Digital healthcare content provision system using health examination data and method thereof

Abstract

The present invention relates to a digital healthcare content provision technology using health examination data, wherein: a plurality of digital healthcare contents are analyzed through a digital content generation model to generate new digital healthcare contents for each disease; and according to disease prediction information obtained by inputting regular health examination data of a user into a disease prediction model, at least one of the digital healthcare contents is selected as personalized digital healthcare content to be provided to the user. The digital healthcare content provision technology using health examination data can automatically generate a plurality of additional digital healthcare contents by analyzing existing digital healthcare contents, provide personalized digital healthcare content according to the health status of each user, and provide a function to preemptively review potential copyright infringement in the generated digital healthcare content.

Description

Digital healthcare content provision system and method using health checkup data

[0001] The present invention relates to a technology for providing digital healthcare content using health checkup data, and more specifically, to a system and method for providing digital healthcare content using health checkup data, which analyzes a plurality of digital healthcare contents through a digital content creation model to newly create digital healthcare contents for each disease, inputs a user's regular health checkup data into a disease prediction model, and selects at least one of the digital healthcare contents as personal digital healthcare contents to be provided to the user based on the disease prediction information outputted therefrom.

[0002] While ongoing research is being conducted on the health impacts of digital healthcare content, including sound, on human health, services that involve healthcare professionals (doctors, pharmacists, etc.) in the development and that utilize digital therapeutic content with proven medical efficacy through actual clinical trials are still in the early stages of development.

[0003] To expand this market, there is a growing need for technology that allows healthcare professionals to create digital therapeutic content developed and verified directly based on evidence derived and verified through Systematic Review & Meta Analysis, which is utilized to establish the highest level of evidence in the healthcare field, and much research is being conducted.

[0004] [Prior patent]

[0005] Korean Patent Application No. 10-2022-0143833: Device and method for generating user-tailored health management content based on a user's health questionnaire and brain wave signals.

[0006] Research on technologies for generating digital therapeutic content based on a user's health information, such as the above-mentioned prior patent, has been conducted in the past. However, research on technologies for analyzing and automatically expanding the generated digital therapeutic content and linking it with predicted disease outbreak information using each user's health checkup information to provide personalized digital healthcare content is currently lacking.

[0007] The purpose of the present invention is to provide a technology for providing digital healthcare content using health checkup data, which analyzes digital healthcare content whose effectiveness has already been proven to generate additional digital healthcare content, and provides personalized digital healthcare content for each user based on their health status by linking predicted disease outbreak information using each user's health checkup information.

[0008] According to one embodiment of the present invention, a digital healthcare content providing system may include: a health checkup data collection unit that directly inputs a user's regular health checkup data or receives it from a connected database; a digital healthcare content management unit that stores a plurality of digital healthcare contents, analyzes the stored digital healthcare contents, and newly generates digital healthcare contents for each disease through a digital content generation model; a disease prediction information generation unit that inputs the received user's regular health checkup data into a disease prediction model to output an onset probability value of each disease for each period, and generates disease prediction information including the type of disease and the onset probability value among the outputted onset probabilities of each disease, the probability of which is greater than a preset reference value; and a personal digital healthcare content selection unit that inputs the disease prediction information into a digital healthcare content selection model to select at least one of the digital healthcare contents stored in the digital healthcare content management unit as personal digital healthcare contents to be provided to the user.

[0009] According to one embodiment of the present invention, the digital healthcare content management unit may further include: a health checkup data collection unit that directly inputs a user's regular health checkup data or receives it from a connected database; a treatment characteristic information generation unit that stores a plurality of digital healthcare contents, inputs the stored digital healthcare contents into a characteristic extraction model to output sound characteristic information including sound essential parameter numerical information and sound pattern information for each digital healthcare contents, analyzes the sound characteristic information of the digital healthcare contents for each disease, and generates disease-specific treatment characteristic information including range information on the variation range of the sound essential parameter numerical information and sound pattern information included in the sound characteristic information; and a digital content generation unit that inputs the treatment characteristic information into a digital content generation model, and outputs digital healthcare contents for each disease by changing the sound essential parameter numerical information and sound pattern information within each range using the range information of each item of the sound characteristic information.

[0010] According to one embodiment of the present invention, the digital content generation unit can perform learning using sound characteristic information included in the treatment characteristic information so that the digital content generation model is a neural vocoder based on a normalizing flow, and a Gaussian distribution is derived from a sound dataset using a transformation function having reversibility.

[0011] According to one embodiment of the present invention, the digital content generation unit can change the sound essential parameter numerical information and sound pattern information within each range by using the range information of each item information in a manner that the learned digital content generation model synthesizes a sound output from a sample of a Gaussian distribution using the inverse function of the transformation function.

[0012] According to one embodiment of the present invention, the digital content generation unit divides the generated digital treatment content and the stored digital treatment content into blocks of 15 seconds each, and inputs the divided plurality of blocks into the feature extraction model to output sound characteristic information including sound essential parameter numerical information and sound pattern information for each block.

[0013] According to one embodiment of the present invention, the digital content generation unit can generate copyright-related information by comparing the sound characteristic information of each block of the generated digital therapeutic content with the sound characteristic information of a plurality of blocks of stored digital therapeutic content, and depending on whether there is similar sound characteristic information of each block within a preset standard.

[0014] According to one embodiment of the present invention, the disease prediction information generation unit may be a model trained to preprocess the input regular health checkup data into individual time series data, analyze the numerical change pattern for each checkup item among the preprocessed data, and output the onset probability value of each disease according to the numerical change pattern.

[0015] According to one embodiment of the present invention, the disease prediction information generation unit may be a model trained to construct a recurrent neural network using LSTM (Long short-term memory), preprocess the input regular health checkup data into individual time series data, analyze the numerical change pattern of each checkup item among the preprocessed data, and output the onset probability value of each disease according to the numerical change pattern.

[0016] According to one embodiment of the present invention, a method for providing digital healthcare content may include: receiving a user's regular health checkup data directly from a connected database; storing a plurality of digital healthcare contents, analyzing the stored digital healthcare contents, and newly generating digital healthcare contents for each disease through a digital content generation model; inputting the received user's regular health checkup data into a disease prediction model to output an onset probability value for each disease at each time period, and generating disease prediction information including the type and onset probability value of a disease having a probability greater than a preset reference value among the output onset probabilities of each disease; and inputting the disease prediction information into a digital healthcare content selection model to select at least one of the stored digital healthcare contents as personal digital healthcare contents to be provided to the user.

[0017] According to one embodiment of the present invention, the step of newly generating the digital healthcare content may further include the steps of: directly inputting the user's regular health checkup data or receiving it from a connected database; storing a plurality of digital healthcare contents, inputting the stored digital healthcare contents into a feature extraction model to output sound characteristic information including sound essential parameter numerical information and sound pattern information for each digital healthcare contents; analyzing the sound characteristic information of the digital healthcare contents for each disease to generate disease-specific treatment characteristic information including range information on the variation range of the sound essential parameter numerical information and sound pattern information included in the sound characteristic information; and inputting the treatment characteristic information into a digital content generation model to change the sound essential parameter numerical information and sound pattern information within each range using the range information of each item of the sound characteristic information, and outputting the generated digital healthcare contents for each disease.

[0018] According to one embodiment of the present invention, in the step of outputting the digital healthcare content, the digital content generation model is a neural vocoder based on a normalizing flow, and learning can be performed using sound characteristic information included in the treatment characteristic information so that a Gaussian distribution is derived from the sound dataset using a transformation function having reversibility.

[0019] According to one embodiment of the present invention, the step of outputting the digital healthcare content is such that the learned digital content generation model can change the sound essential parameter numerical information and sound pattern information within each range by using the range information of information for each item in a manner of synthesizing a sound output from a sample of a Gaussian distribution using the inverse function of the transformation function.

[0020] According to one embodiment of the present invention, the step of outputting the digital healthcare content may include dividing the generated digital treatment content and the stored digital treatment content into blocks of 15 seconds each, inputting the divided plurality of blocks into the feature extraction model, and outputting sound characteristic information including sound essential parameter numerical information and sound pattern information for each block.

[0021] According to one embodiment of the present invention, the step of outputting the digital healthcare content may compare the sound characteristic information of each block of the generated digital treatment content with the sound characteristic information of a plurality of blocks of the stored digital treatment content, and generate copyright-related information based on whether there is similar sound characteristic information of each block within a preset standard.

[0022] According to one embodiment of the present invention, the step of generating the disease prediction information may be a model that is trained to preprocess the input regular health checkup data into individual time series data, analyze the numerical change pattern of each checkup item among the preprocessed data, and output the onset probability value of each disease according to the numerical change pattern.

[0023] According to one embodiment of the present invention, the step of generating the disease prediction information may be a model trained to construct a recurrent neural network using LSTM (Long short-term memory), preprocess the input regular health checkup data into individual time series data, analyze the numerical change pattern of each checkup item among the preprocessed data, and output the onset probability value of each disease according to the numerical change pattern.

[0024] According to a technology for providing digital healthcare content using health checkup data according to one embodiment of the present invention, a plurality of digital healthcare contents can be automatically generated by analyzing existing digital healthcare contents, and personalized digital healthcare contents can be provided according to the health status of each user. Furthermore, a function for reviewing in advance the possibility of copyright infringement of the generated digital healthcare contents can be provided.

[0025] Figure 1 is a configuration diagram of a digital healthcare content provision system using health checkup data according to one embodiment of the present invention.

[0026] Figure 2 is a detailed configuration diagram of the digital healthcare content management unit illustrated in Figure 1.

[0027] FIG. 3 is a flowchart of a method for providing digital healthcare content using health checkup data according to one embodiment of the present invention.

[0028] Below, with reference to the attached drawings, embodiments of the present invention are described in detail so that those skilled in the art can easily implement them. However, the present invention may be implemented in various different forms and is not limited to the embodiments described herein.

[0029] The terms used in this invention are used solely to describe specific embodiments and are not intended to limit the invention. Singular expressions include plural expressions unless the context clearly dictates otherwise.

[0030] In the present invention, it should be understood that terms such as “include” or “have” are intended to specify the presence of a feature, number, step, operation, component, part or combination thereof described in the specification, but do not exclude in advance the possibility of the presence or addition of one or more other features, numbers, steps, operations, components, parts or combinations thereof.

[0031] Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which the present invention belongs.

[0032] Terms defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and will not be interpreted in an idealized or overly formal sense unless expressly defined herein.

[0033] It will also be appreciated that each block of the drawings and combinations of the flowchart drawings can be implemented by computer program instructions, which can be implemented in a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus, such that the instructions, when executed by the processor of the computer or other programmable data processing apparatus, create a means for implementing the functions described in the flowchart block(s).

[0034] These computer program instructions may also be stored in a computer-available or computer-readable memory that can direct a computer or other programmable data processing device to implement a function in a particular manner, so that the instructions stored in the computer-available or computer-readable memory can also produce an article of manufacture that includes instruction means for performing the function described in the flowchart block(s).

[0035] Since the computer program instructions may be installed on a computer or other programmable data processing device, a series of operational steps may be performed on the computer or other programmable data processing device to create a computer-executable process, and the instructions that cause the computer or other programmable data processing device to perform the steps for performing the functions described in the flowchart block(s) may also provide steps for performing the functions described in the flowchart block(s).

[0036] Additionally, each block may represent a module, segment, or portion of code that contains one or more executable instructions for performing a specified logical function(s).

[0037] It should also be noted that in some alternative embodiments, the functions described in the blocks may occur out of order. For example, two blocks depicted in succession may actually be executed substantially simultaneously, or the blocks may sometimes be executed in reverse order, depending on their respective functions.

[0038] At this time, the term '~ part' used in this embodiment means software or hardware components such as FPGA (field-programmable gate array) or ASIC (application specific integrated circuit), and '~ part' performs certain roles.

[0039] However, the term "~bu" is not limited to software or hardware. The "~bu" may be configured to reside on an addressable storage medium or may be configured to trigger one or more processors.

[0040] Thus, as an example, '~part' includes components such as software components, object-oriented software components, class components, and task components, processes, functions, attributes, procedures, subroutines, segments of program code, drivers, firmware, microcode, circuits, data, databases, data structures, tables, arrays, and variables. The functionality provided within the components and '~parts' may be combined into a smaller number of components and '~parts' or further separated into additional components and '~parts'. In addition, the components and '~parts' may be implemented to regenerate one or more CPUs within a device or a secure multimedia card.

[0041] In specifically describing embodiments of the present invention, examples of specific systems will be primarily used, but the main points claimed in this specification can be applied to other communication systems and services having similar technical backgrounds without significantly departing from the scope disclosed in this specification, and this can be done at the discretion of a person skilled in the relevant technical field.

[0042] Hereinafter, a digital healthcare content providing system and device using health checkup data according to an embodiment of the present invention will be described with reference to the drawings.

[0043] Figure 1 is a configuration diagram of a digital healthcare content provision system using health checkup data according to one embodiment of the present invention.

[0044] Referring to FIG. 1, a digital healthcare content provision system (1000) using health checkup data according to one embodiment of the present invention may include a health checkup data collection unit (100), a digital healthcare content management unit (200), a disease prediction information generation unit (300), and a personal digital healthcare content selection unit (400).

[0045] The health checkup data collection unit (100) can directly input the user's regular health checkup data or receive it from a connected database.

[0046] Here, regular health checkup data may refer to data collected in a time series from the results of a health checkup performed on a user's body at least once, and the health checkup items may be based on items included in the National Health Insurance Corporation's regular health checkup, which is generally provided free of charge, but are not limited thereto, and may include all test results on the body's health status.

[0047] The digital healthcare content management unit (200) stores a large number of digital healthcare contents, analyzes the stored digital healthcare contents, and can create new digital healthcare contents for each disease through a digital content creation model.

[0048] The digital healthcare content stored here refers to digital content developed directly by experts in the healthcare field and verified for its effectiveness in alleviating and treating diseases. While audio may be the main format of the content, it is not limited to this and may be content in a format that combines audio and video.

[0049] According to one embodiment of the present invention, verification of the disease alleviation and treatment effect of digital healthcare content may be performed using clinical trials, etc., but is not limited thereto.

[0050] According to one embodiment of the present invention, digital healthcare content may include, in the data header, name information on diseases for which alleviation and treatment effects are effective, alleviation or treatment effect information, etc.

[0051] According to one embodiment of the present invention, the digital content generation model may be a model formed by a plurality of convolutional layers based on an artificial neural network, and may be trained to input a plurality of stored digital contents and output digital healthcare content for each disease by changing sound essential parameter numerical information and sound pattern information within a certain change range.

[0052] Here, the essential parameter numerical information may refer to parameter values ​​of essential sources required for producing voice sounds included in digital healthcare content, and according to one embodiment, may be numerical parameter values ​​for Hz, beats, melody, repetition section pattern, repetition section length, etc.

[0053] Additionally, sound pattern information may refer to information about a pattern generated by dividing voice sounds included in digital healthcare content into blocks with preset lengths and analyzing the development of melody, BPM, and beat information items for each block.

[0054] According to one embodiment of the present invention, a certain change range can be selected as a change range by dividing voice sounds included in stored digital healthcare content into blocks having a preset length, collecting sound essential parameter numerical information and sound pattern information for each block, and generating range information based on the numerical values ​​of each item existing for each item of sound characteristic information.

[0055] The detailed configuration of the digital healthcare content management unit (200) will be described in more detail with reference to Figure 2.

[0056] The disease prediction information generation unit (300) inputs the received user's regular health checkup data into a disease prediction model to output the onset probability value of each disease for each period, and can generate disease prediction information including the type and onset probability value of a disease with a probability higher than a preset reference value among the outputted onset probabilities of each disease.

[0057] According to one embodiment of the present invention, a disease prediction model may be a model formed by a plurality of convolutional layers based on an artificial neural network, and may be trained to input a user's regular health checkup data and output disease prediction information including the type of disease with a possibility of developing and the probability of developing the disease.

[0058] According to one embodiment of the present invention, a disease prediction model may be composed of a first operation module and a second operation module formed by a plurality of convolutional layers.

[0059] According to the above embodiment, the first operation module can be trained to analyze the user's regular health checkup data, which is input data of the disease prediction model, and output a preset probability value of each disease at each time point after a preset period of time.

[0060] According to one embodiment of the present invention, a plurality of prior literature contributed on topics related to health checkup result items and onset for each predetermined disease is input into a meta-analysis model, and an effect size corresponding to the health checkup result items for each of the plurality of prior literature can be calculated.

[0061] According to one embodiment of the present invention, the inverse of the variance of the calculated effect size can be applied as a weight to the effect size of each prior literature to measure the disease onset impact score for each health checkup result item, and the disease onset probability value for each disease can be calculated based on the disease onset impact score calculated for each health checkup result item.

[0062] For example, the higher the disease incidence impact score calculated for each health checkup result item, the higher the probability of disease incidence can be calculated.

[0063] According to one embodiment of the present invention, the effect size corresponding to the topic related to the examination result item and the onset of disease for each of a plurality of prior literatures is calculated by calculating the odds ratio and confidence interval for each prior literature, and the effect size of the genetic variation for each prior literature on the target disease can be estimated based on the odds ratio and confidence interval.

[0064] According to one embodiment of the present invention, there may be various methods for extracting an effect size, and the type of effect size to be extracted may vary depending on the subject, and an effect size based on a standardized mean difference, an effect size based on a correlation coefficient, and an effect size based on an odds ratio, etc. may be used.

[0065] According to one embodiment of the present invention, in order to calculate the effect size based on the odds ratio, the effect size can be estimated based on the odds ratio (OR) and 95% confidence interval (CI), which are indicators of the magnitude of influence (magnitude of association) on the disease for each genetic variant. The odds ratios of each individual document can be combined to calculate the overall effect size (overall OR).

[0066] According to one embodiment of the present invention, a generic inverse variance estimation method can be used to measure a disease incidence influence score for each screening result item using the calculated effect size.

[0067] Inverse variance estimation is a method used to assign weights in meta-analysis, and the inverse of the variance of the estimated effect size can be used as the weight of each prior literature.

[0068] According to one embodiment of the present invention using the inverse variance estimation method, prior literature on studies with a large sample size will have a small variance, and the inverse of the variance will become large, so it can be used to give greater weight to prior literature on studies with a large sample size.

[0069] According to the above example, the natural logarithm of the odds ratio of each prior literature is taken. and produce, Calculate the standard error (SE) for , and weight the reciprocal of the square of the calculated standard error. It can be calculated as, and the weight of each preceding literature calculated as in mathematical formula 1 is multiplied by the odds ratio and the combined effect size (OR) is added up. pooled ) can be calculated.

[0070]

[0071] According to one embodiment of the present invention, the disease prediction model may be a model trained to preprocess input regular health checkup data into individual time series data, analyze the numerical change pattern of each checkup item among the preprocessed data, and output the onset probability value of each disease according to the numerical change pattern.

[0072] According to one embodiment of the present invention, a disease prediction model may be trained to analyze regular health checkup data preprocessed as time series data to obtain a numerical change pattern for each checkup item, determine the similarity between the type of the numerical change pattern and a pre-stored disease onset pattern, and output a probability value of onset of each disease according to the similarity value when the similarity is greater than a preset value.

[0073] According to one embodiment of the present invention, a disease prediction model can calculate an incidence probability value for each disease based on a similarity value between a numerical change pattern for each examination item obtained by analyzing regular health examination data and a pre-stored disease incidence pattern, and a disease incidence impact score output by inputting multiple prior literature contributed on topics related to health examination result items and incidence for each pre-set disease into a meta-analysis model.

[0074] According to the above embodiment, the disease prediction model can be trained by setting weights for each of the similarity value and the disease invention influence score, and adjusting each weight so that the accuracy is relatively high.

[0075] According to one embodiment of the present invention, a disease prediction model may be a model trained to construct a recurrent neural network using LSTM (Long short-term memory), preprocess the input regular health checkup data into individual time series data, analyze the numerical change pattern of each checkup item among the preprocessed data, and output the onset probability value of each disease according to the numerical change pattern.

[0076] According to one embodiment of the present invention, a disease prediction model can be trained to analyze regular health checkup data preprocessed into time series data using LSTM (Long short-term memory) to obtain a numerical change pattern for each checkup item, determine the similarity between the type of numerical change pattern and a pre-stored disease onset pattern, and output a probability value of onset of each disease according to the similarity value when the similarity is higher than a preset value.

[0077] The personal digital healthcare content selection unit (400) can input disease prediction information into a digital healthcare content selection model and select at least one of the digital healthcare contents stored in the digital healthcare content management unit as personal digital healthcare contents to be provided to the user.

[0078] According to one embodiment of the present invention, a digital healthcare content selection model analyzes input disease prediction information to extract diseases with an incidence probability higher than a preset reference value, and then selects digital healthcare content having information on the effect of alleviation or treatment of the extracted disease in its header among the stored digital healthcare content as personal digital healthcare content.

[0079] Figure 2 is a detailed configuration diagram of the digital healthcare content management unit illustrated in Figure 1.

[0080] Referring to FIG. 2, the digital healthcare content management unit (200) illustrated in FIG. 1 may include a health checkup data collection unit (210), a treatment characteristic information generation unit (220), and a digital content generation unit (230).

[0081] The health checkup data collection unit (210) can directly input the user's regular health checkup data or receive it from a connected database.

[0082] According to one embodiment of the present invention, a user can directly input his / her regular health checkup data in the form of a file, and input it by connecting to a database configured as a health checkup data storage server.

[0083] The treatment characteristic information generation unit (220) stores a plurality of digital healthcare contents, inputs the stored digital healthcare contents into a characteristic extraction model, outputs sound characteristic information including sound essential parameter numerical information and sound pattern information for each digital healthcare contents, and analyzes the sound characteristic information of the digital healthcare contents for each disease to generate disease-specific treatment characteristic information including range information on the variation range of sound essential parameter numerical information and sound pattern information included in the sound characteristic information.

[0084] According to one embodiment of the present invention, a feature extraction model may be configured with a first operation module and a second operation module formed of a plurality of convolution layers, and the first operation module may be trained to analyze input digital healthcare content and output sound characteristic information including sound essential parameter numerical information and sound pattern information for each digital healthcare content.

[0085] According to the above embodiment, the numerical parameter values ​​for Hz, beat, melody, repetition section pattern, and repetition section length, which are essential sources required for producing voice sounds included in digital healthcare content, can be extracted and output as essential parameter numerical information.

[0086] In addition, the voice sound included in the digital healthcare content can be divided into blocks with a preset length, and the sound pattern information can be output by analyzing the development of melody, BPM, and beat information items for each block.

[0087] According to one embodiment of the present invention, the second module of the digital content creation model analyzes sound characteristic information including input essential parameter numerical information and sound pattern information, and analyzes sound characteristic information of digital healthcare content for each disease to generate disease-specific treatment characteristic information including range information on the variation range of sound essential parameter numerical information and sound pattern information.

[0088] The digital content generation unit (230) inputs treatment characteristic information into a digital content generation model, and uses the range information of each item of sound characteristic information to change the sound essential parameter numerical information and sound pattern information within each range to output digital healthcare content for each disease.

[0089] According to one embodiment of the present invention, a digital content creation model receives a plurality of pieces of treatment characteristic information, collects sound essential parameter numerical information and sound pattern information for each disease, and generates range information based on the numerical values ​​of each item existing for each item of the disease-specific sound characteristic information, thereby selecting a range within the range information as a change range.

[0090] According to one embodiment of the present invention, the digital content generation model may be a neural vocoder based on a normalizing flow, and may perform learning using sound characteristic information included in the treatment characteristic information so that a Gaussian distribution is derived from a sound dataset using a transformation function having reversibility.

[0091] According to one embodiment of the present invention, a normalizing flow-based neural vocoder constituting a digital content generation model can train voice sound data to derive a Gaussian distribution from a voice sound dataset using a reversible transformation function, and after the training is completed, a voice sound output can be synthesized from a sample of the Gaussian distribution using an inverse function of the transformation function to generate digital healthcare content for each disease.

[0092] According to one embodiment of the present invention, the learned digital content generation model can change the sound essential parameter numerical information and sound pattern information within each range by using the range information of each item information in a manner of synthesizing a sound output from a sample of a Gaussian distribution using the inverse function of the transformation function.

[0093] According to one embodiment of the present invention, the digital content generation unit (230) divides the generated digital treatment content and the stored digital treatment content into blocks of 15 seconds each, and inputs the divided plurality of blocks into a feature extraction model to output sound characteristic information including sound essential parameter numerical information and sound pattern information for each block.

[0094] According to one embodiment of the invention, the digital content generation unit (230) can generate copyright-related information by comparing the sound characteristic information of each block of the generated digital therapeutic content with the sound characteristic information of multiple blocks of the stored digital therapeutic content, and depending on whether there is similar sound characteristic information of each block within a preset standard.

[0095] The reason for comparing the sound characteristic information of blocks divided into 15-second units of the digital treatment content generated here and the stored digital treatment content is to determine the similarity of the section corresponding to the 4-bar section, which is generally the standard for similarity in determining whether or not there has been copyright infringement.

[0096] According to one embodiment of the present invention, when comparing block-by-block sound characteristic information, a relatively higher weight can be applied to essential parameter numerical information included in the sound characteristic information and to the similarity of the sound pattern degree among the sound pattern information.

[0097] According to another embodiment, when comparing block-by-block sound characteristic information, weights are set for essential parameter numerical information and sound pattern information included in the sound characteristic information, and each weight can be learned so that accuracy is relatively high.

[0098] FIG. 3 is a flowchart of a method for providing digital healthcare content using health checkup data according to one embodiment of the present invention.

[0099] The user's regular health checkup data is directly entered or received from a connected database (S10).

[0100] According to one embodiment of the present invention, the user's regular health checkup data can be directly input or received from a connected database.

[0101] According to one embodiment of the present invention, regular health checkup data may mean data collected in a time series from data obtained when a user has performed a health checkup on his or her body at least once, and the health checkup items are generally based on items included in the National Health Insurance Corporation's regular health checkup, which is provided free of charge, but are not limited thereto and may include all test results on the health status of the body.

[0102] A large number of digital healthcare contents are stored, the stored digital healthcare contents are analyzed, and new digital healthcare contents for each disease are generated through a digital content creation model (S20).

[0103] According to one embodiment of the present invention, a plurality of digital healthcare contents can be stored, the stored digital healthcare contents can be analyzed, and new digital healthcare contents for each disease can be generated through a digital content generation model.

[0104] According to one embodiment of the present invention, verification of the disease alleviation and treatment effect of digital healthcare content may be performed using clinical trials, but is not limited thereto.

[0105] According to one embodiment of the present invention, digital healthcare content may include, in the data header, name information on diseases for which alleviation and treatment effects are effective, alleviation or treatment effect information, etc.

[0106] According to one embodiment of the present invention, the digital content generation model may be a model formed by a plurality of convolutional layers based on an artificial neural network, and may be trained to input a plurality of stored digital contents and output digital healthcare content for each disease by changing sound essential parameter numerical information and sound pattern information within a certain change range.

[0107] According to one embodiment of the present invention, a certain change range can be selected as a change range by dividing voice sounds included in stored digital healthcare content into blocks having a preset length, collecting sound essential parameter numerical information and sound pattern information for each block, and generating range information based on the numerical values ​​of each item existing for each item of sound characteristic information.

[0108] According to one embodiment of the present invention, the user's regular health checkup data can be directly input or received from a connected database.

[0109] According to one embodiment of the present invention, a user can directly input his / her regular health checkup data in the form of a file, and input it by connecting to a database configured as a health checkup data storage server.

[0110] According to one embodiment of the present invention, a plurality of digital healthcare contents are stored, and the stored digital healthcare contents are input into a feature extraction model to output sound characteristic information including sound essential parameter numerical information and sound pattern information for each digital healthcare contents, and the sound characteristic information of the digital healthcare contents for each disease is analyzed to generate disease-specific treatment characteristic information including range information on the variation range of sound essential parameter numerical information and sound pattern information included in the sound characteristic information.

[0111] According to one embodiment of the present invention, a feature extraction model may be configured with a first operation module and a second operation module formed of a plurality of convolution layers, and the first operation module may be trained to analyze input digital healthcare content and output sound characteristic information including sound essential parameter numerical information and sound pattern information for each digital healthcare content.

[0112] According to the above embodiment, the numerical parameter values ​​for Hz, beat, melody, repetition section pattern, and repetition section length, which are essential sources required for producing voice sounds included in digital healthcare content, can be extracted and output as essential parameter numerical information.

[0113] In addition, the voice sound included in the digital healthcare content can be divided into blocks with a preset length, and the sound pattern information can be output by analyzing the development of melody, BPM, and beat information items for each block.

[0114] According to one embodiment of the present invention, the second module of the digital content creation model analyzes sound characteristic information including input essential parameter numerical information and sound pattern information, and analyzes sound characteristic information of digital healthcare content for each disease to generate disease-specific treatment characteristic information including range information on the variation range of sound essential parameter numerical information and sound pattern information.

[0115] According to one embodiment of the present invention, by inputting treatment characteristic information into a digital content generation model, the sound characteristic information is used to change the sound essential parameter numerical information and sound pattern information within each range, thereby outputting digital healthcare content for each disease.

[0116] According to one embodiment of the present invention, a digital content creation model receives a plurality of pieces of treatment characteristic information, collects sound essential parameter numerical information and sound pattern information for each disease, and generates range information based on the numerical values ​​of each item existing for each item of the disease-specific sound characteristic information, thereby selecting a range within the range information as a change range.

[0117] According to one embodiment of the present invention, the digital content generation model may be a neural vocoder based on a normalizing flow, and may perform learning using sound characteristic information included in the treatment characteristic information so that a Gaussian distribution is derived from a sound dataset using a transformation function having reversibility.

[0118] According to one embodiment of the present invention, a normalizing flow-based neural vocoder constituting a digital content generation model can train voice sound data to derive a Gaussian distribution from a voice sound dataset using a reversible transformation function, and after the training is completed, a voice sound output can be synthesized from a sample of the Gaussian distribution using an inverse function of the transformation function to generate digital healthcare content for each disease.

[0119] According to one embodiment of the present invention, the learned digital content generation model can change the sound essential parameter numerical information and sound pattern information within each range by using the range information of each item information in a manner of synthesizing a sound output from a sample of a Gaussian distribution using the inverse function of the transformation function.

[0120] According to one embodiment of the present invention, generated digital treatment content and stored digital treatment content are divided into blocks of 15 seconds each, and a plurality of divided blocks are input into a feature extraction model to output sound characteristic information including sound essential parameter numerical information and sound pattern information for each block.

[0121] According to one embodiment of the invention, the sound characteristic information of each block of the generated digital therapeutic content can be compared with the sound characteristic information of multiple blocks of the stored digital therapeutic content, and copyright-related information can be generated based on whether there is similar sound characteristic information of each block within a preset standard.

[0122] The reason for comparing the sound characteristic information of blocks divided into 15-second units of the digital treatment content generated here and the stored digital treatment content is to determine the similarity of the section corresponding to the 4-bar section, which is generally the standard for similarity in determining whether or not there has been copyright infringement.

[0123] According to one embodiment of the present invention, when comparing block-by-block sound characteristic information, a relatively higher weight can be applied to essential parameter numerical information included in the sound characteristic information and to the similarity of the sound pattern degree among the sound pattern information.

[0124] According to another embodiment, when comparing block-by-block sound characteristic information, weights are set for essential parameter numerical information and sound pattern information included in the sound characteristic information, and each weight can be learned so that accuracy is relatively high.

[0125] The user's regular health checkup data is input into a disease prediction model to generate disease prediction information including the type of disease and the probability of occurrence (S30).

[0126] According to one embodiment of the present invention, regular health checkup data of a received user is input into a disease prediction model to output the probability of occurrence of each disease at each time period, and disease prediction information including the type and probability of occurrence of a disease having a probability greater than a preset reference value among the output probability of occurrence of each disease can be generated.

[0127] According to one embodiment of the present invention, a disease prediction model may be a model formed by a plurality of convolutional layers based on an artificial neural network, and may be trained to input a user's regular health checkup data and output disease prediction information including the type of disease with a possibility of developing and the probability of developing the disease.

[0128] According to one embodiment of the present invention, a disease prediction model may be composed of a first operation module and a second operation module formed by a plurality of convolutional layers.

[0129] According to the above embodiment, the first operation module can be trained to analyze the user's regular health checkup data, which is input data of the disease prediction model, and output a preset probability value of each disease at each time point after a preset period of time.

[0130] According to one embodiment of the present invention, a plurality of prior literature contributed on topics related to health checkup result items and onset for each predetermined disease is input into a meta-analysis model, and an effect size corresponding to the health checkup result items for each of the plurality of prior literature can be calculated.

[0131] According to one embodiment of the present invention, the inverse of the variance of the calculated effect size can be applied as a weight to the effect size of each prior literature to measure the disease onset impact score for each health checkup result item, and the disease onset probability value for each disease can be calculated based on the disease onset impact score calculated for each health checkup result item.

[0132] For example, the higher the disease incidence impact score calculated for each health checkup result item, the higher the probability of disease incidence can be calculated.

[0133] According to one embodiment of the present invention, the effect size corresponding to the topic related to the examination result item and the onset of disease for each of a plurality of prior literatures is calculated by calculating the odds ratio and confidence interval for each prior literature, and the effect size of the genetic variation for each prior literature on the target disease can be estimated based on the odds ratio and confidence interval.

[0134] According to one embodiment of the present invention, there may be various methods for extracting an effect size, and the type of effect size to be extracted may vary depending on the subject, and an effect size based on a standardized mean difference, an effect size based on a correlation coefficient, and an effect size based on an odds ratio, etc. may be used.

[0135] According to one embodiment of the present invention, in order to calculate the effect size based on the odds ratio, the effect size can be estimated based on the odds ratio (OR) and 95% confidence interval (CI), which are indicators of the magnitude of influence (magnitude of association) on the disease for each genetic variant. The odds ratios of each individual document can be combined to calculate the overall effect size (overall OR).

[0136] According to one embodiment of the present invention, a generic inverse variance estimation method can be used to measure a disease incidence influence score for each screening result item using the calculated effect size.

[0137] Inverse variance estimation is a method used to assign weights in meta-analysis, and the inverse of the variance of the estimated effect size can be used as the weight of each prior literature.

[0138] According to one embodiment of the present invention using the inverse variance estimation method, prior literature on studies with a large sample size will have a small variance, and the inverse of the variance will become large, so it can be used to give greater weight to prior literature on studies with a large sample size.

[0139] According to the above example, the natural logarithm of the odds ratio of each prior literature is taken. and produce, Calculate the standard error (SE) for , and weight the reciprocal of the square of the calculated standard error. It can be calculated as, and the weight of each preceding literature calculated as in mathematical formula 1 is multiplied by the odds ratio and the combined effect size (OR) is added up. pooled ) can be calculated.

[0140] According to one embodiment of the present invention, the disease prediction model may be a model trained to preprocess input regular health checkup data into individual time series data, analyze the numerical change pattern of each checkup item among the preprocessed data, and output the onset probability value of each disease according to the numerical change pattern.

[0141] According to one embodiment of the present invention, a disease prediction model may be trained to analyze regular health checkup data preprocessed as time series data to obtain a numerical change pattern for each checkup item, determine the similarity between the type of the numerical change pattern and a pre-stored disease onset pattern, and output a probability value of onset of each disease according to the similarity value when the similarity is greater than a preset value.

[0142] According to one embodiment of the present invention, a disease prediction model can calculate an incidence probability value for each disease based on a similarity value between a numerical change pattern for each examination item obtained by analyzing regular health examination data and a pre-stored disease incidence pattern, and a disease incidence impact score output by inputting multiple prior literature contributed on topics related to health examination result items and incidence for each pre-set disease into a meta-analysis model.

[0143] According to the above embodiment, the disease prediction model can be trained by setting weights for each of the similarity value and the disease invention influence score, and adjusting each weight so that the accuracy is relatively high.

[0144] According to one embodiment of the present invention, a disease prediction model may be a model trained to construct a recurrent neural network using LSTM (Long short-term memory), preprocess the input regular health checkup data into individual time series data, analyze the numerical change pattern of each checkup item among the preprocessed data, and output the onset probability value of each disease according to the numerical change pattern.

[0145] According to one embodiment of the present invention, a disease prediction model can be trained to analyze regular health checkup data preprocessed into time series data using LSTM (Long short-term memory) to obtain a numerical change pattern for each checkup item, determine the similarity between the type of numerical change pattern and a pre-stored disease onset pattern, and output a probability value of onset of each disease according to the similarity value when the similarity is higher than a preset value.

[0146] By inputting disease prediction information into a digital healthcare content selection model, at least one of the stored digital healthcare contents is selected as personal digital healthcare content to be provided to the user (S40).

[0147] According to one embodiment of the present invention, disease prediction information may be input into a digital healthcare content selection model to select at least one of the digital healthcare contents stored in the digital healthcare content management unit as personal digital healthcare contents to be provided to the user.

[0148] According to one embodiment of the present invention, a digital healthcare content selection model analyzes input disease prediction information to extract diseases with an incidence probability higher than a preset reference value, and then selects digital healthcare content having information on the effect of alleviation or treatment of the extracted disease in its header among the stored digital healthcare content as personal digital healthcare content.

[0149] Embodiments of the present invention are not implemented only through the devices and / or methods described above, and although embodiments of the present invention have been described in detail above, the scope of the present invention is not limited thereto, and various modifications and improvements made by those skilled in the art using the basic concept of the present invention defined in the following claims also fall within the scope of the present invention.

Claims

1. A health checkup data collection unit that directly inputs the user's regular health checkup data or receives it from a connected database; A digital healthcare content management department that stores a large amount of digital healthcare content, analyzes the stored digital healthcare content, and creates new digital healthcare content for each disease through a digital content creation model; A disease prediction information generation unit that inputs the regular health checkup data of the received user into a disease prediction model to output the probability of occurrence of each disease for each period, and generates disease prediction information including the type of disease and the probability of occurrence of the disease with a probability higher than a preset reference value among the probability of occurrence of each disease output; and A digital healthcare content provision system using health checkup data, including a personal digital healthcare content selection unit that inputs the above disease prediction information into a digital healthcare content selection model and selects at least one of the digital healthcare contents stored in the digital healthcare content management unit as personal digital healthcare content to be provided to the user.

2. In paragraph 1, the digital healthcare content management unit, A health checkup data collection unit that directly inputs the user's regular health checkup data or receives it from a connected database; A treatment characteristic information generation unit that stores a plurality of digital healthcare contents, inputs the stored digital healthcare contents into a characteristic extraction model to output sound characteristic information including sound essential parameter numerical information and sound pattern information for each digital healthcare contents, analyzes the sound characteristic information of each disease-specific digital healthcare contents, and generates disease-specific treatment characteristic information including sound essential parameter numerical information and range information on the variation range of sound pattern information included in the sound characteristic information; and A system for providing digital healthcare content using health checkup data, further comprising a digital content generation unit that inputs the above treatment characteristic information into a digital content generation model, changes the sound essential parameter numerical information and sound pattern information within each range using the range information of each item of sound characteristic information, and outputs digital healthcare content for each disease.

3. In the second paragraph, the digital content creation unit, The above digital content generation model is a neural vocoder based on a normalizing flow, and is a digital healthcare content provision system using health checkup data, characterized in that it performs learning using sound characteristic information included in treatment characteristic information so that a Gaussian distribution is derived from a sound dataset using a transformation function having reversibility.

4. In the third paragraph, the digital content creation unit, A digital healthcare content provision system using health checkup data, characterized in that the learned digital content generation model synthesizes sound output from samples of a Gaussian distribution using the inverse function of a transformation function, and changes sound essential parameter numerical information and sound pattern information within each range using the range information of each item information.

5. In paragraph 4, the digital content creation unit, A system for providing digital healthcare content using health checkup data, characterized in that the generated digital treatment content and the stored digital treatment content are divided into blocks of 15 seconds each, and the divided multiple blocks are input into the above-mentioned feature extraction model to output sound characteristic information including sound essential parameter numerical information and sound pattern information for each block.

6. In paragraph 5, the digital content creation unit, A digital healthcare content provision system using health checkup data, characterized in that the system compares the sound characteristic information of each block of the generated digital treatment content with the sound characteristic information of multiple blocks of stored digital treatment content, and generates copyright-related information based on whether there is similar sound characteristic information of each block within a preset standard.

7. In paragraph 1, the disease prediction information generation unit The above disease prediction model is a digital healthcare content provision system characterized in that it is a model trained to preprocess the input regular health checkup data into individual time series data, analyze the numerical change pattern of each checkup item among the preprocessed data, and output the onset probability value of each disease according to the numerical change pattern.

8. In paragraph 7, the disease prediction information generation unit The above disease prediction model constructs a recurrent neural network using LSTM (Long short-term memory), A digital healthcare content provision system characterized by a model trained to preprocess input regular health checkup data into individual time series data, analyze numerical change patterns for each checkup item among the preprocessed data, and output a probability of onset of each disease based on the numerical change patterns.

9. Step of directly entering the user's regular health checkup data or receiving it from a connected database; A step of storing a large amount of digital healthcare content, analyzing the stored digital healthcare content, and creating new digital healthcare content for each disease through a digital content creation model; A step of inputting the regular health checkup data of the received user into a disease prediction model to output the probability of occurrence of each disease for each period, and generating disease prediction information including the type and probability of occurrence of the disease having a probability higher than a preset reference value among the probability of occurrence of each disease output; and A method for providing digital healthcare content using health checkup data, comprising a step of inputting the above disease prediction information into a digital healthcare content selection model and selecting at least one of the stored digital healthcare contents as personal digital healthcare content to be provided to the user.

10. In the step of newly creating the digital healthcare content in paragraph 9, A step of directly entering the user's regular health check-up data or receiving it from a connected database; A step of storing a plurality of digital healthcare contents, inputting the stored digital healthcare contents into a feature extraction model to output sound characteristic information including sound essential parameter numerical information and sound pattern information for each digital healthcare contents, analyzing the sound characteristic information of the digital healthcare contents for each disease to generate disease-specific treatment characteristic information including range information on the variation range of sound essential parameter numerical information and sound pattern information included in the sound characteristic information; and A method for providing digital healthcare content using health checkup data, further comprising a step of inputting the above treatment characteristic information into a digital content creation model, and outputting digital healthcare content for each disease by changing sound essential parameter numerical information and sound pattern information within each range using the range information of each item of sound characteristic information.

11. In the step of outputting the digital healthcare content in paragraph 10, The above digital content generation model is a neural vocoder based on a normalizing flow, and is a digital healthcare content provision system using health checkup data, characterized in that it performs learning using sound characteristic information included in treatment characteristic information so that a Gaussian distribution is derived from a sound dataset using a transformation function having reversibility.

12. In the step of outputting the digital healthcare content in paragraph 11, A method for providing digital healthcare content using health checkup data, characterized in that the learned digital content generation model synthesizes sound output from samples of a Gaussian distribution using the inverse function of a transformation function, and changes sound essential parameter numerical information and sound pattern information within each range using the range information of information for each item.

13. In the step of outputting the digital healthcare content in paragraph 12, A method for providing digital healthcare content using health checkup data, characterized in that the generated digital treatment content and the stored digital treatment content are divided into blocks of 15 seconds each, and the divided multiple blocks are input into the above-mentioned feature extraction model to output sound characteristic information including sound essential parameter numerical information and sound pattern information for each block.

14. In the step of outputting the digital healthcare content in paragraph 13, A method for providing digital healthcare content using health checkup data, characterized in that the method compares the sound characteristic information of each block of the generated digital treatment content with the sound characteristic information of multiple blocks of stored digital treatment content, and generates copyright-related information based on whether there is similar sound characteristic information of each block within a preset standard.

15. In the step of generating the disease prediction information in paragraph 9, A method for providing digital healthcare content, characterized in that the disease prediction model is a model trained to preprocess the input regular health checkup data into individual time series data, analyze the numerical change pattern of each checkup item among the preprocessed data, and output the onset probability value of each disease according to the numerical change pattern.

16. In the 16th paragraph, the step of generating the disease prediction information comprises: The above disease prediction model constructs a recurrent neural network using LSTM (Long short-term memory), A method for providing digital healthcare content, characterized in that the model is trained to preprocess input regular health checkup data into individual time series data, analyze the numerical change pattern of each checkup item among the preprocessed data, and output the probability of occurrence of each disease according to the numerical change pattern.

Images