Construction method of chemotherapy-related oral mucositis nursing knowledge base

Abstract

The invention relates to the technical field of nursing knowledge base construction, in particular to a construction method of a chemotherapy-related oral mucositis nursing knowledge base. The method comprises the steps of firstly obtaining a knowledge base and access data; further obtaining an initial index weight according to the recent access trend and the total access frequency of each keyword in combination with the distribution of the keywords in the clinical nursing data; further obtaining nursing pertinence according to the difference between clinical nursing data related to each keyword in combination with a recovery effect; and finally, fusing the initial index weight and nursing pertinence to obtain a final index weight of each keyword. According to the scheme, the keyword index weight is dynamically optimized by fusing the universality, the timeliness and the nursing pertinence of the keywords, wide requirements and individual curative effects are considered, hot low-efficiency word deficiency is avoided, and the retrieval accuracy and the application value of the database are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of nursing knowledge base construction, and in particular to a method for constructing a nursing knowledge base for chemotherapy-related oral mucositis. Background Art

[0002] Oral mucositis caused by chemotherapy is called chemotherapy-induced oral mucositis (CTOM). Using appropriate nursing knowledge to care for cancer patients during CTOM can effectively promote the healing of CTOM, prevent it from further deterioration, and affect the treatment course of the tumor.

[0003] Prior art uses clinical nursing data from patients undergoing chemotherapy with CTOM to construct a knowledge base for chemotherapy-related oral mucositis nursing care. This allows medical staff and patients to better understand relevant nursing knowledge and effectively promote CTOM healing. However, due to the sheer volume of this knowledge base, indexing nursing knowledge often involves listing all matching nursing knowledge. This makes it difficult for users to quickly find the nursing knowledge they desire, resulting in low indexing efficiency. Summary of the Invention

[0004] In order to solve the technical problem of low efficiency of the indexing method of the existing chemotherapy-related oral mucositis nursing knowledge base, the present invention aims to provide a method for constructing a chemotherapy-related oral mucositis nursing knowledge base. The technical solution adopted is as follows: Obtaining a knowledge base constructed from clinical care data of historical patients and access data, and extracting keywords and recovery effects from the clinical care data; Obtaining an initial index weight for each keyword based on a recent access trend and a total number of accesses for each keyword in the access data and in combination with the distribution of the keyword in the clinical nursing data; According to the differences between the clinical nursing data involved in each keyword and combined with the recovery effect, the nursing targeting of each keyword is obtained; and the final index weight of each keyword is obtained by fusing the initial index weight and the nursing targeting.

[0005] Furthermore, the method for obtaining the initial index weight includes: Obtaining a distribution weight according to the distribution of each keyword in the clinical nursing data; The access data also includes the number of visits per day for each keyword; recent access intensity is obtained based on the change in the number of visits per day for each keyword; The distribution weight, the recent access intensity and the total number of accesses are integrated to obtain an initial index weight for each keyword.

[0006] Furthermore, the method for obtaining the distribution weight includes: The number of times each keyword appears in the knowledge base is used as its respective distribution weight.

[0007] Furthermore, the method for obtaining the recent access intensity includes: For each keyword, the recent visit intensity is obtained based on the slope of the fitted straight line of the single-day visit count in the preset historical neighborhood of the current day, combined with the prominent feature of the maximum single-day visit count in the preset historical neighborhood of the current day.

[0008] Furthermore, the method for obtaining targeted nursing care includes: vectorizing the patient data in the clinical care data, and forming a tuple of the data vector and the recovery effect; For each keyword, among the binary groups of all the clinical nursing data involved, the binary group with the greatest recovery effect is selected as a target binary group, and a nursing targeting factor is obtained based on the difference in recovery effects between the target binary group and the other binary groups, combined with the difference between the data vectors; The nursing targeting factor and the recovery effect of the target binary are integrated to obtain the nursing targeting of the corresponding keyword.

[0009] Furthermore, the method for obtaining the data vector includes: The patient data is vectorized using one-hot encoding to obtain data vectors.

[0010] Furthermore, the method for obtaining the final index weight includes: The product of the initial index weight and the nursing targeting of each keyword is used as the final index weight.

[0011] Furthermore, after obtaining the final index weight of each keyword, the following steps are further included: The knowledge base system obtains the index term entered by the current user and obtains a series of key field forms that match the index term entered by the user through a string matching algorithm; the keyword forms are arranged from large to small according to the final index weight, and the arranged keyword forms are output.

[0012] Furthermore, the length of the preset historical neighborhood is 7 days.

[0013] Furthermore, the keyword acquisition method includes: The nursing knowledge topic field in the preset nursing knowledge sheet is set as a key field, and the key field involved in the clinical nursing data is identified and used as a keyword.

[0014] The present invention has the following beneficial effects: The present invention first obtains the knowledge base and access data to provide a basis for subsequent analysis; further, based on the recent access trend and total number of accesses of each keyword, combined with the distribution of keywords in clinical nursing data, while taking into account the usage of historical nursing data, the universality of keywords in actual use, and their timeliness in recent clinical nursing needs, the initial index weight is obtained to improve the accuracy of the initial index weight; further, in order to avoid the inflated invalid access of some keywords, the nursing pertinence is obtained based on the differences between the clinical nursing data involved in each keyword, combined with the recovery effect, and the basis for correcting the initial index weight from the perspective of the pertinence of nursing knowledge; finally, the initial index weight and nursing pertinence are integrated to obtain the final index weight of each keyword. This solution dynamically optimizes the keyword index weight by integrating the universality, timeliness and nursing pertinence of keywords, taking into account broad needs and individual efficacy, avoiding the inflated popularity of inefficient words, and improving the retrieval accuracy and application value of the database. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. 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 any creative work.

[0016] Figure 1 A flowchart of a method for constructing a chemotherapy-related oral mucositis nursing knowledge base provided by one embodiment of the present invention; Figure 2 A flowchart of a method for obtaining an initial index weight provided by one embodiment of the present invention; Figure 3 A flowchart of a method for obtaining targeted nursing care provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0017] To further illustrate the technical means and effectiveness of the present invention in achieving its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the method for constructing a chemotherapy-related oral mucositis nursing knowledge base, including its specific implementation, structure, features, and effectiveness. In the following description, references to different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.

[0018] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0019] The specific scheme of the method for constructing a chemotherapy-related oral mucositis nursing knowledge base provided by the present invention is described in detail below with reference to the accompanying drawings.

[0020] See also Figure 1 , which shows a flow chart of a method for constructing a chemotherapy-related oral mucositis nursing knowledge base provided by one embodiment of the present invention, specifically comprising: Step S1: Obtain a knowledge base constructed from clinical care data of historical patients and access data, and extract keywords and recovery effects from the clinical care data.

[0021] In one embodiment of the present invention, the clinical nursing data includes at least the patient's nursing drugs, the types of drugs used in chemotherapy, the onset and recurrence time of oral mucositis (relative to the time after the first treatment), the effect of nursing on CTMO recovery, the patient's medical history, and the history of drug allergies. The clinical nursing data are arranged in a corresponding order, for example, [Kangfuxin Liquid; Paclitaxel, Cyclophosphamide; 6d; 17d; 0.65; Breast Cancer; No History of Allergies], which means that the patient's nursing drugs are Kangfuxin Liquid; the types of drugs used in chemotherapy are Paclitaxel and Cyclophosphamide; oral mucositis occurs on the 6th day after the first treatment and recurs on the 17th day; the recovery effect is 0.65, there is a history of breast cancer, and there is no history of drug allergies.

[0022] The clinical nursing data were converted into structured objects (e.g., JSON, dictionary) and stored in the database to construct a chemotherapy-related oral mucositis nursing knowledge base, referred to as the knowledge base, and the access data of the knowledge base were obtained.

[0023] Furthermore, in order to facilitate users to index the database, keywords in the clinical nursing data are extracted.

[0024] Preferably, in one embodiment of the present invention, the nursing knowledge topic field in the preset nursing knowledge sheet is set as a key field, and the key fields involved in the clinical nursing data are identified and used as keywords.

[0025] As an example, the preset nursing knowledge sheet is "Expert Consensus on the Prevention and Treatment of Radiotherapy-Induced Oral Mucositis" (T / CHSA 074-2024). It contains a variety of recommended detailed nursing knowledge, such as oral management and the use of medications like Kangfuxin solution. The nursing knowledge title field is set as a key field for index matching, identifying key fields involved in clinical nursing data and using them as keywords. Keywords are then matched with the corresponding detailed nursing knowledge and the clinical nursing data that uses this nursing knowledge.

[0026] Then, index weights are established for the keywords, and the index module of the knowledge base is constructed based on this.

[0027] It should be noted that the recovery effect is a specific data value. The higher the data value, the better the patient's recovery. The historical patient data used have been desensitized and do not involve sensitive information such as the patient's name and personal basic information. The knowledge base construction method is already an existing technology and will not be repeated here.

[0028] In other embodiments of the present invention, the implementer can adjust the structure and content of the clinical nursing data on his own; he can also set the "MASCC / ISOO clinical practice guidelines for the management of mucositis secondary to cancer therapy" as a preset nursing knowledge sheet.

[0029] Step S2: According to the recent access trend and total access times of each keyword in the access data, combined with the distribution of the keyword in the clinical nursing data, the initial index weight of each keyword is obtained.

[0030] Considering that the distribution of keywords in clinical nursing data reflects the selection in historical nursing records; after the knowledge base is put into use, users can access the nursing knowledge mapped by different keywords through keyword indexing. The access data records the total number of visits, reflecting the universality of the selected keywords. Knowledge is also time-sensitive, and patients' symptoms are also undergoing new changes. The recent access trend of the keyword can reflect the applicability of the keyword and the nursing knowledge it maps in the recent period. Therefore, based on the recent access trends and total number of visits for each keyword in the access data, combined with the distribution of keywords in clinical nursing data, the initial index weight of each keyword is obtained. At the same time, the usage of historical nursing data, the universality of keywords in actual use, and their timeliness in recent clinical nursing needs are taken into account, so that the index results are more in line with the current user's search intention, significantly improving the accuracy of keyword matching and the indexing efficiency of the knowledge base.

[0031] Preferably, in one embodiment of the present invention, see Figure 2 , which shows a flow chart of a method for obtaining an initial index weight provided by an embodiment of the present invention, specifically comprising: Step S201: Obtain distribution weights based on the distribution of each keyword in clinical nursing data.

[0032] Considering that the more times a keyword appears in the clinical nursing data of the knowledge base, the more times its corresponding nursing operation is selected in the chemotherapy-related oral mucositis nursing plan and the stronger its universality, the number of times each keyword appears in the knowledge base is used as its respective distribution weight, and the number of appearances is used to show the distribution characteristics of the keyword in the clinical nursing data.

[0033] In another embodiment of the present invention, the number of times each keyword appears in the database is used as a proportion of the total number of times all keywords appear, which is used as the distribution weight; it should be noted that the final ranking of the index weights is based on the relative size of the keywords, so the calculation method of the distribution weights of all keywords can be unified.

[0034] Step S202: The access data also includes the number of visits per day for each keyword; recent access intensity is obtained based on changes in the number of visits per day for each keyword.

[0035] In one embodiment of the present invention, daily visits are recorded on a daily basis. The access data also includes the number of daily visits for each keyword. The change in the number of daily visits is used to show the recent visit trend of the keyword, obtain the recent visit intensity of the keyword, and characterize the trend intensity of the recent visit.

[0036] Considering that the stronger the increasing trend of the number of daily visits in the current historical neighborhood is, the trend intensity is rising, and the recent visit intensity is higher; at the same time, the more prominent the maximum number of daily visits in the historical neighborhood is, the more there are recent visits pointing to the corresponding keywords, and the higher the recent visit intensity is. Based on this, for each keyword, the recent visit intensity is obtained according to the slope of the fitted straight line of the number of daily visits in the preset historical neighborhood of the current day, combined with the prominent characteristics of the maximum number of daily visits in the preset historical neighborhood of the current day.

[0037] As an example, the time length of the preset historical neighborhood is 7 days. For each keyword, the number of single-day visits in the 7 historical days adjacent to the current day is taken, and the fitting straight line of the single-day visit number is obtained based on the least squares method and the slope is obtained. The maximum single-day visit number in the preset historical neighborhood of the current day is used as the numerator, and the maximum single-day visit number in all historical days before the preset historical neighborhood of the current day is used as the denominator. The slope is taken as the product of the mapping value of the independent variable after mapping by the exp(x) function and the fractional ratio, which is used as the recent access intensity of the corresponding keyword.

[0038] Among them, in order to prevent the slope from being zero or negative and affecting the logical relationship, it is mapped through the exponential function exp(x) with the natural constant e as the base, and x is the independent variable; based on the maximum number of single-day visits in all historical days before the preset historical neighborhood of the current day, the highlight feature of the maximum number of single-day visits in the preset historical neighborhood of the current day is expressed in the form of a fractional ratio. The larger the fractional ratio, the closer the recent visits to the keyword are to or even exceed the historical peak visits, and the stronger the highlight feature is.

[0039] The least squares method for fitting a straight line and obtaining the slope of the straight line are both existing technologies. When a certain keyword is a newly added word and has a history of less than 8 days but at least 2 days, that is, when it is impossible to analyze the prominent characteristics of the maximum number of single-day visits within the preset historical neighborhood of the current day, the slope of the fitted straight line of the number of single-day visits in the existing historical days is directly used as the independent variable, and the mapping value after mapping by the exp(x) function is used as the recent visit intensity of the corresponding keyword.

[0040] When the historical days are less than 2 days, the average of the latest recent visit intensities of other keywords is taken as the recent visit intensity of the current keyword.

[0041] Step S203: The distribution weight, recent access intensity and total access times are integrated to obtain the initial index weight of each keyword.

[0042] Finally, we consider that the greater the total number of historical visits to a keyword, the greater the possibility of being visited again, and the greater the initial index weight; the greater the distribution weight, the stronger the universality, and the greater the initial index weight; the greater the recent visit intensity, the more the recent visits are on an upward trend, the higher the possibility of being visited recently, and the greater the initial index weight. Finally, the three are integrated to obtain the initial index weight of each keyword.

[0043] As an example, the product of the distribution weight, recent access intensity and total access times of each keyword on the current day is used as the initial index weight of each keyword.

[0044] Among them, the total number of visits and the distribution weight are the data features of all historical days before the current day. As another example, the distribution weight, recent visit intensity and total number of visits are linearly normalized in their respective corresponding data dimensions, and the three normalized results are fused by weighted summation, with the weights being 1 / 3, 1 / 3 and 1 / 3.

[0045] It should be noted that the method for obtaining the initial index weight of each keyword is the same. Only one example is described here and no further explanation is given.

[0046] Step S3: Based on the differences between the clinical nursing data involved in each keyword and combined with the recovery effect, the nursing targeting of each keyword is obtained; the final index weight of each keyword is obtained by fusing the initial index weight and the nursing targeting.

[0047] However, after indexing, most users will give priority to accessing keywords that are ranked at the top (i.e., have a high initial index weight) and have a high total number of views. The nursing knowledge corresponding to these keywords is usually more basic nursing methods, which are highly universal but often have poor effects.

[0048] For certain keywords with a large base of visits within a short period of time, the snowball effect can lead to an increase in invalid user visits, gradually inflating their historical visit counts. This can lead to an inflated index weight based solely on historical visit data. This, in turn, can lead to invalid user visits to the keyword after indexing, further increasing its visit count, creating a vicious cycle. Therefore, it is necessary to further consider the universal validity of keywords and further revise the initial index weight.

[0049] In clinical nursing data, different nursing knowledge promotes different CTOM recovery effects in different individual patients. Individuals with large differences often require different targeted nursing knowledge. The more targeted the patient, the more likely it is to be a nursing knowledge means that some users are not aware of, and the more likely it is to solve difficulties for users and become the user's expected index target, which needs to increase the probability of effective user access.

[0050] Therefore, based on the differences between the clinical nursing data involved in each keyword, we analyze the targeted characteristics of nursing knowledge for patients, analyze the nursing effect in combination with the recovery effect, obtain the nursing targetedness of each keyword, and quantify the targetedness of the nursing support corresponding to the keyword, providing a basis for revising the initial index weight.

[0051] Preferably, in one embodiment of the present invention, see Figure 3 , which shows a flow chart of a method for obtaining targeted nursing care provided by an embodiment of the present invention, specifically including: Step S301: vectorize the patient data in the clinical nursing data, and form a tuple of the data vector and the recovery effect.

[0052] In order to facilitate the comparison of differences between clinical nursing data, the patient data is vectorized. At the same time, the recovery effect is also an important component of clinical nursing data, so the data vector and the recovery effect are combined into a binary group to facilitate the subsequent comparison of the differences between clinical nursing data from the two perspectives of the patient's data vector and the recovery effect.

[0053] As an example, one-hot encoding is used to vectorize patient data to obtain a data vector. The patient data includes nursing medications, types of chemotherapy drugs, oral mucositis onset and recurrence time (relative to the time since the first treatment), patient medical history, and drug allergy history.

[0054] It should be noted that in other embodiments of the present invention, patient data can also be vectorized by means of a bag-of-words model, a Word2Vec model, etc. The data vector can be a multidimensional vector, that is, each type of data corresponds to one dimension. When the similarity or difference between the vectors is subsequently analyzed, all the same dimensions are compared and the average of all dimensions is taken. This and one-hot encoding are technical means well known to those skilled in the art and will not be described in detail.

[0055] Step S302: For each keyword, among all the binary groups of clinical nursing data involved: select the binary group with the largest recovery effect as the target binary group, and obtain the nursing targeted factor based on the difference in recovery effect between the target binary group and other binary groups, combined with the difference between the data vectors.

[0056] For nursing knowledge, if it has a special effect only on certain special patients (for example, dexamethasone ointment can effectively reduce the incidence of severe oral mucositis in patients with squamous cell carcinoma), its frequency of occurrence may not be high, but it has a strong targeted effect on special patients, which is reflected in a high recovery effect. Therefore, the dyad with the greatest recovery effect is selected as the target dyad, and the differences between the target dyad and other dyads are compared; Taking into account the two aspects of recovery effect and data vector, the greater the difference between the target binary group and other binary groups, the more targeted the nursing care is, so the nursing targeted factor is obtained.

[0057] As an example, for each keyword, among all the binary groups of clinical care data involved, the target binary group is compared with each other binary group one by one. For each other binary group: the difference between the recovery effect of the target binary group and the recovery effect of the other binary groups is used as the numerator, the sum of the cosine similarity between the data vector of the target binary group and the data vector of the other binary groups and the preset correction parameter 1.01 is used as the denominator, and the fraction ratio is used as the difference factor between the target binary group and the corresponding other binary groups; The average value of the difference factor of each keyword is used as the nursing targeting factor.

[0058] Among them, the absolute numerical difference of the recovery effect is expressed by the difference between the recovery effects, and the difference between the data vectors is expressed by the cosine similarity. The smaller the cosine similarity, the greater the vector difference. At the same time, the cosine similarity is adjusted with the help of the preset correction parameters. 1 is added to make the value range of the cosine similarity 0-2 to avoid affecting the logical relationship, and 0.01 is added to prevent the denominator from being zero. In this way, the nursing targeting factor is obtained, which reflects the difference between the clinical nursing data of a keyword and characterizes the intensity of nursing targeting.

[0059] Step S303: integrating the nursing targeting factor and the recovery effect of the target binary to obtain the nursing targeting of the corresponding keyword.

[0060] Considering that the greater the recovery effect of the target binary group, the more ideal the nursing effect is and the stronger the reference for targeted nursing is, the nursing targeting factor and the recovery effect of the target binary group are finally integrated to obtain the nursing targeting of the corresponding keyword.

[0061] As an example, the product of the care targeting factor of each keyword and the recovery effect of the target bigram is taken as the care targeting of each keyword.

[0062] The initial index weight reflects the keyword's universality in historical clinical nursing records, overall user attention, and recent trends in timeliness, but it may not necessarily reflect the actual effectiveness of the nursing knowledge corresponding to the keyword in a specific patient population. Nursing targetedness analyzes the differences in the recovery effects of nursing knowledge on different individuals and quantifies its effectiveness for specific patients. The final index weight for each keyword is obtained by combining the initial index weight and nursing targetedness, establishing a balance between breadth and precision, improving the effective hit rate of user searches, and enhancing the efficiency and reference value of the knowledge base.

[0063] Preferably, in one embodiment of the present invention, when some highly targeted nursing knowledge appears in the user index keywords and needs to be targeted at the patient's special condition, these nursing knowledge should be placed at the top of the resulting form, so the product of the initial index weight of each keyword and the nursing targeting is used as the final index weight.

[0064] In another embodiment of the present invention, if the initial index weight is normalized, the nursing targeting also needs to be normalized to avoid the nursing targeting dominating and ignoring the initial index weight. At this time, the initial index weight and the nursing targeting can also be fused by addition or weighted summation.

[0065] It should be noted that the analysis process for the nursing relevance and final index weight of each keyword is the same. Only one example is described here and no repetition is required.

[0066] In another embodiment of the present invention, after obtaining the final index weight of each keyword and providing the final index weight of each keyword to the indexing module, the method further includes: The knowledge base system obtains the index term entered by the current user and obtains a series of key field forms that match the index term entered by the user through a string matching algorithm; the keyword forms are arranged from large to small according to the final index weight, and the arranged keyword forms are output.

[0067] It should be noted that the string matching algorithm is a well-known technology and will not be described in detail. The update frequency of the final index weight of the keyword can be set to once a day. In other embodiments of the present invention, the implementer can adjust it at will.

[0068] In summary, in response to the technical problem of low efficiency of the indexing method of the existing chemotherapy-related oral mucositis nursing knowledge base, the present invention provides a method for constructing a chemotherapy-related oral mucositis nursing knowledge base. The present invention first obtains the knowledge base and access data; further, based on the recent access trend and total access times of each keyword, combined with the distribution of keywords in clinical nursing data, an initial index weight is obtained; further, based on the differences between the clinical nursing data involved in each keyword, combined with the recovery effect, nursing targeting is obtained; finally, the initial index weight and nursing targeting are integrated to obtain the final index weight of each keyword. This solution dynamically optimizes the keyword index weight by integrating the universality, timeliness and nursing targeting of keywords, taking into account both broad needs and individual efficacy, avoiding the false high index of popular and inefficient words, and improving the retrieval accuracy and application value of the database.

[0069] It should be noted that the order in which the embodiments of the present invention are described above is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0070] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments.

Claims

1. A method for constructing a knowledge base for chemotherapy-related oral mucositis care, characterized in that: The method comprises: Obtaining a knowledge base constructed from clinical care data of historical patients and access data, and extracting keywords and recovery effects from the clinical care data; Obtaining an initial index weight for each keyword based on a recent access trend and a total number of accesses for each keyword in the access data and in combination with the distribution of the keyword in the clinical nursing data; According to the differences between the clinical nursing data involved in each keyword and combined with the recovery effect, the nursing targeting of each keyword is obtained; and the final index weight of each keyword is obtained by fusing the initial index weight and the nursing targeting.

2. The method for constructing a chemotherapy-related oral mucositis nursing knowledge base according to claim 1, characterized in that: The method for obtaining the initial index weight includes: Obtaining a distribution weight according to the distribution of each keyword in the clinical nursing data; The access data also includes the number of visits per day for each keyword; recent access intensity is obtained based on the change in the number of visits per day for each keyword; The distribution weight, the recent access intensity and the total number of accesses are integrated to obtain an initial index weight for each keyword.

3. The method for constructing a chemotherapy-related oral mucositis nursing knowledge base according to claim 2, characterized in that: The method for obtaining the distribution weight includes: The number of times each keyword appears in the knowledge base is used as its respective distribution weight.

4. The method for constructing a chemotherapy-related oral mucositis nursing knowledge base according to claim 2, characterized in that: The method for obtaining the recent access intensity includes: For each keyword, the recent visit intensity is obtained based on the slope of the fitted straight line of the single-day visit count in the preset historical neighborhood of the current day, combined with the prominent feature of the maximum single-day visit count in the preset historical neighborhood of the current day.

5. The method for constructing a chemotherapy-related oral mucositis nursing knowledge base according to claim 1, characterized in that: The method for obtaining the targeted nursing care includes: vectorizing the patient data in the clinical care data, and forming a tuple of the data vector and the recovery effect; For each keyword, among the binary groups of all the clinical nursing data involved, the binary group with the greatest recovery effect is selected as a target binary group, and a nursing targeting factor is obtained based on the difference in recovery effects between the target binary group and the other binary groups, combined with the difference between the data vectors; The nursing targeting factor and the recovery effect of the target binary are integrated to obtain the nursing targeting of the corresponding keyword.

6. The method for constructing a chemotherapy-related oral mucositis nursing knowledge base according to claim 5, characterized in that: The method for obtaining the data vector includes: The patient data is vectorized using one-hot encoding to obtain data vectors.

7. The method for constructing a chemotherapy-related oral mucositis nursing knowledge base according to claim 1, characterized in that: The method for obtaining the final index weight includes: The product of the initial index weight and the nursing targeting of each keyword is used as the final index weight.

8. The method for constructing a chemotherapy-related oral mucositis nursing knowledge base according to claim 1, characterized in that: After obtaining the final index weight of each keyword, the following steps are also included: The knowledge base system obtains the index term entered by the current user and obtains a series of key field forms that match the index term entered by the user through a string matching algorithm; the keyword forms are arranged from large to small according to the final index weight, and the arranged keyword forms are output.

9. The method for constructing a chemotherapy-related oral mucositis nursing knowledge base according to claim 4, characterized in that: The length of the preset historical neighborhood is 7 days.

10. The method for constructing a chemotherapy-related oral mucositis nursing knowledge base according to claim 1, characterized in that: The method for obtaining the keywords includes: The nursing knowledge topic field in the preset nursing knowledge sheet is set as a key field, and the key field involved in the clinical nursing data is identified and used as a keyword.

Images