Clustering processing-based postoperative wound abnormality monitoring system for vascular malformation
By constructing a three-in-one objective function and a dynamic adjustment mechanism, the problems of misjudgment and adaptation of the wound monitoring system after vascular malformation surgery were solved, the accuracy and adaptability of monitoring were improved, and refined wound condition judgment was achieved.
Patent Information
- Application Number
- CN202511526482.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-10-24
- Publication Date
- 2025-11-21
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Existing wound abnormality monitoring systems after vascular malformation surgery lack effective interference elimination mechanisms, are prone to misjudging wound status, and suffer from misjudgment due to single-point measurement errors. They also cannot adapt to dynamic changes in postoperative sensitivity, resulting in poor monitoring reliability and effectiveness.
A three-in-one objective function of normal recovery, anomaly detection, and interference elimination is constructed. Spatial correlation factors and dynamic interference deviation are introduced, and weight enhancement terms are designed. Combined with the dynamic transition of rigidity and elasticity, the accuracy and adaptability of the monitoring system are improved through intra-cluster fine-tuning, cross-cluster fusion, and random adjustment.
It improves the accuracy and effectiveness of postoperative wound abnormality monitoring, reduces the impact of interfering data, avoids misjudgment due to single-point errors, dynamically matches the needs of different postoperative stages, and achieves refined judgment.
Smart Images
Figure CN120995149A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The application relates to the field of data processing, in particular to a postoperative wound anomaly monitoring system for vascular malformation based on clustering processing. BACKGROUND
[0002] The postoperative wound anomaly monitoring system for vascular malformation is used for collecting physiological and healing process index data of postoperative wounds of patients with vascular malformation, and monitoring abnormal states such as wound infection and poor healing. However, the general postoperative wound anomaly monitoring system for vascular malformation has the problems of lacking an effective interference elimination mechanism, being prone to misjudgment of the wound state, single-point measurement error leading to misjudgment, and being unable to adapt to postoperative dynamic sensitivity changes, thereby resulting in poor abnormal monitoring reliability. The general postoperative wound anomaly monitoring system for vascular malformation has the problems of being unable to adapt to the demand differences in different postoperative stages, being difficult to match postoperative dynamic sensitive features, and thereby resulting in poor postoperative wound anomaly monitoring effect. SUMMARY
[0003] In view of the above problems, in order to overcome the defects of the prior art, the application provides a postoperative wound anomaly monitoring system for vascular malformation based on clustering processing. In view of the problems of the general postoperative wound anomaly monitoring system for vascular malformation, such as lacking an effective interference elimination mechanism, being prone to misjudgment of the wound state, single-point measurement error leading to misjudgment, and being unable to adapt to postoperative dynamic sensitivity changes, thereby resulting in poor abnormal monitoring reliability, the present application constructs a trinity target function of normal recovery-abnormal judgment-interference elimination, which contains an exclusive interference elimination term and a dynamic interference deviation degree, reduces the influence of postoperative wound interference data on clustering, and avoids misjudgment. A spatial correlation factor is introduced to capture the spatial diffusion characteristics of postoperative wounds and avoid single-point error misjudgment. A weight reinforcement term is designed to highlight key abnormal indicators of postoperative wounds and ensure that abnormal judgment in different stages focuses on core indicators, thereby improving accuracy. Based on the dynamic interference deviation degree, individual differences are accurately adapted to avoid normal fluctuations of postoperative wounds being classified as interference or abnormality. The postoperative wound anomaly monitoring accuracy is improved. In view of the problems of the general postoperative wound anomaly monitoring system for vascular malformation, such as being unable to adapt to the demand differences in different postoperative stages, being difficult to match postoperative dynamic sensitive features, and thereby resulting in poor postoperative wound anomaly monitoring effect, the present application combines dynamic transition from rigid fit degree to elastic fit degree, dynamically matches postoperative stage requirements, and optimizes indicator sensitivity. An interference distance is introduced to finely distinguish critical states and realize fine judgment. Based on dynamic index weight updating, the core indicators in different stages are highlighted to improve the judgment pertinence. Based on the combination of intra-cluster fine tuning, cross-cluster fusion and random adjustment, the adaptability of the cluster center to recovery differences is improved, and the continuous transition characteristics of normal and abnormal are captured. The postoperative wound anomaly effect is improved.
[0004] The technical scheme adopted by the present application is as follows: The postoperative wound anomaly monitoring system based on clustering processing provided by the present application comprises a data acquisition module, a clustering parameter initialization module, a target function definition module, an adaptive wound anomaly fitness updating module, a cluster center adjustment module, and a wound state monitoring module.
[0005] The data acquisition module acquires historical postoperative wound data to obtain a postoperative wound sample set.
[0006] The clustering parameter initialization module sets initial clustering parameters.
[0007] The target function definition module constructs a basic target function containing a spatial correlation factor and adds a label penalty term to the historical postoperative wound data to obtain a total target function.
[0008] The adaptive wound anomaly fitness updating module combines rigid fitness and elastic fitness by using a transition coefficient to obtain total fitness.
[0009] The cluster center adjustment module completes cluster center adjustment through intra-cluster fine tuning, cross-cluster adjustment, and random adjustment.
[0010] The wound state monitoring module clusters the postoperative wound sample set and monitors the state of real-time postoperative wound data based on the clustering result.
[0011] Further, the data acquisition module acquires historical postoperative wound data; the historical postoperative wound data includes physiological index data, healing process data, and biochemical index data; the historical postoperative wound data is labeled with wound monitoring labels and preprocessed; and a postoperative wound sample set is obtained.
[0012] Further, the clustering parameter initialization module sets initial clustering parameters; specifically, a fixed clustering number C=3 is set, representing interference data, normal recovery, and clear anomaly; cluster centers and index weights are initialized; the cluster centers are initialized as the mean value of each label; and the index weights are initialized without bias.
[0013] Further, the target function definition module introduces a spatial correlation factor to construct a target function, including a normal anomaly clustering term and an interference deviation; and a label penalty term is added during target function iteration.
[0014] Further, the adaptive wound anomaly fitness updating module defines total fitness including rigid fitness and elastic fitness, and updates index weights based on target function minimization.
[0015] Further, the cluster center adjustment module is to perform intra-cluster adjustment, fine-tune the cluster center in the same cluster, adjust based on the current monitoring data in the cluster, introduce a risk factor, dynamically adjust the fine-tuning amplitude based on the mean of the current abnormal cluster fitting degree, then perform cross-cluster adjustment, and finally perform random adjustment.
[0016] Further, the wound state monitoring module is to, for a postoperative wound sample set, set initial clustering parameters based on the clustering parameter initialization module, determine the iteration convergence condition based on the total objective function, assign samples in the postoperative wound sample set based on the total fitting degree, update the cluster center based on the cluster center adjustment module, and obtain the clustering result of the postoperative wound sample set; based on the clustering result, monitor the wound state of the real-time postoperative wound.
[0017] The above-mentioned scheme has the following beneficial effects:
[0018] (1) For the general postoperative wound abnormality monitoring system of vascular malformations, there is a lack of effective interference elimination mechanism, easy to misjudge the wound state, single-point measurement error misjudgment, and unable to adapt to the dynamic sensitivity change after surgery, thereby leading to poor reliability of abnormal monitoring. The present scheme constructs a normal recovery-abnormality judgment-interference elimination trinity objective function, including a dedicated interference elimination term and a dynamic interference deviation, reduces the influence of postoperative wound interference data on clustering, and avoids misjudgment. A spatial correlation factor is introduced to capture the spatial diffusion characteristics of postoperative wounds and avoid single-point error misjudgment. A weight reinforcement term is designed to highlight key abnormal indicators of postoperative wounds and ensure that abnormal judgments at different stages focus on core indicators, improving accuracy. Based on the dynamic interference deviation, individual differences are accurately adapted to avoid classifying normal fluctuations in postoperative wounds as interference or abnormalities. Thus, the accuracy of postoperative wound abnormality monitoring is improved.
[0019] (2) For the general postoperative wound abnormality monitoring system of vascular malformations, it cannot adapt to the demand differences at different stages after surgery, and it is difficult to match the dynamic sensitivity characteristics after surgery, thereby leading to poor postoperative wound abnormality monitoring effect. The present scheme combines the dynamic transition from rigid fitting degree to elastic fitting degree, dynamically matches the postoperative stage requirements, and optimizes the index sensitivity. A fine distinction between critical states is introduced to achieve fine judgment. Based on dynamic index weight updating, the core indicators at different stages are highlighted to improve the judgment specificity. Based on the combination of intra-cluster fine-tuning, cross-cluster fusion and random adjustment, the adaptability of the cluster center to recovery differences is improved, and the continuous transition characteristics of normal and abnormal are captured. Thus, the postoperative wound abnormality effect is improved. BRIEF DESCRIPTION OF DRAWINGS
[0020] Figure 1 The flowchart of the postoperative wound abnormality monitoring system of vascular malformations based on clustering processing provided by the present application is shown.
[0021] The accompanying drawings are included to provide a further understanding of the application, and are incorporated in and constitute a part of the specification, illustrate embodiments of the application, and are used to explain the present application, but are not intended to limit the present application. DETAILED DESCRIPTION
[0022] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of the present application.
[0023] In the description of the present application, it should be understood that the terms "upper", "lower", "front", "back", "left", "right", "top", "bottom", "inner", "outer" and the like indicate the orientation or positional relationship shown in the drawings based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation of the present application.
[0024] Embodiment one, refer to Figure 1 The present application provides a postoperative wound anomaly monitoring system based on cluster processing of vascular malformation, comprising a data acquisition module, a cluster parameter initialization module, a target function definition module, an adaptive wound anomaly fitness updating module, a cluster center adjustment module and a wound state monitoring module.
[0025] The data acquisition module acquires historical postoperative wound data to obtain a postoperative wound sample set, and sends the data to the cluster parameter initialization module.
[0026] The cluster parameter initialization module sets initial cluster parameters, and sends the data to the target function definition module.
[0027] The target function definition module constructs a basic target function containing a spatial correlation factor, adds a label penalty term to the historical postoperative wound data to obtain a total target function, and sends the data to the adaptive wound anomaly fitness updating module.
[0028] The adaptive wound anomaly fitness updating module combines the rigid fitness and the elastic fitness by using a transition coefficient to obtain a total fitness, and sends the data to the cluster center adjustment module.
[0029] The cluster center adjustment module completes cluster center adjustment through intra-cluster fine tuning, cross-cluster adjustment and random adjustment, and sends the data to the wound state monitoring module.
[0030] The wound state monitoring module clusters the postoperative wound sample set and performs state monitoring on real-time postoperative wound data based on the clustering result.
[0031] Embodiment Two, see Figure 1 , based on the above embodiment, the data acquisition module acquires historical postoperative wound data; the historical postoperative wound data includes physiological index data, healing process data and biochemical index data; the physiological index data includes local wound skin temperature, patient body temperature, wound peripheral tissue oxygen saturation and heart rate; the healing process data includes wound exudate volume, swelling range and epithelial coverage; the biochemical index data includes peripheral blood leukocyte count, C-reactive protein and procalcitonin; the historical postoperative wound data is labeled with wound monitoring labels and preprocessed, including missing value processing and normalization processing; the wound monitoring labels include interference data, normal recovery and clear abnormality; the interference data refers to non-real wound state data caused by instrument error, improper operation or physiological temporary fluctuation, including skin temperature measurement value > 40 degrees caused by instrument failure, transient heart rate increase within 2 hours after operation (not caused by infection), and single abnormal high value caused by residual dressing change during exudate volume measurement; a postoperative wound sample set is obtained.
[0032] Embodiment Three, see Figure 1 , based on the above embodiment, the clustering parameter initialization module is to set the initial clustering parameters; the specific operation is: setting the fixed clustering number C = 3, representing interference data, normal recovery and clear abnormality respectively; and initializing the cluster center and index weight; the cluster center is initialized as the mean value of each label; the index weight is initialized without bias.
[0033] Embodiment Four, see Figure 1 , based on the above embodiment, the target function definition module is to construct a trinity of normal recovery- abnormality judgment- interference exclusion target function for the postoperative data of vascular malformation surgery with strong interference and dynamic characteristics, update the wound abnormality fitness and index weight at the same time, and realize accurate clustering of each round of monitoring data; the specific operation is:
[0034] Construct the target function to distinguish normal and abnormal data, exclude interference data including instrument error and optimize index weight, and highlight key abnormal indicators; and introduce a spatial correlation factor to strengthen the weight of regional gradient change, capture the spatial diffusion characteristics of swelling and skin temperature, and avoid misjudgment caused by single-point measurement error; the basic target function is represented as: ; ; ; is the interference deviation degree of the jth postoperative wound sample; is the intensity correction parameter; is the change gradient of the jth postoperative wound sample at the kth index, taking the mean of the difference rate of the measurement point and each neighborhood measurement value; when the objective function is iterated, a label penalty term is added to the historical labeled data, represented as: ; wherein, is the basic objective function; i is the cluster index, i=1 corresponds to the normal recovery cluster, i=2 corresponds to the explicit abnormal cluster; N is the total number of samples, j is the sample index; D is the total dimension number, k and are dimension indexes; is the fit degree of the jth postoperative wound sample to the ith cluster; m is the elasticity coefficient; and are the weights of the kth and the th index in the ith cluster, respectively; is the weight reinforcement index; is the value of the kth index of the jth postoperative wound sample; is the mean value of the kth index of the ith cluster, std represents standardization; p is the deviation coefficient; J is the total objective function; is the label correction coefficient; is the fit degree of the jth postoperative wound sample to the cluster corresponding to the true label, if the true label is interference data, then ;
[0035] The first part is the normal-abnormal clustering term, which strengthens the key indicators through the weight term , calculates the deviation of the data from the cluster center through the distance term , and realizes the coordinated judgment of the index weight-deviation degree; the second part is the interference exclusion term, which identifies the interference data that neither belongs to normal nor to abnormal through , and reduces its influence on clustering through the interference deviation , to avoid misjudgment of abnormality due to instrument error;
[0036] The interference deviation is calculated, which is dynamically adjusted based on the clinical normal range to avoid poor adaptability caused by fixed threshold, represented as: ; the weighted deviation of the data from the normal center is calculated based on the index weight of the normal recovery cluster, and then the threshold is expanded through the interference coefficient to ensure that only the data far beyond the normal fluctuation range is classified as interference, which adapts to the postoperative basic state of different patients; is the interference coefficient, which dynamically expands the tolerance threshold of normal fluctuation; is the weight of the kth index of the normal recovery cluster; is the mean value of the kth index of the normal recovery cluster.
[0037] By performing the above operations, this solution addresses the problems of general vascular malformation postoperative wound abnormality monitoring systems, which lack effective interference elimination mechanisms, are prone to misjudging wound status, suffer from misjudgment due to single-point measurement errors, and cannot adapt to postoperative dynamic sensitivity changes, thus leading to poor reliability of abnormality monitoring. This solution constructs a three-in-one objective function of normal recovery, abnormality judgment, and interference elimination, including a dedicated interference elimination term and dynamic interference deviation, to reduce the impact of postoperative wound interference data on clustering and avoid misjudgment. It introduces a spatial correlation factor to capture the spatial diffusion characteristics of postoperative wounds, avoiding misjudgment due to single-point errors; designs a weighted enhancement term to highlight key postoperative wound abnormal indicators, ensuring that abnormality judgments at different stages focus on core indicators and improve accuracy; and accurately adapts to individual differences based on dynamic interference deviation, avoiding classifying normal postoperative wound fluctuations as interference or abnormalities; thereby improving the accuracy of postoperative wound abnormality monitoring.
[0038] Example 5, see Figure 1 This embodiment is based on the above embodiment. The adaptive wound abnormal fit update module has different requirements for judgment speed and accuracy at different stages after surgery: in the early stage (1-5 days), obvious abnormalities need to be identified quickly, and in the later stage (after 5 days), critical abnormalities need to be distinguished. Therefore, a dynamic transition from rigid fit to elastic fit is adopted.
[0039] The specific steps are as follows: Define the total fit, expressed as: t represents the number of days of postoperative monitoring. It is the transition coefficient. ;in, It refers to the degree of elasticity and fit. It represents the rigidity of the fit; t represents the number of postoperative monitoring days; MaxDay represents the maximum number of monitoring days. It is a transition adjustment parameter; when t=1, The overall fit is rigid clustering, enabling rapid screening of obvious anomalies; when t= hour, The overall fit is the elastic fit, enabling precise judgment of critical states;
[0040] Rigid fit is expressed as: ; It is a weighted distance; The degree of elastic fit is expressed as: ;in, and These are the k-th index of the q-th cluster and the index of the k-th cluster. The weight of the indicator; It is the weighting highlighting coefficient; It is the weighted distance between the j-th postoperative wound sample and the i-th cluster; interference distance is introduced to avoid misjudgment of critical data;
[0041] Indicator weight updating; to dynamically adapt to the sensitivity changes of postoperative indicators, update the weight based on the minimization of the objective function, denoted as: ; wherein, is a smoothing term; is the updated indicator weight.
[0042] Embodiment six, see Figure 1 , this embodiment is based on the above embodiment, the cluster center adjustment module is that the postoperative abnormalities of vascular malformations may be caused by single indicator abnormalities (only CRP elevation) or multi-indicator coordinated abnormalities (CRP elevation + exudate pH abnormality + blood flow velocity slowing down), in order to avoid falling into local optimum (misjudgment of abnormality only by high skin temperature) due to unreasonable initial cluster center; Therefore, parameter adjustment is carried out to balance the accuracy of local indicators and the comprehensiveness of global judgment;
[0043] The specific operation is:
[0044] In-cluster adjustment, the cluster center is fine-tuned in the same cluster, based on the current monitoring data adjustment, to improve the judgment accuracy in the cluster, denoted as: ; is a related fine-tuning factor, a risk factor is introduced, based on the mean of the current abnormal cluster fit degree, the fine-tuning amplitude is dynamically adjusted, so that the cluster center better adapts to the subtle differences of individual recovery, ; ; wherein, is the center value of the kth indicator of the ith cluster after in-cluster adjustment; is the cluster center value before in-cluster adjustment, taking the mean of the cluster indicators; is a random number; is a sigmoid function; K is the steepness parameter, which controls the steepness of the Sigmoid function; is the mean of the abnormal cluster fit degree;
[0045] Cross-cluster adjustment, the cluster center performs cross-cluster parameter optimization, the normal recovery after vascular malformation surgery and the clear abnormality are not two absolutely separate states, but a continuous spectrum with dynamic transition, critical elasticity and significant individual differences. The normal cluster parameters may focus too much on the physiological stable state, ignoring the critical transition from slight abnormality to clear abnormality. The abnormal cluster parameters may pay too much attention to significant abnormal features, ignoring individualized differences related to the patient's basic state. Therefore, cross-cluster adjustment is carried out to achieve information complementation through weighted fusion, denoted as: ; ; wherein, is the cluster center value after cross-cluster adjustment; is the cluster center value before cross-cluster adjustment; is the fusion coefficient; is different from the cluster center value after in-cluster adjustment; ;
[0046] Random adjustment; in order to prevent falling into local optimum, random adjustment is performed every 5 iterations, denoted as: , execute ; is the cluster center value after random adjustment; is the adjustment probability threshold.
[0047] By performing the above operations, in view of the problem that the postoperative wound abnormality monitoring system for general vascular deformity cannot adapt to the demand differences in different postoperative stages, it is difficult to match the postoperative dynamic sensitive features, and thus the postoperative wound abnormality monitoring effect is poor, the present scheme combines the dynamic transition from rigid fit degree to elastic fit degree, dynamically matches the postoperative stage demand, and optimizes the index sensitivity; The interference distance is introduced to distinguish the critical state, and the fine judgment is realized; Based on dynamic index weight update, ensure that the core index in different stages is highlighted, improve the judgment pertinence; Based on the combination of intra-cluster fine tuning, cross-cluster fusion and random adjustment, improve the adaptability of cluster center to recovery difference, capture the continuous transition features of normal and abnormal; And improve the postoperative wound abnormality effect.
[0048] Embodiment seven, see Figure 1 , based on the above embodiment, the wound state monitoring module is for the postoperative wound sample set, based on the initial clustering parameter setting module, based on the total objective function to determine the iteration convergence condition, based on the total fit degree to distribute the postoperative wound sample, based on the cluster center adjustment module to update the cluster center, set the maximum iteration number and the convergence threshold, when the target function change is less than the convergence threshold or reaches the maximum iteration number, the clustering is determined, if the label proportion of the most number in the cluster is less than 30%, the clustering parameter is adjusted, optimized by particle swarm search algorithm, otherwise the clustering is ended, and the clustering result of the postoperative wound sample set is obtained; The clustering parameters include index weight, objective function parameter, total fit degree parameter and cluster center adjustment parameter; Based on the clustering result, the abnormal cluster fit degree of real-time postoperative wound data and index fluctuation range are calculated, and the grading result is output; The specific operation is: real-time postoperative wound data is obtained and pretreated, the fit degree of real-time postoperative wound data to clear abnormal cluster is calculated; And wound state grading, divided into normal recovery, suspicious abnormality and clear abnormality; Normal recovery is: , and all index standardized values are in [-1, 1]; Suspicious abnormality is: , or 1-2 indexes exceed normal fluctuation; Clear abnormality is: , or not less than 3 indexes exceed normal fluctuation; is the total fit degree of real-time postoperative wound data to clear abnormal cluster (i=2 corresponding cluster); If the grading result of real-time postoperative wound data is clear abnormality, warning is carried out, if it is suspicious abnormality, sampling frequency is increased.
[0049] While embodiments of the application have been shown and described, it is to be understood that the embodiments described are merely exemplary and that changes can be made in the embodiments without departing from the spirit and scope of the application.
[0050] The above description of the application and its embodiments is not restrictive, and the embodiments shown in the drawings are only one of the embodiments of the application, and the actual structure is not limited thereto. In general, if a person skilled in the art is inspired by it, without departing from the purpose of the application, without creative design, similar structure and embodiments of the technical solution should belong to the protection scope of the application.
Claims
1. A post-surgical wound abnormality monitoring system based on clustering processing for vascular malformations, characterized in that: The system comprises a data acquisition module, a clustering parameter initialization module, a target function definition module, an adaptive wound abnormality fitness updating module, a cluster center adjustment module and a wound state monitoring module. The data acquisition module acquires historical postoperative wound data to obtain a postoperative wound sample set. The clustering parameter initialization module sets initial clustering parameters. The target function definition module constructs a basic target function containing a spatial correlation factor and adds a label penalty term to the historical postoperative wound data to obtain a total target function. The adaptive wound abnormality fitness updating module combines the rigid fitness and the elastic fitness by using a transition coefficient to obtain a total fitness. The cluster center adjustment module completes cluster center adjustment through intra-cluster fine-tuning, cross-cluster adjustment and random adjustment. The wound state monitoring module clusters the postoperative wound sample set and monitors the state of real-time postoperative wound data based on the clustering result.
2. The cluster processing based post-surgical wound abnormality monitoring system for vascular malformations of claim 1, wherein: The target function definition module introduces a spatial correlation factor to construct a target function, which includes a normal-abnormal clustering term and a disturbance deviation degree.
3. The cluster processing based post-surgical wound abnormality monitoring system of claim 2, wherein: The adaptive wound abnormality fitness updating module defines the total fitness including the rigid fitness and the elastic fitness and updates the index weight based on the minimization of the target function.
4. The cluster processing based post-surgical wound abnormality monitoring system of claim 3, wherein: The cluster center adjustment module adjusts the cluster center in the same cluster, adjusts the cluster center based on the current monitoring data in the cluster, introduces a risk factor, dynamically adjusts the fine-tuning amplitude based on the mean value of the current abnormal cluster fitness, then performs cross-cluster adjustment, and finally performs random adjustment.
5. The cluster processing based post-surgical wound abnormality monitoring system of claim 4, wherein: The data acquisition module acquires historical postoperative wound data; the historical postoperative wound data includes physiological index data, healing process data and biochemical index data; the historical postoperative wound data is labeled with wound monitoring labels and preprocessed; and a postoperative wound sample set is obtained.
6. The cluster processing based post-surgical abnormality monitoring system for vascular malformations of claim 5, wherein: The clustering parameter initialization module sets initial clustering parameters; specifically, a fixed clustering number C=3 is set, representing disturbance data, normal recovery and clear abnormality; cluster centers and index weights are initialized; the cluster centers are initialized as the mean value of each label; and the index weights are initialized without bias.
7. The cluster processing based post-surgical abnormality monitoring system for vascular malformations of claim 6, wherein: The wound state monitoring module clusters the postoperative wound sample set based on the initial clustering parameter setting of the clustering parameter initialization module, determines the convergence condition based on the total target function, distributes the samples in the postoperative wound sample set based on the total fitness, updates the cluster center based on the cluster center adjustment module, and obtains the clustering result of the postoperative wound sample set. Based on the clustering result, the wound state of real-time postoperative wound is monitored.