interconnected via a bus to communicate with each other. This bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. This bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 5The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0139] Optionally, in a specific implementation, if the memory 510, processor 520, and communication interface 530 are integrated on a single chip, then the memory 510, processor 520, and communication interface 530 can communicate with each other through an internal interface.

[0140] It should be understood that the aforementioned processor can be a Central Processing Unit (CPU), or other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. General-purpose processors can be microprocessors or any conventional processor. It is worth noting that the processor can be a processor supporting Advanced Reduced Instruction Set Machines (ARM) architecture.

[0141] Further, optionally, the aforementioned memory may include read-only memory and random access memory, and may also include non-volatile random access memory. The memory may be volatile or non-volatile, or may include both. Non-volatile memory may include read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory may include random access memory (RAM), which serves as an external cache. Many forms of RAM are available by way of example, but not limitation. Examples include Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDR SDRAM), Enhanced Synchronous DRAM (ESDRAM), Synchlink DRAM (SLDRAM), and Direct RAMBUS RAM (DR RAM).

[0142] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions. When the computer instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this disclosure are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another via wired (e.g., coaxial cable, fiber optic, Digital Subscriber Line, DSL) or wireless (e.g., infrared, Bluetooth, microwave, etc.) means. The computer-readable storage medium can be any available medium accessible to a computer, or a data storage device such as a server or data center that integrates one or more available media. The available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., Digital Versatile Discs (DVDs)), or semiconductor media (e.g., Solid State Disks (SSDs)). It is worth noting that the computer-readable storage media mentioned in this disclosure can be non-volatile storage media; in other words, it can be non-transient storage media.

[0143] Those skilled in the art will understand that all or part of the steps of the above embodiments can be implemented by hardware or by a program instructing related hardware. The program can be stored in a computer-readable storage medium, such as a read-only memory, a disk, or an optical disk.

[0144] In the description of the embodiments of this disclosure, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this disclosure. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of those different embodiments or examples.

[0145] In the description of the embodiments disclosed herein, unless otherwise stated, " / " means "or". For example, A / B can mean A or B. The "and / or" in this document is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone.

[0146] In the description of embodiments of this disclosure, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of embodiments of this disclosure, unless otherwise stated, "a plurality of" means two or more.

[0147] The above description is merely an exemplary embodiment of this disclosure and is not intended to limit this disclosure. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this disclosure should be included within the protection scope of this disclosure.

Claims

1. A monitoring and early warning method for the observation period after vaccination, characterized in that, Includes the following steps: Continuously acquire multi-source raw observation vectors of the recipient during the observation period; wherein, the multi-source raw observation vectors include physiological observation vectors output by the vital signs acquisition device, visual observation vectors output by the visual acquisition device, and spatial posture observation vectors output by the spatial positioning device; Simultaneously extract multiple preset external confounding factors, calculate the confounding offset, and perform causal intervention correction on the multi-source original observation vector to obtain the deconfounded signal; The validity is determined based on the coupling logic between the decontamination signal and the external contamination factor, and a validity label is obtained. For decontamination signals marked as valid, perform multimodal weighted risk scoring and, in conjunction with preset dual-path judgment rules, output graded early warning instructions; The dual-path determination rule includes: a first path triggered when the visual observation vector extracts abnormal skin and mucous membrane features and the physiological observation vector contains abnormal feature signals in the cardiovascular, respiratory, or gastrointestinal dimensions; and a second path triggered when there are no abnormal skin and mucous membrane features and the physiological observation vector contains acute abnormal feature signals in the cardiovascular or respiratory dimensions. Prior to obtaining the multi-source raw observation vectors, the method also includes the step of establishing multimodal reference data: Obtain the recipient's resting physiological and visual baselines before vaccination; After the recipient enters the observation area and is determined to be seated and stationary for a preset time threshold, the initial spatial coordinates are automatically locked and a positioning baseline is established. During the observation period, if the spatial body posture observation vector indicates that the recipient has moved without triggering abnormal posture features, it is determined to be a physiological activity confounding factor, the positioning baseline is updated and no abnormal alarm signal is triggered. The multiple external confounding factors include at least stress and anxiety factors; the simultaneous extraction of preset multiple external confounding factors includes: Based on the physiological observation vector, skin impedance, systolic blood pressure and heart rate features are extracted in real time. When a decreasing trend in skin impedance and an increasing trend in heart rate are detected, a stress response is determined, and the trend characteristics of systolic blood pressure are further extracted for reverse verification. If the systolic blood pressure is higher than the resting physiological baseline, it is determined to be an emotional stress dominated by sympathetic excitation, and the quantitative value of the tension and anxiety factor is increased; if the systolic blood pressure is lower than the baseline, it is determined to be a non-simple sympathetic hyperactivity state and non-emotional stress, and the quantitative value of the tension and anxiety factor is limited.

2. The method according to claim 1, characterized in that, The validity determination based on the coupling logic between the decontamination signal and the external contamination factor, to obtain a validity marker, includes: When there are abnormal feature signals in the physiological observation vector or the visual observation vector, the associated external confounding factor is invoked for verification. The validity flag of the abnormal feature signal is set to valid only if the quantization assignment of the associated external confounding factor does not reach the preset interference threshold, so as to obtain the validity flag. If the quantitative value of the tension and anxiety factor reaches the interference threshold, the validity marker of the abnormal characteristic signal triggered by the simultaneous increase in heart rate, respiratory rate and systolic blood pressure will be forcibly set to invalid, and it will be prohibited from being input into the calculation step of the multimodal weighted risk score.

3. The method according to claim 1, characterized in that, When acquiring the physical fall features from the spatial posture observation vector, a three-mode cross-fusion determination mechanism is adopted, including: Simultaneously acquire the triaxial acceleration weightlessness mutation features in the physiological observation vector, the vertical axis skeletal point drop features based on key point detection in the visual observation vector, and the coordinate mutation features output by the spatial positioning device; A confirmed physical fall marker is output only if the weightlessness mutation feature, the vertical axis skeletal point drop feature, and the coordinate mutation feature simultaneously meet the triggering conditions within the same time window.

4. The method according to claim 1, characterized in that, The multimodal weighted risk scoring is calculated by summing the preset basic weights of each dimension feature with their corresponding scores.

5. The method according to claim 3, characterized in that, During the execution of the multimodal weighted risk scoring, a severe schizophrenia cross-validation and circuit breaker mechanism are executed in parallel, including: When the physical fall marker or postural imbalance feature is detected, the physiological observation vector or the visual observation vector is forcibly invoked to perform cross-validation of tissue hypoperfusion features; If the cross-validation conditions are met, such as a sudden drop in systolic blood pressure, a drop in pulse oxygen to the critical threshold, or visual characteristics of a pale, clammy, and cold face, then the condition is confirmed as an effective collapse state, triggering the circuit breaker mechanism. Once the circuit breaker mechanism is triggered, the linear weighted calculation step of the multimodal weighted risk score is bypassed, the extreme value of the multimodal weighted risk score is assigned, and the highest level of graded early warning instruction is directly output.

6. The method according to claim 1, characterized in that, The extraction of the visual observation vector combines features from multiple systems and incorporates actively reported observation vectors for verification: The neural network model was used to extract erythema diffusion features in the skin and mucous membrane dimension, nausea and vomiting features in the digestive dimension, and respiratory dyspnea postural features in the respiratory dimension in parallel. When the visual system captures the itching and rubbing motion, and simultaneously detects a sudden drop in skin impedance in the physiological observation vector, a high-risk confidence level is output. Receive subjective symptom codes from the active reporting module. If the codes include symptoms such as a foreign body sensation in the throat or hoarseness, assign them an independent high weight in the multimodal weighted risk score.

7. The method according to claim 1, characterized in that, The specific logic of the dual-path determination rule includes: First path: When the confidence level of the abnormal skin and mucous membrane features output by the visual observation vector is greater than or equal to the preset warning threshold, and there is any valid abnormal feature in the corresponding dimension of the respiratory, digestive or circulatory system, it is determined that the first type of high-level warning instruction is triggered. Second path: When the confidence level of the abnormal skin and mucous membrane features is less than the preset warning threshold, and the physiological observation vector detects a surge in respiratory rate, an acute drop in systolic blood pressure, or an acute drop in pulse oxygen saturation reaching the corresponding critical threshold, a second type of high-level warning instruction is triggered.

8. The method according to claim 1, characterized in that, The calculation of the hybrid offset includes: The confounding offset is obtained by summing the products of the pre-calibrated weight coefficients of each confounding factor and the corresponding confounding factor characteristic offset functions.

9. The method according to claim 1, characterized in that, The output hierarchical early warning instructions include: The overall risk score obtained from the multimodal weighted risk scoring is compared with the preset segmented critical threshold, and a yellow warning, orange warning, or red warning instruction is output. When a red warning command is issued or the circuit breaker mechanism is triggered, the system automatically locks and pushes the spatial location coordinates of the recipient to the monitoring terminal, and simultaneously generates a preset emergency resource allocation prompt.

10. A monitoring and early warning system for the observation period after vaccination, characterized in that, include: The data acquisition module is used to continuously acquire multi-source raw observation vectors of the recipient during the observation period. The multi-source raw observation vectors include physiological observation vectors output by the vital sign acquisition device, visual observation vectors output by the visual acquisition device, and spatial posture observation vectors output by the spatial positioning device. The edge computing module includes a promiscuous correction unit and a validity determination gateway. It is used to simultaneously extract multiple preset external promiscuous factors, calculate the promiscuous offset, and perform causal intervention correction on the multi-source original observation vector to obtain a depromiscuous signal. Based on the coupling logic between the depromiscuous signal and the external promiscuous factors, it performs validity determination and outputs a validity label. The comprehensive analysis module is used to perform multimodal weighted risk scoring on decontamination signals marked as valid, and output graded early warning instructions in combination with preset dual-path judgment rules; The dual-path determination rule includes: a first path triggered when the visual observation vector extracts abnormal skin and mucous membrane features and the physiological observation vector contains abnormal feature signals in the cardiovascular, respiratory, or gastrointestinal dimensions; and a second path triggered when there are no abnormal skin and mucous membrane features and the physiological observation vector contains acute abnormal feature signals in the cardiovascular or respiratory dimensions. Prior to obtaining the multi-source raw observation vectors, the method also includes the step of establishing multimodal reference data: Obtain the recipient's resting physiological and visual baselines before vaccination; After the recipient enters the observation area and is determined to be seated and stationary for a preset time threshold, the initial spatial coordinates are automatically locked and a positioning baseline is established. During the observation period, if the spatial body posture observation vector indicates that the recipient has moved without triggering abnormal posture features, it is determined to be a physiological activity confounding factor, the positioning baseline is updated and no abnormal alarm signal is triggered. The multiple external confounding factors include at least stress and anxiety factors; the simultaneous extraction of preset multiple external confounding factors includes: Based on the physiological observation vector, skin impedance, systolic blood pressure and heart rate features are extracted in real time. When a decreasing trend in skin impedance and an increasing trend in heart rate are detected, a stress response is determined, and the trend characteristics of systolic blood pressure are further extracted for reverse verification. If the systolic blood pressure is higher than the resting physiological baseline, it is determined to be an emotional stress dominated by sympathetic excitation, and the quantitative value of the tension and anxiety factor is increased; if the systolic blood pressure is lower than the baseline, it is determined to be a non-simple sympathetic hyperactivity state and non-emotional stress, and the quantitative value of the tension and anxiety factor is limited.

11. A monitoring and early warning device for the observation period after vaccination, characterized in that, include: The acquisition module is used to continuously acquire multi-source raw observation vectors of the recipient during the observation period; wherein, the multi-source raw observation vectors include physiological observation vectors output by the vital signs acquisition device, visual observation vectors output by the visual acquisition device, and spatial posture observation vectors output by the spatial positioning device. The decontamination module is used to simultaneously extract multiple preset external contamination factors, calculate the contamination offset, and perform causal intervention correction on the multi-source original observation vector to obtain the decontamination signal. The determination module is used to determine the validity based on the coupling logic between the decontamination signal and the external contamination factor, and obtain a validity label; The output module is used to perform multimodal weighted risk scoring on decontamination signals marked as valid, and output graded early warning instructions in combination with preset dual-path judgment rules; The dual-path determination rule includes: a first path triggered when the visual observation vector extracts abnormal skin and mucous membrane features and the physiological observation vector contains abnormal feature signals in the cardiovascular, respiratory, or gastrointestinal dimensions; and a second path triggered when there are no abnormal skin and mucous membrane features and the physiological observation vector contains acute abnormal feature signals in the cardiovascular or respiratory dimensions. Prior to obtaining the multi-source raw observation vectors, the method also includes the step of establishing multimodal reference data: Obtain the recipient's resting physiological and visual baselines before vaccination; After the recipient enters the observation area and is determined to be seated and stationary for a preset time threshold, the initial spatial coordinates are automatically locked and a positioning baseline is established. During the observation period, if the spatial body posture observation vector indicates that the recipient has moved without triggering abnormal posture features, it is determined to be a physiological activity confounding factor, the positioning baseline is updated and no abnormal alarm signal is triggered. The multiple external confounding factors include at least stress and anxiety factors; the simultaneous extraction of preset multiple external confounding factors includes: Based on the physiological observation vector, skin impedance, systolic blood pressure and heart rate features are extracted in real time. When a decreasing trend in skin impedance and an increasing trend in heart rate are detected, a stress response is determined, and the trend characteristics of systolic blood pressure are further extracted for reverse verification. If the systolic blood pressure is higher than the resting physiological baseline, it is determined to be an emotional stress dominated by sympathetic excitation, and the quantitative value of the tension and anxiety factor is increased; if the systolic blood pressure is lower than the baseline, it is determined to be a non-simple sympathetic hyperactivity state and non-emotional stress, and the quantitative value of the tension and anxiety factor is limited.

12. An electronic device, comprising: At least one processor; as well as A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-9.

13. A non-transitory computer-readable storage medium storing computer instructions, wherein, The computer instructions are used to cause the computer to perform the method according to any one of claims 1-9.

14. A computer program product comprising a computer program that, when executed by a processor, implements the method according to any one of claims 1-9.

Images