An intelligent management method for skin plastic surgery data

Abstract

The present invention relates to the technical field of data processing, and proposes an intelligent management method for skin plastic surgery data, including: obtaining word vectors of the patient's skin medical data, historical user behavior data and user behavior data of users who have browsing rights to the skin medical data; determining the privacy of the user under browsing rights and the potential hidden dangers of the user; determining the adjusted real-time behavior data difference of the same type of data of the user according to the fluctuation amplitude of the same type of data in the user's historical user behavior data, and the difference between the user's user behavior data and the same type of data in the historical user behavior data, determining the user's real-time abnormality index according to all the adjusted real-time behavior data differences of the user and the user's potential hidden dangers; managing skin plastic surgery data according to the user's real-time abnormality index. The present invention can improve the accuracy of abnormal access data identification and realize the reasonable management of skin plastic surgery data.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to an intelligent management method for skin plastic surgery data. Background Art

[0002] Skin plastic surgery hospitals will generate and save corresponding skin medical data for each patient who undergoes skin plastic surgery, while protecting the patient's personal privacy while retaining the patient's skin medical data. In order to ensure the smooth progress of medical work and improve the medical level, different doctors, nurses and other personnel within the plastic surgery hospital have certain access and browsing permissions to all skin medical data stored in the skin plastic surgery hospital. However, in order to protect the patient's personal privacy and avoid the leakage of skin medical data, the specific restrictions on browsing permissions are different. In order to further ensure the security of skin medical data, it is necessary to conduct data security monitoring on all users with access rights to skin medical data, and promptly restrict user permissions when abnormal access data occurs, reduce the permission level, and limit access content.

[0003] In the process of data security monitoring of users, user behavior is generally directly compared with historical user behavior. When a user's behavior differs greatly from historical user behavior, it is determined that abnormal access data has occurred. However, when a user browses skin medical data, it is reasonable for the user to view all skin medical data within the scope of authority. If the user only determines whether abnormal access data has occurred based on the difference between the user behavior and the historical user behavior, the obtained judgment result is often inaccurate. Summary of the invention

[0004] The present invention provides an intelligent management method for skin plastic surgery data to solve the problem that abnormal access data is easily misjudged based on the difference between user behavior and historical user behavior, resulting in unreasonable management of skin plastic surgery data. The technical solution adopted is as follows:

[0005] An embodiment of the present invention provides an intelligent management method for skin plastic surgery data, the method comprising the following steps:

[0006] Collect the patient's skin medical data, historical user behavior data and user behavior data of users who have browsing rights to the patient's skin medical data, and extract word vectors of the patient's skin medical data;

[0007] Based on the similarity between all different word vectors, similar case clusters are determined. Based on the similarity between the meanings of sentences corresponding to all different word vectors, similar sentence clusters are determined. Based on the relationship between the skin medical data included in the user's browsing permissions and similar case clusters and similar sentence clusters, the user's privacy under browsing permissions is determined. Based on the user's privacy under all browsing permissions, the user's potential risks are determined.

[0008] Determine the adjusted real-time behavior data difference of the same type of data of the user according to the fluctuation range of the same type of data in the historical user behavior data of the user, and the difference between the user behavior data of the user and the same type of data in the historical user behavior data, and determine the real-time abnormality index of the user according to all the adjusted real-time behavior data differences of the user and the potential hidden dangers of the user;

[0009] Manage skin plastic surgery data based on the user's real-time abnormality index.

[0010] Furthermore, the method of determining similar case clusters based on the similarities between all different word vectors includes the following specific methods:

[0011] Any word vector of any patient's skin medical data is recorded as the target word vector, the average value of the similarity between the target word vector and each other word vector is recorded as the word vector similarity of the target word vector, and the normalized value of the cumulative sum of the word vector similarities of all word vectors of the same patient is recorded as the case similarity of the patient's skin medical data;

[0012] Based on the case similarities of all patients, clusters of similar cases were identified.

[0013] Furthermore, the method of determining similar case clusters based on the case similarities of all patients includes the following specific methods:

[0014] The ISOdata clustering algorithm was used to cluster the case similarities of all patients to obtain similar case clusters.

[0015] Furthermore, the method of determining similar sentence clusters according to the similarities between the sentence meanings corresponding to all different word vectors includes the following specific methods:

[0016] The patient's word vector is converted into the patient's sentence vector, and the sentence vectors of all patients are clustered to obtain similar sentence clusters.

[0017] Furthermore, the privacy of the user under browsing authority is obtained in the following specific method:

[0018] The mean of the case similarities contained in the similar case cluster is recorded as the similarity distribution of the similar case cluster, the cumulative sum of the similarity distributions of the similar case clusters corresponding to all skin medical data that the user has browsing permissions is recorded as the user's browsing permission similarity distribution, the opposite of the user's browsing permission similarity distribution is taken as the exponent value of the exponential function with a natural constant as the base, and the calculated value of the exponential function is recorded as the user's privacy under browsing permissions.

[0019] Furthermore, the method of determining the potential risk of the user includes:

[0020] The maximum privacy value of the user under all browsing permissions is recorded as the user's potential risk.

[0021] Further, the method of determining the adjusted real-time behavior data difference of the same type of data of the user according to the fluctuation range of the same type of data in the historical user behavior data of the user and the difference between the user behavior data of the user and the same type of data in the historical user behavior data includes the following specific methods:

[0022] Any kind of historical user behavior data in the historical user behavior data of the user is recorded as the target historical kind data, the similarity between the word vectors of any two target historical kind data is calculated, and the average value of the similarity between the word vectors of all the target historical kind data is recorded as the fluctuation stability of the target historical kind data; the opposite number of the fluctuation stability of the target historical kind data is taken as the exponent value of the exponential function with the natural constant as the base, and the calculated value of the exponential function is recorded as the fluctuation amplitude of the target historical kind data;

[0023] According to the difference between the fluctuation range of the user behavior data of the user and the historical data of the same category, and the data distribution of the same category in the historical user behavior data of the user, the adjusted real-time behavior data difference of the same category of the user is determined.

[0024] Further, the method of determining the adjusted real-time behavior data difference of the same type of data of the user according to the difference between the fluctuation range of the user behavior data of the user and the historical data of the same type, and the data distribution of the same type in the historical user behavior data of the user, includes the following specific methods:

[0025] The mean of the word vectors of the target category data in all the historical user behavior data of the user is recorded as the historical standard data of the user. For the target historical category data, the absolute value of the difference between the word vector of the user's user behavior data and the user's historical standard data is recorded as the real-time behavior data difference of the user under the target historical category data;

[0026] The ratio of the real-time behavior data difference of the user under the target historical category data to the historical standard data of the user is recorded as the adjusted real-time behavior data difference of the target historical category data of the user.

[0027] Furthermore, the specific method of determining the real-time abnormality index of the user includes:

[0028] The real-time abnormality index of the user is the product of the cumulative sum of the differences in the user's adjusted real-time behavior data in all historical types of data and the user's potential hidden dangers.

[0029] Furthermore, the specific method of managing skin plastic surgery data according to the user's real-time abnormality index is as follows:

[0030] When the user's real-time abnormality index is greater than the abnormality threshold, all access and browsing permissions of the user are closed;

[0031] When the user's real-time anomaly index is less than or equal to the anomaly threshold, the user's permissions are not adjusted.

[0032] The beneficial effects of the present invention are:

[0033] According to the characteristics that the information expressed by conventional skin medical data is universal, does not involve pathological problems and has weak privacy, the application evaluates the privacy degree of information in the user's browsing permissions according to the similarity between word vectors and the similarity between the meanings of sentences corresponding to different word vectors, obtains the privacy degree of the user under the browsing permissions, and considers that when the user's account is abnormal or data leakage occurs, the harmful consequences should be determined by the maximum value of the privacy degree under the browsing permissions, and obtains the potential hidden dangers of the user; since it is reasonable for the user to view all skin medical data within the scope of the permission, in order to avoid the problem that the abnormal access evaluation result is prone to misjudgment based on the difference between the user's user behavior data and the historical user behavior data, the difference between the user's user behavior data and the historical user behavior data is combined with the stability of the data itself for analysis, the difference in the user's adjusted real-time behavior data is determined, and combined with the user's potential hidden dangers, the user's real-time abnormal index is determined, and the user's real-time abnormal index is comprehensive evaluation of the abnormal degree of the user's current browsing behavior; finally, the skin plastic surgery data is managed according to the user's real-time abnormal index, so as to solve the problem that the abnormal access data is prone to misjudgment based only on the difference between the user behavior and the historical user behavior, resulting in unreasonable management of skin plastic surgery data, improve the accuracy of abnormal access data identification, and realize reasonable management of skin plastic surgery data. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0035] Figure 1 A schematic flow chart of an intelligent management method for skin plastic surgery data provided by one embodiment of the present invention;

[0036] Figure 2A flowchart for obtaining potential hazards provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0037] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. 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.

[0038] See also Figure 1 , which shows a flow chart of an intelligent management method for skin plastic surgery data provided by an embodiment of the present invention, the method comprising the following steps:

[0039] Step S001, collecting the patient's skin medical data, historical user behavior data and user behavior data of users who have browsing authority for the patient's skin medical data, and extracting word vectors of the patient's skin medical data.

[0040] Extract the skin medical data of each patient in the dermatology and plastic surgery hospital, extract the historical user behavior data of all users of the dermatology and plastic surgery hospital who have the right to browse the skin medical data, and extract the user's user behavior data in real time when the user browses the skin medical data.

[0041] Among them, it can be understood that the types of data contained in the user's historical user behavior data and user behavior data are the same, both of which include IP address, device identification, access time, access geographic location, account permissions and browsed skin medical data. The difference lies in the time when the data is generated, that is, the historical user behavior data is generated by the user at previous historical moments, while the user behavior data is generated in real time when the user browses skin medical data.

[0042] The NLP natural language algorithm is used to process the patient's skin medical data, the user's historical user behavior data and user behavior data, and all text data is converted into numerical data. The Jieba word segmentation algorithm is used to decompose the patient's skin medical data into phrases, and meaningless function words, prepositions and other interference words are removed. The BERT algorithm model of contextual embeddings is used to obtain the word vector of each patient's skin medical data.

[0043] At this point, the word vectors of the skin medical data of all patients in the skin plastic surgery hospital, the historical user behavior data and user behavior data of each user who has the permission to browse the skin medical data are obtained.

[0044] Step S002: Determine similar case clusters based on the similarities between all different word vectors; determine similar sentence clusters based on the similarities between the sentence meanings corresponding to all different word vectors; determine the privacy of the user under browsing permissions based on the relationship between the skin medical data included in the user's browsing permissions and the similar case clusters and similar sentence clusters; determine the potential risks of the user based on the privacy of the user under all browsing permissions.

[0045] The skin medical data of different patients are different, and the degree of privacy they present is different. For example, if a patient only has common and mild skin management problems, such as uneven skin color, peeling caused by seasonal changes, rough skin, etc., the corresponding skin medical data is universal and does not involve pathological problems. In this case, the patient's skin medical data is more routine and less private. Therefore, it is necessary to determine the universality of the patient's skin medical data.

[0046] Based on the similarity between all different word vectors, similar case clusters are determined.

[0047] Any word vector of any patient's skin medical data is recorded as the target word vector, the average value of the similarity between the target word vector and each other word vector is recorded as the word vector similarity of the target word vector, and the normalized value of the cumulative sum of the word vector similarities of all word vectors of the same patient is recorded as the case similarity of the patient's skin medical data.

[0048] It can be understood that the other word vectors include other word vectors corresponding to the same patient as the target word vector, and also include other word vectors corresponding to different patients as the target word vector; it should be noted that this embodiment uses the Z-Score standard normalization method to calculate the normalized value, and the similarity between word vectors is expressed using cosine similarity. In actual application, implementers can use other methods of the prior art such as the maximum and minimum normalization method to calculate the normalized value, and implementers can also use other methods in the prior art such as the Pearson correlation coefficient to obtain the similarity between word vectors, which is not limited here.

[0049] When the similarity between the target word vector and all other word vectors is greater, the target word vector appears more frequently, the meaning of the words corresponding to the target word vector is more universal and the privacy is weaker, and at this time, the word vector similarity of the target word vector is greater. Furthermore, when the word vector similarity of all word vectors of a patient is greater, the universality of the patient's skin medical data is more significant and the privacy is weaker, and at this time, the patient's case similarity is greater.

[0050] The ISOdata clustering algorithm was used to cluster the case similarities of all patients to obtain similar case clusters.

[0051] The similar case cluster is a cluster composed of case similarities corresponding to skin medical data with high similarity. Therefore, the similarity of the skin medical data of patients corresponding to all case similarities in the same similar case cluster is high.

[0052] The average pooling algorithm is used for all word vectors of the skin medical data of the same patient to convert the patient's word vector into a sentence vector and obtain the patient's sentence vector. The DBSCAN algorithm is used to perform density clustering on the sentence vectors of all patients to obtain similar sentence clusters.

[0053] Similar sentence clusters are clusters composed of sentence vectors with high similarity.

[0054] Among them, this embodiment uses the ISOdata clustering algorithm to cluster case similarities, and uses the DBSCAN algorithm to cluster sentence vectors. In actual application, on the basis of achieving the purpose of clustering, implementers can use other existing methods such as DBSCAN clustering algorithm, Mean Shift clustering algorithm, OPTICS clustering algorithm, Gaussian Mixture Models clustering algorithm, Spectral Clustering spectral clustering, etc. for clustering, and this application does not make any special restrictions.

[0055] According to the relationship between the skin medical data included in the user's browsing permissions and the similar case clusters and similar sentence clusters, the privacy degree of the user under the browsing permissions is determined.

[0056] Preferably, as an embodiment of the present application, the mean of the case similarities contained in the similar case clustering cluster is recorded as the similarity distribution of the similar case clustering cluster, the cumulative sum of the similarity distributions of the similar case clustering clusters corresponding to all skin medical data that the user has browsing permissions is recorded as the user's browsing permission similarity distribution, the opposite of the user's browsing permission similarity distribution is taken as the exponential value of an exponential function with a natural constant as the base, and the calculated value of the exponential function is recorded as the user's privacy under browsing permissions.

[0057] When the similarity of cases contained in similar case clusters is greater, the similarity of skin medical data corresponding to the similarity of cases contained in similar case clusters is higher, the universality is stronger, and the privacy is weaker. The homogeneity of skin medical data that can be viewed within the user's browsing permissions is more obvious, and the privacy of the user account is weaker. When an abnormality occurs in the user account, the loss is less. At this time, the user's privacy under browsing permissions is smaller.

[0058] When a user has multiple different browsing permissions, the privacy levels of skin medical data corresponding to different browsing permissions may differ. If an abnormality occurs in the user's account or data leakage occurs, the harmful consequences should be determined by the maximum value of the privacy level under the browsing permissions. The larger the maximum value of the privacy level under all browsing permissions, the greater the security risks.

[0059] Determine the user's potential risk based on the user's privacy under all browsing permissions.

[0060] Preferably, as an embodiment of the present application, the maximum value of the privacy of the user under all browsing permissions is recorded as the potential risk of the user.

[0061] At this point, the potential risks of the user are obtained. The potential risk acquisition flow chart is as follows: Figure 2 shown.

[0062] Step S003: determine the adjusted real-time behavior data difference of the same type of data of the user based on the fluctuation range of the same type of data in the user's historical user behavior data and the difference between the user's user behavior data and the same type of data in the historical user behavior data; determine the user's real-time abnormality index based on all the user's adjusted real-time behavior data differences and the user's potential hidden dangers.

[0063] The greater the difference between the user's user behavior data and the historical user behavior data, the greater the possibility that the user's account has been accessed abnormally. However, it is also necessary to consider that it is reasonable for the user to view all skin medical data within the scope of the permission, so as to avoid misjudgment of abnormal access data. It is understandable that the fluctuation range of the different types of data contained in the user's historical user behavior data and the user behavior data is different. For example, the geographical location of the access is generally set within the scope of the skin plastic surgery hospital, while the access time may cover every moment during the user's working hours. Therefore, it is necessary to determine the stability of the data itself based on the fluctuation range of the same type of data contained in the historical user behavior data and the user behavior data, and analyze the difference between the user's user behavior data and the historical user behavior data in combination with the stability of the data itself to determine whether the user has had abnormal access.

[0064] According to the fluctuation range of the same type of data in the historical user behavior data of the user, and the difference between the user behavior data of the user and the same type of data in the historical user behavior data, the adjusted real-time behavior data difference of the same type of data of the user is determined.

[0065] Preferably, as an embodiment of the present application, any one type of historical user behavior data in the user's historical user behavior data is recorded as the target historical type data, the similarity between the word vectors of any two target historical type data is calculated, and the average value of the similarity between the word vectors of all the target historical type data is recorded as the fluctuation stability of the target historical type data. The opposite number of the fluctuation stability of the target historical type data is taken as the exponential value of an exponential function with a natural constant as the base, and the calculated value of the exponential function is recorded as the fluctuation amplitude of the target historical type data.

[0066] The same method may be used to obtain the fluctuation range of any type of historical user behavior data of the user.

[0067] It is understandable that, since the user's historical user behavior data and the user behavior data contain the same data type, the user's user behavior data also contains target historical type data.

[0068] The mean of the word vectors of the target category data in all the user's historical user behavior data is recorded as the user's historical standard data. For the target historical category data, the absolute value of the difference between the word vector of the user's user behavior data and the user's historical standard data is recorded as the user's real-time behavior data difference under the target historical category data. The ratio of the user's real-time behavior data difference under the target historical category data to the user's historical standard data is recorded as the user's adjusted real-time behavior data difference of the target historical category data.

[0069] Furthermore, the real-time abnormality index of the user is determined based on all the adjusted real-time behavior data differences of the user and the potential hidden dangers of the user.

[0070] Preferably, as an embodiment of the present application, the product of the cumulative sum of the differences in the user's adjusted real-time behavior data in all historical types of data and the user's potential hidden danger is recorded as the user's real-time abnormality index.

[0071] At this point, the user's real-time abnormality index is obtained.

[0072] Step S004: managing skin plastic surgery data according to the user's real-time abnormality index.

[0073] Compare the user's real-time abnormal index with the abnormal threshold. When the user's real-time abnormal index is greater than the abnormal threshold, it is determined that the user has abnormal access and all access and browsing permissions of the user are closed. When the user's real-time abnormal index is less than or equal to the abnormal threshold, it is determined that the user's access is normal and the user's permissions are not adjusted.

[0074] In this embodiment, the value of the abnormal threshold is 0.7.

[0075] At this point, the management of skin plastic surgery data is completed.

[0076] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An intelligent management method for skin plastic surgery data, characterized in that: The method comprises the following steps: Collect the patient's skin medical data, historical user behavior data and user behavior data of users who have browsing rights to the patient's skin medical data, and extract word vectors of the patient's skin medical data; Based on the similarity between all different word vectors, similar case clusters are determined. Based on the similarity between the meanings of sentences corresponding to all different word vectors, similar sentence clusters are determined. Based on the relationship between the skin medical data included in the user's browsing permissions and similar case clusters and similar sentence clusters, the user's privacy under browsing permissions is determined. Based on the user's privacy under all browsing permissions, the user's potential risks are determined. Determine the adjusted real-time behavior data difference of the same type of data of the user according to the fluctuation range of the same type of data in the historical user behavior data of the user, and the difference between the user behavior data of the user and the same type of data in the historical user behavior data, and determine the real-time abnormality index of the user according to all the adjusted real-time behavior data differences of the user and the potential hidden dangers of the user; Manage skin plastic surgery data based on the user's real-time abnormality index.

2. The intelligent management method of skin plastic surgery data according to claim 1, characterized in that: The specific method of determining similar case clusters based on the similarities between all different word vectors is as follows: Any word vector of any patient's skin medical data is recorded as the target word vector, the average value of the similarity between the target word vector and each other word vector is recorded as the word vector similarity of the target word vector, and the normalized value of the cumulative sum of the word vector similarities of all word vectors of the same patient is recorded as the case similarity of the patient's skin medical data; Based on the case similarities of all patients, clusters of similar cases were identified.

3. The intelligent management method of skin plastic surgery data according to claim 2, characterized in that: The specific method of determining similar case clusters based on the case similarity of all patients is as follows: The ISOdata clustering algorithm was used to cluster the case similarities of all patients to obtain similar case clusters.

4. The intelligent management method of skin plastic surgery data according to claim 1, characterized in that: The specific method of determining similar sentence clusters according to the similarity between the sentence meanings corresponding to all different word vectors is as follows: The patient's word vector is converted into the patient's sentence vector, and the sentence vectors of all patients are clustered to obtain similar sentence clusters.

5. The intelligent management method of skin plastic surgery data according to claim 3, characterized in that: The specific method of obtaining the privacy of the user under browsing authority is as follows: The mean of the case similarities contained in the similar case cluster is recorded as the similarity distribution of the similar case cluster, the cumulative sum of the similarity distributions of the similar case clusters corresponding to all skin medical data that the user has browsing permissions is recorded as the user's browsing permission similarity distribution, the opposite of the user's browsing permission similarity distribution is taken as the exponent value of the exponential function with a natural constant as the base, and the calculated value of the exponential function is recorded as the user's privacy under browsing permissions.

6. The intelligent management method of skin plastic surgery data according to claim 1, characterized in that: The specific method of determining the potential risk of the user includes: The maximum privacy value of the user under all browsing permissions is recorded as the user's potential risk.

7. The intelligent management method of skin plastic surgery data according to claim 1, characterized in that: The specific method of managing skin plastic surgery data according to the user's real-time abnormality index is as follows: When the user's real-time abnormality index is greater than the abnormality threshold, all access and browsing permissions of the user are closed; When the user's real-time anomaly index is less than or equal to the anomaly threshold, the user's permissions are not adjusted.

Images