Artificial intelligence-based server for managing patients by underlying disease using medical records, and method therefor

Abstract

Disclosed are an artificial intelligence-based server for managing patients by underlying disease using medical records, and a method therefor. A patient management server, according to one embodiment of the present invention, comprises: a memory; and a processor for managing information of a patient on the basis of initial data of the patient, wherein the processor: collects the initial data including medical records and performs pre-processing on the initial data, the pre-processing being performed according to pre-processing criteria including text processing and text data vectorization on the initial data; performs training processing by inputting the pre-processed initial data into a pre-trained artificial intelligence model so as to obtain at least one of classification by underlying disease for the patient, precautions for medical treatment, and whether or not an interdepartmental consultation is necessary; and provides training processing results including the at least one of classification by underlying disease for the patient, precautions for medical treatment, and whether or not an interdepartmental consultation is necessary.

Description

Patient management server and method for each underlying disease using artificial intelligence-based medical records

[0001] The present disclosure relates to a method for managing patients by underlying disease. More specifically, the present disclosure relates to a server for managing patients by underlying disease using artificial intelligence-based medical records and a method thereof.

[0002] Before undergoing dental treatment, it is essential to identify the patient's underlying disease and decide on precautions for treatment and whether to consult with other departments during the first visit.

[0003] A first-time dental visit involves a series of preliminary examinations, including a medical history, oral examination, and imaging tests. Furthermore, a diagnosis of dental disease is required, and a treatment plan that takes into account the patient's overall condition must be developed. Therefore, compared to a regular follow-up visit, the procedure is more complex and error-free. This principle also applies to other departments.

[0004] Meanwhile, medical school hospitals have a high proportion of patients with underlying conditions, requiring the development of a system that can assist in the assessment of patients' various underlying conditions during initial consultations and other treatment. Therefore, various methods are being explored to facilitate more efficient patient care.

[0005] The purpose of the embodiment disclosed in the present disclosure is to provide a server and method for managing patients by underlying disease using artificial intelligence-based medical records that can classify patients by underlying disease at the preliminary examination stage by using natural language processing of initial medical records including medical records.

[0006] The problems to be solved by the present disclosure are not limited to the problems mentioned above, and other problems not mentioned will be clearly understood by those skilled in the art from the description below.

[0007] In order to achieve the above-described technical task, according to one aspect of the present disclosure, a patient management server using artificial intelligence-based medical records for each underlying disease comprises: a memory; and a processor for managing patient information based on initial data of the patient; wherein the processor collects the initial data including the medical records, performs preprocessing on the initial data, and performs preprocessing according to preprocessing criteria including text processing and text data vectorization on the initial data; inputs the preprocessed initial data into a pre-trained artificial intelligence model to perform learning processing, thereby obtaining at least one or more of a classification by underlying disease, precautions during treatment, and whether or not collaboration with other departments is required for the patient; and provides a learning processing result including at least one or more of a classification by underlying disease, precautions during treatment, and whether or not collaboration with other departments is required for the patient.

[0008] The above medical records may include a questionnaire including data in at least one format of categorical data, text data, and image file data.

[0009] The above processor can extract the visit address and underlying disease from the questionnaire when collecting the initial data.

[0010] The above medical records may include an initial electronic medical record including data in at least one format among categorical data, text data, and image file data.

[0011] The above processor, when collecting the initial data, can extract at least one of the visit address, underlying disease, patient condition, diagnosis, and treatment plan from the initial electronic medical record.

[0012] The processor may normalize the initial data according to the preprocessing criteria including criteria for at least one of language, numbers, symbols, typos, and abbreviations.

[0013] The processor may, when the initial data includes a first language and a second language, tag only the part of speech of a sentence in the second language and then normalize it by translating only the corresponding word into the first language.

[0014] When the above processor processes the text data vectorization when the initial data includes categorical data and text data, the above processor may separately vectorize the categorical data and the text data and then merge them, or may merge the categorical data and the text data to create a single text and then vectorize them.

[0015] The above processor, when performing the learning process, can determine whether there is a need for treatment precautions and collaboration with other departments based on the underlying disease identified from the initial data.

[0016] The processor can recognize the need for consultation with the circulatory internal medicine department when the underlying disease is hypertension and aspirin use is recognized, recognize the need for consultation with the hematology department when the underlying disease is aplastic anemia and bleeding, identify precautions for blood transfusion and blood test before surgery, recognize the need for consultation with the endocrinology department when the patient is receiving bone injection treatment, and identify precautions for medication-related osteonecrosis of the jaw (MRONJ).

[0017] In addition, a method for managing patients by underlying disease using artificial intelligence-based medical records according to another aspect of the present disclosure may include a step of collecting initial data including medical records, performed by a patient management server; a step of performing preprocessing on the initial data, wherein the preprocessing is performed according to preprocessing criteria including text processing and text data vectorization on the initial data; a step of inputting the preprocessed initial data into a pre-trained artificial intelligence model to perform learning processing to obtain at least one of a classification by underlying disease, precautions during treatment, and whether or not collaboration with other departments is required for the patient; and a step of providing a learning processing result including at least one of a classification by underlying disease, precautions during treatment, and whether or not collaboration with other departments is required for the patient.

[0018] The above medical records may include data in at least one format among categorical data, text data, and image file data.

[0019] The above patient management method can extract the visit address and underlying disease from the medical record at the stage of collecting the initial data.

[0020] The above-mentioned first-time electronic medical record may include data in at least one format among categorical data, text data, and image file data.

[0021] The above patient management method can extract at least one of the following: the visit address, underlying disease, patient condition, diagnosis, and treatment plan from the first-visit electronic medical record in the step of collecting the initial data.

[0022] The above patient management method can, in the step of performing the preprocessing, normalize the initial data according to the preprocessing criteria including criteria for at least one of language, numbers, symbols, typos, and abbreviations.

[0023] In the above patient management method, in the step of performing the preprocessing, if the initial data includes a first language and a second language, only the part of speech of the sentence in the second language is tagged, and only the corresponding word is translated into the first language to be normalized.

[0024] In the above patient management method, in the step of performing the preprocessing, if the initial data includes categorical data and text data, when processing the text data vectorization, the categorical data and text data may be vectorized separately and then merged, or the categorical data and text data may be merged to create a single text and then vectorized.

[0025] The above patient management method can, in the step of performing the learning process, determine whether there is a need for treatment precautions and collaboration with other departments based on the underlying disease identified from the initial data.

[0026] The above patient management method can recognize, in the step of performing the learning process, the need for circulatory internal medicine consultation when the underlying disease is hypertension and aspirin use is recognized, recognize the need for hematology internal medicine consultation when the underlying disease is aplastic anemia and bleeding, identify precautions for blood transfusion and blood tests before surgery, recognize the need for endocrinology internal medicine consultation when the patient receives bone injection treatment, and identify precautions for medication-related osteonecrosis of the jaw (MRONJ).

[0027] The above patient management method may further include a step of generating a collaborative treatment referral based on at least one of the underlying disease and treatment plan using a pre-stored conversational artificial intelligence model; a step of uploading the generated collaborative treatment referral to a hospital's administrative system that can be checked by a specific attending physician; and a step of performing collaborative treatment and making a reservation based on the results of the review and approval of the collaborative treatment referral by the specific attending physician.

[0028] In addition, a computer program stored in a computer-readable recording medium for executing the present disclosure may be further provided.

[0029] In addition, a computer-readable recording medium recording a computer program for executing a method for implementing the present disclosure may be further provided.

[0030] According to the aforementioned problem solving means of the present disclosure, it is expected that the classification by underlying disease, precautions for treatment accordingly, and prediction of collaboration with other departments can be efficiently identified through artificial intelligence learning using natural language processing of initial data such as medical records of patients visiting medical institutions.

[0031] The effects of the present disclosure are not limited to the effects mentioned above, and other effects not mentioned will be clearly understood by those skilled in the art from the description below.

[0032] Figure 1 is a block diagram of the patient management server of the present disclosure.

[0033] Figure 2 is a block diagram of a specific configuration of the patient management server of Figure 1.

[0034] Figures 3 to 6 are exemplary diagrams for explaining the patient management method for each underlying disease of the present disclosure.

[0035] Figure 7 is a flowchart for explaining the patient management method for each underlying disease of the present disclosure.

[0036] Figure 8 is a flowchart for explaining a part of Figure 7 in detail.

[0037] Throughout this disclosure, the same reference numerals denote the same components. This disclosure does not describe all elements of the embodiments, and any content that is common in the technical field to which this disclosure pertains or that overlaps between embodiments is omitted. The terms "part, module, element, block" used in the specification may be implemented in software or hardware, and depending on the embodiments, multiple "parts, modules, elements, blocks" may be implemented as a single component, or a single "part, module, element, block" may include multiple components.

[0038] Throughout the specification, when a part is said to be "connected" to another part, this includes not only direct connection but also indirect connection, and indirect connection includes connection via a wireless communication network.

[0039] Additionally, when a part is said to "include" a component, this does not mean that it excludes other components, but rather that it may include other components, unless otherwise specifically stated.

[0040] Throughout the specification, when we say that an element is "on" another element, this includes not only cases where the element is in contact with the other element, but also cases where another element exists between the two elements.

[0041] The terms first, second, etc. are used to distinguish one component from another, and the components are not limited by the aforementioned terms.

[0042] Singular expressions include plural expressions unless the context clearly indicates otherwise.

[0043] The identification codes for each step are used for convenience of explanation and do not describe the order of each step. Each step may be performed in a different order than specified unless the context clearly indicates a specific order.

[0044] The operating principle and embodiments of the present disclosure are described below with reference to the attached drawings.

[0045] In this specification, the term "patient management server according to the present disclosure" encompasses various devices capable of performing computational processing and providing results to a user. For example, the patient management server according to the present disclosure may include a computer, a server device, and a portable terminal, or may be any one of them.

[0046] Here, the computer may include, for example, a notebook, desktop, laptop, tablet PC, slate PC, etc. equipped with a web browser.

[0047] The above server device is a server that processes information by communicating with an external device, and may include an application server, a computing server, a database server, a file server, a game server, a mail server, a proxy server, and a web server.

[0048] The above portable terminal may include, for example, a wireless communication device that ensures portability and mobility, and may include all kinds of handheld-based wireless communication devices such as a PCS (Personal Communication System), GSM (Global System for Mobile communications), PDC (Personal Digital Cellular), PHS (Personal Handyphone System), PDA (Personal Digital Assistant), IMT (International Mobile Telecommunication)-2000, CDMA (Code Division Multiple Access)-2000, W-CDMA (W-Code Division Multiple Access), WiBro (Wireless Broadband Internet) terminal, a smart phone, and a wearable device such as a watch, ring, bracelet, anklet, necklace, glasses, contact lens, or head-mounted-device (HMD).

[0049] FIG. 1 is a block diagram of a patient management server of the present disclosure, and FIG. 2 is a block diagram of a specific configuration of the patient management server of FIG. 1.

[0050] Hereinafter, the method for managing patients by underlying disease of the present disclosure will be described with reference to FIGS. 3 to 6, which are exemplary diagrams.

[0051] Referring to Fig. 1, the patient management server (100) collects initial data, including the health records of patients visiting a medical institution, and through natural language processing, builds an artificial intelligence model capable of classifying patients by underlying disease at the preliminary examination stage. The medical records may include all types of records, including questionnaires in table or short answer format, as well as descriptive or non-restricted format records. In this case, the medical records may also include electronic medical records (EMRs) in electronic form. In addition, the initial data may include at least one semantic record of the patient's first or follow-up examination.

[0052] Referring to FIG. 2, the patient management server (100) includes a processor (110), a memory (130), and a communication unit (150). The components illustrated in FIG. 2 are not essential for implementing the patient management server (100) according to the present disclosure, and thus, the patient management server (100) described in this specification may have more or fewer components than the components listed above.

[0053] The processor (110) can manage patient information based on the patient's initial data.

[0054] Specifically, the processor (110) can collect initial data, including medical records. At this time, the initial data is not limited to medical records, and any information that can be used to determine the patient's condition can be included.

[0055] The above medical records may include a questionnaire containing data in at least one of the following formats: categorical data, text data, and image file data. For example, categorical data may be in the form of yes / no questions, which frequently appear in questionnaires, but is not limited thereto. Text data in the present disclosure refers to data consisting solely of letters, and may include both descriptive data and word-based data.

[0056] If the questionnaire is in the form of a picture file, the processor (110) can obtain the contents disclosed in the questionnaire through an OCR (Optical Character Recognition) function, or can obtain data from an OCR (Optical Character Recognition) system that is stored and managed in a text file format corresponding to the picture file data, but is not limited thereto.

[0057] The above medical records may include initial electronic medical records containing data in at least one of the following formats: categorical data, text data, and image file data. For example, initial electronic medical records and collaborative consultation records may be obtained from, but are not limited to, a medical institution's Clinical Data Warehouse (CDW) system and medical records room data.

[0058] When collecting initial data, the processor (110) can extract the visit address and underlying disease from the questionnaire. The visit address may refer to the most important complaint among the patient's complaints. Specifically, the visit address is the first thing on the patient's mind, and thus, in most cases, can fall into one of the following categories: pain, dysfunction, change in condition, or observations made by the patient.

[0059] When collecting initial data, the processor (110) may extract at least one of the following: the visit address, underlying disease, patient condition, diagnosis, and treatment plan from the initial electronic medical record. Additionally, the processor (110) may also extract data on consultation and precautions from the initial electronic medical record.

[0060] The processor (110) performs preprocessing on the initial data, and may perform preprocessing according to preprocessing criteria including text processing and text data vectorization on the initial data.

[0061] The processor (110) can perform preprocessing, including digitizing categorical data through simple number substitution and vectorizing text data using natural language processing, in the case of text data.

[0062] The processor (110) may normalize the initial data according to preprocessing criteria including criteria for at least one of language, numbers, symbols, typos, and abbreviations. The language may refer to a case where the language includes languages ​​of different nationalities.

[0063] Text data (e.g., medical records including initial electronic medical records and questionnaires) may include a first language (e.g., Korean) and a second language (e.g., English, etc.), and the processor (110) of the present disclosure may perform a deep learning process through word embedding of the first language and the second language.

[0064] The processor (110) can translate a second language into a first language through a translation application.

[0065] When the initial data includes a first language (e.g., Korean) and a second language (e.g., English, etc.), the processor (110) can normalize the data by tagging only the part of speech of the sentence in the second language and then translating only the corresponding word into the first language.

[0066] When a processor (110) detects a typo in the initial data, it can replace the found typo with a preset word used in medical records among words similar to the found typo. To this end, the processor (110) can pre-register words frequently used in medical records and match and register a number of typos therewith. Since typo words are used with similar meanings or formats to existing words, the processor (110) can extract similarly used words through deep learning-based word embedding to pre-create its own typo dictionary.

[0067] The processor (110) can utilize information from medical terminology searchable sites such as KMLE, MedlinePlus, openmd, etc., in relation to abbreviations and acronyms, to extract all forms in which a specific word can be used, and replace the extracted abbreviations and acronyms with complete words or sentences.

[0068] Specifically, the processor (110) can set various formats that abbreviations and acronyms can have as rules based on regular expressions, extract abbreviations and acronyms from initial data, and then replace them with complete words or sentences. For example, the various formats that abbreviations and acronyms can have based on regular expressions can include formats that include consecutive capital letters, periods between capital letters, etc.

[0069] The present disclosure can also apply a method for improving the performance of an artificial intelligence model by adding a procedure for comparing the performance of an artificial intelligence model based on text data processed based on the above-described methods with the performance of an artificial intelligence model based on existing text data.

[0070] The processor (110) can generate a stopword dictionary of words that are frequently used, such as particles, but do not play a significant role in understanding the patient's condition, and perform stopword deletion processing from the initial data using the generated stopword dictionary. For example, in the case of Korean, stopwords can be particles (~eun / neun, ~i / ga, etc.).

[0071] The processor (110) may pre-set processing rules for cases where the categorical data (yes / no) of a medical record (e.g., a questionnaire) and the text data in the remarks column do not match, and may apply these rules during the preprocessing of the initial data. For example, if the categorical data of the medical record is written as “yes”, but the text data in the remarks column does not match the “yes”, the processor (110) may mark the corresponding portion as an abnormal portion. In this case, the marked abnormal portion may be re-examined by the operator or processed to be reflected when upgrading the artificial intelligence model.

[0072] The processor (110) can process the initial data into text data vectorization.

[0073] Specifically, in the case of medical records (e.g., a medical questionnaire), since the part that checks underlying diseases and past medical history includes both categorical data and text data, the processor (110) can generate the categorical data and text data into one vector.

[0074] For example, when the initial data (e.g., questionnaire) includes categorical data and text data, the processor (110) may, when processing text data vectorization, separately vectorize the categorical data and the text data and then merge them.

[0075] As another example, if the initial data (e.g., questionnaire) includes categorical data and text data, the processor (110) may merge the categorical data and text data to create a single text and then vectorize it.

[0076] Referring to FIG. 3, in the underlying disease question of the questionnaire, the processor (110) extracts the content of “yes / no” from the categorical data, converts it into text data, merges it with the text data of the question item and the remark column to create a single text data, and then processes the vectorization process. At this time, when the processor (110) merges the text data into one and then creates a sentence, the question item text that is not revealed in the binary encoding of the categorical data (yes / no) can also be vectorized.

[0077] As another example, the processor (110) may perform text embedding on each item (e.g., chief complaint, current medical history, past medical history, etc.) of the initial data (e.g., initial electronic medical record), and then vectorize the embedding vectors by multiplying them by weights. At this time, the processor (110) may apply a relatively greater weight to an item that is to be viewed as more important in the initial data (e.g., initial electronic medical record). At this time, the item that is to be viewed as more important may be arbitrarily determined by the operator.

[0078] As another example, the processor (110) can merge the entire text of the initial electronic medical record into a single vectorized text. In this case, the processor (110) can perform learning processing by applying the corresponding importance within the artificial intelligence model itself, without setting individual weights for each item in the initial electronic medical record. To this end, the artificial intelligence model can reflect the preset importance for specific items.

[0079] As another example, the processor (110) can vectorize the initial data by converting text data into a numeric vector format using various embedding models used in natural language processing.

[0080] Specifically, embedding models applied to AI models include word frequency-based techniques like Bag-of-Words and TF-IDF, but neural network-based techniques like Word2Vec and FastText are also relatively widely used. Furthermore, AI models can be generated through comparative analysis with methods applied to BERT-based models like BioBERT and ClinicalBERT, which are pretrained on large medical open datasets like Pubmed and MIMIC-III clinical notes.

[0081] The processor (110) inputs the preprocessed initial data into a pre-trained artificial intelligence model and performs learning processing to obtain at least one of classification by underlying disease, precautions during treatment, and whether or not collaboration with other departments is necessary for the patient. In this case, inputting the data into the artificial intelligence model and performing learning processing may mean a processing process for obtaining a result value by inputting the preprocessed initial data into the pre-trained artificial intelligence model so as to determine classification by underlying disease, precautions during treatment, and whether or not collaboration with other departments is necessary based on the patient's initial data.

[0082] The processor (110) can classify the patient's visit address and underlying disease based on initial data (medical records including a questionnaire and first-time medical records), and then determine precautions for dental treatment (e.g., discontinuation of medication, blood test on the day before surgery, etc.) and whether to consult with other departments (cardiology, neurosurgery, gastroenterology, endocrinology, hematology, general surgery, etc.).

[0083] Meanwhile, the processor (110) may perform an external validation procedure on the initial data and learning processing results. At this time, the processor (110) may utilize data from other medical institutions using CDW.

[0084] The processor (110) can provide a learning processing result including at least one of classification by underlying disease of the patient, precautions during treatment, and whether or not collaboration with other departments is necessary.

[0085] When the processor (110) performs learning processing, it can determine whether there is a need for treatment precautions and collaboration with other departments based on the underlying disease identified from the initial data.

[0086] The processor (110) can pre-create a rule set that matches precautions and consultations for specific words so that it can determine whether precautions and consultations are necessary through frequently used specific words and sentences.

[0087] Specifically, referring to FIG. 4, if the processor (110) recognizes that the underlying disease is hypertension and that aspirin is being taken, it can recognize that consultation with a circulatory internal medicine specialist is necessary.

[0088] The processor (110) can recognize the need for consultation with a hematologist when the underlying disease is aplastic anemia and bleeding, and can identify precautions for blood transfusion and blood tests before surgery.

[0089] The processor (110) can recognize endocrinology consultation when a patient receives bone injection treatment and can identify precautions for medication-related osteonecrosis of the jaw (MRONJ).

[0090] Meanwhile, for the learning process described above, the processor (110) of the present disclosure can generate an artificial intelligence model capable of predicting precautions during treatment and collaboration with other departments based on initial data that has undergone a preprocessing process as medical records including a questionnaire or an initial consultation electronic document record. At this time, the artificial intelligence model can include a classification model function.

[0091] The processor (110) can output warnings and whether collaboration with other departments is necessary based on the learning processing results. At this time, the processor (110) can label the results identified from the initial data in various ways. This labeling can be reflected in the subsequent learning processing and upgrade of the artificial intelligence model.

[0092] For example, referring to FIG. 5, the processor (110) can apply a multi-label classification method that labels each of the contents of precautions and whether or not collaboration with other departments is necessary (whether or not collaboration is required in FIG. 5).

[0093] As another example, referring to FIG. 6, the processor (110) can apply a multi-class classification method that groups the precautions and the need for collaboration with other departments (the presence or absence of collaboration in FIG. 6) into one group and recognizes all cases as each class.

[0094] The artificial intelligence model of the present disclosure can be obtained by comparing and analyzing various deep learning models based on CNN (ResNet, DenseNet, etc.) and RNN (LSTM, GRU, etc.).

[0095] In addition, the artificial intelligence model of the present disclosure can evaluate the model based on indicators such as F1 score, AUROC, and AUPRC to ensure greater accuracy, thereby confirming the generalization performance of the model.

[0096] The processor (110) of the present disclosure may be composed of one or more cores, and may include a processor for data analysis and deep learning, such as a central processing unit, a general purpose graphics processing unit, and a tensor processing unit of a computing device. The processor (110) may read a computer program stored in a memory (130) and perform data processing for machine learning according to the present disclosure. According to the present disclosure, the processor (110) may perform operations for learning a neural network. The processor (110) may perform calculations for learning a neural network, such as processing input data for learning in deep learning, extracting features from input data, calculating errors, and updating weights of a neural network using backpropagation.

[0097] The above neural network model may be a deep neural network. In the present disclosure, neural network, network function, and neural network may be used interchangeably. A deep neural network (DNN) may refer to a neural network that includes multiple hidden layers in addition to an input layer and an output layer. Using a deep neural network, it is possible to identify latent structures of data. That is, it is possible to identify latent structures of photos, text, videos, voices, and music (e.g., what objects are in the photo, what the content and emotion of the text are, what the content and emotion of the voice are, etc.). A deep neural network may include a convolutional neural network (CNN), a recurrent neural network (RNN), a restricted Boltzmann machine (RBM), a deep belief network (DBN), a Q network, a U network, a Siamese network, etc.

[0098] Convolutional neural networks (CNNs) are a type of deep neural network that include neural networks containing convolutional layers. CNNs are a type of multilayer perceptron designed to use minimal preprocessing. CNNs can be composed of one or more convolutional layers and artificial neural network layers combined with them. CNNs can additionally utilize weight and pooling layers. This structure allows CNNs to fully utilize two-dimensional input data. CNNs can be used to recognize objects in images. CNNs can process image data by representing it as a matrix with dimensions. For example, in the case of image data encoded in RGB (red-green-blue), each of the R, G, and B colors can be represented as a two-dimensional (for example, in a two-dimensional image) matrix. That is, the color value of each pixel of the image data can be an element of a matrix, and the size of the matrix can be the same as the size of the image. Therefore, the image data can be represented as three two-dimensional matrices (a three-dimensional data array).

[0099] In a convolutional neural network, a convolutional process (input and output of a convolutional layer) can be performed by moving the convolutional filter and multiplying the matrix elements at each location of the image with the convolutional filter. The convolutional filter can be composed of an n*n matrix. The convolutional filter can generally be composed of a fixed-shape filter that is smaller than the total number of pixels in the image. That is, when an m*m image is input to a convolutional layer (for example, a convolutional layer whose convolutional filter has a size of n*n), a matrix representing n*n pixels containing each pixel of the image can be component-wise multiplied with the convolutional filter (i.e., each element of the matrix is ​​multiplied). By multiplying with the convolutional filter, a component matching the convolutional filter can be extracted from the image. For example, a 3*3 convolutional filter for extracting up and down straight line components from an image can be configured as [[0,1,0], [0,1,0], [0,1,0]]. When a 3*3 convolutional filter for extracting up and down straight line components from an image is applied to an input image, up and down straight line components matching the convolutional filter from the image can be extracted and output. A convolutional layer can apply a convolutional filter to each matrix for each channel representing an image (i.e., R, G, B colors in the case of an R, G, B coded image). A convolutional layer can extract features matching the convolutional filter from the input image by applying a convolutional filter to the input image. The filter value of the convolutional filter (i.e., the value of each element of the matrix) can be updated by backpropagation during the learning process of a convolutional neural network.

[0100] A subsampling layer can be connected to the output of a convolutional layer to simplify the output of the convolutional layer, thereby reducing memory usage and computational complexity. For example, when the output of a convolutional layer is input to a pooling layer having a 2*2 max pooling filter, the image can be compressed by outputting the maximum value included in each patch for each 2*2 patch from each pixel of the image. The above-described pooling may be a method of outputting the minimum value in a patch or the average value of a patch, and any pooling method may be included in the present disclosure.

[0101] A convolutional neural network may include one or more convolutional layers and subsampling layers. A convolutional neural network can extract features from an image by repeatedly performing convolutional and subsampling processes (e.g., the aforementioned max pooling). Through repeated convolutional and subsampling processes, the neural network can extract global features of the image.

[0102] The output of a convolutional layer or a subsampling layer can be input to a fully connected layer. A fully connected layer is a layer in which all neurons in one layer are connected to all neurons in the neighboring layer. A fully connected layer can refer to a structure in a neural network in which all nodes in each layer are connected to all nodes in other layers.

[0103] At least one of the CPU, GPGPU, and TPU of the processor (110) can process network function learning. For example, the CPU and GPGPU can jointly process network function learning and data classification using the network function. Furthermore, in one embodiment of the present disclosure, processors of multiple computing devices can be used together to process network function learning and data classification using the network function. Furthermore, a computer program executed on a computing device according to one embodiment of the present disclosure may be a CPU, GPGPU, or TPU executable program.

[0104] The memory (130) can store a computer program for providing a patient management method for each underlying disease, and the stored computer program can be read and operated by the processor (110). The memory (130) can store any form of information generated or determined by the processor (110) and any form of information received by the communication unit (150).

[0105] The memory (130) can store data supporting various functions of the patient management server (100), a program for the operation of the processor (110), can store input / output data, and can store a plurality of application programs (or applications) running on the patient management server (100), data for the operation of the patient management server (100), and commands. At least some of these application programs can be downloaded from an external server via wireless communication.

[0106] The memory (130) may include at least one type of storage medium among a flash memory type, a hard disk type, an SSD type, an SDD type, a multimedia card micro type, a card type memory (e.g., SD or XD memory, etc.), a random access memory (RAM), a static random access memory (SRAM), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), a programmable read-only memory (PROM), a magnetic memory, a magnetic disk, and an optical disk. In addition, the memory may be a database that is separate from the device but is connected by wire or wirelessly.

[0107] The communication unit (150) may include one or more components that enable communication with an external device, and may include, for example, at least one of a broadcast reception module, a wired communication module, a wireless communication module, a short-range communication module, and a location information module.

[0108] Although not shown, the patient management server (100) of the present disclosure may further include an output section and an input section.

[0109] The output unit may display a user interface (UI) for providing patient management results by underlying disease, etc. The output unit may output any form of information generated or determined by the processor (110) and any form of information received by the communication unit (150).

[0110] The output unit may include at least one of a liquid crystal display (LCD), a thin film transistor-liquid crystal display (TFT LCD), an organic light-emitting diode (OLED), a flexible display, and a three-dimensional display (3D display). Some of these display modules may be configured as transparent or light-transmitting so that the outside can be viewed through them. This may be referred to as a transparent display module, and a representative example of the transparent display module is TOLED (Transparent OLED).

[0111] The input unit can receive information input by a user. The input unit can include keys and / or buttons on a user interface, or physical keys and / or buttons, for receiving information input by a user. A computer program for controlling a display according to embodiments of the present disclosure can be executed based on user input through the input unit.

[0112] Fig. 7 is a flowchart for explaining a patient management method for each underlying disease of the present disclosure, and Fig. 8 is a flowchart for explaining a part of Fig. 7 in detail.

[0113] The patient management method for each underlying disease disclosed below can be implemented in the same manner as all methods implemented by the patient management server (100) of FIGS. 1 to 6 described above. However, for the convenience of explanation, redundant detailed descriptions will be omitted.

[0114] Referring to FIG. 7, the processor (110) of the patient management server (100) can collect initial data including medical records (1100).

[0115] The above medical record may include a questionnaire containing data in at least one format among categorical data, text data, and image file data. The above medical record may include an initial electronic medical record containing data in at least one format among categorical data, text data, and image file data.

[0116] In the initial data collection step, the processor (110) can extract the visit address and underlying disease from the questionnaire. In the initial data collection step, the processor (110) can extract at least one of the visit address, underlying disease, patient condition, diagnosis, and treatment plan from the initial electronic medical record.

[0117] The processor (110) performs preprocessing on the initial data, and may perform preprocessing according to preprocessing criteria including text processing and text data vectorization on the initial data (1200).

[0118] Specifically, when performing preprocessing, the processor (110) may normalize the initial data according to preprocessing criteria including criteria for at least one of language, numbers, symbols, typos, and abbreviations.

[0119] For example, if the initial data includes a first language and a second language, the processor (110) can normalize the data by tagging only the part of speech of a sentence in the second language and then translating only the corresponding word into the first language.

[0120] When the processor (110) processes text data vectorization when the initial data includes categorical data and text data, the processor may vectorize the categorical data and text data separately and then merge them, or may merge the categorical data and text data to create a single text and then vectorize them.

[0121] The processor (110) inputs preprocessed initial data into a pre-learned artificial intelligence model to perform learning processing, thereby obtaining at least one of classification by underlying disease for the patient, precautions during treatment, and whether or not collaboration with other departments is necessary (1300).

[0122] Specifically, the processor (110) can classify patients by underlying disease based on initial data (2100).

[0123] Next, the processor (110) can determine whether there is a need for treatment precautions and collaboration with other departments based on the underlying disease identified from the initial data (2200, 2300).

[0124] For example, if the processor (110) recognizes that the underlying disease is hypertension and that aspirin is being taken, it may recognize that consultation with a circulatory internal medicine specialist is necessary.

[0125] The processor (110) can recognize the need for consultation with a hematologist when the underlying disease is aplastic anemia and bleeding, and can identify precautions for blood transfusion and blood tests before surgery.

[0126] The processor (110) can recognize endocrinology consultation when a patient receives bone injection treatment and can identify precautions for medication-related osteonecrosis of the jaw (MRONJ).

[0127] The processor (110) can provide learning processing results (1400) that include at least one of the following: classification of the patient by underlying disease, precautions during treatment, and the need for collaboration with other departments. Although not shown, the processor (110) can display the learning processing results through an output unit or transmit them to a medical staff terminal (not shown).

[0128] The above-described invention utilizes artificial intelligence learning to process natural language data, including medical records including questionnaires and electronic medical records, to build a patient classification model that classifies patients based on their underlying diseases at the time of first consultation and determines whether to provide consultation and precautions.

[0129] In addition, the present invention can reduce the overload of medical staff and establish a treatment assistance system for efficient treatment in an environment where the proportion of patients with underlying diseases visiting dental clinics in medical school hospitals is relatively high and, unlike dental schools, there is a shortage of residents conducting preliminary examinations.

[0130] In addition, the present invention is expected to have the effect of being applicable not only to the dental field but also to the diagnosis of diseases of other departments.

[0131] Meanwhile, as described above, the processor (110) determines the patient's general condition and precautions according to the treatment plan and whether or not to consult based on the patient's medical history and medical records, and then additionally uses an interactive artificial intelligence model (e.g., chatGPT) to create (write) a consultation request form based on at least one of the underlying disease and the treatment plan, uploads the created consultation request form to a hospital system (e.g., an administrative system) that can be checked by the attending physician, and performs consultation and reservation based on the results of the attending physician's review and approval of the consultation request form.

[0132] In general hospitals, many patients suffer from a variety of complex conditions, often requiring collaboration with other departments for treatment or surgery. Filling out referral requests and scheduling appointments with the appropriate physicians is a time-consuming process.

[0133] Accordingly, the present disclosure links a conversational artificial intelligence model such as chatGPT with a medical record and administrative system, so that the attending physician decides on a collaborative treatment, and the conversational artificial intelligence model generates (writes) a collaborative treatment request form based on common phrases normalized according to the present disclosure, and provides the collaborative treatment request form to the attending physician so that modification and approval can be made, and reservations can be made for the corresponding consulting doctors related to the collaborative treatment request form through the administrative system linked to the conversational artificial intelligence model, and examination feedback can be provided to the corresponding consulting doctors.

[0134] As described above, the present disclosure utilizes a clinical assistance system that integrates conversational AI, such as chatGPT, with a clinical administration system. This system can significantly reduce the workload of healthcare providers, minimize omissions and errors, and provide efficient clinical services. Furthermore, it can improve work efficiency in hospitals with limited resident or assistant staff.

[0135] Meanwhile, the method according to the present disclosure described above can be implemented as a program (or application) and stored in a medium to be executed in combination with a hardware server.

[0136] The disclosed embodiments may be implemented in the form of a recording medium storing computer-executable instructions. The instructions may be stored in the form of program code, and when executed by a processor, may generate program modules to perform the operations of the disclosed embodiments. The recording medium may be implemented as a computer-readable recording medium.

[0137] Computer-readable storage media include all types of storage media that store instructions that can be deciphered by a computer. Examples include read-only memory (ROM), random access memory (RAM), magnetic tape, magnetic disks, flash memory, and optical data storage devices.

[0138] The disclosed embodiments have been described with reference to the attached drawings as described above. Those skilled in the art will understand that the present disclosure can be implemented in forms other than the disclosed embodiments without altering the technical spirit or essential features of the present disclosure. The disclosed embodiments are illustrative and should not be construed as limiting.

Claims

1. Memory; and A processor for managing patient information based on the patient's initial data; The above processor, Collect the above initial data including medical records, Preprocessing is performed on the above initial data, and preprocessing is performed according to preprocessing criteria including text processing and text data vectorization for the initial data, The above preprocessed initial data is input into a pre-trained artificial intelligence model to perform learning processing to obtain at least one of classification by underlying disease for the patient, precautions during treatment, and necessity of collaboration with other departments, and A patient management server for each underlying disease using artificial intelligence-based medical records, which provides learning processing results including at least one of classification by underlying disease for the above patient, precautions during treatment, and necessity of collaboration with other departments.

2. In paragraph 1, The above medical records include a questionnaire containing data in at least one format among categorical data, text data, and image file data, The above processor, A patient management server for each underlying disease using artificial intelligence-based medical records that extracts the visit address and underlying disease from the questionnaire when collecting the above initial data.

3. In paragraph 1, The above medical records include an initial electronic medical record containing data in at least one format among categorical data, text data, and image file data, The above processor, A patient management server for each underlying disease using an artificial intelligence-based medical record that extracts at least one of the following from the initial electronic medical record: the visit address, underlying disease, patient condition, diagnosis, and treatment plan when collecting the above initial data.

4. In paragraph 1, The above processor, A patient management server for each underlying disease using an artificial intelligence-based medical record, which normalizes the above initial data according to the above preprocessing criteria including criteria for at least one of language, numbers, symbols, typos, and abbreviations.

5. In paragraph 4, The above processor, A patient management server for each underlying disease using an artificial intelligence-based medical record, which, when the above initial data includes a first language and a second language, tags only the part of speech of sentences in the second language, and then translates and normalizes only the corresponding words into the first language.

6. In paragraph 1, The above processor, A patient management server for each underlying disease using medical records based on artificial intelligence, in which, when the above initial data includes categorical data and text data, the categorical data and text data are vectorized separately and then merged, or the categorical data and text data are merged to create a single text and then vectorized.

7. In paragraph 1, The above processor, A patient management server for each underlying disease using medical records based on artificial intelligence, which determines whether there is a need for consultation with other departments and precautions for treatment based on the underlying disease identified from the initial data when performing the above learning process.

8. In paragraph 7, The above processor, If the above underlying disease is hypertension and you are aware of taking aspirin, you should be aware of the need for consultation with a cardiology specialist. If the above underlying disease is aplastic anemia and bleeding, recognize the need for consultation with the hematology department and understand precautions for blood transfusion and blood tests before surgery. A patient management server based on medical records based on underlying diseases that recognizes endocrinology consultation when the above patient receives bone injection treatment and identifies precautions for medication-related osteonecrosis of the jaw (MRONJ).

9. In a method performed by a patient management server, The step of collecting the above initial data including medical records; A step of performing preprocessing on the initial data, wherein the preprocessing is performed according to preprocessing criteria including text processing and text data vectorization on the initial data; A step of inputting the above-mentioned pre-processed initial data into a pre-trained artificial intelligence model to perform learning processing to obtain at least one of classification by underlying disease for the patient, precautions during treatment, and necessity of collaboration with other departments; and A method for managing patients by underlying disease using artificial intelligence-based medical records, comprising a step of providing a learning processing result including at least one of classification by underlying disease for the patient, precautions during treatment, and necessity of collaboration with other departments.

10. In paragraph 9, The above medical records include a questionnaire containing data in at least one format among categorical data, text data, and image file data, A method for managing patients by underlying disease using an artificial intelligence-based medical record, extracting the visit address and underlying disease from the questionnaire in the step of collecting the above initial data.

11. In paragraph 9, The above medical records include an initial electronic medical record containing data in at least one format among categorical data, text data, and image file data, A method for managing patients by underlying disease using an artificial intelligence-based medical record, wherein at least one of the following is extracted from the first-visit electronic medical record: the address of the visit, the underlying disease, the patient's condition, the diagnosis, and the treatment plan, in the step of collecting the initial data.

12. In paragraph 9, In the step of performing the above preprocessing, A method for managing patients by underlying disease using artificial intelligence-based medical records, wherein the initial data is normalized according to the preprocessing criteria including criteria for at least one of language, numbers, symbols, typos, and abbreviations.

13. In paragraph 12, In the step of performing the above preprocessing, A method for managing patients by underlying disease using medical records based on artificial intelligence, in which, when the above initial data includes a first language and a second language, only the part of speech of a sentence in the second language is tagged, and only the corresponding words are translated into the first language for normalization.

14. In paragraph 9, In the step of performing the above preprocessing, A method for managing patients by underlying disease using medical records based on artificial intelligence, wherein when the above initial data includes categorical data and text data, when processing the text data vectorization, the categorical data and text data are vectorized separately and then merged, or the categorical data and text data are merged to create a single text and then vectorized.

15. In paragraph 9, In the step of performing the above learning process, A method for managing patients by underlying disease using artificial intelligence-based medical records, which identifies precautions for matching treatment and the need for collaboration with other departments based on the underlying disease identified from the above initial data.

Images