Method for detecting abnormality in central nervous system

Abstract

The present invention relates to a method for detecting an abnormality in the central nervous system, wherein the presence of an abnormality in the central nervous system is determined on the basis of probabilities generated by at least two models. The present invention relates to a method for detecting an abnormality in the central nervous system, wherein the presence of an abnormality in the central nervous system is determined on the basis of probabilities generated by at least two models. The method comprises the steps of: obtaining a first signal reflecting electrical activity of a heart; generating a second signal reflecting at least some characteristics of the first signal; determining at least two first probabilities of detecting an abnormality in the central nervous system from the second signal by using at least two models, respectively; and calculating a second probability of detecting an abnormality in the central nervous system on the basis of the at least two first probabilities of detecting an abnormality in the central nervous system.

Description

Methods for detecting central nervous system abnormalities

[0001] The present invention relates to a method for detecting central nervous system abnormalities, which determines whether there is an abnormality in the central nervous system based on probabilities generated by at least two models.

[0002] An electrocardiogram (ECG) measuring device, commonly measured from a subject's body, is a device that records the tiny electrical currents generated within the heart and is used to diagnose various types of heart disease.

[0003] When the heart muscle contracts and relaxes, an action potential is generated, which causes an electric current to spread from the heart to the entire body. This electric current creates a potential difference depending on the location of the body. This potential difference can be detected and recorded through a surface electrode attached to the skin of the human body.

[0004] Electrocardiogram measuring devices are used to check for abnormalities in the heart, and are also used to diagnose cardiovascular diseases such as angina, myocardial infarction, and arrhythmia, making them very important medical devices.

[0005] Cerebral T waves, as seen on known electrocardiograms, are known to be caused by rapid increases in intracranial pressure following cerebral hemorrhage or cerebral infarction, and brain-derived proteins flowing into the bloodstream following destruction of brain tissue and the blood-brain barrier, as well as changes in sympathetic and parasympathetic nerves. Cerebral T waves can also appear in severe central nervous system damage or lesions.

[0006] The present invention aims to detect abnormalities in a patient's central nervous system based on an electrocardiogram.

[0007] In particular, the present invention aims to calculate the probability of abnormality in the central nervous system using a rule-based model and an artificial neural network-based model, respectively, and to determine a more accurate probability of abnormality based on the probability of abnormality.

[0008] In addition, the present invention provides the user with the probability of abnormality in the central nervous system calculated by each model, thereby providing a more specific basis for the user's judgment.

[0009] A method for detecting a central nervous system abnormality, which determines whether there is an abnormality in the central nervous system based on probabilities generated by at least two or more models according to one embodiment of the present invention, may include the steps of: obtaining a first signal reflecting electrical activity of the heart; generating a second signal reflecting at least some characteristics of the first signal; determining at least two first central nervous system abnormality probabilities from the second signal using each of at least two or more models; and calculating a second central nervous system abnormality probability based on the at least two first central nervous system abnormality probabilities.

[0010] The step of generating the second signal may include a step of generating the second signal based on one or more of a first component of the first signal of less than 10 Hz, a second component of the first signal of 10 Hz or more and less than 150 Hz, and a third component of the first signal of 150 Hz or more.

[0011] The step of generating the second signal may include a step of generating a 2-1 signal corresponding to a reference point in time from the first signal, and at least one 2-2 signal corresponding to at least one past point in time based on the reference point in time.

[0012] The at least two models may include a first model that determines whether there is an abnormality in the central nervous system based on a value in a portion of the second signal and a predetermined threshold value, and the step of determining the first central nervous system abnormality probability may include: a step of extracting at least one judgment target segment from the second signal by considering the periodicity of electrical activity of the heart; a step of extracting a section corresponding to a T wave from the judgment target segment; and a step of determining that the first central nervous system abnormality probability for the first model is equal to or greater than a predetermined threshold probability when the minimum value of the T wave in the section corresponding to the T wave is less than the predetermined threshold value.

[0013] The at least two models may include a second model that determines whether there is an abnormality in the central nervous system based on the similarity between a plurality of electrical activity signals of the heart acquired when there is an abnormality in the central nervous system and the second signal, and the step of determining the first central nervous system abnormality probability may include the step of calculating the similarity between a plurality of electrical activity signals of the heart acquired when there is an abnormality in the central nervous system and the second signal; and the step of determining the first central nervous system abnormality probability for the second model based on the similarity.

[0014] The at least two models may include a third model which is an artificial neural network that has learned a correlation between an image corresponding to an electrical activity signal of the heart and a probability of a central nervous system abnormality, and the step of determining the probability of a first central nervous system abnormality may include a step of generating an input image reflecting the characteristics of the second signal, wherein the second signal is a signal including a second-first signal corresponding to a reference point in time of the first signal; and a step of determining the probability of a first central nervous system abnormality for the third model by inputting the input image into the third model.

[0015] The at least two models may include a fourth model which is an artificial neural network that has learned a correlation between an image corresponding to an electrical activity signal of the heart at each of at least two points in time and a probability of a central nervous system abnormality, and the step of determining the first central nervous system abnormality probability may include a step of generating one or more input images reflecting characteristics of each of the second-1 signal and at least one second-2 signal, wherein the second-1 signal is a signal including a signal corresponding to a reference point in time of the first signal, and the second-2 signal is a signal including a signal corresponding to each of at least one past point in time based on the reference point in time; and a step of determining the first central nervous system abnormality probability for the fourth model by inputting the one or more input images into the fourth model.

[0016] The step of calculating the second central nervous system abnormality probability may include a step of calculating the second central nervous system abnormality probability based at least in part on at least one of the first central nervous system abnormality probability for the first model, the first central nervous system abnormality probability for the second model, the first central nervous system abnormality probability for the third model, and the first central nervous system abnormality probability for the fourth model.

[0017] A method for detecting a central nervous system abnormality according to one embodiment of the present invention may further include, after the step of calculating the second central nervous system abnormality probability, a step of providing at least two first central nervous system abnormality probabilities and the second central nervous system abnormality probability.

[0018] At this time, the step provided above can provide the probability of a first central nervous system abnormality for each model.

[0019] According to the present invention, it is possible to detect abnormalities in a patient's central nervous system based on an electrocardiogram.

[0020] Additionally, the probability of abnormality in the central nervous system can be calculated using a rule-based model and an artificial neural network-based model, respectively, and a more accurate probability of abnormality can be determined based on this.

[0021] Additionally, by providing users with the probability of abnormalities in the central nervous system produced by each model, we can provide more specific grounds for users' judgment.

[0022] FIG. 1 is a schematic diagram illustrating the configuration of a central nervous system abnormality detection system according to one embodiment of the present invention.

[0023] FIG. 2 is a diagram schematically illustrating the configuration of a server (100) according to one embodiment of the present invention.

[0024] FIG. 3 is a diagram schematically illustrating the configuration of a user terminal (200) according to one embodiment of the present invention.

[0025] Figure 4 is a diagram illustrating an exemplary first signal (510).

[0026] FIG. 5 is a diagram illustrating a process in which a server (100) according to one embodiment of the present invention generates a second signal (531, 532, 533) from a portion of a signal (510-1) of a first signal (510) using at least one frequency filter (521, 522, 523).

[0027] FIG. 6 is a drawing for explaining a process in which a server (100) according to one embodiment of the present invention generates second signals (511, 512, 513) by varying the extraction time from a first signal (510).

[0028] FIG. 7 is a diagram illustrating a series of processes in which a server (100) according to one embodiment of the present invention inputs a second signal (540) to each of at least two models (610, 620, 630, 640) to ultimately calculate a second central nervous system abnormality probability (650).

[0029] FIG. 8 is a drawing for explaining a first model (610) according to one embodiment of the present invention.

[0030] FIG. 9 is a drawing for explaining a second model (620) according to one embodiment of the present invention.

[0031] FIG. 10 is a drawing for explaining a third model (630) according to one embodiment of the present invention.

[0032] FIG. 11 is a drawing for explaining a fourth model (640) according to one embodiment of the present invention.

[0033] FIG. 12 is a drawing showing an exemplary screen (1300) showing the probability of an abnormality in the central nervous system on a user terminal (200).

[0034] FIG. 13 is a flowchart for explaining a central nervous system abnormality detection method performed by a server (100) according to one embodiment of the present invention.

[0035] The present invention is capable of various modifications and embodiments. Specific embodiments are illustrated in the drawings and described in detail in the detailed description. The effects and features of the present invention, as well as the methods for achieving them, will become clearer with reference to the embodiments described in detail below, along with the drawings. However, the present invention is not limited to the embodiments disclosed below and can be implemented in various forms.

[0036] Hereinafter, embodiments of the present invention will be described in detail with reference to the attached drawings. When describing with reference to the drawings, identical or corresponding components are given the same drawing reference numerals, and redundant descriptions thereof will be omitted.

[0037] In the following examples, terms such as first, second, etc. are not used in a limiting sense, but are used for the purpose of distinguishing one component from another. In the following examples, singular expressions include plural expressions unless the context clearly indicates otherwise. In the following examples, terms such as include or have mean that a feature or component described in the specification exists, and do not exclude in advance the possibility that one or more other features or components may be added. In the drawings, the sizes of components may be exaggerated or reduced for convenience of explanation. For example, the sizes and shapes of each component shown in the drawings have been arbitrarily shown for convenience of explanation, and therefore, the present invention is not necessarily limited to what is shown.

[0038] FIG. 1 is a schematic diagram illustrating the configuration of a central nervous system abnormality detection system according to one embodiment of the present invention.

[0039] A central nervous system abnormality detection system according to one embodiment of the present invention can determine whether there is an abnormality in the central nervous system based on probabilities generated by at least two models. For example, a central nervous system abnormality detection system according to one embodiment of the present invention can determine whether there is an abnormality in the central nervous system based on a first model that determines whether there is an abnormality in the central nervous system based on a value in a portion of a judgment target signal and a predetermined threshold value, a second model that determines whether there is an abnormality in the central nervous system based on the similarity between a plurality of electrical activity signals of the heart acquired when there is an abnormality in the central nervous system and the judgment target signal, and a third model that is an artificial neural network that outputs a probability of a central nervous system abnormality based on an input of an image corresponding to the judgment target signal. Here, the 'judgment target signal' may mean a signal reflecting the electrical activity of the heart. However, such a model configuration is exemplary and the spirit of the present invention is not limited thereto.

[0040] A central nervous system abnormality detection system according to one embodiment of the present invention may include a server (100), a user terminal (200), a heart activity signal acquisition device (300), and a communication network (400) as illustrated in FIG. 1.

[0041] According to one embodiment of the present invention, a server (100) can determine whether there is an abnormality in the central nervous system based on probabilities generated by at least two models. Furthermore, the server (100) can provide the user terminal (200) with information on whether there is an abnormality in the central nervous system, calculated based on the probabilities.

[0042] FIG. 2 is a diagram schematically illustrating the configuration of a server (100) according to one embodiment of the present invention. Referring to FIG. 2, the server (100) according to one embodiment of the present invention may include a communication unit (110), a first processor (120), a memory (130), and a second processor (140). In addition, although not illustrated in the drawing, the server (100) according to one embodiment of the present invention may further include an input / output unit, a program storage unit, etc.

[0043] The communication unit (110) may be a device including hardware and software necessary for the server (100) to transmit and receive signals such as control signals or data signals through wired or wireless connections with other network devices such as user terminals (200).

[0044] The first processor (120) may be a device that controls a series of processes for detecting abnormalities in the central nervous system based on data received from a user terminal (200) and / or a heart activity signal acquisition device (300).

[0045] Here, the processor may refer to a data processing device built into hardware that has a physically structured circuit to perform a function expressed by a code or command included in a program, for example. Examples of such data processing devices built into hardware include processing devices such as a microprocessor, a central processing unit (CPU), a processor core, a multiprocessor, an application-specific integrated circuit (ASIC), and a field programmable gate array (FPGA), but the scope of the present invention is not limited thereto.

[0046] The memory (130) performs the function of temporarily or permanently storing data processed by the server (100). The memory may include a magnetic storage medium or a flash storage medium, but the scope of the present invention is not limited thereto. For example, the memory (130) may temporarily and / or permanently store data and / or signals received from a cardiac activity signal acquisition device (300).

[0047] The second processor (140) may refer to a device that performs operations under the control of the first processor (120) described above. In this case, the second processor (140) may be a device having higher computational capabilities than the first processor (120) described above. For example, the second processor (140) may be configured as a GPU (Graphics Processing Unit). However, this is merely exemplary and the spirit of the present invention is not limited thereto. In one embodiment of the present invention, the number of second processors (140) may be plural or singular.

[0048] In one embodiment of the present invention, the second processor (140) may provide resources used by the model and / or artificial neural network implemented by the server (100) for computation. For example, the second processor (140) may provide resources used by the third model, which is an artificial neural network that has learned the correlation between images corresponding to electrical activity signals of the heart and the probability of central nervous system abnormality, to calculate the probability of central nervous system abnormality. However, this is merely exemplary and the scope of the present invention is not limited thereto, and a detailed description of the artificial neural network will be provided later.

[0049] In the present invention, the server (100) may sometimes be described as a 'central nervous system abnormality detection device'.

[0050] A user terminal (200) according to one embodiment of the present invention can display and provide various contents provided by a server (100) to a user.

[0051] FIG. 3 is a diagram schematically illustrating the configuration of a user terminal (200) according to one embodiment of the present invention. Referring to FIG. 3, the user terminal (200) according to one embodiment of the present invention may include a communication unit (210), a third processor (220), a memory (230), and an input / output interface (240). In addition, although not illustrated in the drawing, the server (100) according to one embodiment of the present invention may further include an input / output unit for obtaining a user's input or providing a screen to the user.

[0052] The communication unit (210) may be a device including hardware and software necessary for the user terminal (200) to transmit and receive signals such as control signals or data signals through a wired or wireless connection with another network device such as a server (100).

[0053] The third processor (220) may be a device that controls a series of processes for providing information on whether there is an abnormality in the central nervous system through a web page and / or application-based service provided by the server (100).

[0054] Here, the processor may refer to a data processing device built into hardware that has a physically structured circuit to perform a function expressed by a code or command included in a program, for example. Examples of such data processing devices built into hardware include processing devices such as a microprocessor, a central processing unit (CPU), a processor core, a multiprocessor, an application-specific integrated circuit (ASIC), and a field programmable gate array (FPGA), but the scope of the present invention is not limited thereto.

[0055] The memory (230) performs the function of temporarily or permanently storing data processed by the user terminal (200). The memory may include a magnetic storage medium or a flash storage medium, but the scope of the present invention is not limited thereto.

[0056] The input / output interface (240) may be a means for interfacing with an input / output device (not shown) that obtains user input. In this case, the input device may include a device such as a keyboard or mouse, and the output device may include a device such as a display.

[0057] A user terminal (200) according to one embodiment of the present invention may refer to a portable electronic device (201, 202, 203) or a computer (204). In this specification, the user terminal (200) may sometimes be referred to and described as a 'computing device'.

[0058] A cardiac activity signal acquisition device (300) according to one embodiment of the present invention may be a device attached to the human body that generates a signal reflecting the electrical activity of the heart. For example, the cardiac activity signal acquisition device (300) may be a device that generates a signal corresponding to the electrical activity of the heart using at least one electrode attached to the human body. However, such a configuration is exemplary and the scope of the present invention is not limited thereto.

[0059] A communication network (400) according to one embodiment of the present invention may refer to a communication network that mediates data transmission and reception between each component of a central nervous system abnormality detection system. For example, the communication network (400) may encompass wired networks such as LANs (Local Area Networks), WANs (Wide Area Networks), MANs (Metropolitan Area Networks), and ISDNs (Integrated Service Digital Networks), or wireless networks such as wireless LANs, CDMA, Bluetooth, and satellite communication, but the scope of the present invention is not limited thereto.

[0060] Below, the process by which the server (100) determines whether there is an abnormality in the central nervous system is explained.

[0061] Fig. 4 is a diagram illustrating an exemplary first signal (510). As illustrated in Fig. 4, the first signal (510) may include a signal having periodicity.

[0062] A server (100) according to one embodiment of the present invention can obtain a first signal (510) reflecting electrical activity of the heart. For example, the server (100) can obtain the first signal (510) by receiving it from a cardiac activity signal obtaining device (300).

[0063] Meanwhile, the server (100) according to one embodiment of the present invention may also obtain the first signal (510) by receiving it from the user terminal (200). In this case, the user terminal (200) may store the first signal (510) obtained by the heart activity signal obtaining device (300) in the memory (230) and transmit the stored first signal (510) to the server (100).

[0064] A server (100) according to one embodiment of the present invention can generate a second signal that reflects at least some characteristics of the first signal (510) acquired according to the aforementioned process. Here, the term "signal reflecting at least some characteristics" may be a concept encompassing both a signal that includes only at least some section of the first signal (510) and a signal that includes only some frequency components of the entire first signal (510).

[0065] FIG. 5 is a diagram illustrating a process in which a server (100) according to one embodiment of the present invention generates a second signal (531, 532, 533) from a portion of a signal (510-1) of a first signal (510) using at least one frequency filter (521, 522, 523).

[0066] As illustrated in FIG. 5, a server (100) according to an embodiment of the present invention can generate second signals (531, 532, 533) that reflect at least some characteristics of a portion of a section signal (510-1) by using at least one frequency filter (521, 522, 523). For example, the server (100) can generate a second signal (531) including a first component of less than 10 Hz of the portion of a section signal (510-1), a second signal (532) including a second component of 10 Hz or more and less than 150 Hz of the portion of a section signal (510-1), and a second signal (533) including a third component of 150 Hz or more of the portion of a section signal (510-1).

[0067] In an optional embodiment of the present invention, the server (100) may generate a second signal as a combination of frequency components that have passed through at least one frequency filter (521, 522, 523). For example, the server (100) may generate a second signal including a first component and a third component, or may generate a second signal including a second component and a third component. However, this is merely exemplary, and the scope of the present invention is not limited thereto.

[0068] FIG. 6 is a drawing for explaining a process in which a server (100) according to one embodiment of the present invention generates second signals (511, 512, 513) by varying the extraction time from a first signal (510).

[0069] According to one embodiment of the present invention, the server (100) can generate second signals (511, 512, 513) by varying the extraction time from the first signal (510). For example, the server (100) can generate a second-first signal (511) corresponding to a reference time from the first signal (510), and at least one second-second signal (512, 513) corresponding to at least one past time based on the reference time. Here, the 'reference time' can mean a time at which it is desired to determine whether there is an abnormality in the central nervous system.

[0070] The signals generated at multiple points in time in this manner can be used by the fourth model described below to determine whether there is an abnormality in the central nervous system at a reference point in time or to determine whether there is an abnormality in the central nervous system at a future point in time based on the reference point in time, and a detailed description of this will be described below.

[0071] FIG. 7 is a diagram illustrating a series of processes in which a server (100) according to one embodiment of the present invention inputs a second signal (540) to each of at least two models (610, 620, 630, 640) to ultimately calculate a second central nervous system abnormality probability (650).

[0072] According to one embodiment of the present invention, the server (100) can calculate a first central nervous system abnormality probability (611, 621, 631, 641) for each model using the second signal (540) generated according to the above-described process and at least two models (610, 620, 630, 640). In addition, the server (100) can calculate a second central nervous system abnormality probability (650) based on the first central nervous system abnormality probability (611, 621, 631, 641) for each model. Hereinafter, a process in which each of the two or more models (610, 620, 630, 640) calculates the first central nervous system abnormality probability (611, 621, 631, 641) from the second signal (540) will be described.

[0073] FIG. 8 is a drawing for explaining a first model (610) according to one embodiment of the present invention.

[0074] A first model (610) according to one embodiment of the present invention may be a model that determines whether there is an abnormality in the central nervous system based on a value in a portion of a section (540T) of a second signal (540) and a predetermined threshold value (Vth).

[0075] According to one embodiment of the present invention, the server (100) can extract at least one target segment for judgment from the second signal (540) by considering the periodicity of the electrical activity of the heart. For example, the server (100) can extract a micro-interval including a P wave, a Q wave, an R wave, an S wave, and a T wave from the second signal (540) as a target segment for judgment.

[0076] According to one embodiment of the present invention, the server (100) can extract a section (540T) corresponding to a T wave from the above-described judgment target segment. In addition, the server (100) can determine that the first central nervous system abnormality probability (611) for the first model (610) is greater than or equal to the predetermined threshold probability when the minimum value of the T wave within the section (540T) is less than a predetermined threshold value (Vth).

[0077] In an optional embodiment of the present invention, the server (100) may determine that the first central nervous system abnormality probability (611) for the first model (610) is True if the minimum value of the T wave within the section (540T) is less than a predetermined threshold value (Vth), and False otherwise.

[0078] FIG. 9 is a drawing for explaining a second model (620) according to one embodiment of the present invention.

[0079] A second model (620) according to one embodiment of the present invention may be a model that determines whether there is an abnormality in the central nervous system based on the similarity between a plurality of electrical activity signals (660) of the heart obtained when there is an abnormality in the central nervous system and the second signal (540).

[0080] A server (100) according to one embodiment of the present invention can calculate the similarity (542) between a plurality of electrical activity signals (660) of the heart obtained when there is an abnormality in the central nervous system and a second signal (540).

[0081] Additionally, the server (100) can determine the probability of an abnormality (621) of the first central nervous system for the second model (620) based on the calculated similarity (542).

[0082] FIG. 10 is a drawing for explaining a third model (630) according to one embodiment of the present invention.

[0083] A third model (630) according to one embodiment of the present invention may be an artificial neural network that has learned the correlation between an image corresponding to an electrical activity signal of the heart and the probability of a central nervous system abnormality.

[0084] A third model (630) according to one embodiment of the present invention can be pre-trained based on a plurality of training data (710) for learning the correlation as described above.

[0085] At this time, each of the plurality of learning data (710) may include an image corresponding to an electrical activity signal of the heart as input data and a probability of a central nervous system abnormality as output data.

[0086] For example, the first learning data (711) may include an image (711A) corresponding to a heart electrical activity signal as input data, and a central nervous system abnormality probability (711B) as output data. Similarly, the second learning data (712) and the third learning data (713) may each include the above-described items.

[0087] According to one embodiment of the invention, the server (100) can train the third model (630) using the first processor (120) and / or the second processor (140). For example, the server (100) can input an image (711A) corresponding to an electrical activity signal of the heart into the third model (630), and update at least one coefficient and / or weight constituting the third model (630) so that the probability output at that time approaches the probability of a central nervous system abnormality (711B).

[0088] The server (100) can perform learning in a direction in which the error rate is reduced by repeating the above-described process using multiple learning data (710).

[0089] A server (100) according to one embodiment of the present invention may generate an input image reflecting the characteristics of a second signal (540) to calculate a second central nervous system abnormality probability (631) from a third model (630). At this time, the second signal (540) may be a signal including a second-1 signal corresponding to the reference point of the first signal.

[0090] Additionally, the server (100) can determine the probability of an abnormality (631) of the first central nervous system for the third model (630) by inputting the generated input image into the third model (630).

[0091] FIG. 11 is a drawing for explaining a fourth model (640) according to one embodiment of the present invention.

[0092] A fourth model (640) according to one embodiment of the present invention may be an artificial neural network that learns the correlation between an image corresponding to an electrical activity signal of the heart at each of at least two or more time points and a probability of a central nervous system abnormality.

[0093] A fourth model (630) according to one embodiment of the present invention can be pre-trained based on a plurality of training data (810) for learning the correlation as described above.

[0094] At this time, each of the plurality of learning data (810) may include, as input data, an image corresponding to the electrical activity signal of the heart at a reference point in time and an image corresponding to the electrical activity signal of the heart at each of at least one past point in time based on the reference point in time, and, as output data, a probability of a central nervous system abnormality.

[0095] For example, the first learning data (811) may include an image (811A) corresponding to the electrical activity signal of the heart at a reference point in time as input data and an image (811B, 811C) corresponding to the electrical activity signal of the heart at at least one past point in time based on the reference point in time, and a central nervous system abnormality probability (811D) as output data. Similarly, the second learning data (812) and the third learning data (813) may each include the above-described items.

[0096] According to one embodiment of the invention, the server (100) can train the fourth model (640) using the first processor (120) and / or the second processor (140). For example, the server (100) inputs an image (811A) corresponding to an electrical activity signal of the heart at a reference point in time and images (811B, 811C) corresponding to an electrical activity signal of the heart at at least one past point in time based on the reference point in time into the fourth model (640), and updates at least one coefficient and / or weight constituting the fourth model (640) so that the probability output at that time approaches the central nervous system abnormality probability (811D).

[0097] The server (100) can perform learning in a direction in which the error rate is reduced by repeating the above-described process using multiple learning data (810).

[0098] According to one embodiment of the present invention, the server (100) may generate one or more input images reflecting the characteristics of each of the second-1 signal and at least one second-2 signal to calculate the second central nervous system abnormality probability (641) from the fourth model (640). Referring again to FIG. 6, the second-1 signal may be the second signal (511) at the rightmost point in time, which is the reference point, and the second-2 signal may be the second signals (512, 513) at the left points in time with respect to the reference point.

[0099] According to one embodiment of the present invention, the server (100) can determine the probability of an abnormality (641) of the first central nervous system for the fourth model (640) by inputting one or more input images into the fourth model (640).

[0100] Meanwhile, the abnormality probability (641) of the first central nervous system output by the fourth model (640) may be the abnormality probability of the central nervous system at a reference point in time, or may be the abnormality probability of the central nervous system at a future point in time after the reference point in time.

[0101] In other words, the fourth model (640) may be trained to output the probability of an abnormality in the central nervous system at the present time based on data from a past time, or may be trained to output the probability of an abnormality in the central nervous system at a future time based on data from a past time.

[0102] According to one embodiment of the present invention, the server (100) may calculate the second central nervous system abnormality probability (650) at least partially based on at least one of the first central nervous system abnormality probability (611) for the first model (610), the first central nervous system abnormality probability (621) for the second model (620), the first central nervous system abnormality probability (631) for the third model (630), and the first central nervous system abnormality probability (641) for the fourth model (640). For example, the server (100) may calculate the second central nervous system abnormality probability (650) based on a weighted sum of the first central nervous system abnormality probabilities (611, 621, 631) for the first model (610), the second model (620), and the third model (630). However, this is merely exemplary, and the spirit of the present invention is not limited thereto.

[0103] According to one embodiment of the present invention, the server (100) can provide the probability of a central nervous system abnormality calculated according to the above-described process. For example, the server (100) can provide the probability of a central nervous system abnormality calculated according to the above-described process to the user terminal (200).

[0104] FIG. 12 is a drawing showing an exemplary screen (1300) showing the probability of an abnormality in the central nervous system on a user terminal (200).

[0105] As illustrated in FIG. 12, the screen (1300) may include a region (1310) in which a first signal is displayed, and a region (1320) in which the probability of a central nervous system abnormality calculated by each model at multiple time points included in the first signal is displayed. At this time, data provided to the regions (1310, 1320) may be updated and displayed over time.

[0106] Meanwhile, the server (100) may provide the point in time at which a central nervous system abnormality occurred in a form distinct from other points in time. For example, the server (100) may provide a highlight (1330) for Segment 33 as shown on the screen (1300) and display the data of Segment 33 with emphasis. However, such a display form is exemplary and the spirit of the present invention is not limited thereto.

[0107] In this way, the server (100) according to one embodiment of the present invention provides at least two first central nervous system abnormality probabilities and two second central nervous system abnormality probabilities together, but can provide the first central nervous system abnormality probability separately for each model.

[0108] FIG. 13 is a flowchart illustrating a central nervous system abnormality detection method performed by a server (100) according to one embodiment of the present invention. The method is described below with reference to FIGS. 1 to 12.

[0109] A server (100) according to one embodiment of the present invention can obtain a first signal reflecting the electrical activity of the heart. (S1410)

[0110] Fig. 4 is a diagram illustrating an exemplary first signal (510). As illustrated in Fig. 4, the first signal (510) may include a signal having periodicity.

[0111] For example, the server (100) can obtain the first signal (510) by receiving it from the heart activity signal acquisition device (300).

[0112] Meanwhile, the server (100) according to one embodiment of the present invention may also obtain the first signal (510) by receiving it from the user terminal (200). In this case, the user terminal (200) may store the first signal (510) obtained by the heart activity signal obtaining device (300) in the memory (230) and transmit the stored first signal (510) to the server (100).

[0113] The server (100) according to one embodiment of the present invention can generate a second signal that reflects at least some characteristics of the first signal (510) acquired according to the above-described process (S1420). Here, the 'signal reflecting at least some characteristics' may be a concept encompassing both a signal that includes only at least some section of the first signal (510) and a signal that includes only some frequency components of the entire first signal (510).

[0114] FIG. 5 is a diagram illustrating a process in which a server (100) according to one embodiment of the present invention generates a second signal (531, 532, 533) from a portion of a signal (510-1) of a first signal (510) using at least one frequency filter (521, 522, 523).

[0115] As illustrated in FIG. 5, a server (100) according to an embodiment of the present invention can generate second signals (531, 532, 533) that reflect at least some characteristics of a portion of a section signal (510-1) by using at least one frequency filter (521, 522, 523). For example, the server (100) can generate a second signal (531) including a first component of less than 10 Hz of the portion of a section signal (510-1), a second signal (532) including a second component of 10 Hz or more and less than 150 Hz of the portion of a section signal (510-1), and a second signal (533) including a third component of 150 Hz or more of the portion of a section signal (510-1).

[0116] In an optional embodiment of the present invention, the server (100) may generate a second signal as a combination of frequency components that have passed through at least one frequency filter (521, 522, 523). For example, the server (100) may generate a second signal including a first component and a third component, or may generate a second signal including a second component and a third component. However, this is merely exemplary, and the scope of the present invention is not limited thereto.

[0117] FIG. 6 is a drawing for explaining a process in which a server (100) according to one embodiment of the present invention generates second signals (511, 512, 513) by varying the extraction time from a first signal (510).

[0118] According to one embodiment of the present invention, the server (100) can generate second signals (511, 512, 513) by varying the extraction time from the first signal (510). For example, the server (100) can generate a second-first signal (511) corresponding to a reference time from the first signal (510), and at least one second-second signal (512, 513) corresponding to at least one past time based on the reference time. Here, the 'reference time' can mean a time at which it is desired to determine whether there is an abnormality in the central nervous system.

[0119] The signals generated at multiple points in time in this manner can be used by the fourth model described below to determine whether there is an abnormality in the central nervous system at a reference point in time or to determine whether there is an abnormality in the central nervous system at a future point in time based on the reference point in time, and a detailed description of this will be described below.

[0120] FIG. 7 is a diagram illustrating a series of processes in which a server (100) according to one embodiment of the present invention inputs a second signal (540) to each of at least two models (610, 620, 630, 640) to ultimately calculate a second central nervous system abnormality probability (650).

[0121] According to one embodiment of the present invention, the server (100) can calculate the first central nervous system abnormality probability (611, 621, 631, 641) for each model by using the second signal (540) generated according to the above-described process and at least two models (610, 620, 630, 640). (S1430) In addition, the server (100) can calculate the second central nervous system abnormality probability (650) based on the first central nervous system abnormality probability (611, 621, 631, 641) for each model. Hereinafter, the process of calculating the first central nervous system abnormality probability (611, 621, 631, 641) from the second signal (540) by each of the two or more models (610, 620, 630, 640) will be described.

[0122] FIG. 8 is a drawing for explaining a first model (610) according to one embodiment of the present invention.

[0123] A first model (610) according to one embodiment of the present invention may be a model that determines whether there is an abnormality in the central nervous system based on a value in a portion of a section (540T) of a second signal (540) and a predetermined threshold value (Vth).

[0124] According to one embodiment of the present invention, the server (100) can extract at least one target segment for judgment from the second signal (540) by considering the periodicity of the electrical activity of the heart. For example, the server (100) can extract a micro-interval including a P wave, a Q wave, an R wave, an S wave, and a T wave from the second signal (540) as a target segment for judgment.

[0125] According to one embodiment of the present invention, the server (100) can extract a section (540T) corresponding to a T wave from the above-described judgment target segment. In addition, the server (100) can determine that the first central nervous system abnormality probability (611) for the first model (610) is greater than or equal to the predetermined threshold probability when the minimum value of the T wave within the section (540T) is less than a predetermined threshold value (Vth).

[0126] In an optional embodiment of the present invention, the server (100) may determine that the first central nervous system abnormality probability (611) for the first model (610) is True if the minimum value of the T wave within the section (540T) is less than a predetermined threshold value (Vth), and False otherwise.

[0127] FIG. 9 is a drawing for explaining a second model (620) according to one embodiment of the present invention.

[0128] A second model (620) according to one embodiment of the present invention may be a model that determines whether there is an abnormality in the central nervous system based on the similarity between a plurality of electrical activity signals (660) of the heart obtained when there is an abnormality in the central nervous system and the second signal (540).

[0129] A server (100) according to one embodiment of the present invention can calculate the similarity (542) between a plurality of electrical activity signals (660) of the heart obtained when there is an abnormality in the central nervous system and a second signal (540).

[0130] Additionally, the server (100) can determine the probability of an abnormality (621) of the first central nervous system for the second model (620) based on the calculated similarity (542).

[0131] FIG. 10 is a drawing for explaining a third model (630) according to one embodiment of the present invention.

[0132] A third model (630) according to one embodiment of the present invention may be an artificial neural network that has learned the correlation between an image corresponding to an electrical activity signal of the heart and the probability of a central nervous system abnormality.

[0133] A third model (630) according to one embodiment of the present invention can be pre-trained based on a plurality of training data (710) for learning the correlation as described above.

[0134] At this time, each of the plurality of learning data (710) may include an image corresponding to an electrical activity signal of the heart as input data and a probability of a central nervous system abnormality as output data.

[0135] For example, the first learning data (711) may include an image (711A) corresponding to a heart electrical activity signal as input data, and a central nervous system abnormality probability (711B) as output data. Similarly, the second learning data (712) and the third learning data (713) may each include the above-described items.

[0136] According to one embodiment of the invention, the server (100) can train the third model (630) using the first processor (120) and / or the second processor (140). For example, the server (100) can input an image (711A) corresponding to an electrical activity signal of the heart into the third model (630), and update at least one coefficient and / or weight constituting the third model (630) so that the probability output at that time approaches the probability of a central nervous system abnormality (711B).

[0137] The server (100) can perform learning in a direction in which the error rate is reduced by repeating the above-described process using multiple learning data (710).

[0138] A server (100) according to one embodiment of the present invention may generate an input image reflecting the characteristics of a second signal (540) to calculate a second central nervous system abnormality probability (631) from a third model (630). At this time, the second signal (540) may be a signal including a second-1 signal corresponding to the reference point of the first signal.

[0139] Additionally, the server (100) can determine the probability of an abnormality (631) of the first central nervous system for the third model (630) by inputting the generated input image into the third model (630).

[0140] FIG. 11 is a drawing for explaining a fourth model (640) according to one embodiment of the present invention.

[0141] A fourth model (640) according to one embodiment of the present invention may be an artificial neural network that learns the correlation between an image corresponding to an electrical activity signal of the heart at each of at least two or more time points and a probability of a central nervous system abnormality.

[0142] A fourth model (630) according to one embodiment of the present invention can be pre-trained based on a plurality of training data (810) for learning the correlation as described above.

[0143] At this time, each of the plurality of learning data (810) may include, as input data, an image corresponding to the electrical activity signal of the heart at a reference point in time and an image corresponding to the electrical activity signal of the heart at each of at least one past point in time based on the reference point in time, and, as output data, a probability of a central nervous system abnormality.

[0144] For example, the first learning data (811) may include an image (811A) corresponding to the electrical activity signal of the heart at a reference point in time as input data and an image (811B, 811C) corresponding to the electrical activity signal of the heart at at least one past point in time based on the reference point in time, and a central nervous system abnormality probability (811D) as output data. Similarly, the second learning data (812) and the third learning data (813) may each include the above-described items.

[0145] According to one embodiment of the invention, the server (100) can train the fourth model (640) using the first processor (120) and / or the second processor (140). For example, the server (100) inputs an image (811A) corresponding to an electrical activity signal of the heart at a reference point in time and images (811B, 811C) corresponding to an electrical activity signal of the heart at at least one past point in time based on the reference point in time into the fourth model (640), and updates at least one coefficient and / or weight constituting the fourth model (640) so that the probability output at that time approaches the central nervous system abnormality probability (811D).

[0146] The server (100) can perform learning in a direction in which the error rate is reduced by repeating the above-described process using multiple learning data (810).

[0147] According to one embodiment of the present invention, the server (100) may generate one or more input images reflecting the characteristics of each of the second-1 signal and at least one second-2 signal to calculate the second central nervous system abnormality probability (641) from the fourth model (640). Referring again to FIG. 6, the second-1 signal may be the second signal (511) at the rightmost point in time, which is the reference point, and the second-2 signal may be the second signals (512, 513) at the left points in time with respect to the reference point.

[0148] According to one embodiment of the present invention, the server (100) can determine the probability of an abnormality (641) of the first central nervous system for the fourth model (640) by inputting one or more input images into the fourth model (640).

[0149] Meanwhile, the abnormality probability (641) of the first central nervous system output by the fourth model (640) may be the abnormality probability of the central nervous system at a reference point in time, or may be the abnormality probability of the central nervous system at a future point in time after the reference point in time.

[0150] In other words, the fourth model (640) may be trained to output the probability of an abnormality in the central nervous system at the present time based on data from a past time, or may be trained to output the probability of an abnormality in the central nervous system at a future time based on data from a past time.

[0151] According to one embodiment of the present invention, the server (100) may calculate the second central nervous system abnormality probability (650) at least partially based on at least one of the first central nervous system abnormality probability (611) for the first model (610), the first central nervous system abnormality probability (621) for the second model (620), the first central nervous system abnormality probability (631) for the third model (630), and the first central nervous system abnormality probability (641) for the fourth model (640). (S1440) For example, the server (100) may calculate the second central nervous system abnormality probability (650) based on a weighted sum of the first central nervous system abnormality probabilities (611, 621, 631) for the first model (610), the second model (620), and the third model (630). However, this is exemplary and the spirit of the present invention is not limited thereto.

[0152] A server (100) according to one embodiment of the present invention can provide the probability of a central nervous system abnormality calculated according to the above-described process. (S1450) For example, the server (100) can provide the probability of a central nervous system abnormality calculated according to the above-described process to a user terminal (200).

[0153] FIG. 12 is a drawing showing an exemplary screen (1300) showing the probability of an abnormality in the central nervous system on a user terminal (200).

[0154] As illustrated in FIG. 12, the screen (1300) may include a region (1310) in which a first signal is displayed, and a region (1320) in which the probability of a central nervous system abnormality calculated by each model at multiple time points included in the first signal is displayed. At this time, data provided to the regions (1310, 1320) may be updated and displayed over time.

[0155] Meanwhile, the server (100) may provide the point in time at which a central nervous system abnormality occurred in a form distinct from other points in time. For example, the server (100) may provide a highlight (1330) for Segment 33 as shown on the screen (1300) and display the data of Segment 33 with emphasis. However, such a display form is exemplary and the spirit of the present invention is not limited thereto.

[0156] In this way, the server (100) according to one embodiment of the present invention provides at least two first central nervous system abnormality probabilities and two second central nervous system abnormality probabilities together, but can provide the first central nervous system abnormality probability separately for each model.

[0157] The embodiments of the present invention described above may be implemented in the form of a computer program that can be executed through various components on a computer, and such a computer program may be recorded on a computer-readable medium. At this time, the medium may be something that stores a program that can be executed by a computer. Examples of the medium may include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical recording media such as CD-ROMs and DVDs, magneto-optical media such as floptical disks, and those configured to store program instructions, including ROMs, RAMs, flash memory, and cloud memory.

[0158] Meanwhile, the computer program may be specifically designed and constructed for the present invention, or may be one known and available to those skilled in the computer software field. Examples of computer programs may include not only machine language code, such as that generated by a compiler, but also high-level language code that can be executed by a computer using an interpreter or the like.

[0159] The specific implementations described in the present invention are exemplary embodiments and do not limit the scope of the present invention in any way. For the sake of brevity, descriptions of conventional electronic components, control systems, software, and other functional aspects of the systems may be omitted. In addition, the lines connecting or connecting members between components illustrated in the drawings are merely representative of functional connections and / or physical or circuit connections, and may be replaced or represented as various additional functional connections, physical connections, or circuit connections in an actual device. In addition, unless specifically mentioned as “essential,” “important,” etc., a component may not be absolutely necessary for the application of the present invention.

[0160] Therefore, the idea of ​​the present invention should not be limited to the embodiments described above, and not only the scope of the patent claims described below but also all scopes equivalent to or equivalently modified from the scope of the patent claims are considered to fall within the scope of the idea of ​​the present invention.

[0161] This application is based on the results of the National Research Foundation of Korea (NRF)-funded research and development (R&D) project.

[0162] - Specialized organization: National Research Foundation of Korea

[0163] - Project Name: Establishment of a Foundation for International Cooperation (R&D)

[0164] - Research and Development Project Number: RS-2023-00233632

[0165] - Research and Development Project Name: Development of an AI Algorithm to Assist in Endoscopic Neurosurgery of the Central Nervous System

[0166] - Main Research and Development Organization: Eulji University Industry-Academic Cooperation Foundation

[0167] - Total research period: October 1, 2023 - December 31, 2026

Claims

1. A method for detecting central nervous system abnormalities, which determines whether there is an abnormality in the central nervous system based on probabilities generated by at least two models. A step of acquiring a first signal reflecting the electrical activity of the heart; generating a second signal reflecting at least some characteristics of the first signal; A step of determining at least two first central nervous system abnormality probabilities from the second signal using at least two models respectively; and A method for detecting a central nervous system abnormality, comprising: a step of calculating a second central nervous system abnormality probability based on at least two first central nervous system abnormality probabilities.

2. In claim 1, The step of generating the second signal is A method for detecting a central nervous system abnormality, comprising: generating the second signal based on at least one of a first component of less than 10 Hz of the first signal, a second component of 10 Hz or more and less than 150 Hz of the first signal, and a third component of 150 Hz or more of the first signal.

3. In claim 1, The step of generating the second signal is A method for detecting a central nervous system abnormality, comprising: a step of generating a 2-1 signal corresponding to a reference point in time from the first signal, and at least one 2-2 signal corresponding to at least one past point in time based on the reference point in time.

4. In claim 1, The above at least two models include a first model that determines whether there is an abnormality in the central nervous system based on a value in a part of the second signal and a predetermined threshold value, The step of determining the probability of the above first central nervous system abnormality is A step of extracting at least one judgment target segment by considering the periodicity of electrical activity of the heart in the second signal; A step of extracting a section corresponding to a T wave from the above judgment target segment; and A method for detecting a central nervous system abnormality, comprising: a step of determining that the first central nervous system abnormality probability for the first model is greater than or equal to a predetermined threshold probability when the minimum value of the T wave within the section corresponding to the T wave is less than a predetermined threshold probability; 5. In claim 1, The above at least two models include a second model that determines whether there is an abnormality in the central nervous system based on the similarity between multiple electrical activity signals of the heart obtained when there is an abnormality in the central nervous system and the second signal, The step of determining the probability of the above first central nervous system abnormality is A step of calculating the similarity between the electrical activity signals of multiple hearts obtained in the above central nervous system abnormality and the second signal; and A method for detecting central nervous system abnormalities, comprising: a step of determining an abnormality probability of the first central nervous system for the second model based on the similarity.

6. In claim 1, The above at least two models include a third model which is an artificial neural network that has learned the correlation between images corresponding to electrical activity signals of the heart and the probability of central nervous system abnormality, The step of determining the probability of the above first central nervous system abnormality is A step for generating an input image reflecting the characteristics of the second signal, wherein the second signal is a signal including a 2-1 signal corresponding to a reference point of the first signal; and A method for detecting central nervous system abnormalities, comprising: a step of determining an abnormality probability of a first central nervous system for the third model by inputting the input image into the third model; 7. In claim 1, The above at least two models include a fourth model, which is an artificial neural network that has learned the correlation between images corresponding to electrical activity signals of the heart at each of at least two time points and the probability of central nervous system abnormality, The step of determining the probability of the above first central nervous system abnormality is A step of generating one or more input images reflecting the characteristics of each of the 2-1 signal and at least one 2-2 signal, wherein the 2-1 signal is a signal including a signal corresponding to a reference point in time of the 1 signal, and the 2-2 signal is a signal including a signal corresponding to each of at least one past point in time based on the reference point in time; and A method for detecting central nervous system abnormalities, comprising: a step of determining an abnormality probability of a first central nervous system for the fourth model by inputting one or more input images into the fourth model; 8. In claim 1, The step of calculating the probability of the above second central nervous system abnormality is A method for detecting a central nervous system abnormality, comprising: a step of calculating a second central nervous system abnormality probability based at least in part on at least one of a first central nervous system abnormality probability for a first model, a first central nervous system abnormality probability for a second model, a first central nervous system abnormality probability for a third model, and a first central nervous system abnormality probability for a fourth model.

9. In claim 1, The above method of detecting central nervous system abnormalities is After the step of calculating the probability of the second central nervous system abnormality, A step of providing at least two first central nervous system abnormality probabilities and the second central nervous system abnormality probabilities; further comprising; The steps provided above are A method for detecting central nervous system abnormalities, which provides a first central nervous system abnormality probability for each model.

Images