Respiratory compensation warning method and system based on multi-mode behavior evolution recognition

Abstract

This invention provides a respiratory compensation early warning method and system based on multi-modal behavioral evolution recognition, comprising: continuously acquiring patient behavioral posture data; analyzing the behavioral posture data based on a predefined behavioral evolution knowledge base to identify whether there is a type I posture or a type II posture, wherein the type I posture is a high-frequency non-specific posture that exceeds the individual baseline, and the type II posture is a specific compensatory posture that conforms to a preset specific physiological definition; analyzing the temporal correlation between type I and type II postures, and generating a high-confidence early warning signal if an evolution from type I posture to type II posture is identified. This invention achieves a digital closed loop of clinical cognition: for the first time, it transforms the complete clinical cognition of medical staff—"the patient first becomes restless, then becomes short of breath and sits up"—into a calculable and predictable technical model, which is highly consistent with clinical thinking and easily trusted.

Description

Technical Field

[0001] This invention belongs to the field of respiratory compensation early warning technology, and specifically relates to a respiratory compensation early warning method and system based on multi-modal behavior evolution recognition. Background Technology

[0002] Acute respiratory compensation is an early clinical precursor shared by many acute and critical illnesses (such as acute exacerbation of chronic obstructive pulmonary disease, acute exacerbation of heart failure, acute heart failure, acute upper airway obstruction, severe pulmonary infection, and postoperative airway compression after oral and maxillofacial surgery). If not detected and intervened in a timely manner, it can rapidly progress to respiratory failure and is one of the leading causes of death in the perioperative period and intensive care. Current monitoring technologies have significant limitations: Delayed response: Monitoring systems that use pulse oximetry as the core indicator only issue an alarm when oxygenation has already been impaired, missing the critical period for physiological compensation.

[0003] One-sided perception: The existing concept of "patient monitoring" is limited to vital signs and fall monitoring, completely ignoring the interpretation of the patient's behavioral intentions. Early behavioral changes caused by breathing difficulties (such as restlessness and trying different postures) are classified as "agitation," confused with pain and discomfort, and cannot form an effective early warning. Summary of the Invention

[0004] The purpose of this invention is to provide a respiratory compensation early warning method and system based on multi-modal behavior evolution recognition, so as to solve the above-mentioned problems.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: In a first aspect, the present invention provides a respiratory compensation early warning method based on multi-modal behavioral evolution recognition, comprising: Continuously acquire patient behavior and posture data; Based on a predefined behavioral evolution knowledge base, behavioral posture data is analyzed to identify whether there is a type I or type II posture. The type I posture is a high-frequency non-specific posture that exceeds the individual baseline, and the type II posture is a specific compensatory posture that conforms to a pre-defined specific physiological definition. Analyze the temporal correlation between Category I and Category II body positions. If the evolution from Category I to Category II body positions is identified, a high-confidence warning signal is generated.

[0006] Furthermore, the continuous acquisition of the patient's behavioral posture data includes: By deploying a depth vision sensor above the hospital bed, along with a pressure distribution sensing mattress and an environmental audio sensor, the system simultaneously collects high-precision three-dimensional skeletal joint sequence, body pressure distribution, respiratory micro-movements, and environmental acoustic characteristics of the patient.

[0007] Furthermore, the predefined behavior evolution knowledge base includes: Construct a behavioral semantic understanding model that includes two levels and one association logic; The first level is the exploratory behavior of the precursor: unable to remain still, and engaging in high-frequency, non-purposeful rolling or twisting movements beyond their personal baseline within a limited set of body positions; The second level is specific compensatory posture: a steady-state posture with clear respiratory physiological significance; Association logic: Once a first-level abnormal activity state is detected, if the patient is further identified as stably entering and maintaining one or more second-level specific compensatory positions within a preset specific time window, then it is determined that a complete evolutionary path from exploration to compensation has been completed.

[0008] Furthermore, the quantification and identification of this type of body position: Calculate the real-time position switching frequency and compare it with the individualized baseline. When it exceeds twice the baseline standard deviation and persists, it is marked as an abnormal activity state. Analysis of switching modes: Identifying whether it is an invalid exploration loop; Correlation analysis: Simultaneously analyze whether there is a progressive upward trend in respiratory rate during this period, or whether there are brief attempts at second-level compensatory postures.

[0009] Furthermore, the quantification and identification of the two types of body positions: Upper airway open position: Panting position: Neck extension angle >15° and duration >30 seconds, and associated with the detection of open lips; Flower-smelling position: mandibular-sternal angle <20°, and duration >30 seconds; Positions that activate accessory respiratory muscles: Tripod position: 45° < trunk forward tilt angle < 80°, elbow pressure > threshold, and joint movement and trunk angle stability coupling during breathing; Body positions that reduce cardiopulmonary load: Sitting breathing position: The angle between the torso and the horizontal plane is >60°, which needs to be maintained by back support, and both lower limbs are drooping, and the duration is >60 seconds; Local upper airway obstruction caused by intraoral mass: Its specific postural manifestation is "forced head and neck tilting position", that is, forced lateral lying or head tilting to the affected side. Precursor patterns of decompensation: If a patient experiences orthopnea lasting longer than 60 seconds, accompanied by wheezing sounds, it is considered that the patient has developed acute heart failure. The patient exhibits a "forced head and neck tilt position," and after a short period of changing position, quickly returns to the forced position, accompanied by short-term wheezing-like breathing sounds. After excluding the two types of respiratory decompensation patterns mentioned above, if within the time window, N≥3 of the above-mentioned upper airway open position and the four compensatory positions included in the accessory respiratory muscle activation position alternately, and the average duration of each is <45 seconds, then it is considered that the patient has problems such as acute exacerbation of chronic obstructive pulmonary disease, acute upper airway obstruction, severe pulmonary infection, or airway compression after oral and maxillofacial surgery.

[0010] Furthermore, the analysis of the temporal correlation between type I and type II body positions includes: Two-level parallel identification and association analysis: Parallel identification of first-level activity abnormalities and second-level specific compensatory postures; determination of temporal and logical correlation between the two types of behaviors through association logic.

[0011] Furthermore, if a shift from a type I body position to a type II body position is detected, a high-confidence warning signal is generated, including: Level 1 indication: Only level 1 activity abnormalities were detected, but there was no association with respiratory deterioration or a trend of progression, so mild labeling was performed; Level 2 warning: Meets any of the following conditions: a) Identify any second-level specific compensatory posture that persists for more than the threshold time; b) A strong correlation was identified between first-level activity abnormalities and a progressive increase in respiratory rate; c) The decompensated respiratory positional circulation pattern is identified; a clear pop-up window at the nurses' station and a push notification on the nurse's handheld terminal are triggered; Level 3 Alarm: The complete evolutionary path from exploration to compensation is identified, or the patient's vital signs deteriorate early in the compensatory position, triggering an audible and visual alarm and broadcast. The visitation mode is identified by skeletal counting. In this mode, the threshold for the second-level body position recognition is automatically increased, and the evolution path logic is emphasized to filter out interference from social activities.

[0012] Secondly, the present invention provides a respiratory compensation early warning system based on multi-modal behavioral evolution recognition, comprising: The data acquisition module is used to continuously acquire patients' behavioral and posture data; The identification module is used to analyze behavioral posture data based on a predefined behavioral evolution knowledge base and identify whether there is a type I posture or a type II posture. The type I posture is a high-frequency non-specific posture that exceeds the individual baseline, and the type II posture is a specific compensatory posture that conforms to a pre-defined specific physiological definition. The early warning module is used to analyze the temporal correlation between Category I and Category II body positions. If it detects an evolution from Category I to Category II body position, it generates a high-confidence early warning signal.

[0013] Thirdly, the present invention provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the respiratory compensation early warning method based on multi-modal behavior evolution recognition.

[0014] Fourthly, the present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the respiratory compensation early warning method based on multi-modal behavior evolution recognition.

[0015] Compared with the prior art, the present invention has the following technical effects: This invention realizes a digital closed loop of clinical cognition: for the first time, it transforms the complete clinical cognition of medical staff, such as "the patient is first agitated and then sits up with shortness of breath," into a calculable and predictable technical model that is highly consistent with clinical thinking and is easy to trust.

[0016] This invention identifies "precursor exploratory behavior" and its association with subsequent specific compensation, allowing for an earlier warning time compared to simply identifying fixed body positions. Furthermore, due to the existence of behavioral evolution logic, the false positive rate of the warning is significantly reduced.

[0017] The "from non-specific to specific" evolutionary identification framework of this invention and its constructed extended behavioral knowledge base can cover early warning of acute respiratory compensation caused by various etiologies such as pulmonary, cardiac, and upper airway obstruction, and have good clinical applicability. This framework can also be transferred to the monitoring of other clinical scenarios with prodromal behavioral manifestations such as pain and delirium. Attached Figure Description

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

[0019] The present invention will be further described below with reference to the accompanying drawings: Example 1: This invention provides a respiratory compensation early warning method based on multi-modal behavioral evolution recognition, comprising: Continuously acquire patient behavior and posture data; Based on a predefined behavioral evolution knowledge base, behavioral posture data is analyzed to identify whether there is a type I or type II posture. The type I posture is a high-frequency non-specific posture that exceeds the individual baseline, and the type II posture is a specific compensatory posture that conforms to a pre-defined specific physiological definition. Analyze the temporal correlation between Category I and Category II body positions. If the evolution from Category I to Category II body positions is identified, a high-confidence warning signal is generated.

[0020] This invention provides a method that can identify the complete evolution pattern of a patient's respiratory compensation behavior without contact and intelligently, and issue graded warnings before physiological decompensation occurs, thereby significantly advancing the risk intervention point.

[0021] Example 2: This invention provides a respiratory compensation early warning method based on multi-modal behavior evolution recognition, comprising: S1. Continuously acquire patient behavioral and posture data; S2. Based on a predefined behavioral evolution knowledge base, analyze the data to identify whether: A) high-frequency non-specific body position switching exceeds the individual baseline; and / or B) specific compensatory body positions conforming to a specific physiological definition; S3. Analyze the temporal relationship between A) and B). If a pattern of evolution from A) to B) is identified, a high-confidence warning signal is generated.

[0022] In step S2, identifying “high-frequency non-specific body position switching” includes: establishing a baseline for the frequency of body position switching during the patient’s resting period, and marking it when the real-time frequency continuously exceeds a preset standard deviation multiple of the baseline.

[0023] Further analysis was conducted on the marked "high-frequency non-specific body position switching" to determine whether it was associated with the upward trend of the patient's respiratory rate during the same period.

[0024] The “specific compensatory positions” include: a steady-state position with excessive cervical flexion or extension to open the upper airway, and / or a steady-state position with the trunk leaning forward and supported by the elbows to activate accessory respiratory muscles, and / or a forced sitting position to reduce cardiopulmonary load (such as acute left ventricular failure). And / or a “forced head and neck tilt position” to relieve local upper airway obstruction caused by intraoral masses.

[0025] To identify "forced sitting posture", the following conditions must be met: the angle between the trunk and the horizontal plane is consistently greater than the fourth threshold (e.g., 60°), the patient needs to rely on back support, and the patient often has a drooping posture with both lower limbs.

[0026] To identify "forced head and neck tilting position", the following conditions must be met: continuous lateral lying or tilting the head to the affected side and changing position for less than 60 seconds, followed by a rapid return to the original forced position and a short period of wheezing breathing sounds. To identify a "steady posture with excessive cervical spine extension", the following conditions must be met: the angle between the line connecting the cervical spine and the vertical plane must be continuously greater than the first threshold, and the open lips must be detected simultaneously.

[0027] To identify a "steady body position with the trunk leaning forward and the elbows supporting the body", the following conditions must be met: the angle between the trunk axis and the bed surface is within the second threshold range, and the pressure value in the elbow area is consistently greater than the third threshold.

[0028] The "evolution from A) to B) pattern" is defined as follows: a sustained state of A) is detected first in the first time window, and then the patient is detected to enter and maintain state B) in the second time window.

[0029] Step S2 also includes identifying the “decompensated respiratory position cycle” pattern, which is that the patient rapidly switches between multiple specific compensatory positions and is unable to maintain any position for more than a fourth threshold time.

[0030] Example 3: This invention provides a respiratory compensation early warning method based on multi-modal behavior evolution recognition, comprising: S1. Continuously acquire patient behavioral and posture data; S2. Based on a predefined behavioral evolution knowledge base, analyze the data to identify whether: A) high-frequency non-specific body position switching exceeds the individual baseline; and / or B) specific compensatory body positions conforming to a specific physiological definition; S3. Analyze the temporal relationship between A) and B). If a pattern of evolution from A) to B) is identified, a high-confidence warning signal is generated.

[0031] A depth vision sensor deployed above the hospital bed, along with an optional pressure distribution sensing mattress and environmental audio sensors, is used to simultaneously acquire high-precision three-dimensional skeletal joint sequence, body pressure distribution, respiratory-related micro-movements, and environmental acoustic characteristics of the patient.

[0032] Preferably, a tool for capturing breath sounds is attached above the trachea, which can convert breath sounds into electrical signals. When the patient experiences difficulty breathing, such as wheezing, specific wheezing-like sound waves can be visualized and displayed, which can be corroborated with changes in body position to enhance specificity.

[0033] Core Computing and Intelligent Analysis Module: This serves as the system's brain, including: Data fusion and spatiotemporal alignment unit: Synchronizes and calibrates multi-source heterogeneous data.

[0034] Personalized baseline learning unit: Establishes personalized behavioral and postural baseline models for patients during the stable period.

[0035] Multi-level behavioral semantic understanding engine: The core of this invention, which includes a predefined knowledge base of respiratory compensation behavior evolution and corresponding pattern recognition algorithms.

[0036] Dynamic risk assessment and decision-making unit: Integrates the output of multiple engine levels, calculates real-time risk index and generates early warning instructions.

[0037] Tiered early warning and data service module: Enables comprehensive early warning from local audio and visual prompts to push notifications from remote medical terminals, and provides complete digital medical record archiving.

[0038] A behavioral semantic understanding model with two levels and one association logic was constructed: Level 1: Proactive exploratory behavior – a non-specific but related “high-frequency positional switching” pattern Definition: refers to the patient's inability to remain still and to perform frequent, unintentional turning or twisting movements beyond their personal baseline within a limited set of positions.

[0039] Quantization and recognition methods: Calculate the real-time positional switching frequency and compare it with the individualized baseline. When it exceeds twice the baseline standard deviation and persists, it is marked as "abnormal activity state".

[0040] Analyze switching modes: Identify whether it is an invalid exploration cycle (such as rapidly cycling between supine, left lateral, and right lateral positions without finding a stable and comfortable position).

[0041] Correlation analysis: Simultaneously analyze whether there is a progressive upward trend in respiratory rate during this period, or whether there are brief attempts at second-level compensatory postures.

[0042] Level 2: Specific compensatory postures – steady-state postures with clear respiratory physiological significance. In this embodiment, the specific compensatory posture knowledge base is constructed according to pathophysiological purposes, including but not limited to: (1) Upper airway open position: Panting position: cervical spine backward tilt angle >15°, duration T >30 seconds, and associated with the identification of open lips.

[0043] The olfactory position: mandibular-sternal angle <20° and duration T>30 seconds. The patient attempts to extend their mandible to widen the pharyngeal cavity.

[0044] (2) Position for activating accessory respiratory muscles: Tripod position: trunk forward tilt angle 45° < θ ~ t ~ < 80°, elbow pressure > threshold, and shoulder joint movement during breathing is stably coupled with trunk angle.

[0045] (3) Positions to reduce cardiopulmonary load: Orthopnea position: The angle between the trunk and the horizontal plane is >60°, requiring support from the headboard or backrest, with both lower limbs hanging naturally or extended. This position reduces venous return to the heart, alleviating pulmonary congestion caused by acute left ventricular failure. Quantitative indicators include: a trunk angle consistently greater than a threshold, a stable center of pressure in the back, and the angle between the lower limbs and the trunk.

[0046] (4) Local upper airway obstruction caused by "acute exacerbation of intraoral mass": Its specific postural manifestation is "forced head and neck tilting position", that is, forced lateral lying or head tilting to the affected side. Precursor patterns of decompensation: (1) If the patient has orthopnea lasting more than 60 seconds and is accompanied by wheezing sounds, it is considered that the patient has acute heart failure.

[0047] (2) The patient exhibits a forced lateral decubitus position or a forced tilting of the head to the affected side. The patient returns to the forced position within 60 seconds of changing position, accompanied by short-term wheezing breaths. (3) After excluding the respiratory decompensation patterns of (1) and (2), within the time window, if N≥3 of the above four compensatory positions (1) upper airway opening position and (2) accessory respiratory muscle activation position alternately, and the average duration of each position is <45 seconds, it indicates that the patient cannot achieve stable relief through any compensatory position, which is a strong signal of critical condition. Consider that the patient has serious airway problems such as acute exacerbation of chronic obstructive pulmonary disease, acute upper airway obstruction, severe pulmonary infection, or airway compression after oral and maxillofacial surgery.

[0048] Core Relational Logic: Behavioral Evolution Path Identification Evolutionary identification algorithm: After the system detects a "first-level abnormal activity state," if it further identifies the patient stably entering and maintaining one or more "second-level specific compensatory positions" within a specific time window, then it is determined that a complete evolutionary path "from exploration to compensation" has been completed. The identification of this path is the key basis for triggering high-confidence warnings.

[0049] Intelligent early warning workflow Continuous sensing and baseline comparison: The system runs continuously, comparing real-time data with the individual's baseline.

[0050] Two-level parallel identification and association analysis: Parallel identification of the first level of "abnormal activity" and the second level of "specific compensatory posture".

[0051] Run the "Evolution Path Recognition Algorithm" to determine whether there is a temporal and logical relationship between the two types of behaviors.

[0052] Dynamic risk assessment and tiered early warning: Level 1 alert (attention level): Only "Level 1 activity abnormality" was detected, but there was no association with respiratory deterioration or a trend of progression. The system will mildly mark this on the nurse station interface.

[0053] Level 2 Warning (Intervention Level): Meets any of the following conditions: a) Identify any “second-level specific compensatory position” that persists for more than the threshold time.

[0054] b) A strong association was identified between “Level 1 activity abnormality” and “progressive increase in respiratory rate”.

[0055] c) The "decompensated breathing and postural circulation" pattern was detected. This triggered a pop-up window at the nurses' station and a push notification on the nurse's handheld terminal.

[0056] Level 3 Alert (Emergency): A complete "evolutionary path from exploration to compensation" is identified, or the patient experiences early deterioration of vital signs in a compensated position. This triggers bedside audible and visual alarms and a public address system announcement.

[0057] Context-adaptive adjustment: The system identifies "visiting mode" by counting bones. In this mode, it automatically increases the threshold of the second-level body position recognition and emphasizes the "evolution path" logic to filter out interference from social activities.

[0058] This invention collects patient data non-contactly using depth sensors and performs intelligent analysis based on a unique "two-level behavioral evolution" model. The first level identifies frequent, non-specific postural shifts and other prodromal exploratory behaviors. The second level precisely identifies specific compensatory postures with clear respiratory physiological significance, such as "gasping position," "tripod position," "orthopneic position," and "forced lateral tilt position." The core lies in analyzing the temporal evolutionary relationship between these two levels of behavior. When the system detects that a patient's behavior evolves from "exploration" to "compensation," or when an "ineffective compensation cycle" occurs, it issues a graded warning before physiological decompensation. This invention is the first to achieve digital interpretation of the entire process of respiratory compensation for multiple causes, with significant advantages such as early warning, low false alarm rate, and clinically consistent logic.

[0059] In another embodiment of the present invention, a respiratory compensation early warning system based on multi-modal behavior evolution recognition is provided, which can be used to implement the above-mentioned respiratory compensation early warning method based on multi-modal behavior evolution recognition. Specifically, the system includes: The data acquisition module is used to continuously acquire patients' behavioral and posture data; The identification module is used to analyze behavioral posture data based on a predefined behavioral evolution knowledge base and identify whether there is a type I posture or a type II posture. The type I posture is a high-frequency non-specific posture that exceeds the individual baseline, and the type II posture is a specific compensatory posture that conforms to a pre-defined specific physiological definition. The early warning module is used to analyze the temporal correlation between Category I and Category II body positions. If it detects an evolution from Category I to Category II body position, it generates a high-confidence early warning signal.

[0060] The module division in this embodiment of the invention is illustrative and represents only one logical functional division. In actual implementation, other division methods may be used. Furthermore, the functional modules in the various embodiments of the invention can be integrated into a single processor, exist as separate physical entities, or be integrated into a single module. The integrated modules described above can be implemented in hardware or as software functional modules.

[0061] In another embodiment of the present invention, a computer device is provided, comprising a processor and a memory. The memory stores a computer program, which includes program instructions. The processor executes the program instructions stored in the computer storage medium. The processor may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions in the computer storage medium to achieve a corresponding method flow or corresponding function. The processor described in this embodiment of the present invention can be used for the operation of a respiratory compensation early warning method based on multi-mode behavior evolution recognition.

[0062] In another embodiment of the present invention, a storage medium is provided, specifically a computer-readable storage medium (Memory), which is a memory device in a computer device used to store programs and data. It is understood that the computer-readable storage medium here can include both the built-in storage medium in the computer device and extended storage media supported by the computer device. The computer-readable storage medium provides storage space that stores the terminal's operating system. Furthermore, the storage space also stores one or more instructions suitable for loading and execution by a processor. These instructions can be one or more computer programs (including program code). It should be noted that the computer-readable storage medium here can be high-speed RAM or non-volatile memory, such as at least one disk storage device. The processor can load and execute one or more instructions stored in the computer-readable storage medium to implement the corresponding steps of the respiratory compensation early warning method based on multi-modal behavioral evolution recognition in the above embodiments.

[0063] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0064] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0065] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0066] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0067] 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 it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A respiratory compensation early warning method based on multi-modal behavior evolution recognition, characterized in that, include: Continuously acquire patient behavior and posture data; Based on a predefined behavioral evolution knowledge base, behavioral posture data is analyzed to identify whether there is a type I or type II posture. The type I posture is a high-frequency non-specific posture that exceeds the individual baseline, and the type II posture is a specific compensatory posture that conforms to a pre-defined specific physiological definition. Analyze the temporal correlation between Category I and Category II body positions. If the evolution from Category I to Category II body positions is identified, a high-confidence warning signal is generated.

2. The respiratory compensation early warning method based on multi-modal behavior evolution recognition according to claim 1, characterized in that, The continuous acquisition of patient behavioral posture data includes: By deploying a depth vision sensor above the hospital bed, along with a pressure distribution sensing mattress and an environmental audio sensor, the system simultaneously collects high-precision three-dimensional skeletal joint sequence, body pressure distribution, respiratory micro-movements, and environmental acoustic characteristics of the patient.

3. The respiratory compensation early warning method based on multi-modal behavior evolution recognition according to claim 1, characterized in that, The predefined behavior evolution knowledge base includes: Construct a behavioral semantic understanding model that includes two levels and one association logic; The first level is the exploratory behavior of the precursor: unable to remain still, and engaging in high-frequency, non-purposeful rolling or twisting movements beyond their personal baseline within a limited set of body positions; The second level is specific compensatory posture: a steady-state posture with clear respiratory physiological significance; Association logic: Once a first-level abnormal activity state is detected, if the patient is further identified as stably entering and maintaining one or more second-level specific compensatory positions within a preset specific time window, then it is determined that a complete evolutionary path from exploration to compensation has been completed.

4. The respiratory compensation early warning method based on multi-modal behavior evolution recognition according to claim 1, characterized in that, Quantification and identification of the aforementioned body position: Calculate the real-time position switching frequency and compare it with the individualized baseline. When it exceeds twice the baseline standard deviation and persists, it is marked as an abnormal activity state. Analysis of switching modes: Identifying whether it is an invalid exploration loop; Correlation analysis: Simultaneously analyze whether there is a progressive upward trend in respiratory rate during this period, or whether there are brief attempts at second-level compensatory postures.

5. The respiratory compensation early warning method based on multi-modal behavior evolution recognition according to claim 1, characterized in that, Quantification and identification of the two types of body positions: Upper airway open position: Panting position: Neck extension angle >15° and duration >30 seconds, and associated with the detection of open lips; Smelling flowers: Mandibular-sternal angle < 20°, and duration > 30 seconds; Positions that activate accessory respiratory muscles: Tripod position: 45° < trunk forward tilt angle < 80°, elbow pressure > threshold, and joint movement and trunk angle stability coupling during breathing; Body positions that reduce cardiopulmonary load: Sitting breathing position: The angle between the torso and the horizontal plane is >60°, which needs to be maintained by back support, and both lower limbs are drooping, and the duration is >60 seconds; Local upper airway obstruction caused by intraoral mass: Its specific postural manifestation is "forced head and neck tilting position", that is, forced lateral lying or head tilting to the affected side. Precursor patterns of decompensation: If a patient experiences orthopnea lasting longer than 60 seconds, accompanied by wheezing sounds, it is considered that the patient has developed acute heart failure. The patient exhibits a forced lateral decubitus position, briefly changes position and then quickly returns to the forced position, accompanied by short-term wheezing-like breathing sounds. After excluding the two types of respiratory decompensation patterns mentioned above, if within the time window, N≥3 of the above-mentioned upper airway open position and the four compensatory positions included in the accessory respiratory muscle activation position alternately, and the average duration of each is < 45 seconds, then acute exacerbation of chronic obstructive pulmonary disease, acute upper airway obstruction, severe pulmonary infection, or airway compression problem after oral and maxillofacial surgery are considered.

6. The respiratory compensation early warning method based on multi-modal behavior evolution recognition according to claim 3, characterized in that, The analysis of the temporal correlation between Type I and Type II body positions includes: Two-level parallel identification and association analysis: Parallel identification of first-level activity abnormalities and second-level specific compensatory postures; determination of temporal and logical correlation between the two types of behaviors through association logic.

7. The respiratory compensation early warning method based on multi-modal behavior evolution recognition according to claim 6, characterized in that, If a shift from a type I posture to a type II posture is detected, a high-confidence warning signal is generated, including: Level 1 indication: Only level 1 activity abnormalities were detected, but there was no association with respiratory deterioration or a trend of progression, so mild labeling was performed; Level 2 warning: Meets any of the following conditions: a) Identify any second-level specific compensatory posture that persists for more than the threshold time; b) A strong correlation was identified between first-level activity abnormalities and a progressive increase in respiratory rate; c) The decompensated respiratory positional circulation pattern is identified; a clear pop-up window at the nurses' station and a push notification on the nurse's handheld terminal are triggered; Level 3 Alarm: The complete evolutionary path from exploration to compensation is identified, or the patient's vital signs deteriorate early in the compensatory position, triggering an audible and visual alarm and broadcast. The visitation mode is identified by skeletal counting. In this mode, the threshold for the second-level body position recognition is automatically increased, and the evolution path logic is emphasized to filter out interference from social activities.

8. A respiratory compensation early warning system based on multi-modal behavior evolution recognition, characterized in that, include: The data acquisition module is used to continuously acquire patients' behavioral and posture data; The identification module is used to analyze behavioral posture data based on a predefined behavioral evolution knowledge base and identify whether there is a type I posture or a type II posture. The type I posture is a high-frequency non-specific posture that exceeds the individual baseline, and the type II posture is a specific compensatory posture that conforms to a pre-defined specific physiological definition. The early warning module is used to analyze the temporal correlation between Category I and Category II body positions. If it detects an evolution from Category I to Category II body position, it generates a high-confidence early warning signal.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the respiratory compensation early warning method based on multi-modal behavior evolution recognition as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the respiratory compensation early warning method based on multi-modal behavior evolution recognition as described in any one of claims 1 to 7.

Images