An auxiliary recommendation system for orthopedic postoperative rehabilitation training

Abstract

The present invention relates to the technical field of natural language processing, and specifically to an auxiliary recommendation system for orthopedic postoperative rehabilitation training. The present invention first preliminarily classifies the symptom data by the number of surgeries; further, within each classification, the symptom data is clustered according to the similar features of the characterization segmentation sets between the symptom data; further, the reference coefficients of the historical solutions matched by other surgical categories in the word surgery classification are obtained, and the historical solutions with low reference value are eliminated; further, based on the corresponding similar parameters between the symptom data of the current patient and the same type of symptom data, a recommended orthopedic postoperative rehabilitation training program is provided. The present invention carefully classifies the symptom data and historical solutions, matches the symptom data of the current patient to similar categories, and assists rehabilitation physicians in recommending suitable orthopedic postoperative rehabilitation training programs, helping rehabilitation physicians to develop more suitable orthopedic postoperative rehabilitation training programs for patients.

Description

Technical Field

[0001] The present invention relates to the technical field of natural language processing, and in particular to an auxiliary recommendation system for orthopedic postoperative rehabilitation training. Background Art

[0002] Rehabilitation medicine is a modern medicine that focuses on curing functional disorders and improving and compensating for them, so that patients can improve their functional quality and return to society. Orthopedic rehabilitation is an important branch of rehabilitation medicine, mainly focusing on postoperative treatment of orthopedic diseases and trauma.

[0003] Traditional orthopedic postoperative rehabilitation training mainly relies on the experience of rehabilitation physicians, based solely on subjective judgment and lacking objective data support. It has limitations such as poor pain control and poor recovery of joint range of motion, which affects the patient's entire rehabilitation process and rehabilitation effect. Summary of the invention

[0004] In order to solve the technical problem that traditional orthopedic postoperative rehabilitation training is too dependent on the experience of rehabilitation physicians, the purpose of the present invention is to provide an auxiliary recommendation system for orthopedic postoperative rehabilitation training. The technical solution adopted is as follows:

[0005] An auxiliary recommendation system for orthopedic postoperative rehabilitation training, the system comprising:

[0006] Data acquisition module: acquires historical plans for orthopedic postoperative rehabilitation training and the personal identification and symptom data of the patient corresponding to each of the historical plans; acquires the personal identification and symptom data of the current patient; the historical plans include the type of surgery;

[0007] Surgery classification module: select any of the symptom data of the current patient and the historical plan as the target symptom data; perform word segmentation processing on the target symptom data and extract the segmentation describing the number of surgeries; classify the target symptom data according to the similarity between the set of segmentation describing the number of surgeries corresponding to the target symptom data and the set of segmentation describing multiple surgeries, and the classification result at least includes the classification of a single surgery;

[0008] Symptom classification module: according to the frequency characteristics of each segmentation in the target symptom data, obtain the representative segmentation set of the target symptom data; select any classification as the target classification; within the target classification, according to the similar characteristics of the representative segmentation set between any two symptom data, obtain the similarity parameters between any two symptom data; according to the similarity parameters between the symptom data, cluster the symptom data of historical patients;

[0009] A scheme screening module: matching the historical schemes with the same patient personal identifier, and in the single surgery classification, taking any matched historical scheme as the scheme to be analyzed; obtaining the reference coefficient of the scheme to be analyzed based on the similarity characteristics of the segmentation set of the surgery type of the scheme to be analyzed and the segmentation set of the surgery type of the corresponding matched historical scheme; and eliminating the historical scheme based on the reference coefficient;

[0010] Auxiliary recommendation module: based on the symptom data of the current patient and the similar parameters corresponding to the symptom data in all clusters corresponding to similar historical patients, obtain a recommended orthopedic postoperative rehabilitation training program;

[0011] The method for obtaining the recommended orthopedic postoperative rehabilitation training program includes:

[0012] According to the similarity parameters corresponding to the symptom data of the current patient and the symptom data of the same type, obtaining the overall similarity parameter of each cluster of the symptom data of the current patient and the symptom data of the same type for all historical patients;

[0013] The cluster with the largest overall similarity parameter is selected as the orthopedic postoperative rehabilitation training program cluster for the current patient; and all historical programs in the orthopedic postoperative rehabilitation training program cluster for the current patient are used as recommended orthopedic postoperative rehabilitation training programs.

[0014] Furthermore, the method for obtaining the characterization segmentation set includes:

[0015] The occurrence frequency of each segmentation word in the target symptom data is obtained, and the segmentation words are sorted from large to small according to the occurrence frequency to obtain a segmentation sequence; and the segmentation words with a preset proportion in the segmentation sequence are intercepted to form a segmentation set representing the target symptom data.

[0016] Furthermore, the method for obtaining the similarity parameters includes:

[0017] In the target classification, a symptom data group to be analyzed is formed according to any two of the symptom data; a similarity coefficient between two of the representation segmentation sets corresponding to the symptom data group to be analyzed is obtained as a second similarity coefficient;

[0018] Obtaining a frequency similarity coefficient according to the overall similarity characteristics of the occurrence frequencies of the segmented words in the two representative segmented word sets corresponding to the disease data group to be analyzed;

[0019] The similarity parameter corresponding to the disease data group to be analyzed is obtained according to the second similarity coefficient and the frequency similarity coefficient; the second similarity coefficient and the frequency similarity coefficient are both positively correlated with the similarity parameter.

[0020] Furthermore, the method for classifying the target disease data includes:

[0021] Obtain a similarity coefficient between the set of surgery number description participles corresponding to the target disease data and a preset multiple surgery description participle set as a first similarity coefficient; when the first similarity coefficient is greater than a first preset threshold, classify the target disease data into multiple surgery classification; when the first similarity coefficient is less than or equal to the first preset threshold, classify the target disease data into a single surgery classification.

[0022] Furthermore, the method for obtaining the reference coefficient includes:

[0023] In the single surgery classification, any matched historical scheme corresponds to at least one scheme in the multiple surgery classifications; the scheme matched by the scheme to be analyzed is taken as a matching scheme;

[0024] Obtain a similarity coefficient between the word segment set of the surgical type of the scheme to be analyzed and the word segment set of the surgical type of each corresponding matching scheme as a third similarity coefficient; obtain a reference coefficient of the scheme to be analyzed based on the overall characteristics of the third similarity coefficients of the scheme to be analyzed and all corresponding matching schemes; the overall characteristics of the third similarity coefficient are negatively correlated with the reference coefficient.

[0025] Furthermore, the method of eliminating the historical solution according to the reference coefficient includes:

[0026] When the reference coefficient is less than a second preset threshold, the historical plan corresponding to the single surgery classification is eliminated.

[0027] Furthermore, the algorithm used for word segmentation processing of the target disease data is JieBa word segmentation algorithm.

[0028] Furthermore, the segmentations describing the number of surgeries include at least: first surgery, first operation, second surgery, reoperation and second surgery.

[0029] Furthermore, the algorithm used to cluster the disease data of historical patients is a K-means algorithm.

[0030] The present invention has the following beneficial effects:

[0031] The present invention obtains various data through the data acquisition module to provide an analysis basis for subsequent analysis; further in the surgery classification module, the target disease data is segmented to facilitate the system to more accurately extract and analyze important information related to the disease; considering that the focus and planning of the orthopedic postoperative rehabilitation training program after multiple surgeries are different from those of the orthopedic postoperative rehabilitation training program for the first surgery, the segmentation describing the number of surgeries is extracted, and the target disease data is classified according to the number of surgeries; further in the symptom classification module: the representative segmentation set of the target disease data is obtained, focusing on the most representative part of the disease data, so that the similarity analysis between different diseases is more accurate; within the target classification, the similarity parameters between any two disease data are obtained. , according to the similar parameters between the symptom data, the symptom data of historical patients are clustered, and the symptom data are classified more carefully, so as to facilitate the subsequent more detailed recommendation of solutions; further in the solution screening module, the historical solutions with the same patient personal identification are matched, and in the single surgery classification, any matched historical solution is used as the solution to be analyzed; the reference coefficient of the solution to be analyzed is obtained, and some historical solutions in the word surgery classification are eliminated according to the reference coefficient, so as to avoid recommending postoperative rehabilitation training solutions for customers with poor rehabilitation effects in the future, and improve the reference value of auxiliary recommendation; finally, in the auxiliary recommendation module, the recommended orthopedic postoperative rehabilitation training solution is obtained according to the corresponding similar parameters between the symptom data of the current patient and the same symptom data. The present invention classifies the symptom data and historical solutions in detail, matches the symptom data of the current patient to similar categories, and assists rehabilitation physicians in recommending suitable orthopedic postoperative rehabilitation training solutions, so as to help rehabilitation physicians formulate more suitable orthopedic postoperative rehabilitation training solutions for patients. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are 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 work.

[0033] Figure 1 A system block diagram of an auxiliary recommendation system for orthopedic postoperative rehabilitation training provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0034] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the specific implementation, structure, features and effects of an auxiliary recommendation system for orthopedic postoperative rehabilitation training proposed by the present invention in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0035] 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.

[0036] The following is a detailed description of a specific scheme of an auxiliary recommendation system for orthopedic postoperative rehabilitation training provided by the present invention in conjunction with the accompanying drawings.

[0037] See also Figure 1 , which shows a system block diagram of an auxiliary recommendation system for orthopedic postoperative rehabilitation training provided by an embodiment of the present invention, the system includes: a data acquisition module 101; a surgery classification module 102; a symptom classification module 103; a plan screening module 104; and an auxiliary recommendation module 105.

[0038] Data acquisition module 101: acquires historical plans of orthopedic postoperative rehabilitation training and the personal identification and symptom data of the patient corresponding to each historical plan; acquires the personal identification and symptom data of the current patient; the historical plan includes the type of surgery.

[0039] The recommendation system of the embodiment of the present invention is only used to provide reference and assistance for the actual rehabilitation training program. Ultimately, a rehabilitation physician needs to provide a more accurate and complete orthopedic postoperative rehabilitation training program based on the data obtained from this program.

[0040] First, obtain the historical plans for orthopedic postoperative rehabilitation training and the personal identification and disease data of the patients corresponding to each historical plan to provide a basis for data analysis; obtain the personal identification and disease data of the current patient to facilitate comparison with the data corresponding to the historical plan; in order to facilitate the use of the surgical type information in the historical plan and accurately assist in recommending historical training plans, the historical plan needs to include the surgical type.

[0041] As an example, a rehabilitation program after anterior cruciate ligament reconstruction includes the type of surgery, phased rehabilitation goals, and specific training programs, including:

[0042] Rehabilitation after anterior cruciate ligament reconstruction

[0043] Phase 1: 0-2 weeks after surgery

[0044] Rehabilitation goals: 1. Restore knee extension 2. Restore knee flexion ; 3. Restore normal quadriceps strength; 4. Restore normal gait.

[0045] Training items: (1) Patellar movement exercises can be performed after surgery, especially up and down movement of the patella.

[0046] (2) The range of motion of the knee brace is controlled within - .

[0047] (3) Quadriceps contraction exercises and straight leg raising exercises. Emphasize that the knee joint is fully extended. Perform straight leg raise exercises.

[0048] (4) Perform hamstring contraction exercises in prone or standing positions.

[0049] (5) Passive knee extension exercises can be performed in the prone position or with the heels raised (emphasis on fully extending the knee joint).

[0050] (6) Passive, active, and active-assisted knee flexion exercises: foot sliding along the wall, sitting foot sliding, or using a towel to pull the calf in the prone position.

[0051] (7) Control swelling (compression therapy may be used).

[0052] (8) If the quadriceps are weak, electrical stimulation or combined muscle relearning can be used.

[0053] (9) If the knee brace is locked in the fully extended position, crutches may be used and 50% to 75% of the weight may be placed on the affected limb; alternatively, crutches may be used without the patient's need for weight bearing as tolerated.

[0054] (10) When sleeping, the knee brace is locked in the extended position.

[0055] Phase 2: 2-4 weeks after surgery

[0056] Rehabilitation goals: 1. Restore knee joint ROM (Range of Motion), - ; 2. Do not use crutches, bear full weight, and have no feeling of fatigue.

[0057] Training items: (1) Adjust the knee brace to the full range of joint motion.

[0058] (2) Four weeks after surgery, knee ROM gradually increased to .

[0059] (3) Progressive straight leg raising exercises. Active contraction exercises of the hamstring muscles in the prone or standing position.

[0060] (4) When the knee joint mobility has recovered to a certain range, you can practice riding a bicycle and start low-intensity resistance exercises.

[0061] (5) Pedal exercises.

[0062] It should be noted that, in one embodiment of the present invention, the patient's personal identification and disease data can be obtained through the patient's preoperative diagnosis form or patient statement record. Taking into account the possibility that patients may have the same name, an encrypted identity identifier can be generated by combining the patient's ID number with the patient's name, and the identity identifier can be used as the patient's personal identification data.

[0063] Surgery classification module 102: select any symptom data of the current patient and the historical plan as the target symptom data; perform word segmentation processing on the target symptom data and extract the word segmentation describing the number of surgeries; classify the target symptom data according to the similar features of the set of word segmentation describing the number of surgeries corresponding to the target symptom data and the set of word segmentation describing multiple surgeries that are preset, and the classification result at least includes the classification of a single surgery.

[0064] Considering that the directly obtained disease data is too complex and not conducive to computer understanding of the meaning, word segmentation is a basic and important step in natural language processing, which facilitates the system to more accurately extract and analyze important information related to the disease, and facilitates more accurate classification of disease data in the future, thereby improving the reference of the auxiliary recommendation system. Therefore, any disease data in the current patient and historical plans is selected as the target disease data; the target disease data is segmented; considering that surgery is divided into the first or single surgery, the second or multiple surgeries, a patient may undergo a second or multiple surgeries after orthopedic surgery due to the complexity of the surgery and poor recovery, and the second or multiple surgeries The focus and planning of the orthopedic postoperative rehabilitation training program after the second operation is different from that of the first operation. Therefore, the target disease data is first classified according to the number of operations, so it is necessary to extract the segmentation words describing the number of operations; considering that the weaker the similarity between the set of segmentation words describing the number of operations corresponding to the target disease data and the preset set of segmentation words describing multiple operations, the less likely the target disease data is to be a second or multiple operations, and the more likely it is to correspond to a single operation, so according to the similarity between the set of segmentation words describing the number of operations corresponding to the target disease data and the preset set of segmentation words describing multiple operations, the target disease data is classified, and the classification result includes at least a single operation classification.

[0065] Preferably, in one embodiment of the present invention, the algorithm used for word segmentation of the target disease data is the JieBa word segmentation algorithm; in other embodiments of the present invention, other word segmentation methods such as the Ansj word segmenter may also be used, which are technical means well known to those skilled in the art and will not be described in detail.

[0066] Preferably, in one embodiment of the present invention, considering that participles such as again, again, and second time may be detected in the participles, but these participles may describe the recurrence of a symptom and may not describe the number of surgeries, so these participles need to be combined with the next adjacent participle to form a participle; because the symptom description during the second or multiple surgeries usually includes descriptions such as re-surgery, re-surgery, or second surgery, the participles describing the number of surgeries include at least: first surgery, first surgery, re-surgery, re-surgery, and second surgery.

[0067] It should be noted that the segmentation of the symptom data may only include one of them, or even no segmentation describing the number of surgeries, and may not necessarily include all segmentation describing the number of surgeries. In other embodiments of the present invention, the implementer may also set other segmentation describing the number of surgeries.

[0068] Preferably, in one embodiment of the present invention, considering that the similarity coefficient is a common means of measuring similarity features, the similarity coefficient between the set of surgery number description segmentations corresponding to the target symptom data and the preset multiple surgery description segmentation set is obtained as the first similarity coefficient; the larger the first similarity coefficient, the more obvious the similarity feature between the set of surgery number description segmentations corresponding to the target symptom data and the preset multiple surgery description segmentation set, and the more likely the target symptom data corresponds to multiple surgeries, so when the first similarity coefficient is greater than the first preset threshold, the target symptom data is classified into the multiple surgery classification; when the first similarity coefficient is less than or equal to the first preset threshold, the target symptom data is classified into the single surgery classification.

[0069] As an example, a preset word segmentation set describing multiple surgeries includes: reoperation, re-operation and secondary operation; the first similarity coefficient is obtained by the Jaccard similarity coefficient, the first preset threshold is 0.7, and the target disease data is preliminarily classified.

[0070] In other embodiments of the present invention, the implementer may replace the Jaccard similarity coefficient with the Dice similarity coefficient to obtain the first similarity coefficient, or may set other preset multiple surgical description word sets, or may set other first preset thresholds, which are all technical means well known to those skilled in the art and will not be elaborated herein.

[0071] Symptom classification module 103: according to the frequency characteristics of each segmentation in the target symptom data, obtain the representative segmentation set of the target symptom data; select any classification as the target classification; within the target classification, according to the similar characteristics of the representative segmentation set between any two symptom data, obtain the similarity parameters between any two symptom data; according to the similarity parameters between the symptom data, cluster the symptom data of historical patients.

[0072] Considering that the disease data may contain a large number of participles, directly using all the participles for analysis will increase the computational complexity and may introduce noise. The frequency characteristics of the participles reflect the importance of the participles in describing the corresponding disease. The more times a participle is mentioned and the higher the frequency, the more the participle can represent the patient's disease. Therefore, selecting a part of the most representative participles through the frequency characteristics of the participles can effectively reduce the dimension. In the subsequent similarity analysis process, we can focus on the most representative part of the disease data, making the similarity analysis between different diseases more accurate. Therefore, according to the frequency characteristics of each participle in the target disease data, obtain the representative participle set of the target disease data; select any category as the target category.

[0073] Preferably, in one embodiment of the present invention, the frequency of occurrence of each word in the target symptom data is obtained, and the word segments are sorted from large to small according to the frequency of occurrence to obtain a word segmentation sequence; and the word segments with a preset proportion in the word segmentation sequence are intercepted to form a word segmentation set representing the target symptom data.

[0074] As an example, the preset ratio is 25%, that is, the segmentations whose occurrence frequencies are in the top 25% are selected to form the representative segmentation set.

[0075] It should be noted that the method of acquiring the frequency is already a prior art, and in other embodiments of the present invention, the implementer may also set other preset ratios.

[0076] Considering that the focus and planning of orthopedic postoperative rehabilitation training programs after multiple surgeries are different from those of orthopedic postoperative rehabilitation training programs for the first surgery, the similarity of disease data within the same category is more valuable for analysis; therefore, any category is selected as the target category; within the target category, based on the similar features of the word set characterizing between any two disease data, the similarity parameters between any two disease data are obtained, and the similar features between the disease data are characterized by the similar parameters, which facilitates subsequent clustering operations and prepares for the final auxiliary recommendation.

[0077] Preferably, in one embodiment of the present invention, similar to the method used in classification, similarity coefficients are also used to represent similar features of the word sets representing the symptom data. In the target classification, a symptom data group to be analyzed is formed according to any two symptom data; a similarity coefficient between two word sets representing the symptom data group to be analyzed is obtained as a second similarity coefficient;

[0078] It is also considered that when the disease data are similar, the corresponding diagnosis results and disease descriptions are also relatively similar, which is reflected in the similarity of the frequencies of occurrence of the words in the characterization word set. In order to facilitate the comparison of the two characterization word sets, the overall similarity characteristics of the frequencies of occurrence of the words in the characterization word sets are used for comparison, and each word is no longer compared one by one, so as to improve the analysis efficiency. Based on this, the frequency similarity coefficient is obtained according to the overall similarity characteristics of the frequencies of occurrence of the words in the two characterization word sets corresponding to the disease data group to be analyzed;

[0079] The similarity parameter corresponding to the disease data group to be analyzed is obtained according to the second similarity coefficient and the frequency similarity coefficient; the second similarity coefficient and the frequency similarity coefficient are both positively correlated with the similarity parameter.

[0080] As an example, the calculation formula for similarity parameters includes:

[0081] ;

[0082] in, Indicates that within the target category, and Similarity parameters between disease data, ; represents the Jaccard similarity coefficient; Indicates that within the target category, and The second similarity coefficient between the disease data; Indicates that within the target category, A set of word segments representing disease data; Indicates that within the target category, A set of word segments representing disease data; Indicates that within the target category, The mean frequency of all the words in the word set representing the disease data; Indicates that within the target category, The mean frequency of all the words in the word set representing the disease data; Indicates the preset divide-by-zero positive parameter. In this example, ; Indicates that within the target category, and The frequency similarity coefficient between the disease data.

[0083] In the calculation formula of the similarity parameter, the larger the second similarity coefficient is, the more obvious the similarity characteristics of the word set representing the two disease data are from the perspective of the similarity coefficient. The more similar the two disease data are, the larger the similarity parameter is. The overall characteristics of the frequency of occurrence of the word set representing the mean frequency of occurrence are represented by the mean frequency of occurrence. The smaller it is, the smaller the overall characteristic difference of the frequency is and the higher the similarity is. After negative correlation mapping by taking the inverse, the larger the frequency similarity coefficient is, the more similar the two disease data are from the perspective of the frequency of occurrence of the characterizing segmentation words. The larger the similarity parameter is, and with the help of the preset positive division by zero parameter, the denominator is prevented from being zero. The similarity parameter integrates multiple angles and jointly expresses the similar characteristics of the characterizing segmentation sets between the two disease data, providing a reliable basis for subsequent further clustering operations.

[0084] It should be noted that in other embodiments of the present invention, the implementer may also use other methods such as negative correlation mapping function to perform negative correlation mapping, for example, using Negative correlation mapping function, represents the independent variable, Expressed as a natural constant An exponential function with a base may also be used to obtain similarity parameters by positive correlation methods such as summation, for example, the sum of the second similarity coefficient and the frequency similarity coefficient is used as the similarity parameter; the MinHash algorithm in the prior art may also be used to calculate the similarity between two sets of characterizing word segments, which may be used as a supplement to the second similarity coefficient and the frequency similarity coefficient, for example, the product of the similarity calculated by the MinHash algorithm, the second similarity coefficient and the frequency similarity coefficient is used as the similarity parameter; it may also be used as a replacement for any one of the second similarity coefficient and the frequency similarity coefficient to obtain the similarity parameter.

[0085] After analyzing the similar features between the symptom data, the symptom data of historical patients can be clustered according to the similar parameters between the symptom data to prepare for subsequent auxiliary recommendations.

[0086] Preferably, in one embodiment of the present invention, the algorithm used to cluster the historical patient's symptom data is the K-means algorithm, and the number of clusters is determined using the existing silhouette coefficient method. The larger the similarity parameter, the smaller the cluster distance, which are all existing technologies.

[0087] As an example, similar parameters are passed The negative correlation mapping function is used for mapping, and the mapping value is used as the clustering distance. represents the independent variable, Expressed as a natural constant The exponential function of base .

[0088] In other embodiments of the present invention, the implementer may also perform clustering based on similar parameters using an Affinity Propagation clustering algorithm, which is already a prior art and will not be described in detail.

[0089] It should be noted that in the embodiment of the present invention, the historical plans and their patient personal identification and symptom data are one-to-one corresponding. After the symptom data are clustered, the corresponding historical plans will also be clustered. In addition, the embodiment of the present invention first classifies based on the number of surgeries, and then clusters the same type of surgeries. There are no similar parameters between the symptom data of different surgical classifications, and no clustering is performed.

[0090] Plan screening module 104: Match historical plans with the same patient personal identifier. In the classification of a single surgery, any matched historical plan is used as the plan to be analyzed; obtain the reference coefficient of the plan to be analyzed based on the similar features of the word segmentation set of the surgical type of the plan to be analyzed and the word segmentation set of the surgical type of the corresponding matching historical plan; eliminate the historical plan based on the reference coefficient.

[0091] In the embodiment of the present invention, considering that among the historical plans classified into multiple surgeries, some of the rehabilitation training plans classified into single surgeries have unsatisfactory rehabilitation effects, so that after the first surgery, the surgery is performed again, the rehabilitation effects of these historical plans are poor and cannot be recommended to rehabilitation physicians for reference. Therefore, some historical plans in the single surgery classification need to be eliminated to avoid recommending postoperative rehabilitation training plans to customers with poor rehabilitation effects in the future, thereby improving the reference value of auxiliary recommendations.

[0092] Considering that the patient's personal identification is an important identification of the patient's surgery and rehabilitation training program, when a patient undergoes multiple surgeries, there must be a corresponding single surgery. In the single surgery classification, some patients may only need one surgery and one rehabilitation training to recover to normal, and there may not be corresponding rehabilitation program records or symptom data records for multiple surgeries. Therefore, historical programs with the same patient personal identification are matched. In the single surgery classification, any matched historical program is used as the program to be analyzed;

[0093] Considering that the orthopedic postoperative rehabilitation training plan contains the surgical type information, when the surgical type similarity of the matched historical plans is too high, it means that the effect of the rehabilitation training plan after the first surgery is not ideal and the reference value is low. Therefore, based on the similarity characteristics of the word segmentation set of the surgical type of the plan to be analyzed and the corresponding matching historical plan’s surgical type word segmentation set, the reference coefficient of the plan to be analyzed is obtained to characterize the reference value of the plan to be analyzed, providing a basis for the subsequent elimination of historical plans.

[0094] Preferably, in one embodiment of the present invention, considering that in the single surgery classification, any matched historical scheme may correspond to multiple schemes in the multiple surgery classification, for example, if a patient undergoes 3 surgeries and postoperative rehabilitation, then the historical scheme belonging to the single surgery classification corresponds to two schemes in the multiple surgery classification; for ease of description, in the single surgery classification, any matched historical scheme corresponds to at least one scheme in the multiple surgery classification; the scheme matched by the scheme to be analyzed is taken as the matching scheme;

[0095] Similarly, the similarity coefficient is used to express the similarity characteristics of the word set of the surgical type of the to-be-analyzed scheme and the word set of the surgical type of the corresponding matching historical scheme. The larger the third similarity coefficient, the more obvious the similarity characteristics, the worse the rehabilitation effect of the to-be-analyzed scheme, and the lower the reference coefficient.

[0096] Based on this, the similarity coefficient between the word segmentation set of the surgical type of the scheme to be analyzed and the word segmentation set of the surgical type of each corresponding matching scheme is obtained as the third similarity coefficient; according to the overall characteristics of the third similarity coefficients of the scheme to be analyzed and all corresponding matching schemes, the reference coefficient of the scheme to be analyzed is obtained; the overall characteristics of the third similarity coefficient are negatively correlated with the reference coefficient.

[0097] As an example, the calculation formula of the reference coefficient includes:

[0098] ;

[0099] in, Indicates The reference coefficient of the solution to be analyzed; Expressed as a natural constant The exponential function with base ; Indicates The number of matching solutions for the solutions to be analyzed; Indicates The word segmentation set of the surgical types of the plans to be analyzed; Indicates The solutions to be analyzed are A set of word segments of surgical types matching the solution; represents the Jaccard similarity coefficient; Indicates The scheme to be analyzed and its The third similarity coefficient corresponding to the matching scheme.

[0100] In the calculation formula of the reference coefficient, the third similarity coefficient is obtained through the Jaccard similarity coefficient, and the overall characteristics of the third similarity coefficients of the scheme to be analyzed and all corresponding matching schemes are reflected in the form of the mean. The larger the third similarity coefficient, the more obvious the similarity characteristics, which means that the rehabilitation effect of the scheme to be analyzed is worse and the reference coefficient is lower. The reference coefficient is obtained to characterize the reference value of the scheme to be analyzed, which provides a basis for the subsequent elimination of historical schemes.

[0101] It should be noted that the Jaccard similarity coefficient can also be replaced by the Dice similarity coefficient to obtain the third similarity coefficient, which will not be described in detail.

[0102] It should be noted that, in one embodiment of the present invention, please refer to the anterior cruciate ligament reconstruction postoperative rehabilitation plan shown. The anterior cruciate ligament reconstruction before the word postoperative is the surgery type, so all the participles before the postoperative participle in the first paragraph of each orthopedic postoperative rehabilitation training plan are used as the participle set of the surgery type of the corresponding plan.

[0103] Preferably, in one embodiment of the present invention, when the reference coefficient is less than the second preset threshold, the corresponding historical plan in the single surgery classification is eliminated.

[0104] As an example, the second preset threshold is 0.5. When the reference coefficient is less than 0.5, it is considered that the reference value of the corresponding historical plan in the single surgery classification is too low. In order to improve the reference significance of the subsequent auxiliary recommended plans, the corresponding historical plans are eliminated, and the symptom data corresponding to the clustering cluster of the symptom data and the historical plans of the clustering cluster of the historical plans are all eliminated.

[0105] Auxiliary recommendation module 105: obtains a recommended orthopedic postoperative rehabilitation training program based on the symptom data of the current patient and the similar parameters corresponding to the symptom data in all clusters corresponding to similar historical patients.

[0106] After feature analysis and screening of existing historical plans, the same method can be used to classify the current patient's symptom data, and obtain similar parameters between the current patient's symptom data and the symptom data in all clusters corresponding to similar historical patients, and compare to obtain the cluster that best matches the current patient's symptom data, so as to obtain a recommended orthopedic postoperative rehabilitation training plan.

[0107] Preferably, in one embodiment of the present invention, the patient corresponding to the historical scheme is referred to as a historical patient, and the overall similarity parameter of each cluster of the symptom data of the current patient and the symptom data of the same type is obtained according to the corresponding similarity parameters between the symptom data of the current patient and the symptom data of all historical patients;

[0108] The cluster with the largest overall similarity parameter is selected as the orthopedic postoperative rehabilitation training program cluster for the current patient; all historical programs in the orthopedic postoperative rehabilitation training program cluster for the current patient are used as recommended orthopedic postoperative rehabilitation training programs.

[0109] As an example, assuming that the current patient's symptom data is classified into multiple surgical classifications, the similarity parameters of the symptom data of each historical plan in the multiple surgical classifications and the symptom data of the current patient are obtained, and any clustering cluster corresponding to the multiple surgical classifications is used as the target clustering cluster. The average value of the similarity parameters of all symptom data in the target clustering cluster and the current symptom data is obtained by means of the mean, as the overall similarity parameter of the target clustering cluster, and the overall similarity parameters corresponding to the current patient's symptom data and each clustering cluster in the same surgical classification are obtained in turn, and the clustering cluster with the largest overall similarity parameter is selected as the orthopedic postoperative rehabilitation training plan cluster for the current patient; all historical plans in the orthopedic postoperative rehabilitation training plan cluster for the current patient are used as recommended orthopedic postoperative rehabilitation training plans.

[0110] After obtaining the recommended orthopedic postoperative rehabilitation training plan, the corresponding orthopedic postoperative rehabilitation training plan can be presented to the rehabilitation physician for reference, to assist the rehabilitation physician in formulating a more suitable orthopedic postoperative rehabilitation training plan for the patient and improve the rehabilitation effect.

[0111] In summary, the embodiment of the present invention uses the segmentation description of the number of surgeries in the symptom data to preliminarily classify the symptom data by the number of surgeries; further, within each classification, the symptom data is clustered according to the similar features of the segmentation sets that characterize the symptom data; further, the reference coefficients of the historical solutions matched by other surgery categories in the single word surgery classification are obtained, and the historical solutions with low reference value are eliminated; further, based on the corresponding similar parameters between the current patient's symptom data and similar symptom data, a recommended orthopedic postoperative rehabilitation training program is provided. The present invention classifies the symptom data and historical solutions in detail, matches the current patient's symptom data to similar categories, and assists rehabilitation physicians in recommending suitable orthopedic postoperative rehabilitation training programs, helping rehabilitation physicians to develop more suitable orthopedic postoperative rehabilitation training programs for patients.

[0112] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0113] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. An auxiliary recommendation system for postoperative rehabilitation training in orthopedics, characterized in that: The system comprises: Data acquisition module: acquires historical plans for orthopedic postoperative rehabilitation training and the personal identification and symptom data of the patient corresponding to each of the historical plans; acquires the personal identification and symptom data of the current patient; the historical plans include the type of surgery; Surgery classification module: select any of the symptom data of the current patient and the historical plan as the target symptom data; perform word segmentation processing on the target symptom data and extract the segmentation describing the number of surgeries; classify the target symptom data according to the similarity between the set of segmentation describing the number of surgeries corresponding to the target symptom data and the set of segmentation describing multiple surgeries, and the classification result at least includes the classification of a single surgery; Symptom classification module: according to the frequency characteristics of each segmentation in the target symptom data, obtain the representative segmentation set of the target symptom data; select any classification as the target classification; within the target classification, according to the similar characteristics of the representative segmentation set between any two symptom data, obtain the similarity parameters between any two symptom data; according to the similarity parameters between the symptom data, cluster the symptom data of historical patients; A scheme screening module: matching the historical schemes with the same patient personal identifier, and in the single surgery classification, taking any matched historical scheme as the scheme to be analyzed; obtaining the reference coefficient of the scheme to be analyzed based on the similarity characteristics of the segmentation set of the surgery type of the scheme to be analyzed and the segmentation set of the surgery type of the corresponding matched historical scheme; and eliminating the historical scheme based on the reference coefficient; Auxiliary recommendation module: based on the symptom data of the current patient and the similar parameters corresponding to the symptom data in all clusters corresponding to similar historical patients, obtain a recommended orthopedic postoperative rehabilitation training program; The method for obtaining the recommended orthopedic postoperative rehabilitation training program includes: According to the similarity parameters corresponding to the symptom data of the current patient and the symptom data of the same type, obtaining the overall similarity parameter of each cluster of the symptom data of the current patient and the symptom data of the same type for all historical patients; Selecting the cluster with the largest overall similarity parameter as the orthopedic postoperative rehabilitation training program cluster for the current patient; using all historical programs in the orthopedic postoperative rehabilitation training program cluster for the current patient as recommended orthopedic postoperative rehabilitation training programs; The method for classifying the target disease data comprises: Obtaining a similarity coefficient between the set of the number of surgery description segmentations corresponding to the target disease data and a preset multiple surgery description segmentation set as a first similarity coefficient; when the first similarity coefficient is greater than a first preset threshold, classifying the target disease data into a multiple surgery classification; when the first similarity coefficient is less than or equal to the first preset threshold, classifying the target disease data into a single surgery classification; The method for obtaining the reference coefficient includes: In the single surgery classification, any matched historical scheme corresponds to at least one scheme in the multiple surgery classifications; the scheme matched by the scheme to be analyzed is taken as a matching scheme; Obtain a similarity coefficient between the word segment set of the surgical type of the scheme to be analyzed and the word segment set of the surgical type of each corresponding matching scheme as a third similarity coefficient; obtain a reference coefficient of the scheme to be analyzed based on the overall characteristics of the third similarity coefficients of the scheme to be analyzed and all corresponding matching schemes; the overall characteristics of the third similarity coefficient are negatively correlated with the reference coefficient.

2. The auxiliary recommendation system for orthopedic postoperative rehabilitation training according to claim 1, characterized in that: The method for acquiring the representation segmentation set includes: The occurrence frequency of each segmentation word in the target symptom data is obtained, and the segmentation words are sorted from large to small according to the occurrence frequency to obtain a segmentation sequence; and the segmentation words with a preset proportion in the segmentation sequence are intercepted to form a segmentation set representing the target symptom data.

3. The auxiliary recommendation system for orthopedic postoperative rehabilitation training according to claim 2, characterized in that: The method for obtaining the similarity parameters includes: In the target classification, a symptom data group to be analyzed is formed according to any two of the symptom data; a similarity coefficient between two of the representation segmentation sets corresponding to the symptom data group to be analyzed is obtained as a second similarity coefficient; Obtaining a frequency similarity coefficient according to the overall similarity characteristics of the occurrence frequencies of the segmented words in the two representative segmented word sets corresponding to the disease data group to be analyzed; The similarity parameter corresponding to the disease data group to be analyzed is obtained according to the second similarity coefficient and the frequency similarity coefficient; the second similarity coefficient and the frequency similarity coefficient are both positively correlated with the similarity parameter.

4. The auxiliary recommendation system for orthopedic postoperative rehabilitation training according to claim 1, characterized in that: The method for eliminating the historical solution according to the reference coefficient includes: When the reference coefficient is less than a second preset threshold, the historical plan corresponding to the single surgery classification is eliminated.

5. The auxiliary recommendation system for orthopedic postoperative rehabilitation training according to claim 1, characterized in that: The algorithm used for word segmentation processing of the target disease data is the JieBa word segmentation algorithm.

6. The auxiliary recommendation system for orthopedic postoperative rehabilitation training according to claim 1, characterized in that: The segmentation describing the number of surgeries at least includes: first surgery and second surgery.

7. The auxiliary recommendation system for orthopedic postoperative rehabilitation training according to claim 1, characterized in that: The algorithm used to cluster the disease data of historical patients is the K-means algorithm.

Images