Triage data processing method, device, computer equipment and storage medium

Abstract

The present invention discloses a triage data processing method, apparatus, computer device, and storage medium. The method includes: receiving a triage request and obtaining patient information; obtaining symptom information from the patient information through the maximum word matching method; inputting the symptom information into a combined prediction model, and performing prediction processing on the symptom information through the combined prediction model to obtain a first set of symptom departments; inputting the symptom information into a reinforcement learning triage model to obtain a first triage result containing a first symptom result and a first status result output after performing a first action; when the first status result is a first status, determining the first symptom result in the first triage result as the final symptom result; and the final symptom result is the department where the patient seeks medical treatment. The present invention realizes preprocessing for dimensionality reduction through a probability model and a deep neural network model, and automatic triage based on a reinforcement learning method to determine the department where the patient seeks medical treatment, improving the accuracy of medical treatment and enhancing the patient experience.

Description

Technical Field

[0001] The present invention relates to the field of data processing, and particularly to a triage data processing method, apparatus, computer device, and storage medium. Background Art

[0002] Currently, when patients go to the hospital for medical treatment, they first need to go to the triage desk for manual triage. During this process, patients need to spend a lot of queuing time, and there are high requirements for the depth and breadth of the professional knowledge of the service personnel at the triage desk. If the service personnel mis-triage the patients, they need to re-triage, which greatly wastes the patients' time and seriously affects the patients' experience. Therefore, in the prior art, the process of manual triage for patients takes a long time and it is difficult to assign a reasonable department for treatment, resulting in poor patient experience and low accuracy of medical treatment. Summary of the Invention

[0003] The present invention provides a triage data processing method, apparatus, computer device, and storage medium, which realizes preprocessing for dimensionality reduction through a probability model and a deep neural network model, and automatic triage based on a reinforcement learning method, can quickly and accurately determine the department where the patient needs to seek medical treatment, improves the accuracy of medical treatment, and improves the patient experience.

[0004] A triage data processing method includes:

[0005] Receiving a triage request and obtaining patient information;

[0006] Obtaining symptom information from the patient information through the maximum word matching method;

[0007] Inputting the symptom information into a combined prediction model, and performing prediction processing on the symptom information through the combined prediction model to obtain a first symptom department set output by the combined prediction model;

[0008] Inputting the symptom information into a reinforcement learning triage model to obtain a first triage result output after the reinforcement learning triage model performs a first action; wherein, the first action is selected from a first action space set after the reinforcement learning triage model analyzes and processes the input symptom information, and the first action space set is output after a preset total action space set is activated by the first symptom department set; the first triage result includes a first symptom result and a first state result;

[0009] When the first state result is a first state, determining the first symptom result in the first triage result as the final symptom result; the final symptom result is the department where the patient seeks medical treatment.

[0010] A triage data processing apparatus includes:

[0011] A receiving module, configured to receive a triage request and obtain patient information;

[0012] An obtaining module, configured to obtain symptom information from the patient information by using the maximum word matching method;

[0013] A predicting module, configured to input the symptom information into a combined prediction model, perform prediction processing on the symptom information through the combined prediction model, and obtain a first symptom department set output by the combined prediction model;

[0014] An activation module, configured to input the symptom information into a reinforcement learning triage model, and obtain a first triage result output after the reinforcement learning triage model performs a first action; wherein, the first action is selected from a first action space set after the preset total action space set is activated by the first symptom department set; the first triage result includes a first symptom result and a first status result;

[0015] An output module, configured to, when the first status result is a first status, determine the first symptom result in the first triage result as the final symptom result; the final symptom result is the department where the patient seeks medical treatment.

[0016] A computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps of the above triage data processing method are implemented.

[0017] A computer-readable storage medium, storing a computer program, wherein when the computer program is executed by a processor, the steps of the above triage data processing method are implemented.

[0018] The triage data processing method, device, computer device, and storage medium provided by the present invention obtain accurate symptom information from the patient information through the maximum word matching method, input the symptom information into a combined prediction model composed of a trained symptom prediction probability model and a trained department deep convolutional neural network model, obtain the first symptom department set, activate the total action space set through the first symptom department set to output the first action space set, realize dimensionality reduction processing on the total action space set as the first action space set in the reinforcement learning triage model, and then input the symptom information into the reinforcement learning triage model to obtain the first triage result output after executing the first action. If the first status result in the triage result is the first status, the first symptom result in the triage result is determined as the final symptom result. It realizes preprocessing of dimensionality reduction through a probability model and a deep neural network model, and automatic triage based on the reinforcement learning method, can quickly and accurately determine the department where the patient needs to seek medical treatment, saves the patient's time, improves the accuracy of medical treatment, and enhances the patient experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments of the present invention will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art can also obtain other drawings based on these drawings without creative efforts.

[0020] Figure 1 It is a schematic diagram of the application environment of the triage data processing method in an embodiment of the present invention;

[0021] Figure 2 It is a flowchart of the triage data processing method in an embodiment of the present invention;

[0022] Figure 3 It is a flowchart of the triage data processing method in another embodiment of the present invention;

[0023] Figure 4 It is a flowchart of step S10 of the triage data processing method in an embodiment of the present invention;

[0024] Figure 5 It is a flowchart of step S20 of the triage data processing method in an embodiment of the present invention;

[0025] Figure 6 It is a flowchart of step S30 of the triage data processing method in an embodiment of the present invention;

[0026] Figure 7It is a flowchart of step S301 of the triage data processing method in an embodiment of the present invention;

[0027] Figure 8 It is a flowchart of step S302 of the triage data processing method in an embodiment of the present invention;

[0028] Figure 9 It is a schematic block diagram of the triage data processing device in an embodiment of the present invention;

[0029] Figure 10 It is a schematic diagram of a computer device in an embodiment of the present invention. Detailed implementation manners

[0030] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0031] The triage data processing method provided by the present invention can be applied in an application environment such as Figure 1 , where the client (computer device) communicates with the server through a network. Among them, the client (computer device) includes, but is not limited to, various personal computers, laptop computers, smart phones, tablet computers, cameras, and portable wearable devices. The server can be implemented by an independent server or a server cluster composed of multiple servers.

[0032] In an embodiment, as Figure 2 shown, a triage data processing method is provided, and its technical solution mainly includes the following steps S10 - S50:

[0033] S10, receiving a triage request and obtaining patient information.

[0034] Understandably, after receiving the triage request, the patient information is obtained. The triage request is a request triggered after selecting and confirming the patient information that needs to be triaged, and the triggering method can be set according to requirements. For example, a triggering button that can be triggered by clicking, swiping, etc. is provided on the application program platform interface, or it is automatically triggered after executing a preset program, etc.

[0035] Among them, the patient information is information input by the patient related to the patient's symptoms.

[0036] In an embodiment, as Figure 4 shown, before step S10, that is, before receiving the triage request and obtaining patient information, it includes:

[0037] S101. Receive the patient input instruction and obtain the patient input information.

[0038] Understandably, when receiving the patient input instruction and obtaining the patient input information, the patient input instruction is an instruction triggered after the patient input information is entered on the display interface of the application. After receiving the patient input instruction, the patient input information is obtained, and the obtaining method can be set as needed. For example, the obtaining method can be to obtain the patient input information through the patient input instruction, obtain the patient input information according to the storage path of the patient input information included in the patient input instruction, and so on.

[0039] S102. Input the patient input information into a preset preprocessing model, and the preprocessing model identifies the patient input information to obtain an identification result; where the identification result includes text, voice, and image.

[0040] Understandably, the preprocessing model is a preset model for identifying the patient input information. Input the patient input information into the preprocessing model, and the preprocessing model can determine the identification result according to the format of the patient input information. The identification result includes text, voice, and image.

[0041] S103. Obtain a conversion model corresponding to the identification result.

[0042] Understandably, determine a conversion model corresponding to the identification result according to the identification result. The conversion model includes a text conversion model, a voice conversion model, and an image conversion model. That is, if the identification result is text, then obtain the text conversion model; if the identification result is voice, then obtain the voice conversion model; if the identification result is image, then obtain the image conversion model. The conversion model is a trained neural network model. In this way, obtaining a more targeted conversion model can improve the conversion efficiency and accuracy.

[0043] S104. Input the patient input information into the conversion model, and the conversion model performs text conversion on the patient input information and outputs a conversion result.

[0044] Understandably, input the patient input information into the conversion model corresponding to the identification result, and convert it into the conversion result through the conversion model.

[0045] S105. Determine the conversion result as the patient information.

[0046] Thus, by recognizing the patient input information input by the patient in terms of text, voice, and image, corresponding different conversion models according to different recognition results, and obtaining the patient information from the patient input information, multiple input channels are provided for the patient, enhancing the patient experience.

[0047] S20. Obtain symptom information from the patient information by the maximum word matching method.

[0048] Understandably, by the maximum word matching method, the patient information is split into multiple single texts, and the multiple single texts are maximized, that is, the single text is combined with the corresponding preceding and following single texts to generate a pre-text, a post-text, and a full-text, and the text with the highest matching value between the single text, the pre-text, the post-text, the full-text and the texts in the preset symptom thesaurus is obtained, and the text with the highest matching value is determined as the maximum word group corresponding to the single text, and all the maximum word groups are cleared, that is, the maximum word groups with a matching value of zero are removed, and all the cleared maximum word groups are determined as the symptom information.

[0049] In one embodiment, as Figure 5 shown, in the step S20, that is, obtaining symptom information from the patient information by the maximum word matching method includes:

[0050] S201. Obtain a preset symptom thesaurus; the symptom thesaurus includes multiple symptom words.

[0051] Understandably, the symptom thesaurus is a warehouse for storing all symptom words, that is, the symptom thesaurus contains multiple symptom words, the symptom thesaurus can add symptom words according to needs, and the symptom thesaurus can be added over time. The symptom word is a word given to name a symptom. For example, the symptom thesaurus can screen out symptom words of common symptoms from the Symcat public data.

[0052] S202. Split the patient information into multiple single texts.

[0053] Understandably, the single text can be a single character or a single word group. For example, if the patient information is "coughing continuously from yesterday to today", it is split into "yesterday", "to", "today", "continuously", "coughing".

[0054] S203. Obtain the start position and end position of the single text. Combine the single text before the start position with the single text to generate a pre-text. Combine the single text after the end position with the single text to generate a post-text. Combine the single text before the start position, the single text, and the single text after the end position to generate a full-text.

[0055] Understandably, the start position can be set according to requirements. For example, the start position can be the position reached after counting a certain number of digits from the first digit of the single text from left to right in the patient information, and this number of digits is the start position of the single text. The end position can also be set according to requirements. For example, the end position can be obtained by adding the number of digits of the single text after the start position. Combine the single text before the start position with the single text to generate a pre-text. Combine the single text after the end position with the single text to generate a post-text. Combine the single text before the start position, the single text, and the single text after the end position to generate a full-text. For example, if the patient information is "upper abdominal pain for three days", and the split single texts are "upper", "abdominal pain", and "three days", taking "abdominal pain" as the single text, the pre-text is "upper abdominal pain", the post-text is "abdominal pain for three days", and the full-text is "upper abdominal pain for three days".

[0056] S204. Obtain the matching values of the single text, the pre-text, the post-text, and the full-text with the texts in the symptom thesaurus, and determine the text with the highest matching value as the largest phrase corresponding to the single text.

[0057] Understandably, the matching value is the number of characters that match exactly. Search and match the single text with the texts in the symptom thesaurus to obtain the matching value corresponding to the single text. Search and match the pre-text with the texts in the symptom thesaurus to obtain the matching value corresponding to the pre-text. Search and match the post-text with the texts in the symptom thesaurus to obtain the matching value corresponding to the post-text. Search and match the full-text with the texts in the symptom thesaurus to obtain the matching value corresponding to the full-text. For example, in the above example, the matching value corresponding to "abdominal pain" is 2, the matching value corresponding to "upper abdominal pain" is 3, the matching value corresponding to "abdominal pain for three days" is 0, and the matching value corresponding to "upper abdominal pain for three days" is 0.

[0058] S205. Clear all the largest phrases corresponding to each single text in the patient information, and determine the cleared largest phrases as the symptom information.

[0059] Understandably, all the maximum phrases with a matching value of zero among all the maximum phrases corresponding to each single text are removed, and all the remaining maximum phrases are determined as the symptom information.

[0060] In this way, the maximum word matching method can ensure the accuracy of symptom acquisition, improving the correctness and accuracy of symptom acquisition.

[0061] S30. Input the symptom information into the combined prediction model, and perform prediction processing on the symptom information through the combined prediction model to obtain the first symptom department set output by the combined prediction model.

[0062] Understandably, the combined prediction model is a model composed of a trained symptom prediction probability model and a trained department deep convolutional neural network model. The symptom prediction probability model is obtained by training the first symptom samples input into the Bayesian probability model. The department deep convolutional neural network model is obtained by training the second symptom samples input into the deep neural network model. The first symptom department set is obtained by splicing and normalizing the prediction probability distribution result output by the symptom prediction probability model and the department prediction distribution result output by the department deep convolutional neural network model. The prediction probability distribution result is a distribution map of the probability values corresponding to all the symptom words. The department prediction distribution result is a distribution map of the probability values corresponding to all departments. The department is the department name set according to requirements, such as surgery, internal medicine, orthopedics, etc. The prediction probability distribution is a set of data, and the department prediction distribution result is also a set of data. The first symptom department set is a set of arrays after processing the prediction probability distribution and the department prediction distribution result.

[0063] In one embodiment, as Figure 6 shown, in step S30, that is, inputting the symptom information into the combined prediction model and performing prediction processing on the symptom information through the combined prediction model to obtain the first symptom department set output by the combined prediction model, includes:

[0064] S301. Input the symptom information into the trained symptom prediction probability model, and perform prediction on the symptom information through the symptom prediction probability model to obtain the prediction probability distribution result output by the symptom prediction probability model; wherein, the prediction probability distribution result represents the matching probability distribution of the symptoms related to the symptom information in the symptom set.

[0065] Understandably, the symptom prediction probability model is a trained Bayesian probability model. By inputting the symptom information into the symptom prediction probability model for prior distribution processing, the predicted probability distribution result can be obtained. The predicted probability distribution result is the matching probability distribution of the symptoms related to the symptom information in the symptom set. The predicted probability distribution result is an array composed of percentage values in the range from 0% to 100%. The symptom set is the total set of all symptom words.

[0066] In one embodiment, as Figure 7 shown, before step S301, that is, before inputting the symptom information into the trained symptom prediction probability model, it includes:

[0067] S3011, obtain the first symptom sample; wherein, each of the first symptom samples is associated with a symptom category label.

[0068] Among them, the first symptom sample is the collected symptom information. Each of the first symptom samples is manually determined to be associated with a symptom category label, that is, each of the first symptom samples is labeled. The symptom category label is the symptom word in the symptom set.

[0069] S3012, input the first symptom sample into the Bayesian probability model containing the first initial parameters.

[0070] Understandably, the Bayesian probability model is a model of a neural network structure based on the Bayesian algorithm. The first initial parameters are included in the Bayesian probability model. Among them, the first initial parameters can be set according to requirements, such as setting the first initial parameters to a default preset value, or random parameter values, etc.

[0071] S3013, perform prior distribution processing on the first symptom sample through the Bayesian probability model.

[0072] Understandably, the prior distribution processing is the probability distribution processing in the Bayesian algorithm, and the prior distribution processing is performed on the first symptom sample.

[0073] S3014, obtain the distribution result output by the Bayesian probability model, and determine the first loss value according to the matching degree between the distribution result and the symptom category label.

[0074] Understandably, for the probability distribution after prior distribution processing according to the Bayesian probability model, the probability distribution is a distribution diagram of probability values corresponding to all the symptom words in the symptom set, and the distribution result of the Bayesian probability model is obtained. The distribution result is a distribution diagram of probability values corresponding to all the symptom words. Obtain the symptom words corresponding to all the probability values in the distribution result, and compare the probability values corresponding to all the symptom words with the symptom category labels of the first symptom sample to determine the corresponding loss value, that is, calculate the first loss value through the loss function of the Bayesian probability model. S3015. When the first loss value reaches the preset first convergence condition, record the converged Bayesian probability model as the symptom prediction probability model that has been trained.

[0075] Among them, the preset first convergence condition can be the condition that the loss value is very small and will not decrease after 500 calculations, that is, when the loss value is very small and will not decrease after 500 calculations, stop training and record the converged Bayesian probability model as the symptom prediction probability model that has been trained; the preset first convergence condition can also be the condition that the loss value is less than the set threshold, that is, when the loss value is less than the set threshold, stop training and record the converged Bayesian probability model as the symptom prediction probability model that has been trained.

[0076] In this way, by inputting the first symptom sample to train the Bayesian probability model, the accuracy and reliability of the distribution result can be improved.

[0077] In an embodiment, after step S3014, that is, after obtaining the distribution result output by the Bayesian probability model and determining the loss value according to the matching degree between the distribution result and the symptom category label, it includes:

[0078] S3016. When the first loss value does not reach the preset first convergence condition, iteratively update the first initial parameters of the Bayesian probability model until the first loss value reaches the preset first convergence condition, and record the converged Bayesian probability model as the symptom prediction probability model that has been trained.

[0079] In this way, when the first loss value does not reach the preset first convergence condition, continuously update and iterate the first initial parameters of the Bayesian probability model, which can continuously approach the accurate distribution result and make the accuracy of the distribution result higher and higher.

[0080] S302. Input the symptom information into the trained department deep convolutional neural network model, extract the text features of the symptom information through the department deep convolutional neural network model, and obtain the department prediction distribution result output by the department deep convolutional neural network model. Among them, the department prediction distribution result represents the matching probability distribution of the departments related to the symptom information in the department set.

[0081] Understandably, the department deep convolutional neural network model is a trained deep neural network model. By inputting the symptom information into the deep convolutional neural network model for text feature extraction, the department prediction distribution result can be obtained. The department prediction distribution result is the matching probability distribution of the departments related to the symptom information in the department set. The department set is the total set of all hospital departments, and the department prediction distribution result is an array composed of a set of percentage values in the range of 0% to 100%.

[0082] In one embodiment, as Figure 8 shown, before step S302, that is, before inputting the symptom information into the trained department deep convolutional neural network model, it includes:

[0083] S3021. Obtain the second symptom sample. Each of the second symptom samples is associated with a department label.

[0084] Among them, the second symptom sample is the collected symptom information. Each of the second symptom samples is manually determined and associated with a department label, that is, each of the second symptom samples is tagged. The department label is the corresponding name of all the departments.

[0085] S3022. Input the second symptom sample into the deep neural network model containing the second initial parameters.

[0086] Understandably, the deep neural network model is a model based on the neural network structure of multi-classification recognition. The network structure of the deep neural network model can be set according to requirements. For example, the neural network structure can be selected as VGG, GoogLeNet, etc. The deep neural network model contains the second initial parameters. Among them, the second initial parameters can be set according to requirements. For example, set the second initial parameters to a default preset value, or random parameter values, etc.

[0087] S3023. Extract the text features in the symptom sample through the deep neural network model.

[0088] Understandably, the text feature is to convert text data into a feature vector. A relatively common method for extracting text features is the bag-of-words method. The feature vector corresponding to the text is obtained according to the frequency of occurrence.

[0089] S3024. Obtain the recognition result output by the deep neural network model according to the text feature, and determine the second loss value according to the matching degree between the recognition result and the department label.

[0090] Understandably, according to the text feature extracted by the deep neural network model, the recognition result of the deep neural network model is obtained. The recognition result is a distribution map of probability values corresponding to all the departments. Obtain all the probability values in the recognition result, and compare all the probability values corresponding to the departments with the department label of the second symptom sample to determine the corresponding loss value, that is, calculate the second loss value through the loss function of the deep neural network model.

[0091] S3025. When the second loss value reaches the preset second convergence condition, record the converged deep neural network model as the department deep convolutional neural network model that has been trained.

[0092] Among them, the preset second convergence condition can be the condition that the second loss value is very small and will not decrease after 5000 calculations, that is, when the second loss value is very small and will not decrease after 5000 calculations, stop training, and record the converged deep neural network model as the department deep convolutional neural network model that has been trained; the preset second convergence condition can also be the condition that the second loss value is less than the set threshold, that is, when the second loss value is less than the set threshold, stop training, and record the converged deep neural network model as the department deep convolutional neural network model that has been trained.

[0093] In this way, by according to the document type label value in the training image sample and training the initial neural network model, it is possible to identify by extracting texture features and improve the accuracy and reliability of the recognition result.

[0094] In one embodiment, after the step S3024, it includes:

[0095] S3026. When the second loss value does not reach the preset second convergence condition, iteratively update the second initial parameters of the deep neural network model until the second loss value reaches the preset second convergence condition, and record the converged deep neural network model as the department deep convolutional neural network model that has been trained.

[0096] Thus, when the second loss value does not reach the preset second convergence condition, the second initial parameters of the deep neural network model can be continuously updated and iterated, which can continuously approach the accurate recognition result and make the accuracy of the recognition result higher and higher.

[0097] S303. Concatenate and normalize the predicted probability distribution result and the department prediction distribution result to obtain the first symptom department set.

[0098] Understandably, concatenating the predicted probability distribution result and the department prediction distribution result means concatenating the array corresponding to the predicted probability distribution result and the array corresponding to the department prediction distribution result to obtain a concatenated array with the same dimension as the total action space. Normalize the concatenated array, that is, normalize the percentage values greater than or equal to the preset percentage threshold in the concatenated array to 1, and normalize the percentage values less than the preset percentage threshold in the concatenated array to 0. The normalized concatenated array is determined as the first symptom department set.

[0099] Thus, through the symptom prediction probability model, the predicted probability distribution result of the matching probability distribution of the symptoms related to the symptom information is obtained by predicting the symptom information, and through the department deep convolutional neural network model, the department prediction distribution result of the matching probability distribution of the departments related to the symptom information is obtained by predicting the symptom information. Then, the predicted probability distribution result and the department prediction distribution result are concatenated and normalized to obtain the first symptom department set, realizing reasonable prediction of the symptom information and providing an activation factor for the total action space.

[0100] S40. Input the symptom information into the reinforcement learning triage model to obtain the first triage result output after the reinforcement learning triage model executes the first action; wherein, the first action is selected from the first action space set after the preset total action space is activated by the first symptom department set after the reinforcement learning triage model analyzes and processes the input symptom information; the first triage result includes a first symptom result and a first status result.

[0101] Understandably, the reinforcement learning triage model can set the learning method according to requirements. Preferably, the reinforcement learning triage model is set to the DQN learning method. The symptom information is used as the Agent of the reinforcement learning triage model. The reinforcement learning triage model guides the input symptom information and selects a first action from the first action space set. The first action is the action to be performed on the symptom information. The total action space is a set containing all actions, and the total action space is an array. Among them, the array dimension of the total action space is the same as that of the first symptom department set. After the total action space is activated by the first symptom department set, the first action space set is output. The activation process is obtained by matching the total action space and the first symptom department set according to the activation principle. The activation principle is to match, retain or delete the array contents corresponding one by one in the two arrays, that is, the array contents of the total action space are judged to be retained or deleted according to the corresponding values in the array of the first symptom department set. If the value of the first symptom department set is 1, the array contents in the total action space corresponding to the value of the symptom department set are retained. If the value of the first symptom department set is 0, the array contents in the total action space corresponding to the value of the symptom department set are deleted. The finally obtained first action space set is the preset total action space set activated by the first symptom department set. The first action space set provides a set of all guided actions. The reinforcement learning triage model outputs the first triage result after performing the first action on the symptom information.

[0102] In this way, by activating the total action space with the first symptom department set and outputting the first action space set, the dimensionality reduction process of the total action space is realized and used as the first action space set in the reinforcement learning triage model, which can improve the learning effect of the reinforcement learning triage model and the accuracy of the reinforcement learning triage model.

[0103] S50. When the first state result is the first state, the first symptom result in the first triage result is determined as the final symptom result; the final symptom result is the department where the patient seeks medical treatment.

[0104] Understandably, the first state can be set to the termination state, that is, the final state is reached after interacting with the reinforcement learning triage model. When the first state result is the first state, the first symptom result is determined as the final symptom result, and the symptom result is the department where the patient needs to seek medical treatment.

[0105] The present invention obtains accurate symptom information from the patient information through the maximum word matching method, inputs the symptom information into the combined prediction model (a model composed of a trained symptom prediction probability model and a trained department depth convolutional neural network model), obtains the first symptom department set, activates the total action space set through the first symptom department set to output the first action space set, realizes dimensionality reduction processing of the total action space set as the first action space set in the reinforcement learning triage model (which can improve the learning effect of the reinforcement learning triage model and improve the accuracy of the reinforcement learning triage model), and then inputs the symptom information into the reinforcement learning triage model to obtain the first triage result output after executing the first action (selected from the first action space set after analyzing and processing the symptom information). If the first state result in the triage result is the first state (termination state), the first symptom result in the triage result is determined as the final symptom result (the department where the patient seeks medical treatment).

[0106] In this way, the present invention realizes preprocessing of dimensionality reduction through a probability model and a depth neural network model, and automatic triage based on the reinforcement learning method, can quickly and accurately determine the department where the patient needs to seek medical treatment, saves the patient's time, improves the accuracy of medical treatment, and improves the patient experience.

[0107] In one embodiment, the first triage result further includes a first reward result; as Figure 3 shown, after step S40, that is, after obtaining the first triage result output by the reinforcement learning triage model executing the first action, it includes:

[0108] S60, when the first state result is the second state, use the first symptom result in the first triage result as the next symptom information, and at the same time associate the first reward result with the next symptom information.

[0109] Understandably, the second state can be a non-termination state, that is, the final state has not been reached after interacting with the reinforcement learning triage model. At this time, the first symptom result needs to be used as the input for the next interaction with the reinforcement learning triage model, that is, the first symptom result is used as the next symptom information. The first triage result further includes the first reward result. The first reward result is the reward value given by the reinforcement learning triage model to the first action after executing the first action. The reward value can be the value given each time, or can be the value continuously accumulated according to the number of interaction dialogs. Through the reward value, the reinforcement learning triage model can be provided to approach the accurate department for medical treatment (the direction of maximizing the reward value), and the first reward result is associated with the next symptom information.

[0110] S70. Input the next symptom information into the combined prediction model, perform prediction processing on the symptom information through the combined prediction model, and obtain a second set of symptom departments output by the combined prediction model.

[0111] Understandably, by inputting the next symptom information into the combined prediction model, a second set of symptom departments is obtained. The second set of symptom departments is also an array, and the second set of symptom departments has the same dimension as the first set of symptom departments.

[0112] S80. Input the next symptom information and the first reward result associated with the next symptom information into the reinforcement learning triage model, and obtain a second triage result output after the reinforcement learning triage model performs a second action; wherein, the second action is selected from a second action space set by the reinforcement learning triage model after analyzing and processing the input first symptom information and the first reward result. The second action space set is output after the total action space is activated by the second set of symptom departments; the second triage result includes a second symptom result and a second status result.

[0113] Understandably, the reinforcement learning triage model analyzes the input next symptom information and the first reward result associated with the next symptom information, thereby guiding the next symptom information, and selects a second action from the second action space set. The second action is an action to be performed on the symptom information. Among them, the dimension of the total action space is the same as that of the array of the second set of symptom departments. The total action space is activated and processed by the second set of symptom departments to output the second action space set. The second action space set provides a set of all guiding actions. The reinforcement learning triage model outputs the second triage result after performing the second action on the symptom information.

[0114] S90. When the second status result is the first status, determine the second symptom result in the second triage result as the final symptom result; the final symptom result is the department where the patient seeks medical treatment.

[0115] In this way, by approaching the direction of learning and guiding the optimal strategy from the interactive dialogue through the reinforcement learning triage model, when in the second state (non-terminating state), the reinforcement learning triage model continuously learns and has an interactive dialogue until in the first state (terminating state), thereby outputting the final symptom result and improving the accuracy of medical treatment.

[0116] In one embodiment, a triage data processing device is provided, and the triage data processing device corresponds one-to-one to the triage data processing method in the above embodiment. AsFigure 9 As shown in the figure, the triage data processing device includes a receiving module 11, an obtaining module 12, a prediction module 13, an activation module 14, and an output module 15.

[0117] The detailed description of each functional module is as follows:

[0118] The receiving module 11 is used to receive a triage request and obtain patient information;

[0119] The obtaining module 12 is used to obtain symptom information from the patient information by the maximum word matching method;

[0120] The prediction module 13 is used to input the symptom information into a combined prediction model, perform prediction processing on the symptom information through the combined prediction model, and obtain a first symptom department set output by the combined prediction model;

[0121] The activation module 14 is used to input the symptom information into a reinforcement learning triage model, and obtain a first triage result output after the reinforcement learning triage model executes a first action; wherein, the first action is selected from the first action space set after the preset total action space set is activated by the first symptom department set; the first triage result includes a first symptom result and a first state result;

[0122] The output module 15 is used to, when the first state result is the first state, determine the first symptom result in the first triage result as the final symptom result; the final symptom result is the department where the patient seeks medical treatment.

[0123] In an embodiment, the activation module 14 includes:

[0124] The reward unit is used to, when the first state result is the second state, use the first symptom result in the first triage result as the next symptom information, and at the same time associate the first reward result with the next symptom information;

[0125] The input unit is used to input the next symptom information into the combined prediction model, perform prediction processing on the symptom information through the combined prediction model, and obtain a second symptom department set output by the combined prediction model;

[0126] An activation unit, configured to input the next symptom information and the first reward result associated with the next symptom information into the reinforcement learning triage model, and obtain a second triage result output after the reinforcement learning triage model performs a second action; wherein, the second action is selected from a second action space set by the reinforcement learning triage model after analyzing and processing the input first symptom information and the first reward result, and the second action space set is output after the total action space set is activated by the second symptom department set; the second triage result includes a second symptom result and a second status result;

[0127] A determination unit, configured to, when the second status result is the first status, determine the second symptom result in the second triage result as the final symptom result; the final symptom result is the department where the patient seeks medical treatment.

[0128] In one embodiment, the receiving module 11 includes:

[0129] A receiving unit, configured to receive a patient input instruction and obtain patient input information;

[0130] An identification unit, configured to input the patient input information into a preset preprocessing model, and the preprocessing model identifies the patient input information to obtain an identification result; wherein the identification result includes text, voice, and image;

[0131] A selection unit, configured to obtain a conversion model corresponding to the identification result;

[0132] A conversion unit, configured to input the patient input information into the conversion model, and the conversion model performs text conversion on the patient input information and outputs a conversion result;

[0133] A result output unit, configured to determine the conversion result as the patient information.

[0134] In one embodiment, the obtaining module 12 includes:

[0135] A first obtaining unit, configured to obtain a preset symptom thesaurus; the symptom thesaurus includes a plurality of symptom words;

[0136] A splitting unit, configured to split the patient information into a plurality of single texts;

[0137] A second obtaining unit, configured to obtain the start position and the end position of the single text, combine the single text before the start position with the single text to generate a preposition text, combine the single text after the end position with the single text to generate a postposition text, and combine the single text before the start position, the single text, and the single text after the end position to generate a full position text;

[0138] A matching unit, configured to obtain matching values of the single text, the pre-text, the post-text, and the full-text with the text in the symptom thesaurus, and determine the text with the highest matching value as the largest phrase corresponding to the single text;

[0139] A clearing unit, configured to clear all the largest phrases corresponding to each single text in the patient information, and determine all the largest phrases after the clearing process as the symptom information.

[0140] In one embodiment, the prediction module 13 includes:

[0141] A first model unit, configured to input the symptom information into a trained symptom prediction probability model, predict the symptom information through the symptom prediction probability model, and obtain a prediction probability distribution result output by the symptom prediction probability model; wherein, the prediction probability distribution result represents the matching probability distribution of symptoms related to the symptom information in the symptom set;

[0142] A second model unit, configured to input the symptom information into a trained department deep convolutional neural network model, extract text features from the symptom information through the department deep convolutional neural network model, and obtain a department prediction distribution result output by the department deep convolutional neural network model; wherein, the department prediction distribution result represents the matching probability distribution of departments related to the symptom information in the department set;

[0143] A splicing unit, configured to splice and normalize the prediction probability distribution result and the department prediction distribution result to obtain the first symptom department set.

[0144] In one embodiment, the first model unit includes:

[0145] A first obtaining subunit, configured to obtain first symptom samples; wherein each of the first symptom samples is associated with a symptom category label;

[0146] A first input subunit, configured to input the first symptom samples into a Bayesian probability model including first initial parameters;

[0147] A first processing subunit, configured to perform prior distribution processing on the first symptom samples through the Bayesian probability model;

[0148] A first output subunit, configured to obtain the distribution result output by the Bayesian probability model, and determine a first loss value according to the matching degree between the distribution result and the symptom category label;

[0149] The first convergence subunit is configured to record the converged Bayesian probability model as a symptom prediction probability model that has been trained when the first loss value reaches a preset first convergence condition.

[0150] In one embodiment, the second model unit includes:

[0151] A second acquisition subunit, configured to acquire second symptom samples; wherein each of the second symptom samples is associated with a department label;

[0152] A second input subunit, configured to input the second symptom samples into a deep neural network model including second initial parameters;

[0153] A second processing subunit, configured to extract text features in the symptom samples through the deep neural network model;

[0154] A second output subunit, configured to obtain an identification result output by the deep neural network model according to the text features, and determine a second loss value according to the matching degree between the identification result and the department label;

[0155] A second convergence subunit, configured to record the converged deep neural network model as a trained department deep convolutional neural network model when the second loss value reaches a preset second convergence condition.

[0156] For the specific limitations of the triage data processing device, reference may be made to the limitations of the triage data processing method in the foregoing text, which will not be elaborated here. Each module in the foregoing triage data processing device may be implemented in whole or in part by software, hardware, and their combination. The foregoing modules may be embedded in the processor of the computer device in hardware form or independent thereof, or may be stored in the memory of the computer device in software form, so as to facilitate the processor to call and execute the operations corresponding to the foregoing modules.

[0157] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as Figure 10 shown. The computer device includes a processor, a memory, a network interface, and a database connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a triage data processing method is implemented.

[0158] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the method for processing triage data in the above embodiment is implemented.

[0159] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the method for processing triage data in the above embodiment is implemented.

[0160] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, storage, database, or other medium used in the various embodiments provided by the present invention can include non-volatile and / or volatile memories. Non-volatile memories can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memories can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and Rambus dynamic RAM (RDRAM), etc.

[0161] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the above division of each functional unit and module is used as an example. In actual applications, the above functions can be allocated to different functional units and modules according to needs, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above.

[0162] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A triage data processing method, characterized in that, it includes: Receiving a triage request and obtaining patient information; Obtaining symptom information from the patient information by the maximum word matching method; Inputting the symptom information into a combined prediction model, and performing prediction processing on the symptom information through the combined prediction model to obtain a first symptom department set output by the combined prediction model; Inputting the symptom information into a reinforcement learning triage model to obtain a first triage result output after the reinforcement learning triage model executes a first action; wherein, the first action is selected from a first action space set after the reinforcement learning triage model analyzes and processes the input symptom information, and the first action space set is output after a preset total action space set is activated by the first symptom department set; the first triage result includes a first symptom result and a first status result; When the first status result is the first status, determining the first symptom result in the first triage result as the final symptom result; the final symptom result is the department where the patient seeks medical treatment; The step of inputting the symptom information into a combined prediction model, and performing prediction processing on the symptom information through the combined prediction model to obtain a first symptom department set output by the combined prediction model includes: Inputting the symptom information into a trained symptom prediction probability model, and performing prediction on the symptom information through the symptom prediction probability model to obtain a prediction probability distribution result output by the symptom prediction probability model; wherein, the prediction probability distribution result represents the matching probability distribution of symptoms related to the symptom information in a symptom set; Inputting the symptom information into a trained department deep convolutional neural network model, and extracting text features from the symptom information through the department deep convolutional neural network model to obtain a department prediction distribution result output by the department deep convolutional neural network model; wherein, the department prediction distribution result represents the matching probability distribution of departments related to the symptom information in a department set; Performing splicing and normalization processing on the prediction probability distribution result and the department prediction distribution result to obtain the first symptom department set.

2. The triage data processing method according to claim 1, characterized in that, the first triage result further includes a first reward result; After obtaining the first triage result output after the reinforcement learning triage model executes the first action, it includes: When the first status result is the second status, using the first symptom result in the first triage result as the next symptom information, and associating the first reward result with the next symptom information; Inputting the next symptom information into the combined prediction model, and performing prediction processing on the symptom information through the combined prediction model to obtain a second symptom department set output by the combined prediction model; Input the next symptom information and the first reward result associated with the next symptom information into the reinforcement learning triage model to obtain a second triage result output after the reinforcement learning triage model performs a second action; wherein, the second action is selected from the second action space set by the reinforcement learning triage model after analyzing and processing the input next symptom information and the first reward result, and the second action space set is output after the total action space is activated by the second symptom department set; the second triage result includes a second symptom result and a second status result; When the second status result is the first status, determine the second symptom result in the second triage result as the final symptom result; the final symptom result is the department where the patient seeks medical treatment.

3. The triage data processing method according to claim 1, characterized in that, before receiving the triage request and obtaining patient information, it includes: Receiving a patient input instruction and obtaining patient input information; Input the patient input information into a preset preprocessing model, and the preprocessing model identifies the patient input information to obtain an identification result; wherein the identification result includes text, voice, and image; Obtain a conversion model corresponding to the identification result; Input the patient input information into the conversion model, and the conversion model converts the patient input information into text and outputs a conversion result; Determine the conversion result as the patient information.

4. The triage data processing method according to claim 1, characterized in that, obtaining symptom information from the patient information by the maximum word matching method includes: Obtain a preset symptom word library; the symptom word library includes multiple symptom words; Split the patient information into multiple single texts; Obtain the start position and end position of the single text, combine the single text before the start position with the single text to generate a pre-text, combine the single text after the end position with the single text to generate a post-text, and combine the single text before the start position, the single text, and the single text after the end position to generate a full-text; Obtain the matching values of the single text, the pre-text, the post-text, and the full-text with the texts in the symptom word library, and determine the text with the highest matching value as the maximum phrase corresponding to the single text; Clear all the maximum phrases corresponding to each single text in the patient information, and determine all the maximum phrases after the clearing process as the symptom information.

5. The triage data processing method according to claim 1, characterized in that, before inputting the symptom information into the trained symptom prediction probability model, it includes: Obtain a first symptom sample; wherein each first symptom sample is associated with a symptom category label; Input the first symptom sample into a Bayesian probability model containing first initial parameters; Perform a prior distribution process on the first symptom sample through the Bayesian probability model; Obtain the distribution result output by the Bayesian probability model, and determine the first loss value according to the matching degree between the distribution result and the symptom category label; When the first loss value reaches the preset first convergence condition, record the converged Bayesian probability model as the symptom prediction probability model that has been trained.

6. The triage data processing method according to claim 1, characterized in that, before inputting the symptom information into the trained department deep convolutional neural network model, it includes: Obtain a second symptom sample; wherein, each of the second symptom samples is associated with a department label; Input the second symptom sample into a deep neural network model containing second initial parameters; Extract the text features in the symptom sample through the deep neural network model; Obtain the recognition result output by the deep neural network model according to the text features, and determine the second loss value according to the matching degree between the recognition result and the department label; When the second loss value reaches the preset second convergence condition, record the converged deep neural network model as the trained department deep convolutional neural network model.

7. A triage data processing device, characterized in that, it includes: A receiving module, configured to receive a triage request and obtain patient information; An obtaining module, configured to obtain symptom information from the patient information by using the maximum word matching method; A prediction module, configured to input the symptom information into a combined prediction model, perform prediction processing on the symptom information through the combined prediction model, and obtain a first symptom department set output by the combined prediction model; An activation module, configured to input the symptom information into a reinforcement learning triage model, and obtain a first triage result output after the reinforcement learning triage model executes a first action; wherein, the first action is selected from the first action space set after the reinforcement learning triage model analyzes the input symptom information, and the first action space set is output after being activated by the first symptom department set from the preset total action space set; the first triage result includes a first symptom result and a first state result; An output module, configured to, when the first state result is the first state, determine the first symptom result in the first triage result as the final symptom result; the final symptom result is the department where the patient seeks medical treatment; The prediction module includes: A first model unit, configured to input the symptom information into the trained symptom prediction probability model, perform prediction on the symptom information through the symptom prediction probability model, and obtain a predicted probability distribution result output by the symptom prediction probability model; wherein, the predicted probability distribution result represents the matching probability distribution of symptoms related to the symptom information in the symptom set; A second model unit, configured to input the symptom information into the trained department deep convolutional neural network model, extract text features from the symptom information through the department deep convolutional neural network model, and obtain a department prediction distribution result output by the department deep convolutional neural network model; wherein, the department prediction distribution result represents the matching probability distribution of departments related to the symptom information in the department set. A splicing unit, configured to splice and normalize the prediction probability distribution result and the department prediction distribution result to obtain the first symptom department set.

8. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. It is characterized in that when the processor executes the computer program, the triage data processing method according to any one of claims 1 to 6 is implemented.

9. A computer-readable storage medium, storing a computer program. It is characterized in that when the computer program is executed by a processor, the triage data processing method according to any one of claims 1 to 6 is implemented.

Images