Pain and sedation assessment method and device

Abstract

The invention relates to a pain and sedation assessment method and device. The method comprises the following steps: acquiring face image data; inputting the face image data into a recognition module to obtain pain index data and sedation index data; inputting the pain index data into a pain scoring module, and analyzing the pain index data based on a first pain scoring rule and a second pain scoring rule in the pain scoring module to obtain a pain scoring result; inputting the sedation index data into a sedation scoring module, and analyzing the sedation index data based on a first sedation scoring rule and a second sedation scoring rule in the sedation scoring module to obtain a sedation scoring result. By acquiring the face image data for analysis, the dependence on the self-report of the patient is eliminated, and the objectivity and accuracy of evaluation are improved. A reliable evaluation result can be provided, medical staff can accurately adjust the medicine dosage according to the accurate pain scoring result and the sedation scoring result, and improper medication caused by inaccurate evaluation is avoided.

Description

Technical Field

[0001] This invention relates to the field of medical technology, and more particularly to a method and apparatus for pain and sedation assessment. Background Technology

[0002] In medical practice, accurate assessment of a patient's pain and sedation levels is crucial. Precise assessment results provide key information for healthcare professionals to develop treatment plans and adjust medication dosages, thereby improving treatment outcomes and ensuring patient safety and comfort.

[0003] Currently, commonly used clinical methods for pain and sedation assessment have some limitations. Traditional subjective assessment methods, such as the visual analog scale (VAS) and verbal description scale, mainly rely on patient self-report. However, these methods are difficult to implement for patients who cannot express their feelings, such as infants, comatose patients, and patients with speech disorders. Moreover, patients' subjective feelings are easily influenced by various factors, such as emotional state, cultural background, and education level, which can interfere with the accuracy and reliability of the assessment results. Summary of the Invention

[0004] The purpose of this invention is to provide a method and apparatus for assessing pain and sedation, overcoming the aforementioned limitations of the prior art. This method and apparatus can achieve an objective and accurate assessment of a patient's pain and sedation levels without relying on patient self-reporting, and is particularly suitable for patients who are unable to express their feelings.

[0005] The specific technical solution of this application is: a method for pain and sedation assessment, comprising:

[0006] Acquire facial image data;

[0007] The facial image data is input into the recognition module to obtain pain index data and sedation index data;

[0008] The pain index data is input into the pain scoring module, and the pain index data is analyzed based on the first pain scoring rule and the second pain scoring rule in the pain scoring module to obtain the pain scoring result.

[0009] The sedation index data is input into the sedation scoring module, and the sedation index data is analyzed based on the first sedation scoring rule and the second sedation scoring rule in the sedation scoring module to obtain the sedation scoring result.

[0010] The beneficial effects of this invention are as follows:

[0011] By analyzing facial image data, reliance on patient self-reporting is eliminated, significantly improving the objectivity and accuracy of assessments. For patients unable to express their feelings, such as infants, comatose patients, and those with speech impairments, reliable assessment results can be provided. Healthcare professionals can then precisely adjust medication dosages based on accurate pain and sedation scores, avoiding inappropriate medication use due to inaccurate assessments, thereby improving treatment outcomes and ensuring patient safety and comfort. Attached Figure Description

[0012] Figure 1 The method flowchart of this application;

[0013] Figure 2 A schematic diagram of the method of this application;

[0014] Figure 3 A system diagram of this application; Detailed Implementation

[0015] The pain and sedation assessment method and apparatus of the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0016] like Figure 1-2 As shown, in some embodiments, a method for assessing pain and sedation is provided, including:

[0017] Acquire facial image data;

[0018] The facial image data is input into the recognition module to obtain pain index data and sedation index data;

[0019] The pain index data is input into the pain scoring module, and the pain index data is analyzed based on the first pain scoring rule and the second pain scoring rule in the pain scoring module to obtain the pain scoring result.

[0020] The sedation index data is input into the sedation scoring module, and the sedation index data is analyzed based on the first sedation scoring rule and the second sedation scoring rule in the sedation scoring module to obtain the sedation scoring result.

[0021] By analyzing facial image data, reliance on patient self-reporting is eliminated, significantly improving the objectivity and accuracy of assessments. For patients unable to express their feelings, such as infants, comatose patients, and those with speech impairments, reliable assessment results can be provided. Healthcare professionals can then precisely adjust medication dosages based on accurate pain and sedation scores, avoiding inappropriate medication use due to inaccurate assessments, thereby improving treatment outcomes and ensuring patient safety and comfort.

[0022] In some embodiments, a method for acquiring facial image data is provided, including: a visual model capturing images through a camera;

[0023] YOLOv8 is used to identify target objects in images, obtain multiple facial key points of the target objects, calculate the distance between each facial key point, and judge the image quality.

[0024] Select a face image that meets the image quality requirements, and crop the face region to obtain face image data.

[0025] When a camera captures images, it may also capture the patient's surroundings, affecting the quality of the facial image data. For example, it may capture medical equipment or accompanying persons nearby.

[0026] In order to accurately identify patients, some embodiments of this application provide a method for accurately identifying patients:

[0027] Let the input image be I∈R H×W×3 Where H represents the image height, W represents the image width, and 3 indicates that the image has RGB three channels. Using YOLOv8 for inference on the image, a set of target bounding boxes is detected:

[0028]

[0029] Where b i =(x 1i ,y 1i ,x 2i ,y 2i ) is the bounding box of the i-th target, c i ∈{0,1,2,3,4} represents the category (bed, nurse, patient, caregiver, doctor), s i ∈[0,1] represents the confidence score. For each target class, only the item with the highest confidence score is retained:

[0030]

[0031] Then determine if the following conditions exist:

[0032]

[0033] 1, 3, and 4 correspond to nurse, caregiver, and doctor, respectively. If the conditions are met, it indicates a medical or nursing intrusion.

[0034] By following the steps above, interference from non-patient factors such as medical staff can be effectively eliminated from facial image data, further improving the accuracy of subsequent assessments. If medical staff intrude, the camera position can be readjusted or the image can be retaken at a suitable time to obtain high-quality patient facial image data.

[0035] In some embodiments, to ensure effective facial recognition, the key point set includes the nose P. nose Left eye P leye Right eye P reye Calculate the Euclidean distance:

[0036] d left =||p nose -p leye ||2

[0037] d right =||p nose -p reye ||2

[0038] In the above formula, d left d represents the Euclidean distance between the nose and the left eye. right This represents the Euclidean distance between the nose and the right eye. By calculating these two Euclidean distances, the positional relationships of key facial points can be quantitatively analyzed. If these two distances are within the normal range of facial proportions, the facial pose in the image can be considered relatively normal, suitable for subsequent pain and sedation assessment.

[0039] If the following conditions are met:

[0040] d left <13 or d right 13:00

[0041] If the facial pose is too skewed or too close, discard the current frame to avoid misjudgment.

[0042] If the calculated d left and d right A value significantly different from the proportion of a normal human face may indicate that the face is tilted or occluded in the image. For example, when d... left Much smaller than the normal range, while d right A value significantly larger than the normal range may indicate that the face is tilted to the left; if d left or d right If the value is abnormally large or small, there may be a situation where part of the face is obscured.

[0043] Try acquiring the next frame and repeating the same analysis process. When reacquiring the image, adjust parameters such as the camera's shooting angle and focal length to improve image quality and increase the probability of obtaining a suitable facial image. Adjusting the shooting angle can prevent excessive tilting of the face, making the distance between the nose and the left and right eyes more consistent with normal facial proportions. Adjusting the focal length can make the face clearer in the image, reducing inaccurate keypoint detection caused by blur. If the calculated Euclidean distance in the next frame acquired after adjusting the parameters still does not conform to normal facial proportions, adjust the parameters again and reacquire the image, trying multiple times to ensure a suitable facial image for pain and sedation assessment is obtained. Throughout the process, record the parameter settings and corresponding Euclidean distance calculation results for each image acquisition attempt for subsequent analysis, experience summarization, and optimization of image acquisition strategies.

[0044] After completing the above processing, the quality of the facial images needs to be evaluated again. This can be done by combining previous image quality assessment methods, such as calculating the distance between facial landmarks, to ensure that the processed facial images meet the requirements. Only facial image data that has undergone rigorous screening and processing can be input into the subsequent recognition module to ensure the accuracy and reliability of pain and sedation assessment results.

[0045] After obtaining facial image data with normal pose and acceptable quality, it is necessary to crop and segment the facial region. The specific solution is as follows: if key point filtering is used, the facial region is extracted:

[0046] R face =I[y:y+h,x:x+w]

[0047] Where x, y, w, and h represent the detected facial region locations, indicating the x-coordinate of the top-left corner, the y-coordinate, and the width and height of the facial region, respectively. This method allows for accurate extraction of the facial region from the original image, avoiding interference from surrounding environmental information.

[0048] These steps enable the accurate acquisition of patients' facial image data, providing high-quality and accurate basic data for subsequent pain and sedation assessments.

[0049] After acquiring the facial image data, the data is input into the recognition module, which then identifies various events using the following methods.

[0050] Head movement: from initial keypoint K0 to current frame keypoint K t Select relevant points in the head:

[0051]

[0052] In the formula, P tThis represents the set of head-related points selected in the current frame. P1 represents the first head-related keypoint in the current frame, and so on; P0 represents the set of head-related keypoints selected from the initial keypoints. This indicates the first head-related key point in the initial frame, and so on.

[0053] The occurrence of the head-shaking event is determined by comparing the movement distance of key head points between consecutive frames; the occurrence of the eye-closing event is determined by calculating the aspect ratio of the eyes.

[0054] Calculate the average Euclidean distance as the head movement amplitude:

[0055]

[0056] If: d head >θ head (θ head =28) is judged as a head-shaking event.

[0057] For the eye closure event: Calculate the EAR separately for the left and right eyes:

[0058]

[0059] If: EAR < θ eye (θ eye If the value is 0.24, then it is determined that an eye-closing event has occurred.

[0060] For mouth opening events: determine whether a mouth opening event has occurred by calculating the aspect ratio of the mouth.

[0061]

[0062] If: MAR>θ mouth (θ mouth If the value is 1.5, then it is determined to be an open-mouth event.

[0063] These steps automatically generate pain and sedation index data, which directly reflect the patient's current state and provide crucial information for subsequent scoring.

[0064] In some embodiments, pain scores are used to more intuitively reflect the patient's current state.

[0065] The system measures the patient's pain level. After pain index data is input into the pain scoring module, the pain score result, analyzed based on the first and second pain scoring rules, can be converted into a specific pain level. Based on this pain level, healthcare professionals can gain a clearer understanding of the patient's pain status.

[0066] In some embodiments, a method for constructing a pain scoring module is provided:

[0067] Primitive variable declarations are defined:

[0068] E hs (t): The number of head-shaking events in [tT,t].

[0069] E om (t): Number of events where the mouth is opened;

[0070] E ct (t): Number of events where eyes are closed;

[0071] E p (t): General pain events (pain=1);

[0072] E p2 (t): Severe pain event (pain=2);

[0073] E n (t): Normal painless event (pain=0);

[0074] Each sub-scoring function for facial pain is F. f :

[0075] F f (t)=max{S hs (t),S om (t),S ct (t),S p (t)}

[0076] In the above formula, F f (t) represents the change in facial pain score over time t, where S hs (t),S om (t),S ct (t),S p (t) represents the sub-scores for the head-shaking event, mouth-opening event, eye-closing event, and general pain event, respectively.

[0077] in:

[0078]

[0079] S om S ct Similarly, classify by quantity

[0080]

[0081] Posture score (body movement) F m Fixed strategy allocation (e.g., combining head shaking with facial expression)

[0082]

[0083] Muscle tone score F μ (t) depends on the magnitude of eye closure and pain events.

[0084] Final score: CPOT Score = F f (t)+F m (t)+F μ (t)

[0085] Wherein: F f Facial expression, F m Body movement, F μ Muscle tone.

[0086] Since only the patient's face was photographed, assessments of body movement and muscle tone were inferred from facial expressions. Indirect inferences can be made through facial expressions. For example, frequent head shaking (i.e., a large number of head-shaking events) may suggest involuntary body movements, which to some extent reflect physical actions caused by pain or discomfort, and can thus serve as a reference for body movement.

[0087] The frequency of eye-closing episodes is an important indicator for assessing muscle tone. Frequent eye-closing may indicate a state of tension and potentially high muscle tone. The frequency of both severe and mild pain events is also related to muscle tone.

[0088] The calculated pain scores can provide strong support for healthcare professionals to develop personalized pain management plans. Different pain scores correspond to different treatment strategies. For example, a lower pain score may only require gentle interventions such as physical therapy and psychological counseling; while a higher pain score may require timely adjustments to the type of medication or an increase in the dosage to effectively alleviate the patient's suffering.

[0089] Similarly, the sedation index data is input into the sedation scoring module.

[0090] Let the time window be T = 5s, and the total number of frames within that time be N.

[0091] definition:

[0092] N open_eye Number of times eyes open

[0093] N close_eye Number of times eyes closed

[0094] N head_shake Number of head shakes

[0095] Nmouth_change Number of mouth movements

[0096] The frequency of the behavior is:

[0097]

[0098] Rule-based reasoning determines RASS level

[0099] Awareness assessment (RASS = +5):

[0100]

[0101] Sleep status (RASS = -5):

[0102]

[0103] Final sedation score:

[0104]

[0105] The final sedation score determines the patient's sedation status. If the final sedation score is within the range for alertness (RASS = +5), it indicates that the patient is conscious and has good perception and responsiveness to the surrounding environment. At this point, medical staff can conduct necessary communication and examinations based on the patient's specific situation to understand the patient's physical condition and needs. For example, they can ask the patient if they experience any discomfort or if the treatment plan needs adjustment.

[0106] If the final sedation score falls within the range for deep sleep (RASS = -5), it indicates that the patient is in a deep sleep state. Close monitoring of the patient's vital signs, such as respiration, heart rate, and blood pressure, is necessary to ensure the patient's safety. Simultaneously, it is important to check for any abnormalities in the patient's body, such as pressure or airway obstruction by secretions.

[0107] For patients in other score ranges, medical staff need to comprehensively consider the patient's condition, treatment, and dynamic changes in the sedation score to determine whether the level of sedation is appropriate. If the score is too high, it may mean that the patient is too deeply sedated, and the dosage of sedative drugs needs to be appropriately reduced; if the score is too low, it may indicate that the patient is not sedated enough, and the patient may feel uncomfortable or agitated, in which case the dosage of sedative drugs needs to be increased or the treatment plan adjusted.

[0108] In some embodiments, the first pain rating rule is developed based on existing guidelines and takes into full account the professional standards and experience of the medical field in pain assessment. It comprehensively considers the frequency of various pain-related events, such as head shaking, mouth opening, eye closing, and general pain events. The occurrence of these events can reflect the patient's pain experience from different perspectives. For example, frequent head shaking may suggest that the patient has difficulty keeping their head still due to pain, and mouth opening events may be related to physical responses to pain.

[0109] The second pain scoring system is likely developed by incorporating feedback from actual clinical applications and research findings. It may supplement and optimize the first pain scoring system, taking into account more specific factors in different scenarios. For example, it might develop more targeted scoring standards for patients of different ages and with different disease types. For elderly patients, the weighting of certain events might be adjusted appropriately, as their physical responses may differ from those of younger patients.

[0110] In some embodiments, a second pain scoring rule is provided, including:

[0111] Mild pain is defined as the number of head shakes per unit time is greater than 3, the standard value of the eyebrow edge detection is greater than 0.16, the aspect ratio of the mouth is greater than 0.2, or the distance between the upper and lower eyelids is less than 4.1px and the aspect ratio is less than 0.13. When two or three of these conditions are met, it is defined as moderate pain. When all four conditions are met, it is defined as severe pain.

[0112] The second pain scoring rule in these embodiments is highly practical and targeted. Through clear quantitative indicators, it can more precisely classify the degree of pain in patients. For healthcare professionals, this clear classification helps to quickly and accurately assess a patient's pain status, thereby enabling timely and appropriate treatment measures.

[0113] In practice, medical staff can use the second pain rating scale (BRCS) to score the patient based on their specific symptoms. For example, if a patient shakes their head more than three times within a given time period and their mouth's aspect ratio is greater than 0.2, the patient can be diagnosed with moderate pain. Based on this assessment, medical staff can appropriately increase the dosage of analgesics or adjust the treatment plan.

[0114] Furthermore, the second pain scoring system can complement the first. The first system focuses on a comprehensive consideration of various pain-related events, while the second emphasizes the quantification of specific indicators. Using both together allows for a more comprehensive and accurate assessment of a patient's pain level, providing more personalized and effective treatment. Moreover, with continuous accumulation of clinical practice, the second pain scoring system can be continuously optimized and improved. For example, the weights of various indicators can be adjusted according to the characteristics of different departments and diseases, or new assessment indicators can be added to improve the accuracy and reliability of the scoring.

[0115] After the pain index data is input into the pain scoring module, the module will perform a detailed analysis based on these two rules. It quantifies the raw variables of various pain events and then performs a weighted calculation according to the rules. For example, for the head-shaking event, a corresponding score may be assigned based on the number of times it occurs within a certain period of time, and then combined with the scores of other events for a comprehensive calculation.

[0116] In some embodiments, a second sedation scoring rule is provided, including:

[0117] Light sedation is defined as the following: the percentage of time with eyes open within 30 seconds is greater than 0.12, the frequency of head shaking within 30 seconds is greater than 0.1, or the frequency of mouth opening and closing within 30 seconds is greater than 0.1, and the heart rate is between 70 and 100 beats per minute and the respiratory rate is greater than 15 breaths per minute.

[0118] Deep sedation is defined as follows: the percentage of time with eyes closed within 30 seconds is greater than 0.88, the head shaking frequency within 30 seconds is less than 0.1, or the mouth opening and closing frequency within 30 seconds is less than 0.1, and the heart rate is between 50 and 80 beats per minute and the respiratory rate is between 8 and 14 beats per minute.

[0119] If the head shaking frequency is greater than 0.4 within 30 seconds or the standardized value of the eyebrow edge detection is greater than 0, it is judged as agitation.

[0120] The first awakening is defined as the first detection of an eye-opening duration greater than 0.12, a head-shaking frequency greater than 0.1, or a mouth-opening / closing change rate greater than 0.1 within 30 seconds.

[0121] In these embodiments, the second sedation scoring rule provides healthcare professionals with a more detailed and accurate basis for assessing a patient's sedation status. By comprehensively considering multiple indicators such as the percentage of time the eyes are open, the frequency of head shaking, the frequency of mouth opening and closing, as well as heart rate and respiratory rate, different sedation states such as light sedation, deep sedation, agitation, and initial awakening can be clearly distinguished.

[0122] The determination of a state of light sedation requires a more rigorous approach through a combination of criteria. When the patient's eye-opening time accounts for more than 0.12% of the total time within 30 seconds, it may indicate a degree of recovery of consciousness and the initiation of spontaneous eye-opening movements. A head-shaking frequency greater than 0.1% within 30 seconds reflects signs of bodily activity. A mouth-opening and closing frequency greater than 0.1% also suggests increased muscle activity. Simultaneously, a heart rate between 70-100 bpm and a respiratory rate above 15 breaths per minute all indicate a relatively light state of sedation. In such cases, healthcare professionals may consider adjusting the dosage of sedative medication to maintain an appropriate level of sedation.

[0123] The assessment of deep sedation is also based on multiple indicators. A period of eye-closed time greater than 0.88% within 30 seconds indicates that the patient spends most of their time with their eyes closed, indicating a deep level of inhibition. A head-shaking frequency and mouth-opening / closing frequency less than 0.1 within 30 seconds indicate minimal physical activity. A heart rate between 50-80 bpm and a respiratory rate between 8-14 bpm are also physiological characteristics consistent with deep sedation. At this time, medical staff need to closely monitor the patient's vital signs to ensure their safety under deep sedation.

[0124] The assessment of agitation is primarily based on head-shaking frequency and the standardized value of the brow ridge detection. A head-shaking frequency greater than 0.4 or a brow ridge detection standardized value greater than 0 within 30 seconds indicates possible involuntary agitation. This may be due to changes in the patient's tolerance to sedative medications or other adverse factors. Healthcare professionals need to promptly investigate the cause and may need to adjust the type or dosage of sedative medications to alleviate the patient's agitation.

[0125] The initial assessment of awakening focuses on the first detected changes in relevant indicators. When the percentage of time the eyes are open within 30 seconds is greater than 0.12, the head-shaking frequency is greater than 0.1, or the mouth-opening / closing rate is greater than 0.1, it indicates that the patient is beginning to awaken from sedation. Medical staff at this time need to make appropriate preparations, such as closely monitoring the patient's recovery of consciousness and ensuring a clear airway.

[0126] In clinical practice, healthcare professionals can conduct accurate assessments based on the Second Sedation Rating Principle and the patient's specific condition. This principle can also complement other sedation assessment methods, further improving the accuracy and reliability of patient sedation status assessment. As clinical practice deepens, the Second Sedation Rating Principle can be optimized and improved based on actual circumstances to better serve patient treatment and care.

[0127] In practical applications, medical staff can adjust treatment plans promptly based on pain and sedation scores. A high pain score may necessitate increasing the dosage of analgesics or switching to a more effective one; conversely, an oversedation score may require reducing the use of sedatives. This objective and accurate assessment method and device better meets patients' treatment needs, improves the quality of medical services, and provides strong support for patient recovery. Furthermore, with continuous technological development and improvement, this method and device can be further optimized, for example, by incorporating more physiological data to make the assessment results more comprehensive and accurate.

[0128] In some embodiments, such as Figure 3 As shown, this application also provides a pain and sedation assessment system, the system including at least one processor, a memory, an input device, and a display device. The input device is used to obtain input from the outside world. The memory stores instructions. When the instructions are executed by at least one processor, the steps of the method described in the method embodiment are implemented, and the running results are displayed on the display device, thus implementing the steps of the method described in the method embodiment.

[0129] The embodiments and functional operations of the subject matter described in this specification can be implemented in the following ways: digital electronic circuits, tangibly implemented computer software or firmware, computer hardware, including the structures disclosed in this specification and their equivalents, or combinations thereof. Embodiments of the subject matter described in this specification can be implemented as one or more computer programs, that is, one or more modules of computer program instructions encoded on one or more tangible non-transitory program carriers, for execution by a data processing device or to control the operation of the data processing device.

[0130] Similarly, although operations are described in the accompanying drawings in a specific order, it should not be construed as requiring that such operations be performed in the specific order shown or in sequential order, or that all illustrated operations be performed, in order to achieve the desired result. In certain cases, multitasking and parallel processing may be advantageous. Furthermore, the separation of various system modules and components in the above embodiments should not be construed as requiring such separation in all embodiments, and it should be understood that program components and systems can generally be integrated into a single software product or packaged into multiple software products.

Claims

1. A method for assessing pain and sedation, characterized in that, include: Acquire facial image data; The facial image data is input into the recognition module to obtain pain index data and sedation index data; The pain index data is input into the pain scoring module, and the pain index data is analyzed based on the first pain scoring rule and the second pain scoring rule in the pain scoring module to obtain the pain scoring result. The sedation index data is input into the sedation scoring module, and the sedation index data is analyzed based on the first sedation scoring rule and the second sedation scoring rule in the sedation scoring module to obtain the sedation scoring result.

2. The pain and sedation assessment method as described in claim 1, characterized in that, Acquiring facial image data includes: The visual model captures images using a camera; YOLOv8 is used to identify target objects in the image, obtain multiple facial key points of the target objects, calculate the distance between each facial key point, and determine the image quality; Select an image that meets the image quality requirements, and crop the face region to obtain face image data.

3. The pain and sedation assessment method as described in claim 1, characterized in that, The pain index data includes facial expression data, body movement data, and muscle tone data. The facial expression data includes head shaking events, mouth opening events, and eye closing events. The body movement data is determined based on the head shaking events. The muscle tone data is determined based on the duration of eye closure. The sedation index data includes facial expression data.

4. The pain and sedation assessment method as described in claim 1, characterized in that, The facial image data is input into the recognition module to obtain pain index data and sedation index data, specifically including: Whether the head-shaking event has occurred is determined by comparing the movement distance of key head points between consecutive frames; The occurrence of an eye-closing event is determined by calculating the aspect ratio of the eyes. Whether an opening-mouth event has occurred is determined by calculating the aspect ratio of the mouth.

5. The pain and sedation assessment method as described in claim 1, characterized in that, The second sedation scoring rule includes: Light sedation is defined as the following: the percentage of time with eyes open within 30 seconds is greater than 0.12, the frequency of head shaking within 30 seconds is greater than 0.1, or the frequency of mouth opening and closing within 30 seconds is greater than 0.1, and the heart rate is between 70 and 100 beats per minute and the respiratory rate is greater than 15 breaths per minute. Deep sedation is defined as follows: the percentage of time with eyes closed within 30 seconds is greater than 0.88, the head shaking frequency within 30 seconds is less than 0.1, or the mouth opening and closing frequency within 30 seconds is less than 0.1, and the heart rate is between 50 and 80 beats per minute and the respiratory rate is between 8 and 14 beats per minute. If the head shaking frequency is greater than 0.4 within 30 seconds or the standardized value of the eyebrow edge detection is greater than 0, it is judged as agitation. The first awakening is defined as the first detection of an eye-opening duration greater than 0.12, a head-shaking frequency greater than 0.1, or a mouth-opening / closing change rate greater than 0.1 within 30 seconds.

6. The pain and sedation assessment method as described in claim 1, characterized in that, The second pain rating system includes: Mild pain is defined as the number of head shakes per unit time is greater than 3, the standard value of the eyebrow edge detection is greater than 0.16, the aspect ratio of the mouth is greater than 0.2, or the distance between the upper and lower eyelids is less than 4.1px and the aspect ratio is less than 0.

13. When two or three of these conditions are met, it is defined as moderate pain. When all four conditions are met, it is defined as severe pain.

7. The pain and sedation assessment method as described in claim 2, characterized in that, The key points include the nose (P). nose Left eye P leye and right eye P reye Calculate P in the left eye leye and nose P nose Euclidean distance d left and right eye P reye and nose P nose Euclidean distance d right If d is satisfied left <13 or d right If the value is less than 13, the facial pose is determined to be too skewed or too close, and the current frame is discarded.

8. The pain and sedation assessment method as described in claim 3, characterized in that, The rules for determining the head-shaking event are as follows: From the initial keypoint K0 and the current frame keypoint K t Select relevant points in the head: In the formula, P t This represents the set of head-related points selected in the current frame. P1 represents the first head-related keypoint in the current frame, and so on; P0 represents the set of head-related keypoints selected from the initial keypoints. This indicates the first head-related key point in the initial frame, and so on. The occurrence of the head-shaking event is determined by comparing the movement distance of key head points between consecutive frames; the occurrence of the eye-closing event is determined by calculating the aspect ratio of the eyes. Calculate the average Euclidean distance as the head movement amplitude: If: d head >θ head (θ head =28) is judged as a head-shaking event.

9. The pain and sedation assessment method as described in claim 3, characterized in that, The rule for determining the closed-eye event is as follows: Calculate the aspect ratio (EAR) of the eyes for both the left and right eyes separately. In the formula, p1, p2, p3, p4, p5, and p6 are key points on the eye contour, respectively. When the EAR values ​​of both the left and right eyes are less than the set threshold θ, eye At that time, it is determined that an eye-closing event has occurred.

10. A pain and sedation assessment device, characterized in that, The system includes at least one processor; and a memory storing instructions that, when executed by the at least one processor, perform the steps of the method according to any one of claims 1-9.

Images