A training emergency state disposal method, system and device based on data perception and linkage for help and medium

Abstract

The application discloses a training emergency state disposal method, system and device based on data sensing and linkage help-seeking and a medium, and relates to the technical field of data processing. The method comprises the following steps: acquiring bone posture data and real-time physiological parameters of a trainee; constructing an individualized posture and physiological baseline model based on historical training data; determining posture and physiological risk levels according to the deviation of real-time data and the baseline; comprehensively evaluating the comprehensive risk level; triggering a disposal process when the emergency threshold is reached, including projecting a spatial warning signal to the trainee and pushing real-time state and auxiliary decision information to the trainer. The application realizes individualized dynamic risk assessment and linkage help-seeking, and improves the accuracy and timeliness of training safety monitoring.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a training method, system, device, and medium for emergency response based on data perception and coordinated distress calls. Background Technology

[0002] In the field of professional training, trainees are highly susceptible to sudden changes in bodily functions, distorted movements, overexertion injuries, and even sudden fainting during high-intensity, high-load training. Traditional training supervision relies heavily on the coach's manual observation and experience-based judgment, which has limitations in covering multiple trainees across the entire training session.

[0003] Current solutions use wearable devices to monitor single indicators such as heart rate and issue alerts based on fixed thresholds. However, this fixed threshold approach cannot accommodate individual differences among participants. For example, a healthy participant with a resting heart rate of 50 beats per minute and a participant with a resting heart rate of 80 beats per minute will have drastically different risk levels when their heart rate reaches 140 beats per minute, often leading to high false alarm rates and a significant risk of missed alarms. Therefore, how to construct an emergency response mechanism that can perceive individual differences and provide real-time two-way alerts and decision support has become a pressing technical problem to be solved in this field. Summary of the Invention

[0004] To address the problem that existing training monitoring methods using fixed thresholds cannot adapt to individual differences among trainees, leading to high false alarm rates and significant missed alarm risks, this invention provides a training emergency response method, system, device, and medium based on data perception and coordinated emergency calls. This system can perform risk assessment based on a trainee's personalized baseline model, combined with skeletal posture and physiological parameters, and trigger coordinated response processes including spatial warnings and auxiliary decision-making pushes according to different risk levels, thereby improving the accuracy of training safety monitoring. The specific technical solution is as follows: In a first aspect, the present invention provides a training method for emergency response based on data perception and coordinated distress calls, comprising: Acquire sensory data of trainees during training, including skeletal posture data obtained by extracting skeletal posture from the video stream of the trainees' training and real-time physiological parameters from the trainees' terminals. Obtain the trainee's historical training data, and construct the trainee's personalized posture baseline model and personalized physiological baseline model based on the historical training data; The posture risk level is determined based on the degree of deviation between the skeletal posture data and the personalized posture baseline model; the physiological risk level is determined based on the degree of deviation between the real-time physiological parameters and the personalized physiological baseline model. A comprehensive risk level is obtained by comprehensively assessing the posture risk level and the physiological risk level. When the overall risk level reaches the preset emergency alarm threshold, the emergency response process is triggered. The emergency response process includes: projecting a perceptible spatial warning signal to the trainee and pushing the trainee's real-time status data and auxiliary decision-making information based on health records to the coach terminal.

[0005] Preferably, determining the posture risk level based on the degree of deviation between the skeletal posture data and the personalized posture baseline model includes: Multiple posture feature values ​​are extracted from the skeletal posture data, and each posture feature value corresponds to the movement state of different body parts of the trainee. Each of the posture feature values ​​is compared with the corresponding standard posture feature range in the personalized posture baseline model to determine whether each of the posture feature values ​​is within the standard posture feature range. For each of the posture feature values ​​that exceeds the standard posture feature range, the excess range is calculated, and the individual risk level corresponding to each posture feature is determined based on the excess range. Input the individual risk level corresponding to all the posture features into a preset posture risk assessment model to obtain the posture risk level; The attitude risk determination model is a multi-input single-output state mapping model, in which a specific combination of multiple individual risk levels corresponds to a predefined attitude risk level.

[0006] Preferably, determining the physiological risk level based on the degree of deviation between the real-time physiological parameters and the personalized physiological baseline model includes: Each of the real-time physiological parameters is compared with the corresponding preset safety threshold in the personalized physiological baseline model; For any of the real-time physiological parameters, when the real-time physiological parameter exceeds the preset safety threshold, the risk increment level corresponding to each physiological parameter is determined according to the magnitude of the exceedance; wherein, the greater the magnitude of the exceedance, the higher the risk increment level. The physiological risk level of the trainee is determined by combining the risk increment levels corresponding to each physiological parameter. The preset safety threshold in the personalized physiological baseline model is determined based on the statistical distribution of the trainee's historical physiological parameters.

[0007] Preferably, the step of comprehensively assessing the overall risk level based on the posture risk level and the physiological risk level includes: The posture risk level and the physiological risk level are input into a preset risk assessment model to obtain the trainee's current comprehensive risk level. The risk assessment model includes multiple state nodes that are defined by a combination of posture risk level and physiological risk level, and each state node is associated with a predefined comprehensive risk level.

[0008] Preferably, the risk assessment model is a risk assessment matrix; The posture risk level and the physiological risk level are input into a preset risk assessment model to obtain the trainee's current comprehensive risk level, which specifically includes: Determine the first index corresponding to the posture risk level in the risk assessment matrix, and the second index corresponding to the physiological risk level in the risk assessment matrix; The risk assessment matrix is ​​queried using the first index and the second index to obtain the pre-stored comprehensive risk level; The risk assessment matrix is ​​a two-dimensional matrix, with row indices corresponding to different posture risk levels and column indices corresponding to different physiological risk levels. The matrix cells pre-store the corresponding comprehensive risk levels.

[0009] Preferably, an emergency response method for training based on data perception and coordinated distress calls further includes: Obtain the training node label currently located by the trainee; the training node label is used to indicate the type of the current training node. Based on the training node labels, a first fusion weight corresponding to the posture risk level and a second fusion weight corresponding to the physiological risk level are determined; wherein, each posture risk level is preset to correspond to a first risk score, and each physiological risk level is preset to correspond to a second risk score. The step of comprehensively assessing the overall risk level based on the posture risk level and the physiological risk level includes: performing weighted fusion based on the first risk score corresponding to the posture risk level, the second risk score corresponding to the physiological risk level, the first fusion weight, and the second fusion weight to obtain the overall risk score, and determining the overall risk level based on the overall risk score; Different training node labels correspond to different first fusion weights and second fusion weights.

[0010] Preferably, an emergency response method for training based on data perception and coordinated distress calls further includes: When the overall risk level reaches the preset warning threshold but does not reach the emergency alarm threshold, the first-level warning process is triggered. The Level 1 early warning process includes: Send a first alarm command to the student terminal, instructing the student terminal to perform tactile vibration in a first vibration mode and play a first voice prompt; A second alarm command is sent to the coach terminal, instructing the coach terminal to perform tactile vibration in a second vibration mode and play a second voice prompt.

[0011] Secondly, the present invention also provides a training emergency response system based on data perception and linkage distress call, including a management platform, camera equipment, projection equipment, trainee terminal and instructor terminal; The management platform is used for: The video stream of the trainees during training is acquired through the camera device, and the real-time physiological parameters of the trainees are acquired through the trainee terminal. Obtain the trainee's historical training data, and construct the trainee's personalized posture baseline model and personalized physiological baseline model based on the historical training data; Skeletal pose data is extracted from the video stream; the pose risk level is determined based on the degree of deviation between the skeletal pose data and the personalized pose baseline model; the physiological risk level is determined based on the degree of deviation between the real-time physiological parameters and the personalized physiological baseline model. A comprehensive risk level is obtained by comprehensively assessing the posture risk level and the physiological risk level. When the overall risk level reaches the preset emergency alarm threshold, the emergency response procedure is triggered. The projection device is used to project perceptible spatial warning signals to the trainees under the control of the management platform. The coach terminal is used to receive and display the real-time status data of the trainees and auxiliary decision-making information based on health records pushed by the management platform.

[0012] Thirdly, the present invention also provides a computer device, comprising: Memory, used to store programs; A processor for loading the program to execute the aforementioned methods.

[0013] Fourthly, the present invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the aforementioned method.

[0014] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention constructs a baseline model using individualized historical data to achieve personalized risk measurement; it integrates skeletal posture and multi-source physiological parameters for comprehensive analysis, overcoming the uncertainty of single indicators; and it provides a multimodal linkage emergency call mechanism involving spatial warnings, terminal vibration / voice, and coach-side decision pushes, ensuring that trainees and coaches can perceive and respond instantly, accurately, and effectively in emergency situations. In particular, the introduction of dynamic weights for training nodes further enhances the system's scenario adaptability and the accuracy of risk characterization. Attached Figure Description

[0015] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn to scale.

[0016] Figure 1 A flowchart illustrating an emergency response method for training based on data perception and coordinated distress calls, provided as an embodiment of this application; Figure 2 A schematic diagram of a process for calculating physiological risk level provided in an embodiment of this application; Figure 3 A schematic diagram of a process for calculating attitude risk level provided in an embodiment of this application; Figure 4 A flowchart illustrating another training emergency response method provided in this application embodiment; Figure 5 A schematic diagram of the composition structure of the training emergency response system provided in the embodiments of this application; Figure 6 This is a schematic diagram illustrating the effect of spatial warning signal projection provided in an embodiment of this application; Figure 7 A schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0017] 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 only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] It should be understood that, when used in this specification, terms include and encompass the presence of the described feature, integral, step, operation, element and / or component, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0019] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms "one," "an," and "this" are intended to include the plural forms unless the context clearly indicates otherwise.

[0020] It should also be further understood that the terms used in this specification refer to and include any combination of one or more of the associated listed items, as well as all possible combinations.

[0021] In the textual descriptions or drawings of the embodiments of this application shown below, the terms "first," "second," "third," and various numerical designations are merely for descriptive convenience and are not intended to limit the scope of the embodiments of this application. For example, "first voice prompt" and "second voice prompt" are both voice prompts, distinguished only by the prompt information for different terminals. "First fusion weight" and "second fusion weight" are both weight coefficients, set separately for different data sources.

[0022] The technical solution of this application will be described in detail below with reference to the accompanying drawings.

[0023] Figure 1 This document illustrates a flowchart of an emergency response training method based on data perception and coordinated distress calls, according to an embodiment of this application. This method can be executed by a management platform or a computer device integrated into the management platform. The following is in conjunction with… Figure 1 The steps are explained in detail.

[0024] Step 101: Obtain the trainees' perception data during training.

[0025] The management platform collects real-time video streams of trainees using cameras deployed in the training area, and extracts skeletal posture data frame by frame. The extraction of skeletal posture data employs human keypoint detection algorithms based on deep neural networks. For example, it detects the coordinates of key points such as the head, neck, left and right shoulders, left and right elbows, left and right wrists, left and right hips, left and right knees, and left and right ankles from each frame, and calculates posture feature values ​​such as the angles, angular velocities, and limb swing amplitudes of each joint based on continuous inter-frame changes. This skeletal posture data constitutes a time series representing the trainees' body movement patterns.

[0026] Simultaneously, the management platform wirelessly communicates with the trainees' worn terminals to receive their real-time physiological parameters. The trainee terminals can be smart bracelets, chest straps, or sensor modules embedded in training uniforms, integrating heart rate sensors, blood oxygen sensors, body temperature sensors, and skin conductivity sensors. Real-time physiological parameters include, but are not limited to: real-time heart rate, heart rate variability, blood oxygen saturation, respiratory rate, body surface temperature, and acceleration. These parameters are reported to the management platform at a preset collection frequency (e.g., once per second) and are timestamped and identified by the trainee.

[0027] Step 102: Obtain the trainee's historical training data, and construct the trainee's personalized posture baseline model and personalized physiological baseline model based on the historical training data.

[0028] The management platform pre-stores each trainee's historical training data. This historical training data consists of skeletal posture data and physiological parameter data recorded during multiple normal and safe training sessions. The process of building a personalized baseline model includes data cleaning, segmentation, standardization, and statistical feature extraction.

[0029] For the personalized physiological baseline model, for each physiological parameter, such as heart rate, the management platform extracts multiple historical time series data for the trainee under the same or similar training types and environmental conditions. After removing anomalous jumps, the statistical distribution of the parameters for the trainee during steady-state exercise and different load phases is calculated, such as the mean μ and standard deviation σ under a normal distribution. Based on this statistical distribution, a personalized safety threshold range is determined. For example, the upper limit of safe heart rate can be set to μ+2.5σ, and the lower limit to μ-2σ. For blood oxygen saturation, an absolute lower limit is usually set, such as 94%, but it can also be adjusted based on the individual's historical lowest value. The safety thresholds are written into the trainee's personalized physiological baseline model. The model also includes correlation rules between parameters, such as the reasonable ratio range of heart rate to exercise acceleration.

[0030] For the personalized posture baseline model, the management platform extracts the trainee's standard movement sequences under different training subjects from historical video streams. After posture feature extraction, standard posture feature ranges are formed for each feature. Posture features include: trunk tilt angle, knee flexion angle, elbow opening and closing angle, head displacement amplitude, and the sway radius of the center of gravity projection trajectory. For each feature, its mean and tolerance range are calculated in historical normal training data, usually using a confidence interval (such as a 95% confidence interval) as the standard posture feature range. For example, if trainee A's trunk-to-ground angle in the crawling subject has a historically reasonable range of 5°~15°, then this range is used as the standard range for that feature. The ranges of all features together constitute the personalized posture baseline model.

[0031] Through the above steps, a baseline model is established for each trainee, dynamically evolving with their physical fitness and skills. This baseline model can be updated periodically or incrementally after each training session to maintain adaptability to the trainee's latest condition.

[0032] Step 103: Determine the posture risk level based on the degree of deviation between the skeletal posture data and the personalized posture baseline model; determine the physiological risk level based on the degree of deviation between the real-time physiological parameters and the personalized physiological baseline model.

[0033] Figure 2 A detailed process for determining the physiological risk level corresponds to the physiological risk assessment section in step 103. Specifically, this includes: Step 201: Compare each real-time physiological parameter with the corresponding preset safety thresholds in the personalized physiological baseline model. Assume the currently acquired real-time physiological parameters include heart rate, blood oxygen saturation, and respiratory rate. The personalized physiological baseline model stores the trainee's personalized upper and lower safety limits for heart rate, lower safety limit for blood oxygen saturation, and safe range for respiratory rate, etc.

[0034] Step 202: For any real-time physiological parameter, when it exceeds the preset safety threshold, determine the risk increment level based on the magnitude of the exceedance.

[0035] Exceeding the safe upper limit can be measured by the percentage or absolute difference. For example, heart rate exceedances can be categorized as follows: 0%–10% above the upper limit is considered mild exceedance, corresponding to risk increment level 1; 10%–25% is moderate exceedance, corresponding to risk increment level 2; and above 25% is severe exceedance, corresponding to risk increment level 3. The same mapping applies to exceedances below the lower limit. For blood oxygen saturation, the absolute value below the safe lower limit can be categorized, for example, 1%–3% below the lower limit is level 1, 3%–6% is level 2, and above 6% is level 3. The same applies to respiratory rate.

[0036] If the parameter is within a safe range, the risk increment level is 0.

[0037] Step 203: Determine the physiological risk level of the trainee by combining the risk increment levels corresponding to each physiological parameter.

[0038] The comprehensive approach can employ methods such as the highest-level method, weighted summation, or predefined rule-based methods. For example, one rule might be: if any parameter's risk increment level reaches 3, the physiological risk level is directly determined as high risk; if multiple levels are 2 and no level 3 exists, it is determined as medium risk; if only one level is 1, it is determined as low risk; the rest are considered normal. Alternatively, each risk increment level can be assigned a value and summed, then mapped back to the physiological risk level. This yields the physiological risk level at that sampling time.

[0039] Figure 3A detailed process for determining the attitude risk level corresponds to the attitude risk assessment section in step 103. This includes: Step 301: Extract multiple posture feature values ​​from the skeletal posture data. The management platform calculates a predefined set of posture feature values ​​from the acquired human keypoint time series. Examples include trunk tilt angle (the angle between the trunk centerline and the vertical line), knee flexion angle, vertical displacement of the head's center of gravity, the sway radius of the center of gravity projected onto the horizontal plane (the standard deviation of the center of gravity coordinates over N consecutive frames), and the change in tilt angle of the line connecting the left and right shoulders. Each feature value reflects the movement state of a specific body part of the trainee. The selected features are typically strongly correlated with common dangerous posture patterns encountered during training (such as joint hyperextension, excessive trunk twisting, and loss of balance).

[0040] Step 302: Compare each posture feature value with the corresponding standard posture feature range in the personalized posture baseline model.

[0041] For example, if trainee A's torso tilt angle is within the standard range of 5° to 15°, and the current value is 18°, then it exceeds the standard range.

[0042] Step 303: For posture feature values ​​that exceed the range, calculate the extent of the exceedance and determine the corresponding individual risk level.

[0043] The extent of the deviation can be defined as the absolute angle exceeding the upper limit or the percentage exceeding the lower limit. Individual risk levels can be similarly divided into 0 (normal), 1 (minor abnormality), 2 (significant abnormality), and 3 (serious abnormality). For example, a trunk tilt exceeding the upper limit by 3° corresponds to level 1, and exceeding it by 10° corresponds to level 3.

[0044] Step 304: Input the individual risk level corresponding to all posture features into the preset posture risk assessment model.

[0045] The posture risk assessment model is a multi-input, single-output state mapping model, which can be an expert rule set or a decision tree. For example, the rules could be: if the individual risk level of the center of gravity sway radius is >= 2 and the individual risk level of the trunk tilt angle is >= 2, then the posture risk level is determined to be high risk; if only a single peripheral joint (such as the knee joint) has an individual risk level >= 2, then it is determined to be medium risk; if multiple features show level 1 but no level 2, then it is determined to be low risk. The posture risk level output is one of the following: normal, low risk, medium risk, or high risk.

[0046] Through hierarchical assessment, the posture risk level can accurately reflect the situation from local movement deformation to overall instability.

[0047] Step 104: Based on the posture risk level and the physiological risk level, a comprehensive risk level is obtained through comprehensive assessment.

[0048] The management platform has a pre-set risk assessment model. Its inputs are postural risk level and physiological risk level, and its output is a comprehensive risk level. The risk assessment model can be a two-dimensional risk assessment matrix. The row index of the matrix corresponds to the postural risk level, and the column index corresponds to the physiological risk level. Both postural and physiological risk levels can be divided into four levels, such as normal, low risk, medium risk, and high risk. The matrix cells pre-store the corresponding comprehensive risk level, which can be divided into four levels: safe, attention, warning, and alarm. For example, when the postural risk is high and the physiological risk is medium, the corresponding comprehensive risk is alarm; when the postural risk is low and the physiological risk is normal, the comprehensive risk is safe.

[0049] The configuration of this matrix is ​​based on the experience of sports medicine experts and statistical analysis of a large amount of historical training data, and can reflect the actual risk probability of postural and physiological abnormalities under different combinations. In use, only the first index (postural risk level) and the second index (physiological risk level) need to be determined, and the comprehensive risk level can be obtained by looking up the table. This implementation method has low computational load and rapid response, making it suitable for real-time processing by edge computing nodes.

[0050] In other embodiments, the overall risk level can also be calculated using the maximum value method (taking the highest risk increment level as the overall physiological risk level) or the weighted summation method (calculating the weights based on the weights of different physiological parameters during training), and finally output the physiological risk level.

[0051] Step 105: When the overall risk level reaches the preset emergency alarm threshold, the emergency response process is triggered.

[0052] The overall risk level must include at least an alarm level (or a similar level indicating imminent danger), which is preset as an emergency alarm threshold. Once an alarm level is determined in the assessment, the emergency response procedure is triggered, which includes two parallel actions: First, project perceptible spatial warning signals to the trainees. For example... Figure 6 As shown, the management platform, based on the trainee's real-time location information (obtained through visual positioning via camera equipment or using the positioning module in the trainee's terminal), controls a projection device (such as a high-brightness laser projector) installed above the training field to project conspicuous warning patterns onto the ground or the space near the trainee's body. These patterns may include bright red flashing circles, arrows, or specific warning text. These spatial warning signals are highly visually intrusive and can be quickly perceived by trainees even in noisy or obstructed environments, prompting them to immediately stop dangerous actions or seek help.

[0053] For example, in one implementation, the student terminal is embedded with an ultra-wideband (UWB) positioning module or Bluetooth beacon. Multiple UWB base stations or Bluetooth gateways deployed around the training ground measure distance and triangulate the student terminal, and calculate the student's two-dimensional / three-dimensional coordinates in real time. The management platform maps these coordinates to the gimbal control angle of the projection device, and sends the drive command containing the angle and warning pattern type to the projection device through the local area network to achieve precise dynamic projection.

[0054] Second, the system pushes real-time status data of the trainees and auxiliary decision-making information based on their health records to the instructors. The instructor's terminal automatically pops up a high-priority alarm window, displaying the trainee's name, real-time video footage, skeletal posture overlay diagram, real-time physiological parameter waveforms, overall risk level, and specific parameters exceeding limits. Simultaneously, the management platform retrieves information such as allergy history, past medical history, and emergency contacts from the trainee's digital health record and generates auxiliary decision-making suggestions based on the current abnormal state. For example, if trainee A's heart rate drops sharply to 40 bpm, suspected of a cardiac vagal nerve reaction, it is recommended to immediately check consciousness and call the medical team. This trainee has no history of heart disease but has a history of hypoglycemia. The instructor can quickly take appropriate action based on this information.

[0055] Figure 4 This paper illustrates the flowchart of another training emergency response method provided in an embodiment of this application. This method introduces training node labels to dynamically adjust weights and includes an early warning process. The specific steps are as follows: Step 501: Obtain the trainee's perceptual data and historical data, construct a personalized baseline model, and determine the posture risk level and physiological risk level. This process is consistent with the previous one.

[0056] Step 502: Obtain the training node tag where the trainee is currently located. The training ground can be divided into multiple areas according to subjects and nodes, and each area is equipped with Bluetooth beacons, RFID readers, or area identification through camera equipment.

[0057] Trainee terminals or management platforms use positioning technology to obtain training node tags corresponding to the trainee's location, such as obstacle course running - climbing low walls, shooting - standing without support, and comprehensive physical fitness - carrying logs. The training node tags indicate the type of the current training node and the focus on posture or physiology.

[0058] Step 503: Based on the training node labels, determine the first fusion weight corresponding to the posture risk level and the second fusion weight corresponding to the physiological risk level. The management platform stores a weight configuration table, as shown in Table 1.

[0059] Table 1 Examples of Risk Weights under Different Training Node Labels Each posture risk level is pre-set with a first risk score, for example, 0 points for normal, 2 points for low risk, 5 points for medium risk, and 10 points for high risk. Each physiological risk level is pre-set with a second risk score, and the score mapping can be the same or different gradients can be set according to domain knowledge.

[0060] Step 504: Perform weighted fusion to obtain the comprehensive risk score.

[0061] That is, the comprehensive risk score = first risk score × first fusion weight + second risk score × second fusion weight. For example, in the precision shooting node, the attitude risk level is medium risk (5 points), the physiological risk level is low risk (2 points), and the fusion weights are 0.7 and 0.3, then the comprehensive risk score = 5 × 0.7 + 2 × 0.3 = 3.5 + 0.6 = 4.1.

[0062] Step 505: Determine the overall risk level based on the overall risk score.

[0063] Multiple threshold ranges can be preset: 0~1.5 is safe, 1.6~4 is watch out, 4.1~7 is warning, and 7.1~10 is alarm. The score of 4.1 falls into the warning range.

[0064] Step 506: Determine whether the overall risk level has reached the emergency alarm threshold. If it has (i.e., an alarm has been triggered), execute the emergency response procedure in Step 507, which is the same as described above. If it has not reached the threshold, proceed to Step 508.

[0065] Step 508: Determine whether the overall risk level has reached the preset warning threshold. If it has, trigger the Level 1 warning process (Step 509); otherwise, return to continue monitoring.

[0066] Step 509: Trigger the Level 1 Alarm Procedure. The management platform sends a first alarm command to the student terminal, instructing the student terminal to provide tactile alerts using a first vibration mode (e.g., a short, long, short vibration rhythm) and play a first voice prompt. Simultaneously, a second alarm command is sent to the instructor terminal, instructing the instructor terminal to vibrate using a second vibration mode (e.g., continuous rapid vibration) and play a second voice prompt.

[0067] The voice prompts are dynamically generated based on the types of parameters in the perceived data that exceed the safety range of the corresponding baseline model. For example, if the dominant factor is a high heart rate, the student's voice prompt might be: "Your heart rate is consistently high. Please adjust to deep breathing and reduce the intensity." The coach's voice prompt might be: "Student Zhang San's heart rate has exceeded the personalized threshold by 15%, currently at 168 beats per minute. Please pay attention." If the dominant factor is abnormal posture, the student's voice prompt might be: "Your body is swaying too much. Please tighten your core and stabilize your center of gravity." These voice prompts are synthesized in real time by a text-to-speech engine to ensure information accuracy.

[0068] Through the above process, a tiered and progressive response strategy was formed. Intervention was carried out through two-way early warning at the initial stage of danger, and emergency calls for help were made through strong spatial warning and information push in critical moments, so as to ensure the safety of training to the greatest extent.

[0069] Figure 5 This is a structural diagram of the training emergency response system provided in an embodiment of this application. The system includes a management platform 610, a camera device 620, a projection device 630, a trainee terminal 640, and a coach terminal 650. The devices are connected via a wireless local area network or a mobile communication network.

[0070] The management platform 610 is the core of the system and can be an edge server deployed on the local training ground or a cloud platform with computing capabilities. The camera equipment 620 can be multiple high-definition network cameras installed at key locations on the training ground, covering the training area and supporting multi-angle image acquisition to overcome obstructions. The projection equipment 630 is an addressable laser projection device that receives location coordinates and control commands from the management platform and projects warning signals to designated areas. The trainee terminal 640 is a wearable device, such as a smartwatch that can monitor heart rate, blood oxygen, and electrocardiogram, equipped with a display screen, vibration motor, speaker, and physiological sensors, capable of receiving commands, displaying information, emitting vibrations, and issuing voice commands. The coach terminal 650 is typically a tablet or ruggedized laptop running a monitoring application that can display a trainee list, real-time video, physiological waveforms, risk levels, and auxiliary decision-making pop-ups.

[0071] It should be noted that, Figure 1 The architecture shown is for illustrative purposes only; hardware such as cameras and projectors can be distributed across multiple physical spaces (e.g., multiple training areas). The management platform 110 can be a locally deployed server or rely on a private cloud platform for computing. Specifically, student terminals and coach terminals can interact with the management platform 110 via mobile cellular networks (such as 5G), Wi-Fi, or Bluetooth relay.

[0072] The student terminal mentioned in this application can be a dedicated watch customized based on smart wearable devices. When the solution in this application is applied to the scenario of companion robots, the student terminal can also be expanded into the core control module of a smart elderly care robot or a personal monitoring terminal for the elderly.

[0073] The projection equipment mentioned in this application can be an engineering projector with high brightness and anti-interference capabilities, or a dedicated laser safety warning light. Since training bases are mostly outdoor environments, traditional sound-based alarms are easily ignored due to environmental noise. Therefore, this application uses projection equipment to project bright red warning boxes, stop signs, or visual signals such as "please rest immediately" onto the ground under the trainees' feet or on the wall in front of them.

[0074] The management platform 610 includes a data acquisition module 611, a model building module 612, a risk assessment module 613, a coordinated response module 614, and a database 615. The data acquisition module 611 connects to the camera device 620 and the trainee terminal 640, receiving video streams and physiological data, and extracting skeletal posture data from the video streams. The model building module 612 reads historical trainee data from the database 615, constructing and updating personalized posture baseline models and personalized physiological baseline models. The risk assessment module 613 includes a posture risk assessment submodule, a physiological risk assessment submodule, and a comprehensive risk assessment submodule, implementing various assessment logics from the aforementioned methods. After assessing the risk level, the coordinated response module 614 generates corresponding alarm commands, driving the projection device 630, trainee terminal 640, and coach terminal 650 to execute corresponding alarm and push actions. The database 615 stores trainee information, historical training data, baseline models, risk judgment matrices, weight configuration tables, and health records.

[0075] In its implementation, after the management platform 610 starts, it continuously receives data streams from each trainee. Taking trainee A as an example, their trainee terminal 640 sends data packets such as heart rate, blood oxygen, and body temperature every second. The video stream from the camera device 620 is sent to the data acquisition module 611 at a frequency of 30 frames per second. The module 611 uses a deployed skeletal estimation algorithm to extract skeletal pose every 5 frames (i.e., 6 times per second) and aligns it with the latest physiological data. When trainee A first enters training, if there is insufficient historical data, the model building module 612 can first use the default parameters of the group baseline model, and automatically switch to a personalized model after accumulating at least 3 complete training data. The risk assessment module 613 executes a risk assessment process every time it receives an aligned pose and physiological data pair, and outputs a comprehensive risk level. When the linkage response module 614 receives a warning or alarm command, it immediately executes the relevant responses in parallel.

[0076] like Figure 6The diagram illustrates an application scenario architecture of a training emergency response system applicable to an embodiment of this application. This architecture diagram demonstrates the projection process of spatial warning signals during emergency response. Specifically, when the management platform determines that the trainee's overall risk level reaches a preset emergency alarm threshold, it immediately generates a projection command containing location coordinates and control parameters and sends it to the projection device 630. Upon receiving the command, the projection device 630 uses beam projection technology to define a specific projection area on the ground of the training area where the trainee is currently located, and simultaneously projects a high-frequency flashing red ring and conspicuous warning text within that area. The trainee (wearing a trainee terminal 640) can directly perceive the spatial warning signal within the projection area by lowering their line of sight, thus achieving visual perception. Because the warning signal is projected directly onto the trainee's feet or surroundings, regardless of whether there is noise in the training environment, it can forcibly interrupt the trainee's current training actions in the most intuitive optical and physical way, compelling the trainee to pay attention to the dangerous situation. Simultaneously, the student terminal 640 worn by the trainees will also respond to the instructions of the management platform, driving the built-in vibration motor and speaker to emit tactile vibrations with specific patterns and matching voice prompts. This works in conjunction with the spatial warning signals projected on the ground, ensuring that trainees can accurately understand the current risk level through multiple sensory channels, including vision, touch, and hearing, and immediately take actions such as stopping, taking shelter, or calling for help. It should be noted that... Figure 6 The architecture shown only illustrates the specific interaction of spatial warning signals. The projection device 630, student terminal 640 and other supporting hardware can interact with the back-end management platform in real time through wireless LAN or mobile communication network to achieve precise linkage.

[0077] This application also provides a computer device, such as... Figure 7 As shown, the computer device includes a processor, a memory, and a communication interface. The processor, memory, and communication interface are interconnected via a bus. The memory stores a program, which the processor loads and executes to implement the functions described in the method embodiments above. This computer device may be a concrete physical implementation of the aforementioned management platform.

[0078] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, can implement the above-described method. The readable storage medium may be a read-only memory, random access memory, magnetic disk, or optical disk, etc.

[0079] Those skilled in the art will recognize that the units of the various examples described in connection with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of the invention.

[0080] In the embodiments provided by the present invention, it should be understood that the division of units is only a logical functional division. In actual implementation, there may be other division methods, such as multiple units can be combined into one unit, one unit can be split into multiple units, or some features can be ignored.

[0081] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0082] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0083] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the specification of the present invention.

Claims

1. A training method for emergency response based on data perception and coordinated distress calls, characterized in that, include: Acquire sensory data of trainees during training, including skeletal posture data obtained by extracting skeletal posture from the video stream of the trainees' training and real-time physiological parameters from the trainees' terminals. Obtain the trainee's historical training data, and construct the trainee's personalized posture baseline model and personalized physiological baseline model based on the historical training data; The posture risk level is determined based on the degree of deviation between the skeletal posture data and the personalized posture baseline model. The physiological risk level is determined based on the degree of deviation between the real-time physiological parameters and the personalized physiological baseline model. A comprehensive risk level is obtained by comprehensively assessing the posture risk level and the physiological risk level. When the overall risk level reaches the preset emergency alarm threshold, the emergency response procedure is triggered. The emergency response process includes: projecting a perceptible spatial warning signal to the trainee, and pushing the trainee's real-time status data and auxiliary decision-making information based on health records to the coach terminal.

2. The emergency response method for training based on data perception and coordinated distress calls according to claim 1, characterized in that, The determination of posture risk level based on the degree of deviation between the skeletal posture data and the personalized posture baseline model includes: Multiple posture feature values ​​are extracted from the skeletal posture data, and each posture feature value corresponds to the movement state of different body parts of the trainee. Each of the posture feature values ​​is compared with the corresponding standard posture feature range in the personalized posture baseline model to determine whether each of the posture feature values ​​is within the standard posture feature range. For each of the posture feature values ​​that exceeds the standard posture feature range, the excess range is calculated, and the individual risk level corresponding to each posture feature is determined based on the excess range. Input the individual risk level corresponding to all the posture features into a preset posture risk assessment model to obtain the posture risk level; The attitude risk determination model is a multi-input single-output state mapping model, in which a specific combination of multiple individual risk levels corresponds to a predefined attitude risk level.

3. The emergency response method for training based on data perception and coordinated distress calls according to claim 1, characterized in that, The determination of physiological risk level based on the degree of deviation between the real-time physiological parameters and the personalized physiological baseline model includes: Each of the real-time physiological parameters is compared with the corresponding preset safety threshold in the personalized physiological baseline model; For any of the real-time physiological parameters, when the real-time physiological parameter exceeds the preset safety threshold, the risk increment level corresponding to each physiological parameter is determined according to the magnitude of the exceedance; wherein, the greater the magnitude of the exceedance, the higher the risk increment level. The physiological risk level of the trainee is determined by combining the risk increment levels corresponding to each physiological parameter. The preset safety threshold in the personalized physiological baseline model is determined based on the statistical distribution of the trainee's historical physiological parameters.

4. The emergency response method for training based on data perception and coordinated distress calls according to claim 1, characterized in that, The comprehensive risk level is obtained by comprehensively assessing the posture risk level and the physiological risk level, including: The posture risk level and the physiological risk level are input into a preset risk assessment model to obtain the trainee's current comprehensive risk level. The risk assessment model includes multiple state nodes that are defined by a combination of posture risk level and physiological risk level, and each state node is associated with a predefined comprehensive risk level.

5. The emergency response method for training based on data perception and coordinated distress calls according to claim 4, characterized in that, The risk assessment model is a risk assessment matrix; The posture risk level and the physiological risk level are input into a preset risk assessment model to obtain the trainee's current comprehensive risk level, which specifically includes: Determine the first index corresponding to the posture risk level in the risk assessment matrix, and the second index corresponding to the physiological risk level in the risk assessment matrix; The risk assessment matrix is ​​queried using the first index and the second index to obtain the pre-stored comprehensive risk level; The risk assessment matrix is ​​a two-dimensional matrix, with row indices corresponding to different posture risk levels and column indices corresponding to different physiological risk levels. The matrix cells pre-store the corresponding comprehensive risk levels.

6. The emergency response method for training based on data perception and coordinated distress calls according to claim 1, characterized in that, Also includes: Obtain the training node label currently located by the trainee; the training node label is used to indicate the type of the current training node. Based on the training node labels, a first fusion weight corresponding to the posture risk level and a second fusion weight corresponding to the physiological risk level are determined; wherein, each posture risk level is preset to correspond to a first risk score, and each physiological risk level is preset to correspond to a second risk score. The step of comprehensively assessing the overall risk level based on the posture risk level and the physiological risk level includes: performing weighted fusion based on the first risk score corresponding to the posture risk level, the second risk score corresponding to the physiological risk level, the first fusion weight, and the second fusion weight to obtain the overall risk score, and determining the overall risk level based on the overall risk score; Different training node labels correspond to different first fusion weights and second fusion weights.

7. A training emergency response method based on data perception and coordinated distress call according to any one of claims 1-5, characterized in that, The method further includes: When the overall risk level reaches the preset warning threshold but does not reach the emergency alarm threshold, the first-level warning process is triggered. The Level 1 early warning process includes: Send a first alarm command to the student terminal, instructing the student terminal to perform tactile vibration in a first vibration mode and play a first voice prompt; A second alarm command is sent to the coach terminal, instructing the coach terminal to perform tactile vibration in a second vibration mode and play a second voice prompt.

8. A training emergency response system based on data perception and coordinated distress calls, characterized in that, The system includes: a management platform, camera equipment, projection equipment, student terminals, and coach terminals; The management platform is used for: The video stream of the trainees during training is acquired through the camera device, and the real-time physiological parameters of the trainees are acquired through the trainee terminal. Obtain the trainee's historical training data, and construct the trainee's personalized posture baseline model and personalized physiological baseline model based on the historical training data; Skeletal pose data is extracted from the video stream; the pose risk level is determined based on the degree of deviation between the skeletal pose data and the personalized pose baseline model; the physiological risk level is determined based on the degree of deviation between the real-time physiological parameters and the personalized physiological baseline model. A comprehensive risk level is obtained by comprehensively assessing the posture risk level and the physiological risk level. When the overall risk level reaches the preset emergency alarm threshold, the emergency response procedure is triggered. The projection device is used to project perceptible spatial warning signals to the trainees under the control of the management platform. The coach terminal is used to receive and display the real-time status data of the trainees and auxiliary decision-making information based on health records pushed by the management platform.

9. A computer device, characterized in that, include: Memory, used to store programs; A processor for loading the program to perform the method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method as described in any one of claims 1-7.

Images