A medical resource recommendation method, device, equipment and storage medium

Abstract

The present invention relates to artificial intelligence technology, and discloses a method for recommending medical resources, including: using historical patient medical treatment data to extract correlation features between historical patient information feature sets and historical medical resource feature sets; using historical patient information feature sets, historical medical resource feature sets and correlation features to construct a historical patient medical recommendation sample set; based on the historical patient medical recommendation sample set, training a pre-constructed medical resource recommendation model to obtain a trained medical resource recommendation model; using the trained medical resource recommendation model to match and analyze the patient information to be analyzed and the available medical resources to obtain recommended medical resource data. In addition, the present invention also relates to blockchain technology, and the historical patient medical recommendation sample set can be stored in a node of the blockchain. The present invention also proposes a medical resource recommendation device, an electronic device, and a storage medium. The present invention can improve the accuracy of medical resource recommendations.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a medical resource recommendation method, device, electronic device and computer-readable storage medium. Background Art

[0002] Seeking medical advice is a common thing in people's lives. Choosing the right hospital and doctor is crucial. However, due to the limitation of medical knowledge, most patients cannot correctly evaluate their own symptoms. At the same time, the Internet information is complicated and patients have no way to distinguish the authenticity. Due to the lack of medical expertise, it is also difficult to find a hospital and doctor who is good at their condition, which reduces medical efficiency and wastes medical resources. By configuring a medical recommendation system, it can help patients and doctors greatly shorten the time of consultation and save manpower and material resources.

[0003] Existing medical recommendation systems often adopt a fixed search method, or simply use the historical interaction information between doctors and patients as input to recommend relevant medical information. The medical resources recommended by this method are not accurate enough. Summary of the invention

[0004] The present invention provides a medical resource recommendation method, device and computer-readable storage medium, the main purpose of which is to solve the problem of low accuracy when recommending medical resources.

[0005] To achieve the above object, the present invention provides a medical resource recommendation method, comprising:

[0006] Acquire a historical patient data set and a historical medical resource set, extract a historical patient information feature set and historical patient medical treatment data from the historical patient data set, and extract a historical medical resource feature set from the historical medical resource set;

[0007] Utilizing the historical patient medical treatment data, extracting correlation features between the historical patient information feature set and the historical medical resource feature set, and constructing a historical patient medical recommendation sample set using the historical patient information feature set, the historical medical resource feature set and the correlation features;

[0008] Based on the historical patient medical recommendation sample set, using a contrast divergence algorithm, training the pre-built medical resource recommendation model to obtain a trained medical resource recommendation model;

[0009] Obtain patient information to be analyzed and available medical resources, use the trained medical resource recommendation model to perform matching analysis on the patient information to be analyzed and the available medical resources, and obtain recommended medical resource data.

[0010] Optionally, extracting a historical patient information feature set from the historical patient data set includes:

[0011] Convert the historical patient data set into a text format to obtain a historical patient text data set;

[0012] Performing word segmentation and part-of-speech tagging on the historical patient text dataset to obtain word segmentation and part-of-speech tagging results;

[0013] Extracting nouns and noun phrases from the segmented words according to the result of the part-of-speech tagging, and obtaining the feature frequencies of historical patient information according to the nouns and noun phrases, and generating a frequent pattern tree according to the feature frequencies of historical patient information;

[0014] Identify features in the frequent pattern tree to obtain a candidate historical patient information feature set;

[0015] The point mutual information value of each feature in the candidate historical patient information feature set is calculated, and historical patient information features whose point mutual information values ​​are less than a preset standard threshold are filtered out from the candidate historical patient information feature set to obtain a historical patient information feature set.

[0016] Optionally, the extracting the correlation features between the historical patient information feature set and the historical medical resource feature set by using the historical patient consultation data includes:

[0017] Performing word segmentation on each historical patient information feature in the historical patient information feature set and each historical medical resource feature in the historical medical resource feature set, and extracting key words and related words from the word segmentation results;

[0018] The key words and relational words irrelevant to the historical patient consultation data are deleted to obtain the correlation features between the historical patient information feature set and the historical medical resource feature set.

[0019] Optionally, the pre-built medical resource recommendation model is trained based on the historical patient medical recommendation sample set using a contrast divergence algorithm to obtain a trained medical resource recommendation model, including:

[0020] Dividing the historical patient medical recommendation sample set into a training set and a test set;

[0021] According to the training set, adjusting the parameters of the medical resource recommendation model using a contrast divergence algorithm, iteratively training the medical resource recommendation model, and obtaining a trained medical resource recommendation model;

[0022] The trained medical resource recommendation model is tested and adjusted using the test set to obtain a trained medical resource recommendation model.

[0023] Optionally, adjusting the parameters of the medical resource recommendation model using a contrast divergence algorithm according to the training set, iteratively training the medical resource recommendation model to obtain a trained medical resource recommendation model includes:

[0024] The historical patient information feature set, the historical medical resource feature set and the associated features in the training set are converted into feature vectors, and the feature vectors are processed by mean, variance and whitening, and the processed feature vectors are uniformly normalized to between 0 and 1 in ascending order to obtain a normalized sample set;

[0025] Initializing the weight values ​​between the visual layer neurons and the hidden layer neurons in the medical resource recommendation model, as well as the bias parameters of the visual layer neurons and the bias parameters of the hidden layer neurons;

[0026] Using the normalized sample set, the weight value, the bias parameter of the visual layer neuron and the bias parameter of the hidden layer neuron, the visual layer neuron and the hidden layer neuron are iterated cyclically, and the activation probability of the hidden layer neuron is calculated;

[0027] Using the hidden layer neurons after the loop iteration, reverse loop iteration of the visual layer neurons, and calculating the activation probability of the visual layer neurons;

[0028] According to the activation probability of the hidden layer neurons and the activation probability of the visual layer neurons, the weight value and the bias parameter are adjusted using a contrast divergence algorithm, and the activation probability of the visual layer neurons is recalculated using the adjusted weight value and the bias parameter, and the output result of the medical resource recommendation model corresponding to the maximum value of the activation probability of the visual layer neurons is used as the predicted recommended medical resource;

[0029] The loss function is used to calculate the loss value between the predicted recommended medical resources and the historical patient visit data, and the parameters of the medical resource recommendation model are adjusted according to the loss value. The step of initializing the weight values ​​between the visual layer neurons and the hidden layer neurons in the medical resource recommendation model is returned until the loss value is less than the preset loss threshold, thereby obtaining a trained medical resource recommendation model.

[0030] Optionally, adjusting the weight value and the bias parameter by using a contrast divergence algorithm according to the activation probability of the hidden layer neurons and the activation probability of the visual layer neurons includes:

[0031] Initializing the state vectors of the visual layer neurons, the hidden layer neurons, and the weight values ​​between the visual layer neurons and the hidden layer neurons using the normalized sample set;

[0032] Using the activation probability of the hidden layer neurons and the activation probability of the visual layer neurons, K-step Gibbs sampling is performed to obtain the state vector of the visual layer neurons and the state vector of the hidden layer neurons corresponding to time t-1, and the state vector of the visual layer neurons and the state vector of the hidden layer neurons corresponding to time t;

[0033] Using the state vector of the visual layer neuron and the state vector of the hidden layer neuron corresponding to the time t-1, and the state vector of the visual layer neuron and the state vector of the hidden layer neuron corresponding to the time t, the normalized sample set is cyclically calculated to obtain a comparison error value;

[0034] The weight value, the bias parameter of the visible layer neuron and the bias parameter of the hidden layer neuron are adjusted using the contrast error value.

[0035] Optionally, the acquiring of the patient information to be analyzed and the available medical resources, and using the trained medical resource recommendation model to perform matching analysis on the patient information to be analyzed and the available medical resources to obtain recommended medical resource data, include:

[0036] Obtain patient information to be analyzed and available medical resources uploaded by the user;

[0037] Extracting features of the patient information to be analyzed from the patient information to be analyzed and extracting features of available medical resources from the available medical resources;

[0038] Using the trained medical resource recommendation model, matching and analyzing the patient information characteristics to be analyzed and the available medical resource characteristics to obtain recommended medical resource data;

[0039] The recommended medical resource data is fed back to the user terminal.

[0040] In order to solve the above problems, the present invention further provides a medical resource recommendation device, the device comprising:

[0041] A feature extraction module is used to obtain a historical patient data set and a historical medical resource set, extract a historical patient information feature set from the historical patient data set and extract historical patient medical treatment data, and extract a historical medical resource feature set from the historical medical resource set;

[0042] A sample construction module, used to use the historical patient medical treatment data to extract the correlation features between the historical patient information feature set and the historical medical resource feature set, and to construct a historical patient medical recommendation sample set using the historical patient information feature set, the historical medical resource feature set and the correlation features;

[0043] A model training module is used to train the pre-built medical resource recommendation model based on the historical patient medical recommendation sample set using a contrast divergence algorithm to obtain a trained medical resource recommendation model;

[0044] The recommendation module is used to obtain the patient information to be analyzed and the available medical resources, and use the trained medical resource recommendation model to match and analyze the patient information to be analyzed and the available medical resources to obtain recommended medical resource data.

[0045] In order to solve the above problem, the present invention further provides an electronic device, the electronic device comprising:

[0046] at least one processor; and,

[0047] a memory communicatively connected to the at least one processor; wherein,

[0048] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the medical resource recommendation method described above.

[0049] In order to solve the above problems, the present invention also provides a computer-readable storage medium, in which at least one computer program is stored. The at least one computer program is executed by a processor in an electronic device to implement the above-mentioned medical resource recommendation method.

[0050] The embodiment of the present invention extracts the historical patient information feature set and the historical medical resource feature set from the historical patient data set and the historical medical resource set respectively; uses the historical patient medical treatment data to extract the correlation features between the historical patient information feature set and the historical medical resource feature set; uses the historical patient information feature set, the historical medical resource feature set and the correlation features to construct a historical patient medical recommendation sample set, which is conducive to improving the matching degree between patients and recommended medical resources, thereby improving the accuracy of recommended medical resources; uses the contrast divergence algorithm to train to obtain a medical resource recommendation model, accelerates the model convergence speed, reduces the number of model iterations, and improves the accuracy of medical resource recommendations. Therefore, the medical resource recommendation method, device, electronic device and computer-readable storage medium proposed in the present invention can solve the problem of low accuracy of medical resource recommendations. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 A schematic diagram of a process for recommending medical resources according to an embodiment of the present invention;

[0052] Figure 2 for Figure 1A schematic diagram of a detailed implementation process of one of the steps in the medical resource recommendation method shown;

[0053] Figure 3 for Figure 1 A schematic diagram of a detailed implementation flow of another step in the medical resource recommendation method shown;

[0054] Figure 4 A functional module diagram of a medical resource recommendation device provided by an embodiment of the present invention;

[0055] Figure 5 A schematic diagram of the structure of an electronic device for implementing the medical resource recommendation method provided in one embodiment of the present invention.

[0056] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0057] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0058] The embodiment of the present application provides a medical resource recommendation method. The execution subject of the medical resource recommendation method includes but is not limited to at least one of the electronic devices such as a server, a terminal, etc. that can be configured to execute the method provided in the embodiment of the present application. In other words, the medical resource recommendation method can be executed by software or hardware installed on a terminal device or a server device, and the software can be a blockchain platform. The server includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, etc. The server can be an independent server, or it can be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms.

[0059] Reference Figure 1 FIG. 1 is a flow chart of a medical resource recommendation method provided in an embodiment of the present invention. In this embodiment, the medical resource recommendation method includes:

[0060] S1. Acquire a historical patient data set and a historical medical resource set, extract a historical patient information feature set and historical patient medical treatment data from the historical patient data set, and extract a historical medical resource feature set from the historical medical resource set.

[0061] In one embodiment of the present invention, the historical patient data set includes: historical patient medical record data and historical patient consultation data.

[0062] In the embodiment of the present invention, the historical patient medical record data includes the patient's basic information, disease type, symptom description, physical condition, previous treatment, pathological type, etc. Among them, the basic information includes: name, age, weight, gender, etc.; the disease types include: cardiovascular disease, lung disease, rheumatic immune system disease, urinary system disease, kidney disease, diabetes and endocrine system disease, otolaryngology disease, digestive system disease, gynecological disease, nervous system disease, ophthalmology disease, orthopedic disease, etc.; the symptom description is the patient's abnormal sensation, such as dizziness, vertigo, asthma, cough, etc.; the physical condition can be obtained through the Eastern Cooperative Oncology Group of the United States. The ECOG score is an indicator of the patient's general health status and tolerance to treatment based on his or her physical strength, where 0 points means that the patient's ability to move is completely normal, with no difference from the ability to move before the onset of the disease; 1 point means that the patient can move freely and engage in light physical activities, including general housework or office work, but cannot engage in heavier physical activities; 2 points means that the patient can move freely and take care of himself or herself, but has lost the ability to work and can get up and move around for at least half of the day; 3 points means that the patient can only partially take care of himself or herself and spend more than half of the day in bed or in a wheelchair; 4 points means that the patient is bedridden and cannot take care of himself or herself; the previous treatment status refers to the patient's previous treatment status, including whether he or she has received surgical treatment.

[0063] Furthermore, the historical patient medical treatment data includes the hospital data and doctor data of the historical patients' final actual medical treatment, such as the hospital name, hospital ranking, hospital grade, doctor's name, doctor's department, doctor's current position, doctor's expertise, etc.

[0064] Furthermore, the historical medical resource set includes hospital data disclosed on official websites of domestic and foreign hospitals and authoritative websites, and doctor data disclosed on official websites of domestic and foreign hospitals and authoritative websites.

[0065] Among them, the hospital data may specifically include hospital rankings, whether the hospital participates in the formulation of disease-related diagnosis and treatment guidelines, whether the hospital has doctors working in disease-related professional academic institutions, the number of disease-related papers published by the hospital, the number of disease-related clinical trials conducted by the hospital, and the number of expert teams in the hospital; the doctor data may include the doctor's name, the doctor's center / project, the doctor's department, the doctor's current position, the doctor's academic title, the doctor's clinical expertise, whether the doctor participates in the formulation of disease-related diagnosis and treatment guidelines, whether the doctor works in disease-related associations and foundations, and the number of disease-related papers published by the doctor.

[0066] In one of the embodiments of the present invention, the historical patient data set may include multiple data forms such as text and images. Therefore, after obtaining the historical patient data set, it is necessary to use existing text conversion technology (such as OCR recognition technology, etc.) to uniformly convert the historical patient data sets in various data forms into text form, so as to extract the historical patient information feature set and historical patient medical data.

[0067] In detail, the step of extracting a historical patient information feature set from the historical patient data set in S1 includes:

[0068] Convert the historical patient data set into a text format to obtain a historical patient text data set;

[0069] Performing word segmentation and part-of-speech tagging on the historical patient text dataset to obtain word segmentation and part-of-speech tagging results;

[0070] Extracting nouns and noun phrases from the segmented words according to the result of the part-of-speech tagging, and obtaining the feature frequencies of historical patient information according to the nouns and noun phrases, and generating a frequent pattern tree according to the feature frequencies of historical patient information;

[0071] Identify features in the frequent pattern tree to obtain a candidate historical patient information feature set;

[0072] The point mutual information value of each feature in the candidate historical patient information feature set is calculated, and historical patient information features whose point mutual information values ​​are less than a preset standard threshold are filtered out from the candidate historical patient information feature set to obtain a historical patient information feature set.

[0073] In one embodiment of the present invention, the following algorithm can be used to calculate the point mutual information value of each feature in the candidate historical patient information feature set: :

[0074]

[0075] Wherein, historical patient represents the name of the historical patient, and feature represents each feature in the candidate historical patient information feature set; Indicates the number of historical patients obtained by searching; Indicates the number of results obtained by searching for the feature; It represents the number obtained by searching the feature and the historical patient; the higher the point mutual information value is, the higher the correlation between the historical patient and the feature is. When the point mutual information value is lower than the preset standard threshold, the corresponding historical patient information feature is filtered out; when the point mutual information value is higher than the preset standard threshold, the corresponding historical patient information feature is retained and combined to obtain the historical patient information feature set.

[0076] Furthermore, the embodiment of the present invention may adopt the same method as the above-mentioned method of extracting the historical patient information feature set to extract the historical patient consultation data and the historical medical resource feature set, which will not be described in detail here.

[0077] S2. Utilize the historical patient medical treatment data to extract the correlation features between the historical patient information feature set and the historical medical resource feature set, and utilize the historical patient information feature set, the historical medical resource feature set and the correlation features to construct a historical patient medical recommendation sample set.

[0078] In detail, the use of the historical patient consultation data in S2 to extract the correlation features between the historical patient information feature set and the historical medical resource feature set includes:

[0079] Performing word segmentation on each historical patient information feature in the historical patient information feature set and each historical medical resource feature in the historical medical resource feature set, and extracting key words and related words from the word segmentation results;

[0080] The key words and relational words irrelevant to the historical patient consultation data are deleted to obtain the correlation features between the historical patient information feature set and the historical medical resource feature set.

[0081] Specifically, the embodiment of the present invention deletes key words and relationship words that are not related to the historical patient treatment data, and deletes features with too much missing data in the historical patient information features and historical medical resource features, thereby screening out features that play a key role in recommending medical resources, such as the patient's pathological type, whether the doctor participates in the formulation of disease-related diagnosis and treatment guidelines, etc., and uses the screened out features that play a key role in recommending medical resources as the association features between the historical patient information feature set and the historical medical resource feature set.

[0082] Furthermore, in an embodiment of the present invention, before constructing a historical patient medical recommendation sample set, it is necessary to unify the representation forms of the features in the historical patient information feature set, the historical medical resource feature set, and the associated features. For example, in the historical patient information feature set, the historical medical resource feature set, and the associated features, some time representation forms are "**** year ** month ** day ** hour ** minute ** second", and some time representation forms are "**** year / ** month / ** day ** hour ** minute ** second". In this case, an embodiment of the present invention can unify the time representation forms into "**** year ** month ** day ** hour ** minute ** second"; for another example, in the features, some hospital names are abbreviations of the hospital names, and some are full names of the hospital names. In this case, all hospital names can be unified into full names of the hospital names.

[0083] S3. Based on the historical patient medical recommendation sample set, the pre-built medical resource recommendation model is trained using a contrast divergence algorithm to obtain a trained medical resource recommendation model.

[0084] In the implementation of the present invention, the medical resource recommendation model is constructed based on a multi-layer restricted Boltzmann machine (RBM). The RBM is a probabilistic generation model based on deep learning, comprising a visual layer and a hidden layer, the visual layer is composed of a plurality of visual layer neurons, the hidden layer is composed of a plurality of hidden layer neurons, there is a connection between the visual layer and the hidden layer, but there is no connection between the units within the layer, and the multi-layer restricted Boltzmann machine network connection is determined by the weight value under self-orientation. In one embodiment of the present invention, the multi-layer restricted Boltzmann machine can be composed of 3 RBMs.

[0085] For details, see Figure 2 As shown, the S3 includes:

[0086] S31, dividing the historical patient medical recommendation sample set into a training set and a test set;

[0087] S32. According to the training set, using a contrast divergence algorithm to adjust the parameters of the medical resource recommendation model, and iteratively train the medical resource recommendation model to obtain a trained medical resource recommendation model;

[0088] S33. Use the test set to test and adjust the trained medical resource recommendation model to obtain a trained medical resource recommendation model.

[0089] In one embodiment of the present invention, 80% of the samples in the historical patient medical recommendation sample set can be divided into a training set, and the remaining 20% ​​of the samples can be divided into a test set.

[0090] The contrastive divergence algorithm (CD algorithm) described in the embodiment of the present invention reconstructs the value of the visual layer neurons through the value of the neurons of the hidden layer, and adjusts the parameters of the medical resource recommendation model according to the error between the initial value of the visual layer neurons and the reconstructed value of the visual layer neurons.

[0091] Further, see Figure 3 As shown, the S32 includes:

[0092] S321, converting the historical patient information feature set, the historical medical resource feature set and the associated features in the training set into feature vectors, performing mean, variance and whitening processing on the feature vectors, and uniformly normalizing the processed feature vectors to between 0 and 1 in ascending order to obtain a normalized sample set;

[0093] S322, initializing the weight values ​​between the visual layer neurons and the hidden layer neurons in the medical resource recommendation model, as well as the bias parameters of the visual layer neurons and the bias parameters of the hidden layer neurons;

[0094] S323, using the normalized sample set, the weight value, the bias parameter of the visual layer neuron and the bias parameter of the hidden layer neuron, iterate the visual layer neurons and the hidden layer neurons in a loop, and calculate the activation probability of the hidden layer neurons;

[0095] In one embodiment of the present invention, the activation probability of the hidden layer neurons can be calculated by the following formula:

[0096]

[0097] Among them, f() represents the sigmoid function, represents the value of the jth neuron in the hidden layer, Represents the value of the visual layer neuron, θ=( , represents the bias parameter of the jth neuron in the hidden layer, represents the weight value connecting the jth hidden layer neuron and the ith visual layer neuron, represents the bias parameter of the i-th neuron in the visual layer, Represents the value of the i-th neuron in the visual layer.

[0098] S324, using the hidden layer neurons after the loop iteration, reversely loop iterate the visual layer neurons, and calculate the activation probability of the visual layer neurons;

[0099] In one embodiment of the present invention, the activation probability of the visual layer neuron can be calculated by the following formula:

[0100]

[0101] Where f() represents the sigmoid function, represents the value of the i-th neuron in the visual layer, represents the value of the hidden layer neuron, θ=( , represents the bias parameter of the jth neuron in the hidden layer, represents the weight value connecting the hidden layer neuron j and the visual layer neuron i, represents the bias parameter of the i-th neuron in the visual layer, Represents the value of the jth neuron in the visual layer.

[0102] S325. According to the activation probability of the hidden layer neurons and the activation probability of the visual layer neurons, the weight values, the bias parameters of the visual layer neurons and the bias parameters of the hidden layer neurons are adjusted using a contrast divergence algorithm, and the activation probability of the visual layer neurons is recalculated using the adjusted weight values ​​and the bias parameters, and the output result of the medical resource recommendation model corresponding to the maximum value of the activation probability of the visual layer neurons is used as the predicted recommended medical resource;

[0103] S326, using a loss function to calculate the loss value between the predicted recommended medical resources and the historical patient visit data, and determining whether the loss value is less than a preset loss threshold;

[0104] When the loss value is greater than or equal to the preset loss threshold, executing S327, adjusting the parameters of the medical resource recommendation model, and returning to the step corresponding to S322;

[0105] When the loss value is less than a preset loss threshold, S328 is executed to obtain and output a trained medical resource recommendation model.

[0106] Furthermore, the step of adjusting the weight value, the bias parameter of the visible layer neuron and the bias parameter of the hidden layer neuron by using the contrast divergence algorithm in S325 includes:

[0107] Step A, using the normalized sample set to initialize the state vectors of the visual layer neurons, the hidden layer neurons, and the weight values ​​between the visual layer neurons and the hidden layer neurons;

[0108] Step B, using the activation probability of the hidden layer neurons and the activation probability of the visual layer neurons, perform K-step Gibbs sampling to obtain the state vector of the visual layer neurons and the state vector of the hidden layer neurons corresponding to time t-1, and the state vector of the visual layer neurons and the state vector of the hidden layer neurons corresponding to time t;

[0109] The Gibbs sampling described in the embodiment of the present invention is an algorithm used in Markov Monte Carlo (MCMC) in statistics, and is used to approximately extract a sample sequence from a multivariate probability distribution when direct sampling is difficult.

[0110] Step C, using the state vector of the visual layer neuron and the state vector of the hidden layer neuron corresponding to the time t-1, and the state vector of the visual layer neuron and the state vector of the hidden layer neuron corresponding to the time t, to perform a cyclic calculation on the normalized sample set to obtain a comparison error value;

[0111] Step D: using the contrast error value to adjust the weight value, the bias parameter of the visible layer neuron and the bias parameter of the hidden layer neuron.

[0112] The embodiment of the present invention uses a medical resource recommendation model obtained by training with a contrast divergence algorithm, so that the update direction of the model parameters depends on the result of the previous iteration, so that the algorithm jumps out of the local optimum, accelerates the model convergence speed, greatly reduces the number of model iterations, greatly improves the recommendation efficiency, and enhances the accuracy of medical resource recommendations.

[0113] S4. Obtain the patient information to be analyzed and the available medical resources, and use the trained medical resource recommendation model to perform matching analysis on the patient information to be analyzed and the available medical resources to obtain recommended medical resource data.

[0114] In the embodiment of the present invention, the available medical resources include hospital data disclosed on official websites of domestic and foreign hospitals and authoritative websites, and doctor data disclosed on official websites of domestic and foreign hospitals and authoritative websites.

[0115] In detail, the S4 includes:

[0116] Obtain patient information to be analyzed and available medical resources uploaded by the user;

[0117] Extracting features of the patient information to be analyzed from the patient information to be analyzed and extracting features of available medical resources from the available medical resources;

[0118] Using the trained medical resource recommendation model, matching and analyzing the patient information characteristics to be analyzed and the available medical resource characteristics to obtain recommended medical resource data;

[0119] The recommended medical resource data is fed back to the user terminal.

[0120] In one implementation of the present invention, the patient to be analyzed can input relevant case data through the label filtering items related to the disease condition set by the user end to obtain the patient information to be analyzed.

[0121] In one implementation of the present invention, the options of hospital data and doctor data in the recommended medical resource data can be multiple, and the current user can filter them again according to his or her actual needs.

[0122] The embodiment of the present invention extracts the historical patient information feature set and the historical medical resource feature set from the historical patient data set and the historical medical resource set respectively; uses the historical patient medical treatment data to extract the correlation features between the historical patient information feature set and the historical medical resource feature set; uses the historical patient information feature set, the historical medical resource feature set and the correlation features to construct a historical patient medical recommendation sample set, which is conducive to improving the matching degree between patients and recommended medical resources, thereby improving the accuracy of recommended medical resources; uses the contrast divergence algorithm to train to obtain a medical resource recommendation model, accelerates the model convergence speed, reduces the number of model iterations, and improves the accuracy of medical resource recommendations. Therefore, the medical resource recommendation method proposed in the present invention can solve the problem of low accuracy of medical resource recommendations.

[0123] like Figure 4 , is a functional module diagram of a medical resource recommendation device provided by an embodiment of the present invention. The medical resource recommendation device 100 of the present invention can be installed in an electronic device. According to the functions implemented, the medical resource recommendation device 100 may include a feature extraction module 101, a sample construction module 102, a model training module 103 and a recommendation module 104. The module of the present invention may also be referred to as a unit, which refers to a series of computer program segments that can be executed by a processor of an electronic device and can complete fixed functions, which are stored in the memory of the electronic device.

[0124] In this embodiment, the functions of each module / unit are as follows:

[0125] The feature extraction module 101 is used to obtain a historical patient data set and a historical medical resource set, extract a historical patient information feature set from the historical patient data set and extract historical patient medical treatment data, and extract a historical medical resource feature set from the historical medical resource set;

[0126] The sample construction module 102 is used to use the historical patient consultation data to extract the correlation features between the historical patient information feature set and the historical medical resource feature set, and to construct a historical patient medical recommendation sample set using the historical patient information feature set, the historical medical resource feature set and the correlation features;

[0127] The model training module 103 is used to train the pre-built medical resource recommendation model based on the historical patient medical recommendation sample set using a contrast divergence algorithm to obtain a trained medical resource recommendation model;

[0128] The recommendation module 104 is used to obtain the patient information to be analyzed and the available medical resources, and use the trained medical resource recommendation model to perform matching analysis on the patient information to be analyzed and the available medical resources to obtain recommended medical resource data.

[0129] In detail, each module described in the medical resource recommendation device 100 in the embodiment of the present invention is used in the same manner as described above. Figures 1 to 3 The technical means are the same as the medical resource recommendation method described in and can produce the same technical effects, so they will not be repeated here.

[0130] like Figure 5 FIG. 1 is a schematic diagram of the structure of an electronic device for implementing a medical resource recommendation method provided by an embodiment of the present invention.

[0131] The electronic device 1 may include a processor 10, a memory 11, a communication bus 12, and a communication interface 13, and may also include a computer program stored in the memory 11 and executable on the processor 10, such as a medical resource recommendation program.

[0132] In some embodiments, the processor 10 may be composed of an integrated circuit, for example, a single packaged integrated circuit, or a plurality of packaged integrated circuits with the same or different functions, including one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and combinations of various control chips. The processor 10 is the control core (ControlUnit) of the electronic device, and uses various interfaces and lines to connect various components of the entire electronic device, and executes various functions of the electronic device and processes data by running or executing programs or modules stored in the memory 11 (for example, executing a medical resource recommendation program, etc.), and calling data stored in the memory 11.

[0133] The memory 11 includes at least one type of readable storage medium, and the readable storage medium includes a flash memory, a mobile hard disk, a multimedia card, a card-type memory (for example, SD or DX memory, etc.), a magnetic memory, a disk, an optical disk, etc. In some embodiments, the memory 11 may be an internal storage unit of an electronic device, such as a mobile hard disk of the electronic device. In other embodiments, the memory 11 may also be an external storage device of an electronic device, such as a plug-in mobile hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the electronic device. Further, the memory 11 may also include both an internal storage unit of the electronic device and an external storage device. The memory 11 can not only be used to store application software and various types of data installed in the electronic device, such as the code of the medical resource recommendation program, but also can be used to temporarily store data that has been output or is to be output.

[0134] The communication bus 12 may be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. The bus is configured to realize connection and communication between the memory 11 and at least one processor 10, etc.

[0135] The communication interface 13 is used for communication between the above-mentioned electronic device and other devices, including a network interface and a user interface. Optionally, the network interface may include a wired interface and / or a wireless interface (such as a WI-FI interface, a Bluetooth interface, etc.), which is generally used to establish a communication connection between the electronic device and other electronic devices. The user interface may be a display (Display), an input unit (such as a keyboard (Keyboard)), and optionally, the user interface may also be a standard wired interface, a wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, and an OLED (Organic Light-Emitting Diode, organic light-emitting diode) touch device, etc. Among them, the display may also be appropriately referred to as a display screen or a display unit, which is used to display information processed in the electronic device and to display a visual user interface.

[0136] Figure 5 Only an electronic device with components is shown, and those skilled in the art will understand that Figure 5 The structure shown does not constitute a limitation on the electronic device 1, and may include fewer or more components than shown in the figure, or combine certain components, or arrange the components differently.

[0137] For example, although not shown, the electronic device may also include a power source (such as a battery) for supplying power to various components. Preferably, the power source may be logically connected to the at least one processor 10 through a power management device, so that the power management device can realize functions such as charging management, discharging management, and power consumption management. The power source may also include one or more DC or AC power sources, recharging devices, power failure detection circuits, power converters or inverters, power status indicators, and other arbitrary components. The electronic device may also include a variety of sensors, Bluetooth modules, Wi-Fi modules, etc., which will not be repeated here.

[0138] It should be understood that the embodiment is for illustration only and the scope of the patent application is not limited to this structure.

[0139] The medical resource recommendation program stored in the memory 11 of the electronic device 1 is a combination of multiple instructions, and when running in the processor 10, it can achieve:

[0140] Acquire a historical patient data set and a historical medical resource set, extract a historical patient information feature set and historical patient medical treatment data from the historical patient data set, and extract a historical medical resource feature set from the historical medical resource set;

[0141] Utilizing the historical patient medical treatment data, extracting correlation features between the historical patient information feature set and the historical medical resource feature set, and constructing a historical patient medical recommendation sample set using the historical patient information feature set, the historical medical resource feature set and the correlation features;

[0142] Based on the historical patient medical recommendation sample set, using a contrast divergence algorithm, training the pre-built medical resource recommendation model to obtain a trained medical resource recommendation model;

[0143] Obtain patient information to be analyzed and available medical resources, use the trained medical resource recommendation model to perform matching analysis on the patient information to be analyzed and the available medical resources, and obtain recommended medical resource data.

[0144] Specifically, the specific implementation method of the processor 10 for the above instructions can refer to the description of the relevant steps in the corresponding embodiment of the accompanying drawings, which will not be repeated here.

[0145] Furthermore, if the module / unit integrated in the electronic device 1 is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. The computer-readable storage medium can be volatile or non-volatile. For example, the computer-readable medium can include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a mobile hard disk, a magnetic disk, an optical disk, a computer memory, and a read-only memory (ROM).

[0146] The present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor of an electronic device, the computer program can implement:

[0147] Acquire a historical patient data set and a historical medical resource set, extract a historical patient information feature set and historical patient medical treatment data from the historical patient data set, and extract a historical medical resource feature set from the historical medical resource set;

[0148] Utilizing the historical patient medical treatment data, extracting correlation features between the historical patient information feature set and the historical medical resource feature set, and constructing a historical patient medical recommendation sample set using the historical patient information feature set, the historical medical resource feature set and the correlation features;

[0149] Based on the historical patient medical recommendation sample set, using a contrast divergence algorithm, training the pre-built medical resource recommendation model to obtain a trained medical resource recommendation model;

[0150] Obtain the patient information to be analyzed and the available medical resources, use the trained medical resource recommendation model to match and analyze the patient information to be analyzed and the available medical resources, and obtain recommended medical resource data. In the several embodiments provided by the present invention, it should be understood that the disclosed equipment, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the modules is only a logical function division, and there may be other division methods in actual implementation.

[0151] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0152] In addition, each functional module in each embodiment of the present invention may be integrated into one processing unit, each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional modules.

[0153] It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0154] Therefore, no matter from which point of view, the embodiments should be regarded as illustrative and non-restrictive, and the scope of the present invention is limited by the appended claims rather than the above description, so it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims are included in the present invention. Any attached figure mark in the claims should not be regarded as limiting the claims involved.

[0155] The blockchain referred to in this invention is a new application mode of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanism, encryption algorithm, etc. Blockchain is essentially a decentralized database, a string of data blocks generated by cryptographic methods. Each data block contains a batch of network transaction information, which is used to verify the validity of its information (anti-counterfeiting) and generate the next block. Blockchain can include the underlying blockchain platform, platform product service layer, and application service layer.

[0156] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Artificial Intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0157] In addition, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices stated in the system claim can also be implemented by one unit or device through software or hardware. The words first, second, etc. are used to indicate names, and do not indicate any particular order.

[0158] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A medical resource recommendation method, characterized in that: The method comprises: Acquire a historical patient data set and a historical medical resource set, extract a historical patient information feature set and historical patient medical treatment data from the historical patient data set, and extract a historical medical resource feature set from the historical medical resource set; Utilizing the historical patient medical treatment data, extracting correlation features between the historical patient information feature set and the historical medical resource feature set, and constructing a historical patient medical recommendation sample set using the historical patient information feature set, the historical medical resource feature set and the correlation features; Divide the historical patient medical recommendation sample set into a training set and a test set, convert the features in the training set into feature vectors and perform mean, variance and whitening processing on the feature vectors, normalize the processed feature vectors to obtain a normalized training set, use contrast divergence algorithm to reconstruct the values ​​of the visual layer neurons of the medical resource recommendation model through the values ​​of the neurons of the hidden layer of the pre-constructed medical resource recommendation model, adjust the parameters of the medical resource recommendation model according to the error between the initial value of the visual layer neuron and the reconstructed value of the visual layer neuron, iteratively train the medical resource recommendation model according to the normalized training set to obtain a trained medical resource recommendation model, use the test set to test and adjust the trained medical resource recommendation model to obtain a trained medical resource recommendation model; Obtain patient information to be analyzed and available medical resources, use the trained medical resource recommendation model to perform matching analysis on the patient information to be analyzed and the available medical resources, and obtain recommended medical resource data.

2. The medical resource recommendation method according to claim 1, characterized in that: The extracting of the historical patient information feature set from the historical patient data set includes: Convert the historical patient data set into a text format to obtain a historical patient text data set; Performing word segmentation and part-of-speech tagging on the historical patient text dataset to obtain word segmentation and part-of-speech tagging results; Extracting nouns and noun phrases from the segmented words according to the result of the part-of-speech tagging, and obtaining the feature frequencies of historical patient information according to the nouns and noun phrases, and generating a frequent pattern tree according to the feature frequencies of historical patient information; Identify features in the frequent pattern tree to obtain a candidate historical patient information feature set; The point mutual information value of each feature in the candidate historical patient information feature set is calculated, and historical patient information features whose point mutual information values ​​are less than a preset standard threshold are filtered out from the candidate historical patient information feature set to obtain a historical patient information feature set.

3. The medical resource recommendation method according to claim 1, characterized in that: The extracting of correlation features between the historical patient information feature set and the historical medical resource feature set using the historical patient consultation data includes: Performing word segmentation on each historical patient information feature in the historical patient information feature set and each historical medical resource feature in the historical medical resource feature set, and extracting key words and related words from the word segmentation results; The key words and relational words irrelevant to the historical patient consultation data are deleted to obtain the correlation features between the historical patient information feature set and the historical medical resource feature set.

4. The medical resource recommendation method according to claim 1, characterized in that: The step of adjusting the parameters of the medical resource recommendation model by using a contrast divergence algorithm according to the training set, and iteratively training the medical resource recommendation model to obtain a trained medical resource recommendation model includes: The historical patient information feature set, the historical medical resource feature set and the associated features in the training set are converted into feature vectors, and the feature vectors are processed by mean, variance and whitening, and the processed feature vectors are uniformly normalized to between 0 and 1 in ascending order to obtain a normalized sample set; Initializing the weight values ​​between the visual layer neurons and the hidden layer neurons in the medical resource recommendation model, as well as the bias parameters of the visual layer neurons and the bias parameters of the hidden layer neurons; Using the normalized sample set, the weight value, the bias parameter of the visual layer neuron and the bias parameter of the hidden layer neuron, the visual layer neuron and the hidden layer neuron are iterated cyclically, and the activation probability of the hidden layer neuron is calculated; Using the hidden layer neurons after the loop iteration, reverse loop iteration of the visual layer neurons, and calculating the activation probability of the visual layer neurons; According to the activation probability of the hidden layer neurons and the activation probability of the visual layer neurons, the weight value and the bias parameter are adjusted using a contrast divergence algorithm, and the activation probability of the visual layer neurons is recalculated using the adjusted weight value and the bias parameter, and the output result of the medical resource recommendation model corresponding to the maximum value of the activation probability of the visual layer neurons is used as the predicted recommended medical resource; The loss function is used to calculate the loss value between the predicted recommended medical resources and the historical patient visit data, and the parameters of the medical resource recommendation model are adjusted according to the loss value. The step of initializing the weight values ​​between the visual layer neurons and the hidden layer neurons in the medical resource recommendation model is returned until the loss value is less than the preset loss threshold, thereby obtaining a trained medical resource recommendation model.

5. The medical resource recommendation method as claimed in claim 4, characterized in that: The step of adjusting the weight value and the bias parameter by using a contrast divergence algorithm according to the activation probability of the hidden layer neurons and the activation probability of the visual layer neurons includes: Initializing the state vectors of the visual layer neurons, the hidden layer neurons, and the weight values ​​between the visual layer neurons and the hidden layer neurons using the normalized sample set; Using the activation probability of the hidden layer neurons and the activation probability of the visual layer neurons, K-step Gibbs sampling is performed to obtain the state vector of the visual layer neurons and the state vector of the hidden layer neurons corresponding to time t-1, and the state vector of the visual layer neurons and the state vector of the hidden layer neurons corresponding to time t; Using the state vector of the visual layer neuron and the state vector of the hidden layer neuron corresponding to the time t-1, and the state vector of the visual layer neuron and the state vector of the hidden layer neuron corresponding to the time t, the normalized sample set is cyclically calculated to obtain a comparison error value; The weight value, the bias parameter of the visible layer neuron and the bias parameter of the hidden layer neuron are adjusted using the contrast error value.

6. The medical resource recommendation method according to any one of claims 1 to 5, characterized in that: The obtaining of the patient information to be analyzed and the available medical resources, and using the trained medical resource recommendation model to perform matching analysis on the patient information to be analyzed and the available medical resources to obtain recommended medical resource data, includes: Obtain patient information to be analyzed and available medical resources uploaded by the user; Extracting features of the patient information to be analyzed from the patient information to be analyzed and extracting features of available medical resources from the available medical resources; Using the trained medical resource recommendation model, matching and analyzing the patient information characteristics to be analyzed and the available medical resource characteristics to obtain recommended medical resource data; The recommended medical resource data is fed back to the user terminal.

7. A medical resource recommendation device, characterized in that: The device comprises: A feature extraction module is used to obtain a historical patient data set and a historical medical resource set, extract a historical patient information feature set from the historical patient data set and extract historical patient medical treatment data, and extract a historical medical resource feature set from the historical medical resource set; A sample construction module, used to use the historical patient medical treatment data to extract the correlation features between the historical patient information feature set and the historical medical resource feature set, and to construct a historical patient medical recommendation sample set using the historical patient information feature set, the historical medical resource feature set and the correlation features; A model training module, used for dividing the historical patient medical recommendation sample set into a training set and a test set, converting the features in the training set into feature vectors and performing mean, variance and whitening processing on the feature vectors, normalizing the processed feature vectors to obtain a normalized training set, using a contrast divergence algorithm to reconstruct the values ​​of the visual layer neurons of the medical resource recommendation model through the values ​​of the neurons of the hidden layer of the pre-constructed medical resource recommendation model, adjusting the parameters of the medical resource recommendation model according to the errors between the initial values ​​of the visual layer neurons and the reconstructed values ​​of the visual layer neurons, iteratively training the medical resource recommendation model according to the normalized training set to obtain a trained medical resource recommendation model, and using the test set to test and adjust the trained medical resource recommendation model to obtain a trained medical resource recommendation model; The recommendation module is used to obtain the patient information to be analyzed and the available medical resources, and use the trained medical resource recommendation model to match and analyze the patient information to be analyzed and the available medical resources to obtain recommended medical resource data.

8. An electronic device, characterized in that: The electronic device comprises: at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the medical resource recommendation method as described in any one of claims 1 to 6.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the medical resource recommendation method according to any one of claims 1 to 6 is implemented.

Images