Method and apparatus for diagnosing lower urinary tract symptoms using uroflowmetry test graph

Abstract

A method for training a prediction model for a lower urinary tract symptom risk score is disclosed. A method for training a prediction model for a lower urinary tract symptom risk score according to an embodiment comprises the steps of: obtaining numerical data of a urinary system; obtaining graph data of the urinary system; obtaining composite graph data in which the numerical data is reflected in the graph data; obtaining a numerical feature map corresponding to the numerical data; obtaining a composite graph feature map corresponding to the composite graph data; and training the prediction model for the lower urinary tract symptom risk score on the basis of the numerical feature map and the composite graph feature map.

Description

Method and device for diagnosing lower urinary tract symptoms using a urinary flow measurement graph

[0001] The following examples relate to a method and device for diagnosing lower urinary tract symptoms, and more specifically, to an artificial intelligence technology that simultaneously analyzes numerical data obtained through a noninvasive simple uroflowmetry test and image data, which is a graph of the results of the uroflowmetry test, to produce a lower urinary tract symptoms (LUTS) prediction score, and relates to a method and device for diagnosing lower urinary tract symptoms using a graph of a uroflowmetry test.

[0002]

[0003] Lower urinary tract symptoms (LUTS) are a collective term for various symptoms related to the storage and emptying of urine, such as difficulty starting, residual urine, frequency, dribbling, straining to urinate, nocturia, urgency, and intermittent urination. Recently, the incidence and severity of LUTS have been increasing due to various factors, such as increased animal fat intake and social complexity, psychological stress, smoking, drinking, weight gain, lack of rest, and lack of exercise. As LUTS become more severe, activities become restricted and a constant state of anxiety and tension occurs, causing significant psychological stress in the patient. In addition to the need to go to the bathroom during sleep and the resulting sleep deprivation, physical fatigue also worsens, leading to various physical problems.

[0004] Lower urinary tract symptoms are diagnosed using a urodynamic study (UDS), which evaluates bladder function. Urodynamic studies are mainly performed to determine whether to perform prostate surgery, and to distinguish between patients with only detrusor underactivity (DUA), for which surgery is unlikely to be effective, and patients with bladder outlet obstruction (BOO), for which surgery is known to be highly effective.

[0005] However, conventional urodynamic testing involves inserting a pressure-measuring tube into the bladder and anus, slowly filling the bladder with saline solution to measure pressure, and then measuring bladder pressure while urinating. In other words, urodynamic testing, currently used to diagnose lower urinary tract symptoms, is not only uncomfortable and embarrassing for patients, but also carries the risk of infection due to the long-term insertion of a catheter, causing pain and shame to the patient.

[0006] Accordingly, there is a continuous demand for the development of diagnostic assistance methods for lower urinary tract symptoms that can alleviate pain and shame for the examinee and accurately determine the condition of the examinee's urinary system.

[0007]

[0008] A method for learning a lower urinary tract symptom risk score prediction model according to one embodiment may include: acquiring urinary system numerical data; acquiring urinary system graph data; acquiring composite graph data in which the numerical data is reflected in the graph data; acquiring a numerical feature map corresponding to the numerical data; acquiring a composite graph feature map corresponding to the composite graph data; and learning a lower urinary tract symptom risk score prediction model based on the numerical feature map and the composite graph feature map.

[0009] The step of obtaining the above complex graph data may include a step of determining a pixel value of the graph data based on the numerical data.

[0010] The step of obtaining the above complex graph data may include the step of extracting a plurality of display areas from the graph data; the step of mapping the numerical data to the plurality of display areas; and the step of obtaining the complex graph data in which the numerical data is reflected in each of the plurality of mapped display areas.

[0011] The step of extracting the plurality of display areas may include a step of extracting a contour from the graph data; and a step of extracting the plurality of display areas based on the contour.

[0012] The step of extracting the plurality of display areas may include the step of extracting a contour from the graph data; and the step of removing a tick mark from the graph data based on the contour.

[0013] The step of learning the risk score prediction model may include a step of generating a dual feature map by concatenating the numerical feature map and the composite graph feature map; and a step of learning the risk score prediction model based on the dual feature map.

[0014] The above urinary system numerical data may include at least one of maximal flow rate, average flow rate, voiding time, flow time, time to maximal flow, voided volume (VV), postvoid residual volume (PVR), bladder filling volume (BFV), and flow rate at t seconds, and the graph data may include at least one of voided volume over time and voiding rate over time.

[0015] The step of obtaining the above complex graph data may include: extracting a contour from the graph data; extracting a first region and a second region from the graph data based on the contour; determining a first pixel value of the first region based on the maximum urine output and the urine output; and determining a second pixel value of the second region based on the maximum post-urination residual amount and the post-urination residual amount.

[0016] An electronic device according to one embodiment comprises a memory storing at least one command; and by executing the command stored in the memory, obtains urinary system numerical data,

[0017] The method may include a processor that obtains urological graph data, obtains composite graph data in which the numerical data is reflected in the graph data, obtains a numerical feature map corresponding to the numerical data, obtains a composite graph feature map corresponding to the composite graph data, and inputs the numerical feature map and the composite graph feature into a prediction model to output a lower urinary tract symptom risk score.

[0018] The processor can determine a pixel value of the graph data based on the numerical data.

[0019] The processor can extract a plurality of display areas from the graph data, map the numerical data to the plurality of display areas, and obtain the composite graph data in which the numerical data is reflected in each of the plurality of mapped display areas.

[0020] The above processor can extract a contour from the graph data and, based on the contour, extract the plurality of display areas.

[0021] The processor may extract a contour from the graph data; and based on the contour, may remove tick marks from the graph data.

[0022] The above processor can generate a dual feature map by concatenating the numerical feature map and the composite graph feature map, and input the dual feature map into the prediction model to output the lower urinary tract symptom risk score.

[0023] The above urinary system numerical data may include at least one of maximal flow rate, average flow rate, voiding time, flow time, time to maximal flow, voided volume (VV), postvoid residual volume (PVR), bladder filling volume (BFV), and flow rate at t seconds, and the graph data may include at least one of voided volume over time and voiding rate over time.

[0024] The processor may include a step of extracting a contour from the graph data; extracting a first region and a second region from the graph data based on the contour; determining a first pixel value of the first region based on the maximum amount of urine voided and the amount of urine voided; and determining a second pixel value of the second region based on the maximum amount of urine voided and the amount of urine voided.

[0025]

[0026] FIG. 1 is a diagram illustrating a lower urinary tract symptom risk score prediction system according to one embodiment.

[0027] Figures 2 and 3 are diagrams for explaining the operation of a prediction model according to one embodiment.

[0028] FIG. 4 is a diagram illustrating a method for generating composite graph data according to one embodiment.

[0029] FIG. 5 is a diagram illustrating an example of a method for generating composite graph data according to one embodiment.

[0030] Figure 6 is a drawing for explaining a contour extraction method according to one embodiment.

[0031] FIG. 7 is a diagram illustrating an example of a process for generating composite graph data according to one embodiment.

[0032] FIGS. 8 to 11 are diagrams illustrating examples of composite graph data according to one embodiment.

[0033]

[0034] The specific structural or functional descriptions disclosed in this specification are merely illustrative for the purpose of explaining embodiments according to technical concepts, and the actually implemented form may have various different appearances and is not limited to the embodiments described in this specification.

[0035] While terms like "first" and "second" may be used to describe various components, these terms should be understood solely to distinguish one component from another. For example, a "first" component may be referred to as a "second" component, and similarly, a "second" component may also be referred to as a "first" component.

[0036] When a component is referred to as being "connected" or "connected" to another component, it should be understood that it may be directly connected or connected to that other component, but that there may be other components in between. Conversely, when a component is referred to as being "directly connected" or "directly connected" to another component, it should be understood that there are no other components in between. Expressions that describe relationships between components, such as "between" and "immediately between," or "adjacent to" and "directly adjacent to," should be interpreted similarly.

[0037] Singular expressions include plural expressions unless the context clearly dictates otherwise. In this specification, the terms "comprises" or "has" should be understood to indicate the presence of a feature, number, step, operation, component, part, or combination thereof, but not to exclude the possibility of the presence or addition of one or more other features, numbers, steps, operations, components, parts, or combinations thereof.

[0038] Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by a person of ordinary skill in the art. Terms defined in commonly used dictionaries should be interpreted to have 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 explicitly defined herein.

[0039] The embodiments can be implemented in various forms of products, such as personal computers, laptop computers, tablet computers, smartphones, televisions, smart home appliances, intelligent vehicles, kiosks, and wearable devices. The embodiments are described in detail below with reference to the attached drawings. Like reference numerals in each drawing represent like elements.

[0040] FIG. 1 is a diagram illustrating a lower urinary tract symptom risk score prediction system according to one embodiment.

[0041] Referring to FIG. 1, a lower urinary tract symptom risk score prediction system according to one embodiment can predict a lower urinary tract symptom risk score based on urinary system numerical data and urinary system graph data.

[0042] Numerical data according to one embodiment is data obtained from a subject for diagnosing lower urinary tract symptoms, and may include at least one of maximal flow rate, average flow rate, voiding time, flow time, time to maximal flow, voided volume (VV), postvoid residual volume (PVR), bladder filling volume (BFV), and flow rate at t seconds. However, the urinary system numerical data is not limited to the examples described above, and may further include various types of numerical data obtained from a subject for diagnosing lower urinary tract symptoms.

[0043] In one embodiment, the urinary system graph data includes voided volume over time or voiding speed over time. In one embodiment, the urinary system graph data may include a Uroflowmetry (UFM) graph. The UFM graph may be a graph obtained through a uroflowmetry test that objectively confirms the state of urination through the discharged urine. Through the uroflowmetry test, when the subject urinates in a test toilet, the measured urine discharge speed may be displayed as a graph. The UFM graph may be referred to as a UFM curve.

[0044] According to one embodiment, a lower urinary tract symptom (ULS) risk score prediction system can generate composite graph data by inserting numerical data into graph data, which is image data. The lower urinary tract symptom (ULS) risk score prediction system can improve performance by inserting key numerical information into a simple graph image. For example, there are blank spaces above and below the UFM graph, and the lower urinary tract symptom (ULS) risk score prediction system can utilize these blank spaces to insert information on PVR and VV, which are key features of numerical data that cannot be provided in the graph. More detailed information on generating composite graph data is described below with reference to FIGS. 4 to 11.

[0045] According to one embodiment, a lower urinary tract symptom risk score prediction system may include a deep learning-based prediction model. Artificial intelligence (AI) algorithms, including deep learning, input data into an artificial neural network (ANN), learn output data through operations such as convolution, and extract features using the learned ANN. An ANN may refer to a computational architecture that models a biological brain. Within an ANN, nodes corresponding to neurons in the brain are interconnected and operate collectively to process input data.

[0046] Examples of various types of neural networks include, but are not limited to, convolutional neural networks (CNNs), recurrent neural networks (RNNs), deep belief networks (DBNs), and restricted Boltzman machines (RBMs). In a feed-forward neural network, neurons in the neural network have connections (links) with other neurons. These connections can extend through the neural network in one direction, for example, in the forward direction. An artificial neural network can be a deep neural network with one or more layers.

[0047] According to one embodiment, a predictive model can receive numerical data and complex graph data to predict a lower urinary tract symptom risk score. More specifically, the predictive model can be trained to predict a lower urinary tract symptom risk score by receiving numerical data and complex graph data through a learning device. According to one embodiment, the learning device corresponds to a computing device having various processing functions, such as functions for generating a neural network of a predictive model, training (or learning) a neural network, or retraining a neural network. For example, the learning device can be implemented as various types of devices, such as a personal computer (PC), a server device, or a mobile device.

[0048] A learning device according to one embodiment can generate one or more trained neural networks by repeatedly training a given initial neural network. Generating one or more trained neural networks may mean determining neural network parameters. Here, the parameters may include various types of data input / output to the neural network, such as input / output activations, weights, and biases of the neural network. As the repeated training of the neural network progresses, the parameters of the neural network may be adjusted to compute more accurate outputs for given inputs.

[0049] A learning device according to one embodiment may transmit one or more trained neural networks to an inference device. The inference device may be included in a mobile device, an embedded device, or the like. The inference device according to one embodiment may be an electronic device that includes at least one of a processor, memory, an input / output (I / O) interface, a display, a communication interface, or a sensor as dedicated hardware for operating a neural network.

[0050] An inference device according to one embodiment may be a concept that includes all digital devices having a memory means, a microprocessor, and a computing capability, such as a tablet PC, a smartphone, a personal computer (e.g., a laptop computer, etc.), an artificial intelligence speaker, a smart TV, a mobile phone, a navigation system, a web pad, a PDA, a workstation, etc.

[0051] According to one embodiment, the inference device can operate one or more trained neural networks as-is, or operate one or more processed (e.g., quantized) neural networks from the trained neural networks. The inference device operating the processed neural networks can be implemented in an independent device separate from the learning device. However, the inference device is not limited thereto, and can also be implemented within the same device as the learning device.

[0052] According to one embodiment, an inference device inputs a subject's numerical data and complex graph data into a trained prediction model, thereby predicting the subject's risk score. The operation of the prediction model according to one embodiment is described in detail with reference to FIGS. 2 and 3 below.

[0053] Figures 2 and 3 are diagrams illustrating the operation of a prediction model according to one embodiment. The contents described with reference to Figure 1 can be equally applied to Figure 2, and any overlapping contents may be omitted.

[0054] Referring to FIG. 2, a lower urinary tract symptom risk score prediction system according to one embodiment can obtain a numeric feature map from numeric data and a composite graph feature map from composite graph data.

[0055] Feature maps can be the output of each layer of a neural network for input data (e.g., numerical data or complex graph data). Each feature map represents the detection of specific features of the input data through the filter of that layer. Multiple feature maps can be generated for each layer, and each feature map can represent various features of that layer. For example, feature maps in early layers may represent simple features of an image (e.g., lines, edges, etc.), while feature maps in deeper layers may represent more abstract, high-level features (e.g., shape of objects, patterns, etc.). Feature maps are automatically generated during the training process of a neural network, and as the filter weights are adjusted during the training process, they can be optimized to detect important features in the input data.

[0056] In other words, the lower urinary tract symptom risk score prediction system can obtain a numeric feature map by inputting numeric data into a neural network corresponding to the numeric data (numerical neural network), and can obtain a complex graph feature map by inputting complex graph data into a neural network corresponding to image data (graph neural network).

[0057] Furthermore, the lower urinary tract symptom risk score prediction system according to one embodiment can generate a dual feature map based on a numerical feature map and a composite graph feature map. For example, the lower urinary tract symptom risk score prediction system can generate a dual feature map by concatenating a numerical feature map and a composite graph feature map.

[0058] More specifically, referring to FIG. 3, the lower urinary tract symptom risk score prediction system can input complex graph data into a graph neural network and obtain a complex graph feature map having a shape of 512*512 generated through layers of the graph neural network, and input numerical data into a numerical neural network and obtain a numerical feature map having a shape of 1*512 generated through layers of the graph neural network. Furthermore, the lower urinary tract symptom risk score prediction system can obtain a dual feature map having a shape of 513*512 by connecting the complex graph feature map having a shape of 512*512 and the numerical feature map having a shape of 1*512.

[0059] Referring back to FIG. 2, the lower urinary tract symptom risk score prediction system can predict a risk score by inputting the dual feature map into a risk score prediction neural network (e.g., CNN).

[0060] The prediction model described above with reference to FIG. 1 may be a concept including a numerical neural network, a graph neural network, and a risk score prediction neural network described in FIG. 2. Therefore, training the prediction model may mean training the numerical neural network, the graph neural network, and the risk score prediction neural network. The numerical neural network, the graph neural network, and the risk score prediction neural network may be trained separately, or may be trained end-to-end at once. For example, when a labeled data set (e.g., a data set consisting of (numeric data, graph data, and correct risk score)) for training is provided, all neural networks included in the prediction model may be trained at once so that the difference between the predicted risk score obtained by inputting the numerical data and graph data into the prediction model and the correct risk score is minimized.

[0061] FIG. 4 is a diagram illustrating a method for generating composite graph data according to one embodiment.

[0062] Referring to FIG. 4, the lower urinary tract symptom risk score prediction system according to one embodiment can generate composite graph data that can provide additional information not available in graph data. For example, the lower urinary tract symptom risk score prediction system can generate composite graph data by determining pixel values ​​of the graph data based on numerical data.

[0063] The lower urinary tract symptom risk score prediction system can extract multiple display areas from graph data, map numerical data to the multiple display areas, and obtain composite graph data in which the numerical data is reflected in each of the mapped multiple display areas. More specifically, the lower urinary tract symptom risk score prediction system can extract contours from the graph data and, based on the contours, extract multiple display areas.

[0064] For example, in the UFM graph, there are blank spaces above and below, and the lower urinary tract symptom risk score prediction system can utilize these blank spaces to insert key features of numerical data (e.g., information on PVR and VV) that are not provided in the graph.

[0065] In the image, RGB values ​​are red, green, and blue, each with a value from 0 to 255. By assigning values ​​proportional to PVR and VV, PVR and VV can be expressed as a color difference.

[0066] For example, at the bottom of the UFM graph, the VV value is multiplied by the value obtained by dividing the maximum RGB value of 255 by the maximum VV value, and then substituted into one of the RGB values ​​(for example, the G value). This means that the larger the VV value, the brighter the green color. At the top of the UFM graph, the PVR value is multiplied by the value obtained by dividing the maximum RGB value of 255 by the maximum VV value, and then substituted into another RGB value (for example, the B value). This means that the smaller the PVR value, the brighter the blue color.

[0067] The curve of the composite graph data generated in this way can represent the measured urine excretion rate, and the green brightness in the first region (e.g., below the curve) can represent the VV value, and the blue brightness in the second region (e.g., above the curve) can represent the PVR value.

[0068] However, the content described with reference to FIG. 4 is merely an example, and the method for generating composite graph data may not be limited thereto. For example, the numerical data displayed in the composite graph data may include various data types other than VV and PVR, and the method for determining RGB values ​​is also not limited to the above-described example.

[0069] FIG. 5 is a diagram illustrating an example of a method for generating composite graph data according to one embodiment.

[0070] Referring to FIG. 5, the lower urinary tract symptom risk score prediction system can delete unit scales by applying a threshold to the RGB values ​​of graph data (e.g., UFM graph). The lower urinary tract symptom risk score prediction system can extract a contour from the graph data, and extract a first region and a second region from the graph data based on the contour. The lower urinary tract symptom risk score prediction system can fill the first region (the lower area of ​​the graph) with an RGB color corresponding to a void volume (VV) value, and can fill the second region (the upper area of ​​the graph) with an RGB color corresponding to a post-void residual volume (PVR) value.

[0071] Figure 6 is a drawing for explaining a contour extraction method according to one embodiment.

[0072] In one embodiment, a lower urinary tract symptom risk score prediction system can blur the scale of a UFM graph image using techniques such as dilation, erosion, and NLmeansdenosing, highlighting the curve, and recognizing only the curve as a contour using the findcontour technique. However, since the type of test strip (UFM graph) varies depending on the type of uroflowmeter, a combination of techniques can be used.

[0073] For example, referring to FIG. 6, a lower urinary tract symptom risk score prediction system according to one embodiment can detect a contour by blurring the scale included in a UFM graph image through an erosion technique and emphasizing a curve through a dilation technique.

[0074] FIG. 7 is a diagram illustrating an example of a process for generating composite graph data according to one embodiment.

[0075] Referring to FIG. 7, a lower urinary tract symptom risk score prediction system according to one embodiment can extract a contour from a UFM graph image through the method described above with reference to FIG. 6, extract a plurality of display areas based on the extracted contour, and then obtain composite graph data in which numerical data is reflected as RGB values.

[0076] FIGS. 8 to 11 are diagrams illustrating examples of composite graph data according to one embodiment.

[0077] Referring to FIG. 8, the lower urinary tract symptom risk score prediction system according to one embodiment can insert backgrounds of blue and green RGB 1 values ​​for color contrast. The lower urinary tract symptom risk score prediction system can insert RGB 1 values ​​into the graph in various ways (e.g., as a background or in a specific section) to recognize them as contrasting reference values. Through this, the learning of colors (or RGB values) in the converted input data (composite graph data) can be strengthened.

[0078] Referring to FIG. 9, the lower urinary tract symptom risk score prediction system according to one embodiment can emphasize the original data size of the cropped input data by pasting a UFM graph image onto the background image. For example, the lower urinary tract symptom risk score prediction system according to one embodiment can convert only the converted original data portion by coloring it, and the remaining section can be blanked out with a reference value such as RGB 1 or RGB 255.

[0079] Referring to FIG. 10, the lower urinary tract symptom risk score prediction system according to one embodiment can reflect the flow rate value at each point. This is a method for emphasizing the numerical value (flow rate) of each point of the UFM graph. Below the UFM graph, the flow rate value at each point of the curve is multiplied by the value obtained by dividing the maximum RGB value of 255 by the maximum flow rate value, and then substituted into one of the RGB values ​​(e.g., the G value). As a result, the green color becomes brighter as the flow rate value increases. As a result, the section under the curve appears to have a gradient due to the color difference at each point. The maximum flow rate can be defined as 25.5 mL / sec.

[0080] Referring to FIG. 11, the lower urinary tract symptom risk score prediction system according to one embodiment may highlight the curve of the UFM graph by making it bold.

[0081] The embodiments described above may be implemented using hardware components, software components, and / or a combination of hardware components and software components. For example, the devices, methods, and components described in the embodiments may be implemented using a general-purpose computer or a special-purpose computer, such as, for example, a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable gate array (FPGA), a programmable logic unit (PLU), a microprocessor, or any other device capable of executing instructions and responding to them. The processing device may execute an operating system (OS) and software applications running on the operating system. Furthermore, the processing device may access, store, manipulate, process, and generate data in response to the execution of the software. For ease of understanding, the processing device is sometimes described as being used alone; however, one of ordinary skill in the art will recognize that the processing device may include multiple processing elements and / or multiple types of processing elements. For example, a processing unit may include multiple processors, or a processor and a controller. Other processing configurations, such as parallel processors, are also possible.

[0082] Software may include computer programs, codes, instructions, or any combination thereof, which may configure a processing device to perform a desired operation or, independently or collectively, command the processing device. The software and / or data may be permanently or temporarily embodied in any type of machine, component, physical device, virtual equipment, computer storage media or devices, or transmitted signal waves, for interpretation by the processing device or for providing instructions or data to the processing device. The software may also be distributed over networked computer systems and stored or executed in a distributed manner. The software and data may be stored on a computer-readable recording medium.

[0083] The method according to the embodiment may be implemented in the form of program commands that can be executed through various computer means and recorded on a computer-readable medium. The computer-readable medium may include program commands, data files, data structures, etc., alone or in combination. The program commands recorded on the medium may be those specially designed and configured for the embodiment or may be those known and available to those skilled in the art of computer software. Examples of the computer-readable recording medium include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical media such as CD-ROMs and DVDs, magneto-optical media such as floptical disks, and hardware devices specially configured to store and execute program commands such as ROMs, RAMs, and flash memories. Examples of program commands include not only machine language codes such as those generated by a compiler, but also high-level language codes that can be executed by a computer using an interpreter, etc.

[0084] Although the embodiments described above have been described with limited drawings, those skilled in the art will appreciate that various technical modifications and variations can be applied based on the above. For example, appropriate results can still be achieved even if the described techniques are performed in a different order than described, and / or components of the described systems, structures, devices, circuits, etc. are combined or combined in a different manner than described, or are replaced or substituted with other components or equivalents.

[0085] Therefore, other implementations, other embodiments, and equivalents to the claims also fall within the scope of the claims described below.

Claims

1. Step of acquiring urinary system numerical data; Step of acquiring urinary system graph data; A step of obtaining composite graph data in which the numerical data is reflected in the graph data; A step of obtaining a numerical feature map corresponding to the above numerical data; A step of obtaining a composite graph feature map corresponding to the composite graph data; and A step of learning a lower urinary tract symptom risk score prediction model based on the above numerical feature map and the above composite graph feature map. A method for learning a prediction model for lower urinary tract symptom risk score including .

2. In paragraph 1, The step of obtaining the above complex graph data is A step of determining a pixel value of the graph data based on the numerical data. A method for learning a lower urinary tract symptom risk score prediction model, including:

3. In paragraph 1, The step of obtaining the above complex graph data is A step of extracting multiple display areas from the above graph data; a step of mapping the numerical data to the plurality of display areas; and A step of obtaining the composite graph data in which the numerical data is reflected in each of the plurality of mapped display areas. A method for learning a lower urinary tract symptom risk score prediction model, including:

4. In paragraph 3, The step of extracting the above multiple display areas is A step of extracting a contour from the above graph data; and A step of extracting the plurality of display areas based on the above contour A method for learning a lower urinary tract symptom risk score prediction model, including:

5. In paragraph 3, The step of extracting the above multiple display areas is A step of extracting a contour from the above graph data; and A step of removing tick marks from graph data based on the above contour. A method for learning a lower urinary tract symptom risk score prediction model, including:

6. In paragraph 1, The step of learning the above risk score prediction model is A step of generating a dual feature map by concatenating the above numerical feature map and the above composite graph feature map; and A step of learning the risk score prediction model based on the above dual feature map. A method for learning a lower urinary tract symptom risk score prediction model, including:

7. In paragraph 1, The above urinary system numerical data are Includes at least one of maximal flow rate, average flow rate, voiding time, flow time, time to maximal flow, voided volume (VV), postvoid residual volume (PVR), bladder filling volume (BFV), and flow rate at t seconds. The above graph data is A method for learning a lower urinary tract symptom risk score prediction model, comprising at least one of time-dependent urine volume and time-dependent urine flow rate.

8. In paragraph 7, The step of obtaining the above complex graph data is A step of extracting a contour from the above graph data; A step of extracting a first region and a second region from the graph data based on the contour; A step of determining a first pixel value of the first area based on the maximum urine volume and the urine volume; and A step of determining a second pixel value of the second area based on the maximum post-urination residual amount and the post-urination residual amount. A method for learning a lower urinary tract symptom risk score prediction model, including:

9. A computer program stored on a medium to execute any one of the methods of claims 1 to 8 in combination with hardware.

10. Memory for storing at least one instruction; and By executing the command stored in the above memory, Obtain urinary tract numerical data, Obtain urinary system graph data, Obtain composite graph data in which the numerical data is reflected in the above graph data, Obtain a numerical feature map corresponding to the above numerical data, Obtain a composite graph feature map corresponding to the above composite graph data, A processor that inputs the above numerical feature map and the above composite graph feature into a prediction model and outputs a lower urinary tract symptom risk score. Electronic devices containing.

11. In paragraph 10, The above processor An electronic device that determines pixel values ​​of the graph data based on the numerical data.

12. In paragraph 10, The above processor Extract multiple display areas from the above graph data, Mapping the above numerical data to the plurality of display areas, An electronic device that obtains the composite graph data in which the numerical data is reflected in each of the plurality of mapped display areas.

13. In paragraph 12, The above processor Extract a contour from the above graph data, An electronic device that extracts the plurality of display areas based on the contour.

14. In paragraph 12, The above processor A step of extracting a contour from the above graph data; and An electronic device for removing tick marks from graph data based on the above contour.

15. In paragraph 10, The above processor Concatenate the above numerical feature map and the above composite graph feature map to create a dual feature map, An electronic device that inputs the above dual feature map into the above prediction model and outputs the lower urinary tract symptom risk score.

16. In paragraph 10, The above urinary system numerical data are Includes at least one of maximal flow rate, average flow rate, voiding time, flow time, time to maximal flow, voided volume (VV), postvoid residual volume (PVR), bladder filling volume (BFV), and flow rate at t seconds. The above graph data is An electronic device comprising at least one of a time-dependent urine volume and a time-dependent urine flow rate.

17. In paragraph 10, The above processor A step of extracting a contour from the above graph data; Based on the above contour, a first region and a second region are extracted from the graph data, Determine the maximum urine volume and the first pixel value of the first area based on the urine volume, An electronic device that determines a second pixel value of the second area based on the maximum post-urination residual amount and the post-urination residual amount.

Images