Method and system for constructing a clinical nurse disaster resilience classification training program

Abstract

This invention discloses a method and system for constructing a disaster resilience classification training program for clinical nurses, belonging to the fields of disaster nursing management and computer image recognition technology. It involves acquiring facial micro-expression image sequences of nurses in standardized disaster simulation scenarios; extracting the activation frequency, activation duration, and left-right facial activation symmetry features of the zygomaticus major and orbicularis oculi muscles and inputting them into a preset resilience classifier; outputting subtype labels based on a fixed mapping relationship between the spatiotemporal features of muscle activity and resilience subtypes; retrieving the corresponding basic training framework based on the labels; comparing the activation duration features with a threshold; and generating reinforcement training labels on the framework using image overlay technology to form a personalized classification training program. This invention achieves objective and real-time subtype identification of clinical nurses' disaster resilience and can automatically generate differentiated training content targeting resource deficiencies in different subtypes, significantly improving the accuracy and automation level of disaster resilience training.

Description

Technical Field

[0001] This invention discloses a method and system for constructing a training program for clinical nurses' disaster resilience classification, belonging to the field of disaster nursing management and computer image recognition technology. Background Technology

[0002] In the field of disaster medicine and nursing management, the disaster resilience level of clinical nurses directly affects the quality of disaster relief and the nurses' own mental health. Existing methods for assessing disaster resilience mainly rely on psychological scales or behavioral self-report questionnaires, such as the General Psychological Resilience Scale or Disaster Resilience Measurement Tools. While these methods have a certain degree of reliability and validity, they are subjective self-reports, easily influenced by recall bias, social desirability, and current emotional state, making it difficult to capture nurses' true stress responses to dynamic disaster stimuli in real time and objectively. Furthermore, current disaster nursing training for nurses often adopts a standardized or seniority-based tiered model, lacking a refined identification of individual differences. This leads to a mismatch between training programs and nurses' actual skill gaps, resulting in limited training effectiveness.

[0003] In recent years, computer image recognition technology has been explored for its application in analyzing the correlation between facial micro-expressions and psychological states. Current technologies in facial expression recognition primarily focus on basic emotion classification, such as the identification and classification of macro-expressions like happiness, sadness, and anger. The extracted features are mostly the amplitude of regional deformation of the entire face or the intensity of overall movement units. However, in disaster relief simulations, nurses' stress responses manifest as extremely short-duration, minute changes in facial expression, particularly concentrated in the subtle contractions of specific muscle groups such as the zygomaticus major and orbicularis oculi muscles, and exhibiting asymmetrical activation on the left and right sides of the face. Traditional facial expression recognition methods lack sufficient resolution for these micro-scale spatial features and, more importantly, lack a mechanism to map them to the complex psychological construct of disaster resilience.

[0004] Furthermore, current training programs are still generated manually, meaning that trainees' macro resilience scores are first obtained through assessment tools, and then trainers manually select corresponding course modules. This approach is not only inefficient but also makes it difficult to automate and refine the transformation of assessment results into training content. Therefore, there is a need for a technology that can automatically identify nurses' disaster resilience subtypes based on objective physiological signals, namely facial micro-expression images, and dynamically generate differentiated training programs for different subtypes based on their resource deficiencies, in order to compensate for the shortcomings of subjective scale assessments and standardized training models. Summary of the Invention

[0005] To achieve the above objectives, this application provides the following technical solution: A method for constructing a training program for clinical nurses' disaster resilience classification includes the following steps: Step 1: Obtain facial micro-expression image sequences of the target clinical nurse in a standardized disaster simulation scenario; Step 2: Use image recognition technology to extract features from the facial micro-expression image sequence to generate a micro-expression spatiotemporal feature set that includes at least the activation frequency, activation duration, and activation symmetry of the left and right sides of the face of the zygomaticus major and orbicularis oculi muscles. Step 3: Input the micro-expression spatiotemporal feature set into a preset resilience classifier. The resilience classifier outputs the disaster resilience subtype label of the target clinical nurse based on the mapping relationship between the spatiotemporal features of micro-expressions and disaster resilience subtypes. The disaster resilience subtype label includes at least resource depletion type, capability dependence type and collaborative buffer type. Step 4: Based on the disaster resilience subtype label, retrieve the corresponding basic training framework from the pre-built training scheme library. The basic training framework includes core capability thresholds and resource replenishment anchors. Step 5: Compare the basic training framework with the activation duration features in the micro-expression spatiotemporal feature set. When the activation duration features exceed the core capability threshold, use image overlay technology to generate reinforcement training labels on the basic training framework to form the disaster resilience classification training program for the target clinical nurse.

[0006] Furthermore, the standardized disaster simulation scenario in step 1 is as follows: In a controlled lighting environment, a holographic image of a disaster scene with gradient pressure is presented to clinical nurses through a head-mounted display. The gradient pressure includes the chaotic audio-visual stimulation in the early stage of the disaster, the visual impact of the mass arrival of the wounded in the middle stage of the disaster, and the static image of resource scarcity in the later stage of the disaster.

[0007] Further, step 2 includes: The continuous frame difference method is used to locate the start and end frames of facial micro-expression changes, and a rapid deformation stage and a slow recovery stage are divided between the start and end frames. The activation frequency of the zygomaticus major and orbicularis oculi muscles is the number of times the rapid deformation phase occurs per unit time, and the activation duration is the time from the start of a single rapid deformation phase to the end of the slow recovery phase.

[0008] Furthermore, the resilience classifier in step 3 establishes a mapping relationship between the spatiotemporal features of micro-expressions and disaster resilience subtypes based on the following logic: Obtain the original parameters of zygomaticus major muscle activation frequency, zygomaticus major muscle activation duration, orbicularis oculi muscle activation frequency, orbicularis oculi muscle activation duration, and left and right facial activation symmetry from the micro-expression spatiotemporal feature set. When the ratio of the activation frequency of the orbicularis oculi muscle to the activation frequency of the zygomaticus major muscle is in the first range and the activation symmetry of the left and right faces is lower than the preset asymmetry threshold, the disaster resilience subtype label is determined to be resource depletion type. When the activation duration of the zygomaticus major muscle is greater than that of the orbicularis oculi muscle, and the activation frequency of the zygomaticus major muscle is higher than the preset high-frequency threshold, the disaster resilience subtype label is determined to be capability-dependent. When the activation symmetry of the left and right sides of the face is higher than the preset symmetry threshold, and the ratio of the activation frequency of the orbicularis oculi muscle to the activation frequency of the zygomaticus major muscle is in the second numerical range, the disaster resilience subtype label is determined to be the collaborative buffer type. The minimum value in the first numerical interval is greater than the maximum value in the second numerical interval.

[0009] Furthermore, in step 4, the core capability threshold corresponding to the resource depletion subtype is the first time threshold of the zygomaticus major muscle activation duration, and the corresponding resource replenishment anchor point is the social support dimension enhancement module. The core competency threshold corresponding to the competency-dependent subtype is the first frequency threshold of the orbicularis oculi muscle activation frequency, and the corresponding resource supply anchor is the self-efficacy dimension construction module. The core capability threshold corresponding to the collaborative buffering subtype is the symmetry threshold of the activation symmetry of the left and right faces, and the corresponding resource supply anchor is the altruism dimension expansion module.

[0010] Furthermore, the image overlay technique in step 5 includes: The enhanced training identifier is dynamically attached above the graphic explanation area corresponding to the resource supply anchor point in the basic training framework in the form of a highlighted color block that is different from the basic training framework. The display duration of the highlighted color block is positively correlated with the amount by which the activation duration feature exceeds the core capability threshold.

[0011] Furthermore, the specific process of generating reinforcement training labels on the basic training framework using image overlay technology in step 5 is as follows: The page layout structure of the basic training framework is analyzed to identify the target layer where the resource supply anchor point is located and the coordinate boundary of the target layer. Obtain the difference between the activation duration feature and the core capability threshold, and generate the pixel density of the reinforcement training label that matches the difference according to the preset difference-label density lookup table; While maintaining the original background texture and text content of the target layer as recognizable, semi-transparent warning dots are rendered in a random distribution manner according to the pixel density within the coordinate boundary of the target layer to form an enhanced training label. The target layer carrying the reinforcement training identifier is re-integrated with other layers of the basic training framework to generate a visual preview image of the classification training scheme.

[0012] Furthermore, the head-mounted display has a built-in eye-tracking module, and step 1 further includes: Simultaneously record the eye movement trajectory of clinical nurses during the stress presentation of holographic images at the disaster site, the eye movement trajectory including the coordinates of the fixation hot zone and the pupil diameter change curve; The micro-expression spatiotemporal feature set in step 2 also includes the stress triggering moment defined by the inflection point of the slope on the pupil diameter change curve. This stress triggering moment is used for temporal alignment verification with the activation start frame of the zygomaticus major muscle.

[0013] Furthermore, a timing alignment verification step is included before step 5: Calculate the time offset between the stress triggering time and the zygomaticus major muscle activation start frame. If the time offset exceeds the preset offset threshold, the current facial micro-expression image sequence is removed, and a re-acquisition command is triggered, instructing step 1 to reacquire the facial micro-expression image sequence of the target clinical nurse. The disaster resilience classification training scheme generated in step 5 is a multi-layer interactive image file. The first layer of the multi-layer interactive image file is a carrier frame layer, which is used to display the complete content of the basic training framework. The second layer is the identifier layer, which is used to carry reinforcement training identifiers; The third layer is the interactive response layer, which is used to capture the click operation of clinical nurses on the reinforcement training icon, and pop up a micro-expression playback window associated with the reinforcement training icon when clicked. The micro-expression playback window is used to loop the original facial micro-expression image sequence fragments collected in step 1 that correspond to the reinforcement training icon.

[0014] According to a second aspect of the present invention, the present invention claims protection for a system for constructing a training program on disaster resilience classification for clinical nurses, comprising: One or more processors; A memory storing one or more programs, which, when executed by one or more processors, enable the one or more processors to implement the method for constructing a clinical nurse disaster resilience classification training program.

[0015] This invention discloses a method and system for constructing a disaster resilience classification training program for clinical nurses, belonging to the fields of disaster nursing management and computer image recognition technology. It involves acquiring facial micro-expression image sequences of nurses in standardized disaster simulation scenarios; extracting the activation frequency, activation duration, and left-right facial activation symmetry features of the zygomaticus major and orbicularis oculi muscles and inputting them into a preset resilience classifier; outputting subtype labels based on a fixed mapping relationship between the spatiotemporal features of muscle activity and resilience subtypes; retrieving the corresponding basic training framework based on the labels; comparing the activation duration features with a threshold; and generating reinforcement training labels on the framework using image overlay technology to form a personalized classification training program. This invention achieves objective and real-time subtype identification of clinical nurses' disaster resilience and can automatically generate differentiated training content targeting resource deficiencies in different subtypes, significantly improving the accuracy and automation level of disaster resilience training. Attached Figure Description

[0016] Figure 1 A flowchart illustrating a method for constructing a disaster resilience classification training program for clinical nurses, which is claimed in an embodiment of the present invention. Figure 2 The second flowchart is a method for constructing a clinical nurse disaster resilience classification training program, which is claimed in an embodiment of the present invention. Figure 3 This is a structural module diagram of a system for constructing a clinical nurse disaster resilience classification training program, which is claimed in an embodiment of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0018] The terms "first," "second," and "third" in this application are for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first," "second," or "third" may explicitly or implicitly include at least one of those features. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified. All directional indications in the embodiments of this application, such as up, down, left, right, front, back, etc., are only used to explain the relative positional relationships and movements between components in a specific orientation as shown in the accompanying drawings. If the specific orientation changes, the directional indications will change accordingly. Furthermore, the terms "including" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or devices.

[0019] References to embodiments herein mean that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a mutually exclusive, independent, or alternative embodiment. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0020] According to the first embodiment of the present invention, referring to Figure 1 This invention claims protection for a method for constructing a training program on disaster resilience classification for clinical nurses, comprising the following steps: Step 1: Obtain facial micro-expression image sequences of the target clinical nurse in a standardized disaster simulation scenario; Step 2: Use image recognition technology to extract features from the facial micro-expression image sequence to generate a micro-expression spatiotemporal feature set that includes at least the activation frequency, activation duration, and activation symmetry of the left and right sides of the face of the zygomaticus major and orbicularis oculi muscles. Step 3: Input the micro-expression spatiotemporal feature set into a preset resilience classifier. The resilience classifier outputs the disaster resilience subtype label of the target clinical nurse based on the mapping relationship between the spatiotemporal features of micro-expressions and disaster resilience subtypes. The disaster resilience subtype label includes at least resource depletion type, capability dependence type and collaborative buffer type. Step 4: Based on the disaster resilience subtype label, retrieve the corresponding basic training framework from the pre-built training scheme library. The basic training framework includes core capability thresholds and resource replenishment anchors. Step 5: Compare the basic training framework with the activation duration features in the micro-expression spatiotemporal feature set. When the activation duration features exceed the core capability threshold, use image overlay technology to generate reinforcement training labels on the basic training framework to form the disaster resilience classification training program for the target clinical nurse.

[0021] In this embodiment, in a workstation configured with a standardized simulation environment, a head-mounted display and a high-definition facial camera device are first activated. The operator selects a pre-stored disaster scene simulation video via a console. This video includes a crowded emergency room scene after an earthquake, audio recordings of injured people calling for help, and emergency instructions from medical staff. The video is then presented to the clinical nurse undergoing the test via the head-mounted display. Simultaneously, the high-definition camera continuously captures facial images of the nurse at a rate of 30 frames per second while she watches the video, forming a raw image sequence.

[0022] Image acquisition lasted for 3 minutes, covering the entire video playback cycle. After acquisition, the workstation automatically transmitted the raw image sequence to the image processing module. This module first located the nurse's facial area and, by comparing the pixel grayscale changes in the corners of the eyes and mouth between two consecutive frames, identified the starting and ending frames of the minute contractions of the zygomaticus major muscle (located in the cheek, controlling smiling) and the orbicularis oculi muscle (located around the eyes, controlling blinking and squinting). For each identified muscle contraction event, the image processing module recorded three values: the number of frames representing the duration of activation from the starting frame to the ending frame, the number of times such an event occurred per minute representing the activation frequency, and the ratio representing the symmetry of activation between the left and right sides of the face in the same time period. These three values ​​constitute a spatiotemporal feature set corresponding to this test.

[0023] Subsequently, the workstation transmits this spatiotemporal feature set to a pre-defined resilience classifier, which has three pre-defined sets of judgment logic. The first set of logic checks whether the ratio of the activation frequency of the orbicularis oculi muscle to the activation frequency of the zygomaticus major muscle is less than 0.5, and whether the activation symmetry of the left and right sides of the face is less than 0.7. If these conditions are met, the label "resource-depletion type" is output. The second set of logic checks whether the activation duration of the zygomaticus major muscle is greater than 50 frames, and whether the activation frequency of the zygomaticus major muscle is greater than 6 times per minute. If these conditions are met, the label "ability-dependent type" is output. The third set of logic checks whether the activation symmetry of the left and right sides of the face is greater than 0.85, and whether the ratio of the activation frequency of the orbicularis oculi muscle to the activation frequency of the zygomaticus major muscle is between 0.8 and 1.2. If these conditions are met, the label "cooperative buffer type" is output. In this example, the classifier outputs "ability-dependent type".

[0024] Based on the competency-dependent tag, the workstation searches for the corresponding basic training framework file in the locally stored training program library. This file is an editable presentation. The top of the first page is labeled with the competency-dependent training framework, and the main body contains three blank text boxes, which are reserved for filling in self-efficacy training cases, skill achievement standards, and emergency decision-making processes, respectively.

[0025] The workstation then compares the zygomaticus major muscle activation duration recorded in step 2 (65 frames in this example) with the core competency threshold corresponding to the competency-dependent type, which is 50 frames. Since 65 is greater than 50, the workstation determines that a reinforcement training indicator needs to be generated. Specifically, the workstation locates the skill achievement standard text box in the basic training framework and automatically generates a red triangle icon in the lower right corner of the text box. The icon contains the number 15, representing the number of frames exceeding the threshold. After completing the above modifications, the workstation saves the presentation as a new file, which is the disaster resilience classification training plan for this clinical nurse.

[0026] Furthermore, the standardized disaster simulation scenario in step 1 is as follows: In a controlled lighting environment, a holographic image of a disaster scene with gradient pressure is presented to clinical nurses through a head-mounted display. The gradient pressure includes the chaotic audio-visual stimulation in the early stage of the disaster, the visual impact of the mass arrival of the wounded in the middle stage of the disaster, and the static image of resource scarcity in the later stage of the disaster.

[0027] In one embodiment, a standardized disaster simulation scenario was constructed. The laboratory was a soundproof room 4 meters long and 3 meters wide, with two sets of adjustable color temperature LED lights installed on the ceiling. The lighting intensity was set to 500 lux and remained constant throughout the test to eliminate the interference of ambient light changes on facial expression acquisition.

[0028] The test nurse sat in a swivel chair in the center of the room, wearing a see-through head-mounted display with a resolution of 1920×1080; the display had built-in stereo headphones that could independently deliver audio to the left and right ears; the workstation sent a holographic video stream with a total duration of 5 minutes to the head-mounted display, which was divided into three consecutive gradient pressure phases in chronological order.

[0029] The first stage is the initial chaotic audio-visual stimulation stage of the disaster, from second 0 to second 60; the video content shows the scene of a strong earthquake in the city streets, with images of buildings shaking violently, glass breaking and flying, and pedestrians running in panic; at the same time, the headphones play high-decibel alarm sounds, the rumbling sound of buildings collapsing, and screams from the crowd; the main purpose of this stage is to induce the nurses' stress response to the sudden chaotic environment.

[0030] The second phase, from 61 to 240 seconds, is the visually impactful phase of the mass arrival of injured people during the mid-disaster period. The holographic image switches to the entrance of the hospital emergency room, showing a scene of stretcher teams continuously pushing in a large number of injured people. Each injured person's virtual image has obvious open wounds or bloodstains on their face and limbs, and some injured person models show groans or expressions of pain; the number of injured people waiting in the upper left corner of the screen continuously jumps to the number 25, simulating the urgency of resources not being able to meet the demand; this phase focuses on testing the nurses' response mode under multiple visual information loads.

[0031] The third stage is a static display of resource scarcity in the post-disaster period, from second 241 to second 300. The holographic image shows an empty treatment preparation room, with most of the first aid kits on the shelves marked as empty, and the only two remaining kits circled in dashed circles; a whiteboard on the wall has a red sign indicating that the intravenous infusion sets are used up; the entire scene remains still, with only the occasional low hum of a generator coming from a distance through the headphones; this stage is used to observe the micro-expressions of nurses' faces when faced with static information about resource depletion, especially muscle activity related to helplessness.

[0032] The three stages switch seamlessly, forming a continuous sequence of increasing stress stimuli throughout the video stream. During video playback, an infrared camera mounted in front of the head-mounted display continuously captures images of the nurse's face, ensuring clear capture of micro-expressions even in low-light conditions.

[0033] Further, step 2 includes: The continuous frame difference method is used to locate the start and end frames of facial micro-expression changes, and a rapid deformation stage and a slow recovery stage are divided between the start and end frames. The activation frequency of the zygomaticus major and orbicularis oculi muscles is the number of times the rapid deformation phase occurs per unit time, and the activation duration is the time from the start of a single rapid deformation phase to the end of the slow recovery phase.

[0034] In this embodiment, the workstation obtains a continuous grayscale image sequence from a high-definition camera device, with each frame having a resolution of 640×480 pixels. For the nth frame and the (n+1)th frame, the system calculates the absolute value of the grayscale difference of each pixel within a 32×32 pixel rectangular window in a pre-marked left corner of the mouth region between the two frames, and then calculates the average value of the grayscale difference of all pixels in that region. When the average value exceeds a preset grayscale change threshold, the system determines the nth frame as a candidate starting frame.

[0035] The system scans frame by frame from the candidate starting frame, recording the cumulative trend of grayscale changes in the corner of the mouth area. Specifically, the system calculates the absolute value of the grayscale difference between the current frame and the previous frame in the same area; when the absolute value of this difference is greater than the corresponding value of the previous frame for five consecutive frames, the system identifies the first frame of these five frames as the starting frame of the rapid deformation stage; then the system continues scanning until it finds that the absolute value of the grayscale difference for three consecutive frames is less than a certain proportion of the difference between the starting frame and the second frame, at which point the first frame of these three frames is marked as the ending frame of the rapid deformation stage.

[0036] Next, the system enters the slow recovery phase. Starting from the frame after the end of the fast deformation phase, the system compares the absolute value of the grayscale difference frame by frame again, but this time a more lenient judgment standard is used. When the absolute value of the grayscale difference of eight consecutive frames is less than half of the peak difference of the fast deformation phase, the system marks the first frame of these eight frames as the starting frame of the slow recovery phase. The system continues to scan until the grayscale value of the pixels in the corner of the mouth area is restored to a value no more than two grayscale levels different from the average grayscale value of the five frames before the starting frame of the fast deformation phase, and marks the first restored frame as the end frame of the slow recovery phase.

[0037] A complete micro-expression event is defined as starting from the beginning frame of the fast deformation phase and ending at the end frame of the slow recovery phase. Its activation duration is equal to the sequence number of the end frame of the slow recovery phase minus the sequence number of the beginning frame of the fast deformation phase, multiplied by the frame interval. For example, if the beginning frame is 1000, the end frame is 1050, and the frame rate is 30fps, then the duration is 50 / 30 ≈ 1.67 seconds.

[0038] The activation frequency is calculated as follows: The system identifies all micro-expression events in a 3-minute image sequence using the method described above. The number of events belonging to the rapid deformation phase is counted, divided by 3 minutes (180 seconds), and then multiplied by 60 to obtain the number of rapid deformation phase occurrences per minute. For example, if 15 events are identified, the activation frequency is 15 / 180 × 60 = 5 times per minute.

[0039] Furthermore, the resilience classifier in step 3 establishes a mapping relationship between the spatiotemporal features of micro-expressions and disaster resilience subtypes based on the following logic: Obtain the original parameters of zygomaticus major muscle activation frequency, zygomaticus major muscle activation duration, orbicularis oculi muscle activation frequency, orbicularis oculi muscle activation duration, and left and right facial activation symmetry from the micro-expression spatiotemporal feature set. When the ratio of the activation frequency of the orbicularis oculi muscle to the activation frequency of the zygomaticus major muscle is in the first range and the activation symmetry of the left and right faces is lower than the preset asymmetry threshold, the disaster resilience subtype label is determined to be resource depletion type. When the activation duration of the zygomaticus major muscle is greater than that of the orbicularis oculi muscle, and the activation frequency of the zygomaticus major muscle is higher than the preset high-frequency threshold, the disaster resilience subtype label is determined to be capability-dependent. When the activation symmetry of the left and right sides of the face is higher than the preset symmetry threshold, and the ratio of the activation frequency of the orbicularis oculi muscle to the activation frequency of the zygomaticus major muscle is in the second numerical range, the disaster resilience subtype label is determined to be the collaborative buffer type. The minimum value in the first numerical interval is greater than the maximum value in the second numerical interval.

[0040] In this embodiment, the resilience classifier pre-stores a judgment rule table, which does not depend on any external training process but is manually set by experts after observing a large number of sample videos based on empirical thresholds. The classifier receives five raw parameters from the image processing module, which are denoted as: A. zygomaticus major muscle activation frequency, unit: times / minute; B. zygomaticus major muscle activation duration, unit: frames; C. orbicularis oculi muscle activation frequency, unit: times / minute; D. orbicularis oculi muscle activation duration, unit: frames; E. left and right facial activation symmetry, dimensionless ratio, with a value range of 0 to 1, where 1 indicates complete symmetry.

[0041] The classifier first calculates the ratio R = C / A. Then it executes the following sequential decision logic: The first step is to determine whether the data belongs to the resource depletion type. The classifier checks whether two conditions are met simultaneously: Condition 1 is whether the value of R is within the first numerical interval, which is defined as greater than or equal to 1.5 and less than or equal to 2.8; Condition 2 is whether the value of E is less than 0.55. If both conditions are met, the subtype label is directly output as resource depletion type, and no further judgment is performed. For example, if A = 2 times / minute and C = 3.5 times / minute, then R = 1.75, falling into [1.5, 2.8]; at the same time, E = 0.52 < 0.55, so it is determined to be resource depletion type.

[0042] The second step involves determining whether a data type is capability-dependent if the conditions in the first step are not met. The classifier checks if two conditions are simultaneously met: Condition 1 is whether the value of B is greater than the value of D; Condition 2 is whether the value of A is greater than 4.5 times / minute. If both conditions are met, the subtype label is output as capability-dependent. For example, if B = 68 frames and D = 40 frames, then B > D is true; A = 5.2 times / minute > 4.5, thus it is determined to be capability-dependent.

[0043] The third step involves determining whether a classifier belongs to the co-buffered type if the conditions in the first two steps are not met. The classifier checks whether two conditions are simultaneously met: Condition 1 is whether the value of E is greater than or equal to 0.82; Condition 2 is whether the value of R is within the second numerical interval, which is defined as greater than or equal to 0.7 and less than or equal to 1.1. Note that the maximum value of the second numerical interval, 1.1, is strictly less than the minimum value of the first numerical interval, 1.5, to ensure that the two intervals do not overlap. If both conditions are met, the subtype label is output as co-buffered. For example, if E=0.88, A=4 times / minute, C=3.6 times / minute, then R=0.9, falling within [0.7, 1.1], and is determined to be co-buffered.

[0044] Fourth, if none of the above three conditions are met, the classifier outputs a default label of mixed unclassified. At this point, the workstation will mark the case as awaiting manual review, and the operator will manually specify the closest subtype.

[0045] After making the determination, the classifier sends the subtype label, such as capability dependency, as a string to the training scheme retrieval module. The entire process does not involve any iterative optimization or parameter learning; it operates solely based on fixed numerical comparison rules.

[0046] Furthermore, in step 4, the core capability threshold corresponding to the resource depletion subtype is the first time threshold of the zygomaticus major muscle activation duration, and the corresponding resource replenishment anchor point is the social support dimension enhancement module. The core competency threshold corresponding to the competency-dependent subtype is the first frequency threshold of the orbicularis oculi muscle activation frequency, and the corresponding resource supply anchor is the self-efficacy dimension construction module. The core capability threshold corresponding to the collaborative buffering subtype is the symmetry threshold of the activation symmetry of the left and right faces, and the corresponding resource supply anchor is the altruism dimension expansion module.

[0047] In this embodiment, the pre-built training scheme library is stored as a directory in a designated folder on the local hard drive. This folder contains three subfolders named type_R, type_C, and type_B, corresponding to resource-intensive, capability-dependent, and collaborative buffering types, respectively.

[0048] After the workstation receives the subtype labels output by the classifier, it performs the following lookup and mapping operation: If the tag is resource-consuming, open the `type_R` folder on the workstation. Within this folder, there is a text file named `threshold_R.txt`, which stores only one value, 55. This value represents the first time threshold for the duration of zygomaticus major muscle activation. The folder also contains a presentation file named `module_social.pptx`, which is the social support dimension enhancement module. The presentation content is as follows: the first page explains how to recognize colleagues' emotional distress signals; the second page lists the dialing procedure for the hospital's internal psychological assistance hotline; and the third page showcases interactive games designed to promote trust during team-building activities.

[0049] If the tag is competency-dependent, open the type_C folder on the workstation. The threshold_C.txt file stores the value 4.8, which is the first frequency threshold unit for orbicularis oculi muscle activation frequency: times / minute. The corresponding module file is module_self_efficacy.pptx, which contains: the first page lists illustrated reviews of successful small-scale disaster care cases; the second page provides a template for completing a self-efficacy self-assessment scale; and the third page provides SMART principle cards for setting progressive skill mastery goals.

[0050] If the tag is collaborative buffering, open the type_B folder on the workstation. The threshold_B.txt file stores the value 0.80, which is the symmetry threshold for left and right facial activation symmetry. The corresponding module file is module_altruism.pptx, which contains: the first page shows a collection of real photos of elderly disaster victims expressing their gratitude after being helped by volunteer rescuers; the second page lists activity planning templates for conducting disaster prevention and science popularization campaigns in the community; and the third page provides a list of guiding questions for writing a rescue diary to strengthen altruistic motivation.

[0051] The workstation then reads the selected threshold file and module file into memory together, preparing for the superposition and comparison in step 5; the values ​​in the threshold file are directly compared with the corresponding parameters measured in step 2, while the module file serves as the backbone of the basic training framework.

[0052] Furthermore, the image overlay technique in step 5 includes: The enhanced training identifier is dynamically attached above the graphic explanation area corresponding to the resource supply anchor point in the basic training framework in the form of a highlighted color block that is different from the basic training framework. The display duration of the highlighted color block is positively correlated with the amount by which the activation duration feature exceeds the core capability threshold.

[0053] In this embodiment, the image overlay technique is implemented as follows: The workstation first parses the basic training framework, i.e., the module file retrieved in step 4, such as the internal structure of module_self_efficacy.pptx; after the presentation file is opened, the workstation traverses all shape objects on each slide and looks for shapes whose alternative text properties contain specific keyword anchors.

[0054] In this example, in the module file corresponding to the competency-dependent type, there is a rounded rectangle text box on the second slide, whose alternative text attribute value is anchor_self-efficacy self-assessment; the workstation records the position coordinates of this text box on the entire slide canvas, specifically the top left corner X=120 pixels, Y=200 pixels, and the bottom right corner X=480 pixels, Y=320 pixels.

[0055] The workstation then calculates the excess; the value of the zygomaticus major muscle activation duration exceeding the core capability threshold measured in step 2 is 65 frames minus 50 frames, which equals 15 frames; the workstation uses these 15 frames as the excess value.

[0056] The workstation then maps the excess value of 15 to a display duration range; the predefined mapping rule is: for every additional frame of excess, the display duration increases by 0.2 seconds, with a minimum display duration of 2 seconds and a maximum of 10 seconds. Therefore, 15 frames correspond to a display duration of 2 + 15 × 0.2 = 5 seconds.

[0057] The workstation then generates reinforcement training labels; in the lower right corner of the target text box, a circular color block with a width of 40 pixels and a height of 40 pixels is created; the fill color of this color block is set to a golden yellow with RGB values ​​(255, 200, 0), which is significantly different from the default blue background of the slides; the color block does not contain any text or icons.

[0058] The workstation exports the slideshow page containing the golden circular color block as a separate temporary image file; in the next 5 seconds, the workstation sends a command to the connected monitor to redraw the temporary image file at a refresh rate of 10 times per second to keep it highlighted; after 5 seconds, the workstation resumes displaying the original slideshow page without the color block.

[0059] If the excess is greater, such as 25 frames, then the mapped display duration is 2 + 25 × 0.2 = 7 seconds, and the golden circular color block will remain on the screen for 7 seconds before disappearing.

[0060] Furthermore, referring to Figure 2 The specific process of generating reinforcement training labels on the basic training framework using image overlay technology in step 5 is as follows: The page layout structure of the basic training framework is analyzed to identify the target layer where the resource supply anchor point is located and the coordinate boundary of the target layer. Obtain the difference between the activation duration feature and the core capability threshold, and generate the pixel density of the reinforcement training label that matches the difference according to the preset difference-label density lookup table; While maintaining the original background texture and text content of the target layer as recognizable, semi-transparent warning dots are rendered in a random distribution manner according to the pixel density within the coordinate boundary of the target layer to form an enhanced training label. The target layer carrying the reinforcement training identifier is re-integrated with other layers of the basic training framework to generate a visual preview image of the classification training scheme.

[0061] In this embodiment, the detailed process of generating reinforcement training labels is as follows: The workstation first calls a local image parsing library, which can read the layer composition of each slide in the presentation file. For the second slide of the capability-dependent module, the parsing library reports that the page contains three independent layers: a bottom white background layer, a middle layer of text and shape content, and a top empty transparent placeholder layer.

[0062] The workstation locates the shape object marked as an anchor point in the middle layer and determines its coordinate boundary to be from pixel (120, 200) to (480, 320); this rectangular area has a width of 360 pixels and a height of 120 pixels. The workstation sets this rectangular area as the target area.

[0063] The workstation obtains the excess value of 15 frames from step 2. The workstation internally stores a difference-identifier density lookup table, which records the mapping relationship of three levels in tabular form: the excess of 1-10 frames corresponds to an identifier density of 1 dot per square centimeter; the excess of 11-20 frames corresponds to 3 dots per square centimeter; the excess of 21 frames or more corresponds to 6 dots per square centimeter. Frame 15 belongs to the second level, so the target density is 3 dots per square centimeter.

[0064] The workstation calculates the area of ​​the target region: the rectangle is 360 pixels wide and 120 pixels high. Assuming the physical resolution of the monitor is 40 pixels per centimeter, the width is 9 centimeters, the height is 3 centimeters, and the area is 27 square centimeters. The total number of dots required is 27 multiplied by 3, which equals 81 dots.

[0065] The workstation generates 81 semi-transparent warning dots. The generation process for each dot is as follows: Two integers are generated using a pseudo-random number generator. The first integer is between 120 and 479, and the second integer is between 200 and 319. These integers serve as the X and Y coordinates of the center point of the circle. The radius of the circle is fixed at 3 pixels. The fill color of the circle is red, and the opacity is set to 75%, which is equivalent to an Alpha channel value of 191, ranging from 0 to 255. The randomly generated center coordinates may overlap, but the system does not perform deduplication, allowing the circles to naturally cluster within the target area, forming a randomly scattered appearance.

[0066] The workstation draws the 81 generated dots on the top transparent placeholder layer; then, it merges the bottom background layer, the middle text and image content layer, and the top layer with dots in order from bottom to top to generate a new bitmap image; this bitmap image is the preview image of the classification training program with reinforcement training labels; the workstation saves the preview image as a PNG file, with the filename containing the test nurse's ID and the current timestamp.

[0067] Furthermore, the head-mounted display has a built-in eye-tracking module, and step 1 further includes: Simultaneously record the eye movement trajectory of clinical nurses during the stress presentation of holographic images at the disaster site, the eye movement trajectory including the coordinates of the fixation hot zone and the pupil diameter change curve; The micro-expression spatiotemporal feature set in step 2 also includes the stress triggering moment defined by the inflection point of the slope on the pupil diameter change curve. This stress triggering moment is used for temporal alignment verification with the activation start frame of the zygomaticus major muscle.

[0068] In this embodiment, in addition to displaying holographic images, the head-mounted display also integrates two small infrared eye-tracking cameras inside its lenses, which are aimed at the nurse's left and right eyes respectively; these cameras record eye movements and pupil diameter changes at a rate of 60 frames per second.

[0069] While presenting the three gradient pressure stages described in claim 2, the eye-tracking module records two sets of data in real time. The first set is the gaze hotspot coordinates. The system divides the display's field of view into a 10×10 grid, with each grid cell representing a hotspot cell. Whenever the nurse's gaze lingers within a grid cell for more than 0.1 seconds, the counter for that cell is incremented by 1. After the video playback ends, the system outputs a 100×100 hotspot matrix, where the value of each element represents the cumulative duration of gaze at the corresponding grid. The second set is the pupil diameter variation curve. The system records the diameter values ​​of the left and right pupils in millimeters for each frame, forming a time-series curve.

[0070] The workstation inputs the curve into a turning point detection module. This module iterates through each point on the curve, calculating the average slope of the five adjacent points to the left and the average slope of the five adjacent points to the right of that point. When the difference between the average slopes on the left and right sides exceeds a preset slope change threshold, the point is marked as a turning point. The system further checks the direction of slope change at the turning point: if the slope changes from positive to negative, the turning point is defined as the stress trigger moment; if the slope changes from negative to positive, it is defined as the stress release moment. In this embodiment, when the video plays to the 35th second, a clear turning point appears on the pupil diameter curve, changing from rapid expansion to slow contraction. The timestamp 35.2 seconds corresponding to this point is recorded as the stress trigger moment.

[0071] The workstation appends the original sequence of the gaze hot zone coordinate matrix, the pupil diameter change curve, and the timestamp of the stress trigger moment to the micro-expression spatiotemporal feature set generated in step 2. At this time, the spatiotemporal feature set not only contains the parameters of the zygomaticus major muscle and the orbicularis oculi muscle, but also the eye movement parameters.

[0072] Furthermore, a timing alignment verification step is included before step 5: Calculate the time offset between the stress triggering time and the zygomaticus major muscle activation start frame. If the time offset exceeds the preset offset threshold, the current facial micro-expression image sequence is removed, and a re-acquisition command is triggered, instructing step 1 to reacquire the facial micro-expression image sequence of the target clinical nurse. The disaster resilience classification training scheme generated in step 5 is a multi-layer interactive image file. The first layer of the multi-layer interactive image file is a carrier frame layer, which is used to display the complete content of the basic training framework. The second layer is the identifier layer, which is used to carry reinforcement training identifiers; The third layer is the interactive response layer, which is used to capture the click operation of clinical nurses on the reinforcement training icon, and pop up a micro-expression playback window associated with the reinforcement training icon when clicked. The micro-expression playback window is used to loop the original facial micro-expression image sequence fragments collected in step 1 that correspond to the reinforcement training icon.

[0073] In this embodiment, the workstation first executes a timing alignment verification subroutine. This subroutine reads two key time points from the extended micro-expression spatiotemporal feature set generated in claim 8: the first is the stress trigger moment recorded in claim 8, for example, 35.2 seconds; the second is the timestamp corresponding to the starting frame of the first rapid deformation phase of the zygomaticus major muscle determined in claim 3. Assuming the starting frame is frame number 1056 in the sequence and the frame rate is 30fps, the corresponding timestamp is 1056 / 30 = 35.2 seconds. The time offset is calculated as 35.2 seconds minus 35.2 seconds equals 0 seconds.

[0074] The workstation's internal preset offset threshold is 0.05 seconds; since 0 seconds is less than 0.05 seconds, the timing alignment check passes, and the workstation continues to execute the subsequent step 5.

[0075] To illustrate the abnormal situation, consider another scenario: the stress trigger time is 35.2 seconds, while the timestamp corresponding to the zygomaticus major muscle activation start frame is 35.8 seconds, resulting in an offset of 0.6 seconds. The workstation determines that 0.6 seconds is greater than 0.05 seconds, causing the verification to fail. At this point, the workstation immediately performs the following operations: First, it clears all facial micro-expression image sequences and eye-tracking data currently stored in temporary memory; second, it records a timing alignment failure event in the log file, noting the offset of 0.6 seconds; third, it sends a re-acquisition command to the control interface of the head-mounted display, which includes the parameter reset=1. Upon receiving the command, the head-mounted display automatically replays the standardized disaster simulation holographic video described in claim 2 from the beginning and starts acquiring facial images and eye-tracking data again, overwriting the previously failed data. The workstation then repeats this process until the timing alignment verification passes, or terminates and reports an error after reaching the maximum number of retries (3).

[0076] The workstation generates a file format that is not a typical single-layer image, but a multi-layered interactive file in PDF format with embedded JavaScript. Each page of this file is an independent object, and the page stacking structure is as follows: The first layer is the carrier framework layer. This layer is directly derived from the basic training framework module called in claim 5, such as the page converted from module_self_efficacy.pptx. This layer presents complete training text content and static charts, which are visible and readable by the user. This layer is set to be non-editable, but can respond to mouse hover and click events.

[0077] The second layer is the identifier layer. This layer is located above the carrier frame layer, but its content is initially completely transparent. This layer stores the position coordinates and radius information of all the semi-transparent warning dots generated in claim 7. When the user moves the mouse pointer to the range of the anchor point area in the carrier frame layer, the second layer makes the dots at the corresponding coordinates visible, that is, the transparency is reduced from 75% to 0%, while keeping the dots in other areas invisible.

[0078] The third layer is the interaction response layer. This layer is a transparent layer that does not contain any visual elements, but registers a click event listener for each semi-transparent alert dot. When the user clicks the position corresponding to a dot on the screen, the interaction response layer captures the event and queries the feature data segment identifier corresponding to that dot. Each dot is assigned a unique ID when it is generated, and this ID is linked to a specific segment in the original facial micro-expression image sequence acquired in step 1 of claim 1.

[0079] Specifically, for capability-dependent solutions, clicking the dot will locate the original frame sequence of the micro-expression event most relevant to the 15 frames exceeding the limit. This segment begins 15 frames before the start frame of the rapid deformation phase of the event and ends 15 frames after the end frame of the slow recovery phase. The workstation creates a new pop-up window, 320×240 pixels in size, located in the upper right corner of the main window. A video player is embedded within the pop-up window, automatically looping the extracted frame sequence segment at the original frame rate. During playback, a timestamp and frame number are overlaid in the lower left corner of the video screen. The user can stop the loop and close the window by clicking the close button in the upper right corner of the pop-up window.

[0080] The multi-layered interactive image file is ultimately saved as a single PDF file with the extension .interactive.pdf. Users can experience the aforementioned click-to-play function by opening the file with a JavaScript-enabled PDF reader such as Adobe Acrobat Reader DC.

[0081] According to a second embodiment of the present invention, referring to Figure 3 This invention claims protection for a system for constructing a training program on disaster resilience classification for clinical nurses, comprising: One or more processors; A memory storing one or more programs, which, when executed by one or more processors, enable the one or more processors to implement the method for constructing a clinical nurse disaster resilience classification training program.

[0082] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, or indirect coupling or communication connection between apparatuses or units, and may be electrical, mechanical, or other forms.

[0083] Furthermore, the functional units in the various embodiments of this application 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 units described above can be implemented in hardware or as software functional units. The above are merely embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made based on the description and drawings of this application, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

[0084] The specific embodiments of the invention have been described in detail above, but they are only examples, and this application is not limited to the specific embodiments described above. For those skilled in the art, any equivalent modifications or substitutions to the invention are also within the scope of this application. Therefore, all equivalent changes, modifications, and improvements made without departing from the spirit and principles of this application should be covered within the scope of this application.

Claims

1. A method for constructing a training program for clinical nurses' disaster resilience classification, characterized in that, Includes the following steps: Step 1: Obtain facial micro-expression image sequences of the target clinical nurse in a standardized disaster simulation scenario; Step 2: Use image recognition technology to extract features from the facial micro-expression image sequence to generate a micro-expression spatiotemporal feature set that includes at least the activation frequency, activation duration, and activation symmetry of the left and right sides of the face of the zygomaticus major and orbicularis oculi muscles. Step 3: Input the micro-expression spatiotemporal feature set into a preset resilience classifier. The resilience classifier outputs the disaster resilience subtype label of the target clinical nurse based on the mapping relationship between the spatiotemporal features of micro-expressions and disaster resilience subtypes. The disaster resilience subtype label includes at least resource depletion type, capability dependence type and collaborative buffer type. Step 4: Based on the disaster resilience subtype label, retrieve the corresponding basic training framework from the pre-built training scheme library. The basic training framework includes core capability thresholds and resource replenishment anchors. Step 5: Compare the basic training framework with the activation duration features in the micro-expression spatiotemporal feature set. When the activation duration features exceed the core capability threshold, use image overlay technology to generate reinforcement training labels on the basic training framework to form the disaster resilience classification training program for the target clinical nurse.

2. The method for constructing a clinical nurse disaster resilience classification training program according to claim 1, characterized in that, The standardized disaster simulation scenario in step 1 is as follows: In a controlled lighting environment, a holographic image of a disaster scene with gradient pressure is presented to clinical nurses through a head-mounted display. The gradient pressure includes the chaotic audio-visual stimulation in the early stage of the disaster, the visual impact of the mass arrival of the wounded in the middle stage of the disaster, and the static image of resource scarcity in the later stage of the disaster.

3. The method for constructing a clinical nurse disaster resilience classification training program according to claim 2, characterized in that, Step 2 includes: The continuous frame difference method is used to locate the start and end frames of facial micro-expression changes, and a rapid deformation stage and a slow recovery stage are divided between the start and end frames. The activation frequency of the zygomaticus major and orbicularis oculi muscles is the number of times the rapid deformation phase occurs per unit time, and the activation duration is the time from the start of a single rapid deformation phase to the end of the slow recovery phase.

4. The method for constructing a clinical nurse disaster resilience classification training program according to claim 3, characterized in that, The resilience classifier in step 3 establishes a mapping relationship between micro-expression spatiotemporal features and disaster resilience subtypes based on the following logic: Obtain the original parameters of zygomaticus major muscle activation frequency, zygomaticus major muscle activation duration, orbicularis oculi muscle activation frequency, orbicularis oculi muscle activation duration, and left and right facial activation symmetry from the micro-expression spatiotemporal feature set. When the ratio of the activation frequency of the orbicularis oculi muscle to the activation frequency of the zygomaticus major muscle is in the first range and the activation symmetry of the left and right faces is lower than the preset asymmetry threshold, the disaster resilience subtype label is determined to be resource depletion type. When the activation duration of the zygomaticus major muscle is greater than that of the orbicularis oculi muscle, and the activation frequency of the zygomaticus major muscle is higher than the preset high-frequency threshold, the disaster resilience subtype label is determined to be capability-dependent. When the activation symmetry of the left and right sides of the face is higher than the preset symmetry threshold, and the ratio of the activation frequency of the orbicularis oculi muscle to the activation frequency of the zygomaticus major muscle is in the second numerical range, the disaster resilience subtype label is determined to be the collaborative buffer type. The minimum value in the first numerical interval is greater than the maximum value in the second numerical interval.

5. The method for constructing a clinical nurse disaster resilience classification training program according to claim 1, characterized in that, In step 4, the core capability threshold corresponding to the resource depletion subtype is the first time threshold of the zygomaticus major muscle activation duration, and the corresponding resource replenishment anchor point is the social support dimension enhancement module. The core competency threshold corresponding to the competency-dependent subtype is the first frequency threshold of the orbicularis oculi muscle activation frequency, and the corresponding resource supply anchor is the self-efficacy dimension construction module. The core capability threshold corresponding to the collaborative buffering subtype is the symmetry threshold of the activation symmetry of the left and right faces, and the corresponding resource supply anchor is the altruism dimension expansion module.

6. The method for constructing a clinical nurse disaster resilience classification training program according to claim 1, characterized in that, The image overlay technique in step 5 includes: The enhanced training identifier is dynamically attached above the graphic explanation area corresponding to the resource supply anchor point in the basic training framework in the form of a highlighted color block that is different from the basic training framework. The display duration of the highlighted color block is positively correlated with the amount by which the activation duration feature exceeds the core capability threshold.

7. The method for constructing a clinical nurse disaster resilience classification training program according to claim 1, characterized in that, The specific process of generating reinforcement training labels on the basic training framework using image overlay technology in step 5 is as follows: The page layout structure of the basic training framework is analyzed to identify the target layer where the resource supply anchor point is located and the coordinate boundary of the target layer. Obtain the difference between the activation duration feature and the core capability threshold, and generate the pixel density of the reinforcement training label that matches the difference according to the preset difference-label density lookup table; While maintaining the original background texture and text content of the target layer as recognizable, semi-transparent warning dots are rendered in a random distribution manner according to the pixel density within the coordinate boundary of the target layer to form an enhanced training label. The target layer carrying reinforcement training identifiers is re-integrated with other layers in the basic training framework to generate a visual preview image of the classification training scheme.

8. The method for constructing a clinical nurse disaster resilience classification training program according to claim 2, characterized in that, The head-mounted display has a built-in eye-tracking module, and step 1 further includes: Simultaneously record the eye movement trajectory of clinical nurses during the stress presentation of holographic images at the disaster site, the eye movement trajectory including the coordinates of the fixation hot zone and the pupil diameter change curve; The micro-expression spatiotemporal feature set in step 2 also includes the stress triggering moment defined by the inflection point of the slope on the pupil diameter change curve. This stress triggering moment is used for temporal alignment verification with the activation start frame of the zygomaticus major muscle.

9. The method for constructing a clinical nurse disaster resilience classification training program according to claim 8, characterized in that, The step 5 is preceded by a timing alignment verification step: Calculate the time offset between the stress triggering time and the zygomaticus major muscle activation start frame. If the time offset exceeds the preset offset threshold, the current facial micro-expression image sequence is removed, and a re-acquisition command is triggered, instructing step 1 to reacquire the facial micro-expression image sequence of the target clinical nurse. The disaster resilience classification training scheme generated in step 5 is a multi-layer interactive image file. The first layer of the multi-layer interactive image file is a carrier frame layer, which is used to display the complete content of the basic training framework. The second layer is the identifier layer, which is used to carry reinforcement training identifiers; The third layer is the interactive response layer, which is used to capture the click operation of clinical nurses on the reinforcement training icon, and pop up a micro-expression playback window associated with the reinforcement training icon when clicked. The micro-expression playback window is used to loop the original facial micro-expression image sequence fragments collected in step 1 that correspond to the reinforcement training icon.

10. A system for constructing a training program on disaster resilience classification for clinical nurses, characterized in that: include: One or more processors; A memory having stored one or more programs that, when executed by one or more processors, cause the one or more processors to implement the method for constructing a clinical nurse disaster resilience classification training program according to any one of claims 1 to 9.

Images