A surgical drainage fluid monitoring method and system based on image recognition

Abstract

The present invention relates to the technical field of medical indicator monitoring, and specifically to a surgical drainage fluid monitoring method and system based on image recognition. The method includes: parsing historical medical plans and corresponding historical monitoring strategies to obtain corresponding first and second features; parsing the patient's medical plan to extract the current feature; based on the similarity between the current feature and the first feature, screening the primary strategy from the historical monitoring strategy; real-time monitoring of drainage fluid parameters based on the primary strategy; comparing the drainage fluid parameters and the warning threshold to determine the drainage intervention measures. By parsing the historical medical plan and the current medical plan to extract multiple features, a targeted primary strategy is screened out for the patient, making the monitoring more targeted, reducing misjudgment, and improving the recognition accuracy.

Description

Technical Field

[0001] The present invention relates to the technical field of medical index monitoring, and in particular to a surgical drainage fluid monitoring method and system based on image recognition. Background Art

[0002] After surgery, monitoring the patient's drainage fluid is an important means of evaluating recovery and judging postoperative complications. The amount, color, turbidity and other indicators of the drainage fluid directly reflect the healing of the surgical site and whether there are complications such as bleeding and infection. Traditional drainage fluid monitoring methods usually rely on regular manual observation by medical staff, which is limited by the experience and subjective judgment of medical staff and may lead to errors in monitoring data. Since the color, turbidity and other characteristics of the drainage fluid often change slightly with the changes in the patient's condition, manual inspection is difficult to capture these subtle changes, thus missing the opportunity for early warning.

[0003] In the case of night duty or limited medical staff resources, manual inspection is difficult to achieve real-time and continuous monitoring, which affects the timeliness and accuracy of monitoring. Therefore, the present invention proposes a surgical drainage fluid monitoring method and system based on image recognition, which realizes automatic data collection and analysis and processing, and timely feeds back the monitoring results to medical staff, thereby achieving accurate and efficient monitoring, solving the problem of low accuracy and lack of real-time monitoring in manual inspection of drainage fluid, and ensuring the treatment effect and medical safety of patients. Summary of the invention

[0004] The purpose of the present invention is to address the problem in the prior art that manual inspection of drainage fluid is difficult to achieve real-time and continuous monitoring, which affects the timeliness and accuracy of monitoring. Through the embodiments of the present application, a surgical drainage fluid monitoring method and system based on image recognition are provided, which realizes the automation of drainage fluid monitoring, which is conducive to improving the overall monitoring efficiency and effect.

[0005] In order to achieve the above-mentioned object of the invention, the present invention provides the following technical solutions: A surgical drainage fluid monitoring method based on image recognition, comprising: Analyze historical medical plans and corresponding historical monitoring strategies to obtain corresponding first and second features; Analyze the patient's medical plan and extract the current features; based on the similarity between the current features and the first features, select the primary strategy from the historical monitoring strategies; Based on the primary strategy, drainage fluid parameters are monitored in real time; drainage fluid parameters are compared with warning thresholds to determine drainage intervention measures.

[0006] As a preferred technical solution of this application, real-time monitoring of drainage fluid parameters based on the primary strategy includes: The drainage fluid flow is measured by a flow sensor; the drainage equipment image is collected by an optical sensor, and the drainage fluid image features are extracted from the drainage equipment image; the drainage fluid properties and concentration are analyzed by a chemical sensor; and the drainage fluid parameters are displayed in real time by a display control submodule.

[0007] As the preferred technical solution of this application, after screening the primary strategy, it also includes: using the first prediction model to predict expected drainage fluid parameters based on the patient's medical regimen; obtaining drainage fluid parameters corresponding to the primary strategy; The similarity between the expected parameters and the drainage fluid parameters is calculated, and the monitoring strategy is adjusted according to the comparison result between the similarity and the first threshold.

[0008] As a preferred technical solution of this application, the patient medical plan includes patient physical sign information, treatment plan, doctor's advice plan and prognosis plan; predicting expected drainage fluid parameters, the method includes: Extract medical history information and treatment information from the treatment plan to obtain wound characteristics; Extract drug information and nursing information from the doctor's prescription plan, and combine it with wound characteristics to obtain daily recovery characteristics; Extract complication information from the prognostic scheme to obtain complication characteristics; Obtain drainage fluid reference parameters based on patient physical information and wound characteristics; Wound characteristics, daily recovery characteristics, complication characteristics and drainage fluid reference parameters are input into the first prediction model, and drainage fluid parameters are output.

[0009] As a preferred technical solution of this application, real-time monitoring of drainage fluid parameters based on the primary strategy includes: Perform limb analysis on the patient's image to extract motion features; analyze the patient's motion intensity based on the motion features to obtain the motion severity; Obtain the movement slow time according to the exercise intensity; adjust the monitoring strategy during the movement slow time; after the movement slow time ends, restore the monitoring strategy before adjustment.

[0010] As the preferred technical solution of this application, it includes: When the patient is exercising and in a slow exercise period, the intensity of the exercise is graded, and the monitoring weight of the drainage fluid parameters is adjusted according to the classification results.

[0011] As a preferred technical solution of this application, limb analysis is performed on the patient's image, including: Extract head key frames and limb key frames from the patient image to form action features; compare the action features with the preset action library to obtain the action type; The action frequency is calculated based on the number of action features, and the duration of the action features is counted; The action type, action frequency and duration are vectorized, and the obtained vectors are combined to obtain the motion features.

[0012] As a preferred technical solution of this application, before vectorization, it also includes: The wound position is obtained and fused with the patient image; the displacement of the motion feature and the wound position is calculated, and the vectorized discount factor is calculated based on the displacement.

[0013] The present application also provides a surgical drainage fluid monitoring system based on image recognition, comprising: The parsing module is used to parse the historical medical plan and the corresponding historical monitoring strategy to obtain the corresponding first feature and second feature; and is used to parse the patient's medical plan and extract the current feature; A parameter configuration module, used to select a primary strategy from historical monitoring strategies based on the similarity between the current feature and the second feature; A monitoring module for real-time monitoring of drainage fluid parameters based on the primary strategy; The adjustment module is used to compare drainage fluid parameters with warning thresholds and determine drainage intervention measures.

[0014] The monitoring module includes: A flow sensing submodule, used for monitoring the drainage fluid flow in real time through a flow sensor; A color analysis submodule, which is used to collect drainage equipment images through optical sensors and extract drainage fluid image features from the drainage equipment images, including color features and turbidity features; A property analysis submodule is used to analyze the properties and concentration of drainage fluid through chemical sensors; The display control submodule is used to receive data from the analysis module, parameter configuration module and monitoring module, and display the data in real time after processing.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. This automated system can not only reduce the workload of medical staff, but also effectively improve the accuracy and real-time performance of monitoring, which helps to respond quickly when postoperative complications occur. By analyzing historical and current medical plans to extract multiple features, we can fully grasp the patient's situation, and then screen the primary strategy through similarity, making monitoring more targeted, reducing misjudgment, and improving recognition accuracy. Effective use of historical data, through similarity search to match similar cases, and obtain corresponding monitoring strategies, quickly formulate appropriate monitoring plans for current patients, and improve overall monitoring efficiency and effectiveness.

[0016] 2. Due to differences in physical conditions, surgical types and other factors, the normal range of drainage fluid parameters for different patients will also be different. After screening out the primary strategy, the present invention uses the first prediction model to predict the expected parameters based on the patient's medical plan, and obtains the drainage fluid parameters corresponding to the primary strategy, and then calculates the similarity between the expected parameters and the drainage fluid parameters, and adjusts the monitoring strategy according to the comparison result of the similarity and the first threshold. Personalized prediction can more accurately determine whether the drainage fluid parameters are abnormal, so that the predicted expected parameters can be more in line with the individual situation of the patient. The dynamic adjustment mechanism can respond to changes in the patient's condition in a timely manner. When there is a large deviation between the actual drainage fluid parameters and the expected parameters, timely adjustment of the monitoring frequency or taking other intervention measures will help improve the accuracy of monitoring and reduce the possibility of missed diagnosis or misdiagnosis.

[0017] 3. Conduct a detailed analysis of each part of the medical plan, extract wound characteristics from the treatment plan, obtain daily recovery characteristics from the doctor's order plan, determine complication characteristics from the prognosis plan, and combine patient vital signs to obtain drainage fluid reference parameters. This comprehensive information integration and analysis can more accurately grasp the patient's disease progression and recovery, and identify potential risk factors in advance. The first prediction model comprehensively considers multiple feature information to predict drainage fluid parameters, which can more accurately predict the changes in drainage fluid that may occur in patients at different stages, so as to prepare for it in advance.

[0018] 4. Dynamically adjust the monitoring strategy based on the prediction results so that monitoring resources can be allocated more reasonably. For patients whose condition is stable and whose drainage fluid parameters meet expectations, the monitoring frequency can be appropriately reduced to reduce unnecessary consumption of medical resources; for patients whose condition changes greatly and who are at higher risk, the monitoring frequency can be increased to ensure timely detection and treatment of problems. This can not only ensure the safety of patients, but also improve the efficiency of medical resource utilization. By providing accurate expected parameters and risk assessments, it helps medical staff to formulate monitoring strategies and intervention measures more scientifically. Medical staff can prepare in advance based on the prediction results, arrange medical resources reasonably, and avoid excessive medical treatment or delayed treatment.

[0019] 5. By performing limb analysis on the patient's image to extract movement features and determine the intensity of movement, the patient's physical activity status can be understood more accurately. Different movement intensities have different effects on drainage fluid parameters. Adjusting the monitoring strategy accordingly can avoid excessive monitoring when the patient's movement is stable or insufficient monitoring when the patient's movement is intense, thereby improving the accuracy of drainage fluid parameter monitoring. Reasonably adjust the monitoring strategy during the slow movement time, such as focusing on key parameters such as flow rate at the beginning of exercise, and then gradually increasing attention to other parameters, and resuming routine monitoring after the slow movement time ends. In this way, the distribution of drainage fluid parameter monitoring weights can be flexibly adjusted according to the patient's actual movement situation, improving the response efficiency of abnormal situations.

[0020] 6. The closer the movement is to the wound, the higher the possibility of affecting the wound, causing the wound tissue to secrete more drainage fluid, so the monitoring attention should be higher. When the patient moves violently and the displacement is large, the discount factor can guide the system to pay more attention to the key parameters of the drainage fluid, such as real-time monitoring of the flow and color changes of the drainage fluid, while reducing the monitoring attention to other parameters. In this way, while ensuring the monitoring effect, key information can be highlighted and the response efficiency of abnormal situations can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 It is a schematic diagram of the overall process of the drainage fluid monitoring method; Figure 2 It is a schematic diagram of the feature extraction process of the first prediction model; Figure 3 Schematic diagram of the structure of the drainage fluid monitoring system. DETAILED DESCRIPTION

[0022] To make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be described clearly and completely in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them.

[0023] Therefore, the following detailed description of the embodiments of the present invention is not intended to limit the scope of the invention claimed for protection, but merely represents some embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0024] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features and technical solutions in the embodiments may be combined with each other.

[0025] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0026] In the description of the present invention, it should be noted that the orientation or positional relationship indicated by the terms "upper", "lower", etc. is based on the orientation or positional relationship shown in the drawings, or the orientation or positional relationship in which the invention product is usually placed when in use, or the orientation or positional relationship commonly understood by those skilled in the art. Such terms are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first", "second", etc. are only used to distinguish the description, and cannot be understood as indicating or implying relative importance.

[0027] Example 1: For example Figure 1-Figure 3 As shown, this embodiment provides a surgical drainage fluid monitoring method based on image recognition, including: Analyze historical medical plans and corresponding historical monitoring strategies to obtain corresponding first and second features; Analyze the patient's medical plan and extract the current features; based on the similarity between the current features and the first features, select the primary strategy from the historical monitoring strategies; Based on the primary strategy, drainage fluid parameters are monitored in real time; drainage fluid parameters are compared with warning thresholds to determine drainage intervention measures.

[0028] Specifically, multiple historical patients whose drainage fluid has been monitored are screened out from the medical database, and their medical plans and corresponding monitoring strategies are obtained, which are recorded as historical medical plans and historical monitoring strategies, respectively. Among them, the medical plan includes but is not limited to patient vital signs information, treatment plan, doctor's order plan and prognosis plan. These data include text data and image data, which can be data from the patient's hospital or from databases of different hospitals, and are usually stored in the form of tables, documents, pictures, and videos.

[0029] In this embodiment, when selecting primary strategies from historical monitoring strategies, the text data in the medical plan is used for parsing. The acquired text data is cleaned to remove noise data, such as irrelevant punctuation marks, repeated information, etc., to ensure the accuracy and consistency of the data.

[0030] As a preferred technical solution of this application, the method for analyzing the patient's medical plan includes: Parse the text data in the medical plan, extract the patient's physical signs, treatment pathway characteristics and prognosis risk characteristics; encode the characteristics into a multi-dimensional feature vector based on graph embedding technology.

[0031] Specifically, the named entity recognition (NER) algorithm in the natural language processing (NLP) technology is used to identify the data in the medical plan text that is related to the patient's vital signs, including but not limited to body temperature, blood pressure, heart rate, and respiratory rate. The identified patient vital signs information is further statistically analyzed and processed, such as calculating the average value, fluctuation range, and other statistical features of the vital signs data in different time periods as patient vital signs features. Similarly, the treatment plan is parsed, for example, the text features such as the type of surgery, medication regimen, treatment cycle, patient's medical history, wound size and location are identified through predefined rules as treatment path features. The doctor's order plan includes text features such as movement restrictions and drainage methods. The prognosis plan includes text features such as underlying diseases, complications, and corresponding risk levels. The text features of the doctor's order plan and the prognosis plan are combined as prognostic risk features.

[0032] The extracted patient physical signs, treatment pathways, and prognosis risk features are used as nodes, and a feature graph is constructed based on the correlation between the features. For example, patients with heavier weight have thicker fat layers, and the wounds caused during surgery are relatively larger, resulting in more tissue damage, which in turn increases the production of postoperative drainage fluid. Older patients have degenerated immune systems, and the likelihood of inflammation in the prognosis increases, which increases the probability of increased drainage fluid. The weights of the correlations between nodes are defined based on the correlation and importance between different features. Using graph embedding algorithms, such as Node2Vec and graph convolutional networks (GCN), the nodes in the feature graph are mapped to vector representations in low-dimensional space to obtain multidimensional feature vectors, which are used to characterize the correlations and semantic information between features.

[0033] Among them, the multidimensional feature vector corresponding to the historical medical plan is recorded as the first feature; the multidimensional feature vector corresponding to the historical monitoring strategy is recorded as the second feature; and the multidimensional feature vector corresponding to the patient's medical plan is recorded as the current feature.

[0034] Furthermore, the similarity between the current feature and the first feature is calculated, a similarity search is performed in the medical database, the similarities are sorted from large to small, a plurality of similar cases with the highest ranking are screened out, and the historical monitoring strategies corresponding to the similar cases are recorded as primary strategies. In this embodiment, the primary strategy ranked first is selected to monitor the drainage fluid parameters of the current patient, including but not limited to flow, color, turbidity, properties and concentration; a weighted average method can also be used to assign different weights to the primary strategies of each similar case according to the similarity between the first feature of the similar case and the current feature of the current patient, and then perform weighted summation to obtain the final primary strategy.

[0035] The monitoring strategy includes the monitored drainage fluid parameters, monitoring frequency, warning thresholds, intervention measures, etc. The generated personalized monitoring parameter set is output to the monitoring system to guide the real-time monitoring and intervention decisions of surgical drainage fluid. In actual applications, the monitoring system monitors the patient's drainage fluid parameters in real time based on the monitoring strategy. When the drainage fluid parameters exceed the warning threshold, the control display module promptly sends out voice and / or text alarm signals to promptly notify medical staff to confirm the patient's wound status.

[0036] This automated system can not only reduce the workload of medical staff, but also effectively improve the accuracy and real-time performance of monitoring, which helps to respond quickly when postoperative complications occur. By analyzing historical and current medical plans to extract multiple features, we can fully grasp the patient's situation, and then screen the primary strategy through similarity, making monitoring more targeted, reducing misjudgment, and improving recognition accuracy. By effectively using historical data, matching similar cases through similarity search, and obtaining corresponding monitoring strategies, we can quickly develop appropriate monitoring plans for current patients and improve overall monitoring efficiency and effectiveness.

[0037] As a preferred technical solution of this application, real-time monitoring of drainage fluid parameters based on the primary strategy includes: The drainage fluid flow is measured by a flow sensor; the drainage equipment image is collected by an optical sensor, and the drainage fluid image features are extracted from the drainage equipment image; the drainage fluid properties and concentration are analyzed by a chemical sensor; and the drainage fluid parameters are displayed in real time by a display control submodule.

[0038] Specifically, before using the present invention, it is necessary to ensure that the connection part of the drainage tube is tightly connected to the surgical drainage tube so that the drainage fluid can flow smoothly into the drainage fluid collection and analysis device. The device uses a high-precision flow sensor to accurately measure the drainage fluid flow and convert the flow data into an electrical signal. The device uses an optical sensor, such as a high-definition camera or a mobile phone camera, to collect images of drainage equipment containing drainage fluid. The drainage equipment includes a drainage bag and / or a drainage tube. Preferably, the drainage equipment is a light-transmitting material. The collected drainage equipment image is analyzed and color features are extracted by a computer vision algorithm, including RGB values, chromaticity values, and brightness values. The device uses a chemical sensor, such as a biosensor, to analyze the properties, concentration, and pH of the drainage fluid. The data collected by these sensors are recorded as drainage fluid parameters, and statistical analysis is performed on them, and the various parameters of the drainage fluid are displayed in an intuitive interface, and finally the drainage fluid parameters are stored as time series data.

[0039] Example 2: For example Figure 2 As shown, this embodiment is further improved on the basis of the technical solution of embodiment 1.

[0040] As the preferred technical solution of this application, after screening the primary strategy, it also includes: using the first prediction model to predict expected drainage fluid parameters based on the patient's medical regimen; obtaining drainage fluid parameters corresponding to the primary strategy; The similarity between the expected parameters and the drainage fluid parameters is calculated, and the monitoring strategy is adjusted according to the comparison result between the similarity and the first threshold.

[0041] Specifically, the first prediction model is constructed based on a time-series evolution algorithm, including but not limited to a recurrent neural network (RNN) and a long short-term memory network (LSTM). By receiving a large amount of historical medical data and drainage fluid parameter data, the time-series features are learned and analyzed, and the correlation between the medical plan and the drainage fluid parameters is extracted, and finally the patient's drainage fluid parameters are predicted. The model can comprehensively consider multiple factors in the patient's medical plan, including but not limited to the patient's physical sign information, treatment plan, doctor's order plan and prognosis plan. Based on the learned correlation, the model is analyzed according to factors such as the patient's surgery type, postoperative recovery stage, and medication status, and combined with the drainage fluid parameter change rules of similar patients in the historical data at this stage, the possible drainage fluid parameter range of the current patient in the future is predicted.

[0042] The drainage fluid parameters corresponding to the primary strategy are recorded as reference drainage fluid parameters. Due to individual differences between patients, it is difficult to reflect the actual situation of the current patient by formulating a monitoring strategy for the current patient based on the drainage fluid parameters of similar cases in the medical database. Therefore, in this embodiment, the primary strategy used for reference is adjusted by training a deep learning model and predicting drainage fluid parameters.

[0043] After obtaining the expected parameters and reference drainage fluid parameters, the similarity calculation between the predicted expected parameters and the actually obtained drainage fluid parameters can be realized by fitting functions and calculating the correlation between the two functions; or by calculating the Euclidean distance through a curve graph. Similarity calculation methods include but are not limited to cosine similarity, dot product similarity and Euclidean distance. The calculated similarity value is compared with the preset first threshold. If the similarity value is lower than the first threshold, it means that there is a large difference between the actual drainage fluid parameters and the expected parameters. At this time, the monitoring strategy needs to be adjusted according to the difference. For example, if the predicted drainage fluid flow rate is 50ml / h, and the actual collected flow rate is 80ml / h, and the similarity is lower than the threshold, then the monitoring frequency can be appropriately increased, and medical staff can be notified to pay attention to the patient's condition so that appropriate intervention measures can be taken in time.

[0044] Due to differences in physical condition, type of surgery and other factors, the normal range of drainage fluid parameters may also be different for different patients. After screening out the primary strategy, the present invention uses the first prediction model to predict the expected parameters based on the patient's medical plan, and obtains the drainage fluid parameters corresponding to the primary strategy, and then calculates the similarity between the expected parameters and the drainage fluid parameters, and adjusts the monitoring strategy according to the comparison result of the similarity and the first threshold. Through personalized prediction, it is possible to more accurately judge whether the drainage fluid parameters are abnormal, so that the predicted expected parameters can be more in line with the individual situation of the patient. The dynamic adjustment mechanism can respond to changes in the patient's condition in a timely manner. When there is a large deviation between the actual drainage fluid parameters and the expected parameters, timely adjusting the monitoring frequency or taking other intervention measures will help improve the accuracy of monitoring and reduce the possibility of missed diagnosis or misdiagnosis.

[0045] Further, predicting expected drainage fluid parameters, the method comprises: Extract medical history information and treatment information from the treatment plan to obtain wound characteristics; Extract drug information and nursing information from the doctor's prescription plan, and combine it with wound characteristics to obtain daily recovery characteristics; Extract complication information from the prognostic scheme to obtain complication characteristics; Obtain drainage fluid reference parameters based on patient physical information and wound characteristics; Wound characteristics, daily recovery characteristics, complication characteristics and drainage fluid reference parameters are input into the first prediction model, and drainage fluid parameters are output.

[0046] Specifically, the medical history information and treatment information are extracted from the treatment plan to obtain wound features. For example, the treatment plan records in detail the type of surgery (such as laparoscopic cholecystectomy), the size, location, depth and other information of the surgical wound. By analyzing this information, wound-related features can be extracted. These wound features will serve as one of the important bases for predicting expected parameters.

[0047] Drug information and nursing information are extracted from the doctor's prescription plan, and daily recovery characteristics are obtained by combining with wound characteristics. The doctor's prescription plan contains the types of drugs, dosages, medication time, and nursing requirements that patients need to use after surgery. For example, the doctor may instruct the patient to use antibiotics to prevent infection after surgery and change the wound dressing every day. Combined with wound characteristics, such as wound size and location, the changes in drainage fluid that may occur during the patient's daily recovery process can be analyzed. For example, a larger wound will heal more slowly, with a greater chance of interference, and more drainage fluid may be discharged.

[0048] Extract complication information from the prognostic plan and obtain complication characteristics. The prognostic plan is a plan to estimate and prevent complications that may occur after surgery. For example, for some surgical patients, there may be a risk of complications such as postoperative bleeding and infection. By analyzing the complication information in the prognostic plan and the individual conditions of the patient, the probability and possible manifestations of complications after surgery can be predicted. For example, if the patient is at risk of postoperative bleeding, then when predicting the expected parameters, it is necessary to take into account that the drainage fluid may become darker in color and the flow may increase.

[0049] Drainage fluid reference parameters are obtained based on the patient's vital signs information and wound characteristics. Patient vital signs information includes body temperature, blood pressure, heart rate, etc. These vital signs can reflect the patient's physical condition and recovery. Combined with wound characteristics, such as wound size and location, similar cases with similar patient signs are obtained from the medical database, and drainage fluid parameters of similar cases are obtained to determine drainage fluid reference parameters for patients at different recovery stages. For example, in the early postoperative period, the patient's body temperature may rise slightly and blood circulation may speed up. At this time, the flow of drainage fluid may be relatively large. As the patient recovers, the body temperature gradually returns to normal, and the drainage fluid flow will decrease accordingly.

[0050] Finally, wound characteristics, daily recovery characteristics, complication characteristics and drainage fluid reference parameters are input into the first prediction model, and drainage fluid parameters are output. The first prediction model can accurately predict the drainage fluid parameters of patients at different stages and time points based on the input of multiple characteristic information by learning and analyzing a large amount of historical medical data, providing a reference for medical staff so that they can adjust monitoring strategies and intervention measures in a timely manner.

[0051] Detailed analysis is performed on each part of the medical plan, wound characteristics are extracted from the treatment plan, daily recovery characteristics are obtained from the doctor's order plan, complication characteristics are determined from the prognosis plan, and drainage fluid reference parameters are obtained in combination with patient vital signs information. This comprehensive information integration and analysis can more accurately grasp the patient's disease progression and recovery, and identify potential risk factors in advance. The first prediction model comprehensively considers multiple characteristic information to predict drainage fluid parameters, which can more accurately predict the changes in drainage fluid that may occur in patients at different stages, so as to prepare for the response in advance.

[0052] Dynamically adjust the monitoring strategy based on the prediction results so that monitoring resources can be allocated more reasonably. For patients with stable conditions and drainage fluid parameters that meet expectations, the monitoring frequency can be appropriately reduced to reduce unnecessary consumption of medical resources; while for patients with large changes in their conditions and higher risks, the monitoring frequency can be increased to ensure timely detection and treatment of problems. This can not only ensure the safety of patients, but also improve the efficiency of medical resource utilization. By providing accurate expected parameters and risk assessments, it helps medical staff to formulate monitoring strategies and intervention measures more scientifically. Medical staff can prepare in advance based on the prediction results, arrange medical resources reasonably, and avoid excessive medical treatment or delayed treatment.

[0053] Embodiment 3: This embodiment is a further improvement based on the technical solution of embodiment 1.

[0054] As a preferred technical solution of this application, real-time monitoring of drainage fluid parameters based on the primary strategy includes: Perform limb analysis on the patient's image to extract motion features; analyze the patient's motion intensity based on the motion features to obtain the motion severity; Obtain the movement slow time according to the exercise intensity; adjust the monitoring strategy during the movement slow time; after the movement slow time ends, restore the monitoring strategy before adjustment.

[0055] Specifically, the system collects patient image data in real time through a camera or a camera. The patient's image can be collected by multiple cameras for full-body or partial shooting. Image acquisition includes static and dynamic images of the patient, which can fully reflect the patient's movements and limb status. The patient's body contour and limb parts are identified through the target detection algorithm, and then the key feature points of the limbs are extracted, including but not limited to joint positions, limb endpoints, and spinal curves. Based on these key feature points, the displacement and angle changes of the limbs between different key frame images are calculated to obtain the motion trajectory and motion state of the limbs.

[0056] Furthermore, the patient's image is subjected to limb analysis, including: Extract head key frames and limb key frames from the patient image to form action features; compare the action features with the preset action library to obtain the action type; The action frequency is calculated based on the number of action features, and the duration of the action features is counted; The action type, action frequency and duration are vectorized, and the obtained vectors are combined to obtain the motion features.

[0057] Specifically, when extracting the head keyframe and limb keyframe from the patient image, a target detection algorithm based on deep learning, such as the YOLOv8 model or the Mask R-CNN model, is used to accurately identify the head and limb parts in the patient image and extract the keyframes. For example, in different patient images, for the limb keyframes, the position of the joints, the angle formed by the joint bending, and the length of the limbs are extracted, and the displacement vector of the limb is obtained by calculating the change in the joint position between adjacent frames. The ratio of the displacement vector and time is used to calculate the movement speed and direction to more accurately describe the movement state of the limb. In the same image, the position before the key feature point is calculated to identify the basic movements of the patient. For example, the vertical distance between the key feature point of the elbow joint and the key feature point of the waist, and the horizontal distance between the key feature point of the knee joint, identify whether the patient is sitting, squatting or standing. By analyzing the position, posture and movement trajectory of the limb parts in the key frame, the action features are formed. These action features are used to reflect the patient's limb movement state and behavioral intention.

[0058] The action features are compared with the preset action library to obtain the action type. The preset action library contains a variety of common limb action patterns, such as turning over, raising hands, walking, etc. The patient's current action type is determined by calculating the similarity between the action features and each action pattern in the action library.

[0059] The action frequency is calculated based on the number and duration of action features. The action frequency reflects the speed of the patient's limb movement. The higher the action frequency, the greater the patient's strength. For example, the more times the arm is raised within a certain period of time, the higher the frequency of hand raising. The action duration can reflect the length of time the patient maintains a certain action. The longer the time, the greater the impact on the wound. By counting the duration of action features, we can have a more comprehensive understanding of the patient's movement.

[0060] The motion type, motion frequency, and duration are vectorized, and the resulting vectors are combined to obtain motion features. For example, the motion type can be represented by a one-hot encoding, and the motion frequency and duration can be represented by real numbers. By combining these vectors together, a comprehensive motion feature vector is formed to describe the patient's motion state and behavioral characteristics.

[0061] By performing limb analysis on the patient's images to extract motion features and determine the intensity of the exercise, we can more accurately understand the patient's physical activity status. Different exercise intensities have different effects on drainage fluid parameters. Adjusting the monitoring strategy accordingly can avoid over-monitoring when the patient's exercise is stable or under-monitoring when the exercise is intense, thereby improving the accuracy of drainage fluid parameter monitoring. Reasonable adjustment of monitoring strategies during the slow exercise time, such as focusing on key parameters such as flow rate at the beginning of exercise, and then gradually increasing attention to other parameters, and resuming routine monitoring after the slow exercise time ends. In this way, the distribution of drainage fluid parameter monitoring weights can be flexibly adjusted according to the patient's actual exercise situation, improving the response efficiency of abnormal situations.

[0062] Furthermore, before vectorization, it also includes: The wound position is obtained and fused with the patient image; the displacement of the motion feature and the wound position is calculated, and the vectorized discount factor is calculated based on the displacement.

[0063] Specifically, the coordinate system of the wound position is aligned with the coordinate system of the patient image, and the pixel information of the wound area is fused with the pixel information of the surrounding normal tissue to generate a new fused image. Based on the recognized motion features and the coordinate information of the wound position in the same frame image, the displacement between them is calculated. The inverse of the displacement is used as the discount factor during vectorization.

[0064] The closer the movement is to the wound, the higher the possibility of affecting the wound, causing the wound tissue to secrete more drainage fluid, so the monitoring attention should be higher. When the patient moves violently and the displacement is large, the discount factor can guide the system to pay more attention to the key parameters of the drainage fluid, such as real-time monitoring of the flow and color changes of the drainage fluid, while reducing the monitoring attention to other parameters. In this way, while ensuring the monitoring effect, key information can be highlighted and the response efficiency of abnormal situations can be improved.

[0065] The extracted motion features are vector normalized, and the modulus of the normalized vector is calculated to obtain the patient's motion intensity. The preset motion slow time is obtained according to the motion intensity. The setting of the motion slow time can be based on clinical experience, or a function fitting is performed on the motion intensity and the motion slow time, and the calculation is performed through the function.

[0066] Furthermore, when the patient is exercising and in a slow exercise period, the exercise intensity is graded, and the monitoring weight of the drainage fluid parameters is adjusted according to the grading results.

[0067] Specifically, in this embodiment, the threshold ranges of different exercise intensity levels are determined based on expert experience, and the exercise intensity is divided into three levels: low, medium, and high. When the patient is in a low exercise intensity state, since the effect of exercise on the drainage fluid parameters is relatively small, the monitoring weight of relatively stable drainage fluid parameters, such as drainage fluid color and turbidity, is reduced; focus on drainage fluid flow. In a medium exercise intensity state, increase the level of attention to the color and turbidity of the drainage fluid. When the patient is in a high exercise intensity state, keep other drainage fluid parameters unchanged, focus on the drainage fluid color, and perform real-time analysis and early warning of the data.

[0068] During the slow movement period, adjust the monitoring strategy. For example, lower the monitoring threshold for key parameters such as drainage fluid flow and color to detect abnormal wound bleeding in time. After the slow movement period is over, resume the monitoring strategy before adjustment. After resuming the monitoring strategy, it is necessary to analyze and summarize the previous monitoring data to evaluate the impact of exercise on the patient's drainage fluid parameters. If abnormal changes are found in the drainage fluid parameters, appropriate intervention measures need to be taken in a timely manner, and the subsequent monitoring strategy needs to be adjusted according to the patient's recovery.

[0069] After grading the intensity of exercise, the monitoring weights are adjusted according to different levels, so that monitoring resources can be reasonably allocated while ensuring the monitoring effect. During the patient's exercise and the slow period after exercise, the attention to key drainage fluid parameters, such as drainage fluid color and flow, is increased. The monitoring strategy can be adjusted accordingly, such as lowering the warning threshold of drainage fluid color. While meeting the monitoring needs of the patient's condition, abnormalities in drainage fluid during exercise can be detected in time, which is conducive to improving monitoring efficiency.

[0070] Embodiment 4: This embodiment provides a surgical drainage fluid monitoring system, device and storage medium based on image recognition.

[0071] For example Figure 3 As shown, a surgical drainage fluid monitoring system based on image recognition comprises: The parsing module is used to parse the historical medical plan and the corresponding historical monitoring strategy to obtain the corresponding first feature and second feature; and is used to parse the patient's medical plan and extract the current feature; A parameter configuration module, used to select a primary strategy from historical monitoring strategies based on the similarity between the current feature and the second feature; A monitoring module for real-time monitoring of drainage fluid parameters based on the primary strategy; The adjustment module is used to compare drainage fluid parameters with warning thresholds and determine drainage intervention measures.

[0072] The monitoring module includes: A flow sensing submodule, used for monitoring the drainage fluid flow in real time through a flow sensor; A color analysis submodule, which is used to collect drainage equipment images through optical sensors and extract drainage fluid image features from the drainage equipment images, including color features and turbidity features; A property analysis submodule is used to analyze the properties and concentration of drainage fluid through chemical sensors; The display control submodule is used to receive data from the analysis module, parameter configuration module and monitoring module, and display the data in real time after processing.

[0073] In actual use, first correctly connect the drainage tube connection to the patient's surgical drainage tube to ensure a good seal. After the system is started, the parameter configuration module checks the drainage fluid parameters, and the medical staff configures the monitoring strategies such as the monitoring frequency and early warning threshold of the drainage fluid, and transmits the configured monitoring strategy to the monitoring module through the data transmission module. After receiving the configured monitoring strategy, the monitoring module monitors the drainage fluid, including: the flow sensing submodule starts to monitor the flow of the drainage fluid in real time; the color analysis submodule collects the drainage fluid equipment image in real time and extracts the drainage fluid image features, including but not limited to the color and turbidity of the drainage fluid; the property analysis submodule simultaneously analyzes and detects the drainage fluid to obtain the properties and concentration of the drainage fluid.

[0074] The data transmission module transmits the collected monitoring strategy and drainage fluid parameters to the display control submodule. The display control submodule integrates and processes the data. On the one hand, it displays the flow, color, properties and other parameter information of the drainage fluid on the display screen for medical staff to view at any time; on the other hand, when the flow of the drainage fluid exceeds the preset safety range or the color or properties change abnormally, such as heavy bleeding causing the drainage fluid to be bright red, a sudden increase in the flow of the drainage fluid, and infection causing the drainage fluid to be turbid and the pH to be abnormal. The display control submodule immediately triggers the alarm device and issues an audible and visual alarm to remind medical staff to take appropriate treatment measures in a timely manner.

[0075] After taking measures, medical staff will determine whether the monitoring strategy needs to be adjusted. If the monitoring strategy is adjusted, it will be transmitted to the medical database for storage through the data transmission module; and the analysis module will extract the features of the medical plan and monitoring strategy, and store them in the medical database according to the adjustment sequence of the monitoring strategy.

[0076] A surgical drainage fluid monitoring device based on image recognition, the device includes: a sensor for collecting various data related to drainage fluid; at least one processor, capable of processing various data of drainage fluid collected by the sensor; at least one memory, for storing various data of drainage fluid collected by the sensor and at least one program; when at least one program is executed by at least one processor, at least one processor implements a surgical drainage fluid monitoring method based on image recognition, the processor and the memory can be connected by a bus or other means for storing at least one program; the memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer executable programs. In addition, the memory can include a high-speed random access memory, and can also include a non-transitory memory, such as at least one disk storage device, a flash memory device or other non-transitory solid-state storage device. In some embodiments, the memory optionally includes a memory remotely arranged relative to the processor, and these remote memories can be connected to the device via a network. Examples of the above-mentioned network include but are not limited to the Internet, an intranet, a local area network, a mobile communication network and a combination thereof, when at least one program is executed by at least one processor, at least one processor implements a surgical drainage fluid monitoring method based on image recognition.

[0077] A computer storage medium stores a program executable by a processor. When the program is executed by the processor, it is used to implement a surgical drainage fluid monitoring method based on image recognition.

[0078] It will be appreciated by those skilled in the art that all or some of the steps and systems in the disclosed method above may be implemented as software, firmware, hardware and appropriate combinations thereof. Some physical components or all physical components may be implemented as software executed by a processor, such as a central processing unit, a digital signal processor or a microprocessor, or may be implemented as hardware, or may be implemented as an integrated circuit, such as an application specific integrated circuit. Such software may be distributed on a computer-readable medium, which may include a computer storage medium (or a non-transitory medium) and a communication medium (or a temporary medium). As known to those skilled in the art, the term computer storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules or other data). Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tapes, disk storage or other magnetic storage devices, or any other medium that may be used to store desired information and may be accessed by a computer. Furthermore, it is well known to those skilled in the art that communication media typically embodies computer readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media.

[0079] The above embodiments are only used to illustrate the present invention and are not intended to limit the technical solutions described in the present invention. Although the present invention has been described in detail with reference to the above embodiments, the present invention is not limited to the above specific implementation methods. Therefore, any modification or equivalent replacement of the present invention; and all technical solutions and improvements thereof that do not depart from the spirit and scope of the invention are included in the scope of the claims of the present invention.

Claims

1. A surgical drainage fluid monitoring method based on image recognition, characterized in that: include: Analyze historical medical plans and corresponding historical monitoring strategies to obtain corresponding first and second features; Parse the patient's medical plan and extract current features; Based on the similarity between the current feature and the first feature, a primary strategy is selected from the historical monitoring strategies; Real-time monitoring of drainage fluid parameters based on primary strategies; Compare drainage fluid parameters and warning thresholds to determine drainage intervention measures.

2. A surgical drainage fluid monitoring method based on image recognition as claimed in claim 1, characterized in that: Real-time monitoring of drainage fluid parameters based on primary strategies, including: The drainage fluid flow is measured by a flow sensor; the drainage equipment image is collected by an optical sensor, and the drainage fluid image features are extracted from the drainage equipment image; the drainage fluid properties and concentration are analyzed by a chemical sensor; and the drainage fluid parameters are displayed in real time by a display control submodule.

3. A surgical drainage fluid monitoring method based on image recognition as claimed in claim 1, characterized in that: After screening the primary strategy, it also includes: using the first prediction model to predict expected drainage fluid parameters based on the patient's medical regimen; obtaining drainage fluid parameters corresponding to the primary strategy; The similarity between the expected parameters and the drainage fluid parameters is calculated, and the monitoring strategy is adjusted according to the comparison result between the similarity and the first threshold.

4. A surgical drainage fluid monitoring method based on image recognition as claimed in claim 3, characterized in that: Patient medical plans include patient vital signs information, treatment plans, doctor's orders and prognosis plans; Predict expected drainage fluid parameters by: Extract medical history information and treatment information from the treatment plan to obtain wound characteristics; Extract drug information and nursing information from the doctor's prescription plan, and combine it with wound characteristics to obtain daily recovery characteristics; Extract complication information from the prognostic scheme to obtain complication characteristics; Obtain drainage fluid reference parameters based on patient physical information and wound characteristics; Wound characteristics, daily recovery characteristics, complication characteristics and drainage fluid reference parameters are input into the first prediction model, and drainage fluid parameters are output.

5. The surgical drainage fluid monitoring method based on image recognition according to claim 1, characterized in that: Real-time monitoring of drainage fluid parameters based on primary strategies, including: Perform limb analysis on the patient's image to extract motion features; analyze the patient's motion intensity based on the motion features to obtain the motion severity; Obtain the movement slow time according to the exercise intensity; adjust the monitoring strategy during the movement slow time; after the movement slow time ends, restore the monitoring strategy before adjustment.

6. A surgical drainage fluid monitoring method based on image recognition as claimed in claim 5, characterized in that: include: When the patient is exercising and in a slow exercise period, the intensity of the exercise is graded, and the monitoring weight of the drainage fluid parameters is adjusted according to the classification results.

7. A surgical drainage fluid monitoring method based on image recognition as claimed in claim 5, characterized in that: Perform limb analysis on patient images, including: Extract head key frames and limb key frames from the patient image to form action features; compare the action features with the preset action library to obtain the action type; The action frequency is calculated based on the number of action features, and the duration of the action features is counted; The action type, action frequency and duration are vectorized, and the obtained vectors are combined to obtain the motion features.

8. A surgical drainage fluid monitoring method based on image recognition as claimed in claim 7, characterized in that: Before vectorization, it also includes: The wound position is obtained and fused with the patient image; the displacement of the motion feature and the wound position is calculated, and the vectorized discount factor is calculated based on the displacement.

9. A surgical drainage fluid monitoring system based on image recognition, characterized in that: include: A parsing module, used for parsing historical medical plans and corresponding historical monitoring strategies to obtain corresponding first and second features; Used to analyze patient medical plans and extract current features; A parameter configuration module, used to select a primary strategy from historical monitoring strategies based on the similarity between the current feature and the second feature; A monitoring module for real-time monitoring of drainage fluid parameters based on the primary strategy; The adjustment module is used to compare drainage fluid parameters with warning thresholds and determine drainage intervention measures.

10. A surgical drainage fluid monitoring system based on image recognition as claimed in claim 9, characterized in that: The monitoring modules include: A flow sensing submodule, used for monitoring the drainage fluid flow in real time through a flow sensor; A color analysis submodule, which is used to collect drainage equipment images through optical sensors and extract drainage fluid image features from the drainage equipment images, including color features and turbidity features; A property analysis submodule is used to analyze the properties and concentration of drainage fluid through chemical sensors; The display control submodule is used to receive data from the analysis module, parameter configuration module and monitoring module, and display the data in real time after processing.

Images