Method for classifying patients with severe cognitive impairment through electroencephalogram data analysis, server and computer program

The method analyzes EEG data to classify users into cognitive impairment stages, predicting dementia risk through a multi-step process, addressing the limitations of conventional platforms by identifying users at risk for dementia.

JP7586450B2Active Publication Date: 2024-11-19IMEDISYNC INC
View PDF 6 Cites 0 Cited by

Patent Information

Application Number
JP2023555743
Authority / Receiving Office
JP · JP
Patent Type
Patents
Current Assignee / Owner
Priority Date
2021-03-12
Filing Date
2021-08-06
Publication Date
2024-11-19
Estimated Expiration
2041-08-06

AI Technical Summary

Technical Problem

Conventional dementia diagnosis platforms can identify patients with dementia but fail to predict the development of dementia in individuals who appear normal, lacking the ability to classify users based on the presence or absence, type, and degree of memory impairment.

Method used

A method and system for analyzing electroencephalogram (EEG) data to classify users into normal, severe cognitive impairment, and pre-dementia groups using a multi-step classification process involving quantified EEG generation, frequency domain analysis, and optimized classification models to predict the likelihood of dementia development.

Benefits of technology

Enables early identification of individuals at risk for dementia by categorizing users into specific cognitive impairment stages, providing a numerical representation of severity and probability of dementia progression.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 0007586450000001
    Figure 0007586450000001
  • Figure 0007586450000002
    Figure 0007586450000002
  • Figure 0007586450000003
    Figure 0007586450000003
Patent Text Reader

Abstract

A method for classifying patients with severe cognitive impairment through electroencephalogram data analysis, a server, and a computer program are provided. The method for classifying patients with severe cognitive impairment through electroencephalogram data analysis according to various embodiments of the present invention is a method performed by a computing device, and includes an electroencephalogram collecting step of collecting electroencephalogram data for a plurality of users, a first classification step of classifying the plurality of users into a severe cognitive impairment group or a non-severe cognitive impairment group by analyzing the collected electroencephalogram data, a second classification step of classifying users included in the non-severe cognitive impairment group into a normal group or an amnestic Mild Cognitive Impairment (aMCI) group, and a third classification step of classifying users included in the normal group into a normal range group or a cognitive impairment latency group, and classifying users included in the amnestic mild cognitive impairment group into a non-Alzheimer's mild cognitive impairment group or a pre-dementia mild cognitive impairment group.
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] Various embodiments of the present invention relate to a method, server and computer program for classifying patients with severe cognitive impairment through electroencephalogram data analysis. [Background technology]

[0002] Dementia refers to a set of symptoms resulting from brain disease.

[0003] Dementia is a condition that results in a decline in cognitive ability and a loss of the ability to perform everyday activities, and as dementia progresses, it can affect the ability to think, act, and carry out daily activities. Doctors diagnose dementia when there is significant impairment in two or more cognitive functions (including, for example, memory, language, information comprehension, spatial function, judgment, and attention).

[0004] People with dementia can have difficulty solving problems and regulating their emotions and may undergo personality changes. The exact symptoms a person with dementia experiences depends on which parts of the brain are damaged by the disease that caused the dementia. In many types of dementia, some of the brain's nerve cells stop functioning, lose connections to other cells and die. Dementia usually progresses steadily, meaning it gradually spreads through the brain and the patient's symptoms worsen over time.

[0005] In Japan, about 290,000 people, or about 9.5% of the population aged 65 or older, suffer from early-onset dementia, and of these, 180,000 people, or 73%, are severe cases such as those who habitually wander the streets. This number is expected to continue to increase as the aging of the population accelerates in the future.

[0006] As no effective treatment has yet been developed for dementia, early detection and prevention is of utmost importance. However, until now, dementia diagnosis has mainly relied on neuropsychological tests, with MRI (testing brain deformation such as cerebral atrophy) and PET (testing amyloid deposition levels) being used as auxiliary methods, but these have the disadvantage of being expensive to test, and cerebrospinal fluid testing (measuring the amount of amyloid or tau protein) is an invasive method that patients dislike.

[0007] To address these issues, a dementia diagnosis platform based on EEG data analysis was previously developed to determine whether a patient has dementia by analyzing the patient's EEG data (e.g., by analyzing whether the EEG data is from a patient with dementia or from a normal person).

[0008] However, such conventional dementia diagnosis platforms analyze EEG data to classify the patient as either normal or demented, i.e., they are merely a technology that simply divides the normal state into two states, namely, they can identify patients who currently have dementia, but they are unable to select users who are normal but may develop dementia. Summary of the Invention [Problem to be solved by the invention]

[0009] The problem to be solved by the present invention is to provide a method, server, and computer program for classifying patients with severe cognitive impairment through EEG data analysis, which not only classifies users in a normal state from those with dementia by analyzing EEG data for multiple users, but also subdivides the possibility of dementia for users in a normal state by considering the presence or absence of memory impairment, and the type and degree of memory impairment, thereby not only identifying patients who currently have dementia, but also early selecting users who are likely to develop dementia from among users in a normal state, in order to improve the problems of the conventional dementia diagnosis platform described above.

[0010] The problems to be solved by the present invention are not limited to those mentioned above, and other problems not mentioned will be clearly understood by those skilled in the art from the following description. [Means for solving the problem]

[0011] A method for classifying patients with severe cognitive impairment through electroencephalogram data analysis according to one embodiment of the present invention for solving the above-mentioned problems is a method performed by a computing device, and includes an electroencephalogram collecting step of collecting electroencephalogram data for a plurality of users, a first classification step of analyzing the collected electroencephalogram data to classify the plurality of users into a severe cognitive impairment group or a non-severe cognitive impairment group, a second classification step of classifying users included in the non-severe cognitive impairment group into a normal group or an amnestic Mild Cognitive Impairment (aMCI) group, and a third classification step of classifying users included in the normal group into a normal range group or a cognitive impairment latency group, and classifying users included in the amnestic mild cognitive impairment group into a non-Alzheimer's mild cognitive impairment group or a pre-dementia mild cognitive impairment group.

[0012] In various embodiments, the EEG collecting step may include a step of collecting a plurality of unit EEG data through a plurality of EEG measurement channels attached to different positions on the first user's head, a step of digitizing and imaging the collected plurality of unit EEG data to generate quantified EEG data (QEEG), and a step of standardizing the generated quantified EEG data based on preset gender and age standards.

[0013] In various embodiments, the step of generating the quantified electroencephalogram data (QEEG) may include the steps of calculating a frequency domain for the measured plurality of unit electroencephalogram data based on a Fast-Fourier Transform, rearranging frequency domains corresponding to a plurality of unit electroencephalogram data measured at the same time point among the measured plurality of unit electroencephalogram data based on the calculated frequency domain according to positions of the plurality of electroencephalogram measurement channels, and accumulating the rearranged plurality of unit electroencephalogram data in a time series to generate quantified electroencephalogram data in the form of a symmetric image.

[0014] In various embodiments, the first classification step may include a step of calculating a probability of severe cognitive impairment and a probability of non-severe cognitive impairment for each of the plurality of users by performing image analysis of the generated quantified electroencephalogram data using a first classification model, and a step of classifying the plurality of users into the severe cognitive impairment group or the non-severe cognitive impairment group using the calculated probability of severe cognitive impairment and the calculated probability of non-severe cognitive impairment.

[0015] In various embodiments, the second classification step may include classifying users with memory impairment among the users included in the non-severe cognitive impairment group into the memory impairment-type mild cognitive impairment group, and classifying users not classified into the memory impairment-type mild cognitive impairment group into the normal group, using a second classification model.

[0016] In various embodiments, the third classification step may include the steps of generating an optimized classification model for user classification, the optimized classification model including a third classification model and a fourth classification model; classifying users without Subjective Cognitive Decline (SCD) among users included in the normal group into the normal range group and classifying users with the subjective cognitive decline into the cognitive impairment latency group using the third classification model; and classifying users with Retrieval Failure among users included in the memory impairment type mild cognitive impairment group into the non-Alzheimer's mild cognitive impairment group and classifying users with Encoding Failure into the pre-dementia mild cognitive impairment group using the fourth classification model.

[0017] In various embodiments, the step of generating the optimized classification model may include the steps of: arbitrarily and repeatedly selecting a predetermined number of feature values ​​from among a plurality of feature values ​​derived by analyzing a plurality of electroencephalogram data collected from each of the plurality of users to generate a plurality of different feature value combinations; generating a plurality of multidimensional models using each of the generated plurality of feature value combinations; calculating accuracy for each of the generated plurality of multidimensional models; and selecting top N multidimensional models in order of the calculated accuracy, counting the number of feature values ​​included in each of the selected N multidimensional models, sequentially selecting M feature values ​​starting from the feature value having the highest count value, and generating the optimized classification model including the selected M feature values.

[0018] In various embodiments, the method may further include a step of scoring the degree of severe cognitive impairment for each of the users included in the normal range group, the cognitive impairment latency group, the non-Alzheimer's mild cognitive impairment group, the pre-dementia mild cognitive impairment group, and the severe cognitive impairment group using the classification results of the first classification step, the classification results of the second classification step, and the classification results of the third classification step.

[0019] In various embodiments, the pre-scoring step may include the steps of: assigning a score within a range of 0 to 50 to users included in the normal range group according to the degree of the severe cognitive impairment; assigning a score within a range of 50 to 60 to users included in the cognitive impairment latency period group according to the degree of the severe cognitive impairment; assigning a score within a range of 60 to 70 to users included in the non-Alzheimer's mild cognitive impairment group according to the degree of the severe cognitive impairment; assigning a score within a range of 70 to 85 to users included in the pre-dementia mild cognitive impairment group according to the degree of the severe cognitive impairment; and assigning a score within a range of 85 to 100 to users included in the severe cognitive impairment group according to the degree of the severe cognitive impairment.

[0020] A server for classifying patients with severe cognitive impairment through EEG data analysis according to another embodiment of the present invention for solving the above-mentioned problems includes a processor, a network interface, a memory, and a computer program loaded into the memory and executed by the processor, and the computer program may include EEG collection instructions for collecting EEG data from a plurality of users, a first classification instruction for analyzing the collected EEG data and classifying the plurality of users into a severe cognitive impairment group or a non-severe cognitive impairment group, a second classification instruction for classifying users included in the non-severe cognitive impairment group into a normal group or an amnestic Mild Cognitive Impairment (aMCI) group, and a third classification instruction for classifying users included in the normal group into a normal range group or a cognitive impairment latency group, and classifying users included in the amnestic mild cognitive impairment group into a non-Alzheimer's mild cognitive impairment group or a pre-dementia mild cognitive impairment group.

[0021] A computer program according to yet another embodiment of the present invention for solving the above-mentioned problems may be stored on a computer-readable recording medium in combination with a computing device to execute an electroencephalogram collecting step of collecting electroencephalogram data for a plurality of users, a first classification step of analyzing the collected electroencephalogram data to classify the plurality of users into a severe cognitive impairment group or a non-severe cognitive impairment group, a second classification step of classifying users included in the non-severe cognitive impairment group into a normal group or an amnestic Mild Cognitive Impairment (aMCI) group, and a third classification step of classifying users included in the normal group into a normal range group or a cognitive impairment latency group, and classifying users included in the amnestic mild cognitive impairment group into a non-Alzheimer's mild cognitive impairment group or a pre-dementia mild cognitive impairment group.

[0022] Further details of the invention are contained in the detailed description and the drawings. Effect of the Invention

[0023] According to various embodiments of the present invention, not only is it possible to analyze EEG data for a plurality of users to classify them into normal users and dementia patients, but also to subdivide the possibility of dementia for normal users by considering the presence or absence of memory impairment, and the type and degree of memory impairment, etc., thereby making it possible to identify patients who currently have dementia, as well as to early screen users who are likely to develop dementia, even among normal users.

[0024] The effects of the present invention are not limited to those mentioned above, and other effects not mentioned will be clearly understood by those skilled in the art from the following description. [Brief description of the drawings]

[0025] [Figure 1] 1 is a diagram illustrating a system for classifying patients with severe cognitive impairment through electroencephalogram data analysis according to an embodiment of the present invention.

[0026] [Diagram 2] FIG. 11 is a hardware configuration diagram of a server for classifying patients with severe cognitive impairment through electroencephalogram data analysis according to another embodiment of the present invention.

[0027] [Diagram 3] 11 is a flowchart of a method for classifying patients with severe cognitive impairment through electroencephalogram data analysis according to yet another embodiment of the present invention.

[0028] [Figure 4] 1 is a diagram illustrating a process of classifying and scoring a user according to a method for classifying patients with severe cognitive impairment through electroencephalogram data analysis in various embodiments.

[0029] [Diagram 5]1 is a diagram illustrating a method for generating quantified electroencephalogram data in various embodiments. [Figure 6] 1 is a diagram illustrating a method for generating quantified electroencephalogram data in various embodiments.

[0030] [Figure 7] 1 is a diagram illustrating an example of quantified electroencephalogram data in the form of a symmetric image generated by a method for generating quantified electroencephalogram data in various embodiments;

[0031] [Figure 8] 1 is a flowchart of a method for generating an optimized classification model according to various embodiments.

[0032] [Figure 9] 10 is a diagram illustrating a process of generating an optimized classification model according to the method of FIG. 9 in various embodiments.

[0033] [Figure 10] 1 is a graph illustrating a process of scoring the degree of severe cognitive impairment for users included in each group in various embodiments.

[0034] [Figure 11] 1 is a diagram illustrating exemplary result data derived from a method for classifying patients with severe cognitive impairment through electroencephalogram data analysis in various embodiments. [Figure 12] 1 is a diagram illustrating exemplary result data derived from a method for classifying patients with severe cognitive impairment through electroencephalogram data analysis in various embodiments. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0035] The advantages and features of the present invention, and the methods for achieving them, will become clear from the following detailed description of the embodiments in conjunction with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below, and may be embodied in various different forms, but the embodiments are provided to complete the disclosure of the present invention and to fully inform those skilled in the art of the present invention of the scope of the present invention, and the present invention is defined only by the scope of the claims.

[0036] The terms used in the present specification are for the purpose of describing the embodiments and are not intended to limit the present invention. In the present specification, the singular includes the plural unless otherwise specified in the text. The terms "comprises" and / or "comprising" used in the specification do not exclude the presence or addition of one or more other elements in addition to the elements mentioned. The same reference numerals refer to the same elements throughout the specification, and "and / or" includes each and every combination of one or more of the elements mentioned. Although "first", "second", etc. are used to describe various elements, it is understood that these elements are not limited by these terms. These terms are used merely to distinguish one element from another element. Therefore, it is understood that the first element mentioned below may be the second element within the technical concept of the present invention.

[0037] Unless otherwise defined, all terms (including technical and scientific terms) used herein may be used in the sense commonly understood by a person of ordinary skill in the art to which the present invention belongs. Furthermore, terms defined in commonly used dictionaries are not to be interpreted ideally or excessively unless specifically and explicitly defined.

[0038] The term "module" or "module" as used herein means a software or hardware component such as an FPGA or ASIC, and the "module" or "module" performs a certain function. However, the term "module" or "module" is not limited to software or hardware. A "module" or "module" may be configured to reside on an addressable storage medium and to execute on one or more processors. Thus, by way of example, a "module" or "module" includes components such as software components, object-oriented software components, class components, and task components, as well as processes, functions, attributes, procedures, subroutines, program code segments, drivers, firmware, microcode, circuits, data, databases, data structures, tables, arrays, and variables. The components and functions provided within a "module" or "module" may be combined into a smaller number of components and "modules" or "modules" or may be further separated into additional components and "modules" or "modules".

[0039] Spatially relative terms such as "below," "beneath," "lower," "above," "upper," and the like, may be used to easily describe the relationship of one component to another, as illustrated in the drawings. Spatially relative terms should be understood to include different orientations of the components in use or operation in addition to the orientation illustrated in the drawings. For example, if a component illustrated in the drawings is turned over, a component described as "below" or "beneath" another component may be placed "above" the other component. Thus, the exemplary term "below" can include both an orientation of below and above. Components may be oriented in other directions, and the spatially relative terms may be interpreted accordingly.

[0040] In this specification, the term "computer" refers to any type of hardware device including at least one processor, and may also include software configurations that operate on the hardware device in accordance with the embodiment. For example, the term "computer" refers to any type of device, including, but not limited to, a smartphone, tablet PC, desktop, notebook computer, and user clients and applications that run on each device.

[0041] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

[0042] Although each step described in this specification is described as being performed by a computer, the subject of each step is not limited to this, and depending on the embodiment, at least some of each step may be performed by different devices.

[0043] Although this specification describes classifying multiple users based on severe cognitive impairment (e.g., Alzheimer's disease dementia (ADD)), the present invention is not limited to this and may be applied in the same or similar form to various diseases that can be classified by analyzing the user's biometric data, including EEG data.

[0044] FIG. 1 is a diagram illustrating a system for classifying patients with severe cognitive impairment through electroencephalogram data analysis according to an embodiment of the present invention.

[0045] Referring to FIG. 1, a system for classifying patients with severe cognitive impairment through electroencephalogram data analysis according to one embodiment of the present invention may include a server 100 for classifying patients with severe cognitive impairment, a user terminal 200 and an external server 300.

[0046] Here, the severe cognitive impairment patient classification system through EEG data analysis illustrated in FIG. 1 is according to one embodiment, and its components are not limited to the embodiment illustrated in FIG. 1, and may be added, modified or deleted as necessary.

[0047] In one embodiment, the severe cognitive impairment patient classification server 100 may collect electroencephalogram data for a plurality of users, analyze the collected electroencephalogram data, and classify each of the plurality of users into one of a severe cognitive impairment group, a Within Normal Limits (WNL) group, a Preclinical AD group, a Non-AD Mild Cognitive Impairment (Non-AD MCI) group, and a Prodromal AD group, but is not limited thereto.

[0048] In various embodiments, the severe cognitive impairment patient classification server 100 may analyze the user's EEG data at the user's request to determine whether the user is a severe cognitive impairment patient, and provide a result report (e.g., FIGS. 11 and 12) including the determination result (e.g., information on the group to which the user belongs, information on the presence or absence of dementia or the probability (score) of dementia occurrence, etc.). For example, the severe cognitive impairment patient classification server 100 may be connected to a user terminal 200 and an EEG measuring device (not shown) through the network 400, and may generate and provide a result report by collecting and analyzing EEG data from the EEG measuring device in response to a diagnosis request received from the user terminal 200.

[0049] Here, the network 400 may refer to a connection structure that allows information exchange between each of a plurality of nodes such as terminals and servers, etc. For example, the network 400 may include a local area network (LAN), a wide area network (WAN), the Internet (WWW), a wired / wireless data communication network, a telephone network, a wired / wireless television communication network, etc.

[0050] In addition, the wireless data communication network may include, but is not limited to, 3G, 4G, 5G, 3GPP (3rd Generation Partnership Project), 5GPP (5th Generation Partnership Project), LTE (Long Term Evolution), WIMAX (World Interoperability for Microwave Access), Wi-Fi, the Internet, LAN (Local Area Network), Wireless LAN (Wireless Local Area Network), WAN (Wide Area Network), PAN (Personal Area Network), RF (Radio Frequency), Bluetooth network, NFC (Near-Field Communication) network, satellite broadcasting network, analog broadcasting network, DMB (Digital Multimedia Broadcasting) network, etc.

[0051] In one embodiment, the user terminal 200 can be connected to the severe cognitive impairment patient classification server 100 via the network 400, and can request the execution of a severe cognitive impairment patient classification process through EEG data analysis through a user interface (UI) provided by the severe cognitive impairment patient classification server 100, and in response thereto, can receive a result report including information on the group to which the user belongs, the presence or absence of dementia, or the probability (score) of dementia onset.

[0052] In various embodiments, the user terminal 200 is a wireless communication device that ensures portability and mobility, and may include, but is not limited to, all kinds of handheld-based wireless communication devices such as navigation, Personal Communication System (PCS), Global System for Mobile communications (GSM), Personal Digital Cellular (PDC), Personal Handyphone System (PHS), Personal Digital Assistant (PDA), International Mobile Telecommunication (IMT)-2000, Code Division Multiple Access (CDMA)-2000, W-Code Division Multiple Access (W-CDMA), Wireless Broadband Internet (Wibro) terminals, smartphones, smartpads, tablet PCs, etc.

[0053] In one embodiment, the external server 300 may be connected to the severe cognitive impairment patient classification server 100 through a network 400, and may store and manage various information / data required for the severe cognitive impairment patient classification server 100 to perform a severe cognitive impairment patient classification process through EEG data analysis, or may store and manage various information / data generated by performing a severe cognitive impairment patient classification process through EEG data analysis. For example, the external server 300 may be, but is not limited to, a storage server separately provided outside the severe cognitive impairment patient classification server 100. Hereinafter, a hardware configuration of the severe cognitive impairment patient classification server 100 will be described with reference to FIG. 2.

[0054] FIG. 2 is a hardware configuration diagram of a server for classifying patients with severe cognitive impairment through electroencephalogram data analysis according to another embodiment of the present invention. Referring to Fig. 2, a severe cognitive impairment patient classification server 100 (hereinafter referred to as "server 100") according to another embodiment of the present invention may include one or more processors 110, a memory 120 for loading a computer program 151 executed by the processor 110, a bus 130, a communication interface 140, and a storage 150 for storing the computer program 151. Here, Fig. 2 illustrates only components related to the embodiment of the present invention. Therefore, a person skilled in the art to which the present invention belongs will understand that other general components may be included in addition to the components illustrated in Fig. 2.

[0055] The processor 110 controls the overall operation of each component of the server 100. The processor 110 may include a central processing unit (CPU), a microprocessor unit (MPU), a microcontroller unit (MCU), a graphic processing unit (GPU), or any other type of processor commonly known in the technical field of the present invention.

[0056] Furthermore, the processor 110 may execute operations for at least one application or program for executing a method according to an embodiment of the present invention, and the server 100 may include one or more processors. In various embodiments, the processor 110 may further include a random access memory (RAM, not shown) and a read-only memory (ROM, not shown) for temporarily and / or permanently storing signals (or data) processed within the processor 110. The processor 110 may also be implemented in the form of a system on chip (SoC) including at least one of a graphics processing unit, a RAM, and a ROM.

[0057] The memory 120 stores various data, instructions and / or information. The memory 120 can load a computer program 151 from the storage 150 to perform the methods / operations according to various embodiments of the present invention. When the computer program 151 is loaded into the memory 120, the processor 110 can execute one or more instructions constituting the computer program 151 to perform the methods / operations. The memory 120 may be embodied as a volatile memory such as a RAM, although the scope of the present disclosure is not limited in this respect.

[0058] The bus 130 provides a communication function between the components of the server 100. The bus 130 may be implemented as various types of buses such as an address bus, a data bus, and a control bus. The communication interface 140 supports wired / wireless Internet communication of the server 100. The communication interface 140 may also support various communication methods other than Internet communication. To this end, the communication interface 140 may be configured to include a communication module that is widely known in the technical field of the present invention. In some embodiments, the communication interface 140 may be omitted.

[0059] The storage 150 can non-temporarily store the computer program 151. When the process of classifying patients with severe cognitive impairment through EEG data analysis is performed through the server 100, the storage 150 can store various information required to provide the process of classifying patients with severe cognitive impairment through EEG data analysis.

[0060] Storage 150 may be configured to include non-volatile memory such as ROM (Read Only Memory), EPROM (Erasable Programmable ROM), EEPROM (Electrically Erasable Programmable ROM), flash memory, etc., a hard disk, a removable disk, or any form of computer-readable recording medium widely known in the technical field to which the present invention belongs.

[0061] The computer program 151 may include one or more instructions that, when loaded into the memory 120, cause the processor 110 to perform the methods / operations according to various embodiments of the present invention. That is, the processor 110 may execute the one or more instructions to perform the methods / operations according to various embodiments of the present invention.

[0062] In one embodiment, the computer program 151 may include one or more instructions for performing a method for classifying patients with severe cognitive impairment through analysis of electroencephalogram data, including an electroencephalogram collecting step of collecting electroencephalogram data for a plurality of users, a first classification step of classifying the plurality of users into a severe cognitive impairment group or a non-severe cognitive impairment group by analyzing the collected electroencephalogram data, a second classification step of classifying the users included in the non-severe cognitive impairment group into a normal group or an amnestic Mild Cognitive Impairment (aMCI) group, and a third classification step of classifying the users included in the normal group into a normal range group or a cognitive impairment latency group, and classifying the users included in the amnestic mild cognitive impairment group into a non-Alzheimer's mild cognitive impairment group or a pre-dementia mild cognitive impairment group.

[0063] The steps of a method or algorithm described in connection with the embodiments of the present invention may be embodied directly in hardware, in a software module executed by hardware, or in a combination thereof. The software module may reside in a Random Access Memory (RAM), a Read Only Memory (ROM), an Erasable Programmable ROM (EPROM), an Electrically Erasable Programmable ROM (EEPROM), a Flash Memory, a hard disk, a removable disk, a CD-ROM, or any other form of computer readable recording medium commonly known in the art to which the present invention pertains.

[0064] The components of the present invention may be embodied as a program (or application) and stored on a medium to be executed in combination with a computer, which is hardware. The components of the present invention may be implemented as software programs or software elements, and similarly, embodiments may be implemented in programming or scripting languages ​​such as C, C++, Java, assembler, etc., including various algorithms embodied in a combination of data structures, processes, routines, or other programming constructs. Functional aspects may be embodied as algorithms executed by one or more processors. Hereinafter, a process for classifying patients with severe cognitive impairment through EEG data analysis performed by the server 100 will be described with reference to FIGS. 3 to 12.

[0065] Figure 3 is a flowchart of a method for classifying patients with severe cognitive impairment through electroencephalogram data analysis in accordance with yet another embodiment of the present invention, and Figure 4 is a diagram illustrating the process of classifying and scoring users according to a method for classifying patients with severe cognitive impairment through electroencephalogram data analysis in various embodiments. 3 and 4, in step S110, the server 100 may perform an electroencephalogram collecting operation for collecting electroencephalogram data for a plurality of users. For example, the server 100 may collect electroencephalogram data for a plurality of users measured in real time through an electroencephalogram measuring device (not shown). However, without being limited thereto, the server 100 may receive pre-stored electroencephalogram data for each of a plurality of users from an external server 300.

[0066] Here, the EEG data may refer to a plurality of unit EEG data (e.g., independent EEG signals measured through each channel) measured through an EEG measuring device (not shown) including a plurality of EEG measuring channels (e.g., a total of 19 channels (e.g., Fp1, Fp2, F3, F4, C3, C4, P3, P4, O1, O2, F7, F8, T3, T4, T5, T6, Fz, Cz, Pz)) attached to different positions on the user's head (scalp).

[0067] In various embodiments, the server 100 can collect, for each of a plurality of users, not only brainwave data measured when the plurality of users are in a normal state and not performing any other actions, but also brainwave data measured in the course of performing various tests (e.g., verbal fluency test, Boston naming test, brief mental status test, word list memory test, constructed behavior test, word list recall test, word list recognition test, constructed recall test, path making test A / B, etc.). In various embodiments, the server 100 may process each of the plurality of electroencephalogram data collected for each of the plurality of users to generate a plurality of quantified electroencephalogram data (QEEG), which will be described below with reference to FIGS.

[0068] 5 and 6 are diagrams illustrating methods for generating quantified electroencephalogram data in various embodiments. 5 and 6, in step S210, the server 100 may calculate frequency domains for a plurality of unit electroencephalogram data collected through a plurality of electroencephalogram measurement channels attached to different positions on the head of a first user.

[0069] In various embodiments, the server 100 may perform frequency conversion based on a predetermined range of frequencies for each of the plurality of unit EEG data to generate one or more spectrograms (e.g., FIG. 6(A)). Here, the spectrogram is for visualizing and understanding sound or waves, and may be related to an image that combines the characteristics of a waveform and a spectrum. The spectrogram may represent the difference in amplitude according to the change in the time axis and the frequency axis as the difference in print density or display hue. However, it is not limited thereto.

[0070] Also, the predetermined frequency range may be, but is not limited to, 1 to 45 Hz. In various embodiments, the server 100 may adjust the size of the spectrogram into a square matrix through overlapping if the time during which the brainwave data is measured is less than a preset time (e.g., 4 seconds). For example, the server 100 may calculate the relative power of each spectrogram composed of a square matrix based on the time axis (calculate the relative intensity of the corresponding time period) and perform scaling between -1 and 1 based on the largest value in each matrix to generate one or more spectrograms. However, the present invention is not limited to this.

[0071] In various embodiments, the server 100 calculates the frequency domain based on the Fast-Fourier Transform, and may calculate the frequency domain using a superlet technique to more efficiently represent changes over time and frequency, but is not limited thereto, and various frequency domain calculation methods (e.g., wavelet transform) may be applied. In step S220, the server 100 can generate multiple image sub-data (e.g., Figure 6(B)) by rearranging the frequency domains corresponding to multiple unit EEG data among multiple unit EEG data measured at the same time through EEG measurement channels according to the positions of the multiple EEG measurement channels based on the frequency domains calculated in step S210.

[0072] For example, the server 100 can generate video sub-data by rearranging multiple unit EEG data determined to have been measured at the same time according to the position (e.g., the position corresponding to the user's scalp surface) of multiple areas (i.e., 19 areas) on the user's scalp surface (e.g., based on the positions of a total of 19 EEG measurement channels, rearranging the multiple unit EEG data into 11 on the left and right sides (including three common channels (Fz, Cz, Pz))).

[0073] In various embodiments, the server 100 may generate multiple video sub-data for different time points. In step S230, the server 100 may store the rearranged unit EEG data in time series. For example, the server 100 may store the multiple image sub-data for different time points generated in step S230 in time series (e.g., FIG. 6B). Here, each of the multiple image sub-data may further include information related to channel connectivity as well as information related to the frequency domain.

[0074] In step S240, the server 100 can generate quantified electroencephalogram data. Here, the frequency used may be 1Hz to 45Hz, and the quantified EEG data may be expressed in the form of a symmetry image as shown in Fig. 6(B). This has the advantage that all information (e.g. frequency, power by frequency, location where the signal occurs, relationship with other signals, signal pattern (regularity, complexity) (time-dynamics)) can be confirmed at once with only the quantified EEG data, since it contains information on the topologic of each channel as well as the entire frequency range.

[0075] Here, the form of the quantified electroencephalogram data is not limited to the above-mentioned symmetric image, but various types of images such as a topology image, a time-frequency image, etc. may be applied.

[0076] In various embodiments, the server 100 may standardize (e.g., standardize standard score) the quantified electroencephalogram data generated in step S240 based on preset gender and age criteria, but is not limited thereto.

[0077] 3 and 4 again, in step S120, the server 100 may perform a first classification operation of analyzing the quantified electroencephalogram data to classify a plurality of users into a severe cognitive impairment group or a non-severe cognitive impairment group.

[0078] In various embodiments, the server 100 may use the first classification model to classify a plurality of users into a severe cognitive impairment group or a non-severe cognitive impairment group. Here, the first classification model is a model (e.g., a Random forest (8:2) 5 fold CV) model) that performs image analysis on quantified EEG data in the form of a symmetric image to extract a symmetry image value as a feature value, and calculates and classifies the probability of being a patient with severe cognitive impairment based on the extracted feature value, and may be a model that has been pre-trained using multiple quantified EEG data labeled with the presence or absence of a patient with severe cognitive impairment as training data.

[0079] Here, the first classification model is composed of one or more network functions, which may be composed of a collection of interconnected computational units that may generally be referred to as "nodes." Such "nodes" may be referred to as "neurons." The one or more network functions are composed of at least one or more nodes. The nodes (or neurons) that make up the one or more network functions may be interconnected by one or more "links." Within the first classification model, one or more nodes connected through links can form a relative input node and output node relationship. The concepts of input node and output node are relative, and any node that is in an output node relationship with one node can also be in an input node relationship with another node, and vice versa. As mentioned above, the input node to output node relationship can be generated around links. One or more output nodes can be connected to an input node through links, and vice versa.

[0080] In the relationship between an input node and an output node connected through a link, the value of the output node may be determined based on data input to the input node. Here, the node connecting the input node and the output node may have a weight. The weight may be variable and may be changed by a user or an algorithm so that the first classification model performs a desired function. For example, when one or more input nodes are connected to one output node through respective links, the output node may determine an output node value based on the value input to the input node connected to the output node and the weight set for the link corresponding to each input node.

[0081] As described above, in the first classification model, one or more nodes are connected to each other through one or more links to form input node and output node relationships within the first classification model. The characteristics of the first classification model may be determined by the number of nodes and links in the first classification model, the correlation between the nodes and links, and the weights assigned to each link. For example, if there are two first classification models that have the same number of nodes and links but different weights between the links, the two first classification models may be recognized as different from each other.

[0082] Some of the nodes constituting the first classification model may constitute one layer based on the distance from the first input node. For example, a set of nodes whose distance from the first input node is n may constitute n layers. The distance from the first input node may be defined by the minimum number of links that must be traversed to reach the corresponding node from the first input node. However, such a definition of layers is arbitrary for the purpose of explanation, and the order of layers in the first classification model may be defined in a manner different from that described above. For example, the layer of nodes may be defined by the distance from the final output node.

[0083] The first input node may refer to one or more nodes to which data is directly input without passing through a link in relation to other nodes in the first classification model. Or, in relation between nodes based on a link in the first classification model network, it may refer to a node that does not have other input nodes connected to a link. Similarly, the final output node may refer to one or more nodes that do not have an output node in relation to other nodes in the first classification model. Also, the hidden node may refer to a node that constitutes the first classification model other than the first input node and the last output node. The first classification model according to an embodiment of the present disclosure may have more nodes in the input layer than the nodes in the hidden layer close to the output layer, and may be a first classification model in which the number of nodes decreases as the input layer progresses to the hidden layer.

[0084] The first classification model may include one or more hidden layers. Hidden nodes in the hidden layers may receive the output of a previous layer and the output of surrounding hidden nodes as input. The number of hidden nodes in each hidden layer may be the same or different. The number of nodes in the input layer may be determined based on the number of data fields of the input data, and may be the same or different from the number of hidden nodes. Input data input to the input layer may be operated by hidden nodes in the hidden layer and may be output by a fully connected layer (FCL), which is an output layer.

[0085] In various embodiments, the first classification model may be a deep neural network (DNN). A deep neural network may refer to a first classification model that includes multiple hidden layers in addition to an input layer and an output layer. A deep neural network can be used to understand latent structures of data. That is, latent structures of photos, text, video, audio, and music (e.g., what objects are in the photo, what is the content and emotion of the text, what is the content and emotion of the audio, etc.) can be understood. Deep neural networks can include convolutional neural networks (CNNs), recurrent neural networks (RNNs), autoencoders, generative adversarial networks (GANs), restricted boltzmann machines (RBMs), deep belief networks (DBNs), Q networks, U networks, Siamese networks, and the like. The above description of deep neural networks is merely exemplary, and the present disclosure is not limited thereto.

[0086] In various embodiments, the network function may include an autoencoder. An autoencoder may be a type of artificial neural network for outputting output data similar to input data. An autoencoder may include at least one hidden layer, and an odd number of hidden layers may be disposed between the input and output layers. The number of nodes in each layer may be reduced from the number of nodes in the input layer to an intermediate layer called a bottleneck layer (encoding), and then expanded from the bottleneck layer to an output layer (symmetrical to the input layer) in a reduced and symmetric manner. The nodes in the dimensionality reduction layer and the dimensionality restoration layer may or may not be symmetrical.

[0087] Autoencoders can perform nonlinear dimensionality reduction. The number of input layers and output layers may correspond to the number of sensors remaining after preprocessing of the input data. In the autoencoder structure, the number of nodes in the hidden layer included in the encoder may decrease the farther it is from the input layer. If the number of nodes in the bottleneck layer (the layer with the fewest nodes located between the encoder and the decoder) is too small, a sufficient amount of information may not be transmitted, so it may be maintained at a certain number or more (e.g., more than half of the input layer). In various embodiments, the first classification model may be a model trained by a supervised learning method. Here, supervised learning generally refers to a method of labeling specific data and information related to the specific data to generate training data (e.g., FIG. 7) and learning using the generated training data, or a method of labeling two data having a causal relationship to generate training data and learning through the generated training data.

[0088] More specifically, the server 100 may use the labeled training data to perform training on one or more network functions constituting a classification model. For example, the server 100 may input each of the training input data to one or more network functions, and derive an error by comparing each of the output data calculated by the one or more network functions with each of the training output data corresponding to the label of each of the training input data. That is, in training the classification model, the training input data may be input to an input layer of one or more network functions, and the training output data may be compared with the output of one or more network functions. The server 100 can train a classification model based on the error between the results of one or more network functions on the training input data and the training output data (labels). In addition, the server 100 may adjust the weights of one or more network functions based on the error in a backpropagation manner. That is, the server 100 may adjust the weights so that the output of one or more network functions approaches the learning output data based on the error between the operation result of one or more network functions on the learning input data and the learning output data.

[0089] The server 100 may determine whether to interrupt the learning process by using the validation data when the learning process of one or more network functions is performed for a predetermined epoch or more. The predetermined epoch may be a part of the overall learning target epoch. The validation data may be composed of at least a portion of the labeled learning data. That is, the server 100 may perform learning of the classification model through the learning data, and after the learning of the classification model is repeated for a predetermined epoch or more, may determine whether the learning effect of the classification model is equal to or higher than a predetermined level using the validation data. For example, when performing learning with a target number of iterative learning times of 10 using 100 pieces of learning data, the server 100 may perform 10 iterative learning, which is a predetermined epoch, and then perform 3 iterative learning using 10 pieces of validation data. If the change in the classification model output during the 3 iterative learning is equal to or lower than a predetermined level, the server 100 may determine that further learning is meaningless and terminate the learning. In other words, the validation data may be used to determine the completion of learning based on whether the effect of epoch-by-epoch learning in the iterative learning of the classification model is above or below a certain level. The numbers of training data and validation data and the number of iterations described above are merely examples, and the present disclosure is not limited thereto.

[0090] The server 100 may generate a classification model by testing the performance of one or more network functions using the test data and determining whether or not to activate one or more network functions. The test data may be used to verify the performance of the classification model and may be composed of at least a portion of the training data. For example, 70% of the training data may be used for training the classification model (i.e., training to adjust weights to output result values ​​similar to the labels), and 30% may be used as test data for verifying the performance of the classification model. The server 100 may input the test data to the classification model that has completed training, measure the error, and determine whether or not to activate the classification model depending on whether the performance is equal to or greater than a predetermined performance. The server 100 verifies the performance of the trained classification model by using test data for the trained classification model, and if the performance of the trained classification model is above a predetermined standard, the server 100 can activate the classification model for use in other applications.

[0091] In addition, the server 100 may deactivate and discard a classification model that has completed training if the performance of the classification model is below a predetermined standard. For example, the server 100 may determine the performance of the generated classification model based on factors such as accuracy, precision, and recall. The above-mentioned performance evaluation criteria are merely examples, and the present disclosure is not limited thereto. According to an embodiment of the present disclosure, the server 100 may generate a plurality of classification models by training each classification model independently, and may evaluate the performance and use only classification models that have a certain level of performance or higher.

[0092] In various embodiments, the server 100 can calculate a probability of severe cognitive impairment and a probability of non-severe cognitive impairment for each of a plurality of users by performing image analysis on the quantified electroencephalogram data using the first classification model, and can classify a plurality of users into a severe cognitive impairment group or a non-severe cognitive impairment group using the calculated probability of severe cognitive impairment and the non-severe cognitive impairment. For example, the server 100 can classify users whose probability of severe cognitive impairment is equal to or greater than a preset first probability and whose probability of non-severe cognitive impairment is less than a preset second probability into a severe cognitive impairment group using the calculated probability of severe cognitive impairment and the non-severe cognitive impairment, and can classify the remaining users who are not classified into the severe cognitive impairment group into a non-severe cognitive impairment group. However, without being limited thereto, the server 100 may, in contrast to the above method, classify users whose probability of non-severe cognitive impairment is equal to or greater than a preset first probability and whose probability of severe cognitive impairment is less than a preset second probability into a non-severe cognitive impairment group, and classify the remaining users who are not classified into the non-severe cognitive impairment group into a severe cognitive impairment group.

[0093] However, in order to subdivide users classified into the non-severe cognitive impairment group into more groups based on the degree of severe cognitive impairment, it is preferable to classify only users whose probability of severe cognitive impairment is equal to or higher than a predetermined first probability, i.e., users who are clearly determined to have a severe cognitive impairment, into the severe cognitive impairment group, and classify the remaining users into the non-severe cognitive impairment group.

[0094] In various embodiments, the server 100 may perform filtering on the EEG data used to train the first classification model based on pre-set EEG data removal conditions (e.g., core clinical criteria for potential Alzheimer's disease dementia). For example, the server 100 can remove, from among a plurality of electroencephalogram data, electroencephalogram data that is determined to have an abnormality based on age and gender, and electroencephalogram data that is normal and has a TBR2 standard score above a certain level.

[0095] In addition, the server 100 can remove, from among the multiple EEG data, EEG data with insufficient objective evidence and atypical and questionable EEG data. However, the present invention is not limited to this. For example, if a user satisfies the core clinical criteria for Alzheimer's disease dementia in terms of the characteristics of cognitive damage, but the cognitive impairment shows a sudden onset or there is insufficient objective (medical history or cognitive test) evidence of gradual decline, the server 100 can remove such user's EEG data. In addition, the server 100 can remove from the learning data the EEG data of users who meet all the core clinical criteria for Alzheimer's disease dementia, but who have a history of strokes temporally associated with the development or weakening of cognitive impairment, users with multiple or large cerebral infarctions, users with a significant level of cerebrovascular disease confirmed by severe white matter hyperintense signal lesions, etc., users with core symptoms of Lewy body dementia excluding the presence of dementia itself, users with active neurological disease, non-neurological coexisting disease, or users who have taken drugs that may seriously affect cognitive function, but are not limited to these.

[0096] In step S130, the server 100 may perform a second classification operation of classifying the users included in the non-severe cognitive impairment group into a normal group or a memory impairment type mild cognitive impairment group. In various embodiments, the server 100 can use the second classification model to classify users with memory impairment among those included in the non-severe cognitive impairment group into a memory impairment-type mild cognitive impairment group, and classify users who are not classified into the memory impairment-type mild cognitive impairment group (e.g., users without memory impairment) into a normal group. Here, the second classification model may be, but is not limited to, Random Forest, LGBM, Gradient Boosting, and SVM, KNN, TREE, and balanced Random forest (BRF) models that extract TBR-related feature values.

[0097] In various embodiments, the server 100 can use the second classification model to preferentially select and classify users who have a memory disorder-related disease from among users included in the non-severe cognitive impairment group, and then perform an operation of selecting and classifying normal users who do not have a memory disorder-related disease. For this purpose, the second classification model may include two or more different classification models (e.g., a classification model specialized in sensitivity and a classification model specialized in specificity) and may perform classification operations using a combination of a classification model specialized in sensitivity and a classification model specialized in specificity.

[0098] In one embodiment, the server 100 may perform a classification operation by combining a sensitivity-specialized classification model and a specificity-specialized classification model regardless of order. For example, the server 100 may perform a classification operation through a sensitivity-specialized classification model according to a user's selection, and then perform a classification operation through a specificity-specialized classification model. As another example, the server 100 may perform a classification operation through a specificity-specialized classification model according to a user's selection, and then perform a classification operation through a sensitivity-specialized classification model.

[0099] In another embodiment, the second classification model may include at least one or more sensitivity-specialized classification models and at least one or more specificity-specialized classification models, and the server 100 may perform a classification operation using at least one or more sensitivity-specialized classification models and at least one or more specificity-specialized classification models. For example, the server 100 may perform a classification operation using a first specificity-specialized classification model, a classification operation using a first sensitivity-specialized classification model, and a classification operation using a second specificity-specialized classification model. The server 100 may also perform a classification operation through a combination of a plurality of classification models as needed. Here, the classification model with specialized sensitivity may be a model trained using EEG data of a user with a memory disorder-related disease as first learning data, i.e., a model that determines whether a user with a memory disorder-related disease has a memory disorder-related disease, and the classification model with specialized specificity may be a model trained using EEG data of a user without a memory disorder-related disease as second learning data, i.e., a model that determines whether a user without a memory disorder-related disease does not have a memory disorder-related disease.

[0100] In various embodiments, the server 100 uses the first learning data to train a classification model specialized in sensitivity, and uses the second learning data to train a classification model specialized in specificity. However, if the first learning data and the second learning data are unbalanced (e.g., if the difference between the number of first learning data and the number of second learning data is greater than a preset value), the ratio of the first learning data for training the classification model specialized in sensitivity and the second learning data for training the classification model specialized in specificity can be corrected (e.g., balanced random forest) to be less than a predetermined ratio, thereby overcoming the imbalance of the learning data between the two classification models.

[0101] In step S140, the server 100 can perform a third classification operation to classify users included in the normal group into a normal range group or a cognitive impairment latent stage group, and classify users included in the memory impairment type mild cognitive impairment group into a non-Alzheimer's mild cognitive impairment group or a pre-dementia mild cognitive impairment group. First, the server 100 may generate an optimized classification model for classifying users. For example, the server 100 may generate a third classification model for classifying users included in the normal group into a normal range group or a cognitive impairment latent stage group, and a fourth classification model for classifying users included in the memory impairment type mild cognitive impairment group into a non-Alzheimer's mild cognitive impairment group or a pre-dementia mild cognitive impairment group. Hereinafter, the method for generating an optimized classification model performed by the server 100 will be described with reference to FIGS. 8 and 9.

[0102] FIG. 8 is a flowchart of a method for generating an optimized classification model in various embodiments, and FIG. 9 is a diagram illustrating a process for generating an optimized classification model using the method of FIG. 9 in various embodiments. 9 and 10, in step S310, the server 100 can generate a plurality of multidimensional models including different feature value combinations. For example, the server 100 can generate a plurality of different feature value combinations by arbitrarily and repeatedly selecting three feature values, which is a preset number, from the ten feature values ​​(e.g., "FT1," "FT2," to "FT10" in FIG. 10), and can generate a plurality of multidimensional models (three-dimensional models) using each of the generated feature value combinations. In this case, the server 100 can select all feature value combinations (e.g., "FT1," "FT2," to "FT10" in FIG. 10) that have different combinations for the ten feature values. 10 It is possible to generate a multidimensional model for each of the three combinations of

[0103] Here, the feature values ​​constituting the multidimensional model may be derived by analyzing a plurality of electroencephalogram data collected from a plurality of users, but are not limited thereto. In step S320, the server 100 may calculate the accuracy of each of the multiple multi-dimensional models generated in step S310. For example, the server 100 may classify multiple users according to preset conditions using multiple multi-dimensional models, and compare the classified results with actual results to calculate the accuracy of the classification.

[0104] At this time, the server 100 may individually calculate classification accuracy according to different conditions to individually generate a third classification model for classifying users in the normal group and a fourth classification model for classifying users in the memory impairment type mild cognitive impairment group. For example, the server 100 may classify users included in the normal group into a normal range group or a cognitive impairment latent period group using a plurality of multidimensional models to generate the third classification model, and may compare the classification result with the actual result to calculate the classification accuracy for the normal group. In addition, the server 100 may classify users included in the memory impairment type mild cognitive impairment group into a non-Alzheimer's mild cognitive impairment group or a pre-dementia mild cognitive impairment group using a plurality of multidimensional models to generate the fourth classification model, and may compare the classification result with the actual result again to calculate the classification accuracy for the memory impairment type mild cognitive impairment group.

[0105] Here, actual results may refer to the results of directly classifying multiple users through various tests (e.g., verbal fluency test, Boston naming test, Mini-Mental Status Examination, word list memory test, constructed behavior test, word list recall test, word list recognition test, constructed recall test, path making test A / B, etc.) conducted on multiple users. In addition, the accuracy may be a numerical representation of the degree of similarity between the result of classification using a plurality of multidimensional models and the actual result, but is not limited thereto.

[0106] In step S330, the server 100 may select the top N multidimensional models in order of the accuracy calculated in step S320, count the number of feature values ​​included in each of the selected N multidimensional models, and sequentially select M feature values ​​starting from the feature value having the highest count value. 10, the server 100 may sort a plurality of multidimensional models in order of the accuracy calculated in step S320, and select the top five multidimensional models from the sorted plurality of multidimensional models. Thereafter, the server 100 may select the top M feature values ​​from the feature values ​​included in the top five multidimensional models.

[0107] In this case, the server 100 may select more feature values ​​than the number of feature values ​​included in the plurality of multidimensional models in order to generate a classification model with higher accuracy than the plurality of multidimensional models. For example, if the plurality of multidimensional models are three-dimensional models including three feature values, the server 100 may select the top four feature values ​​so that the optimized classification model includes four or more feature values.

[0108] In step S340, the server 100 may generate an optimized classification model including the M feature values ​​selected in step S330.

[0109] Referring again to Figures 3 and 4, the server 100 can use the third classification model to classify users who do not have a subjective cognitive decline (SCD) among the users included in the normal group into a normal range group, and classify users who have a subjective cognitive decline into a cognitive impairment latency group.

[0110] In addition, the server 100 can use the fourth classification model to classify users with retrieval failure among users included in the memory impairment type mild cognitive impairment group into the non-Alzheimer's mild cognitive impairment group, and classify users with encoding failure into the pre-dementia mild cognitive impairment group. In step S150, the server 100 can use the results of classifying multiple users through steps S120 to S140 to score the degree of severe cognitive impairment for each of the users included in the normal range group, the cognitive impairment latency group, the non-Alzheimer's mild cognitive impairment group, the pre-dementia mild cognitive impairment group, and the severe cognitive impairment group.

[0111] Here, the degree of severe cognitive impairment may refer to the degree of a factor that is a criterion for classifying each group. For example, the degree of severe cognitive impairment for a user included in the normal group may refer to the degree of subjective cognitive decline, the degree of severe cognitive impairment for a user included in the non-Alzheimer's mild impairment group among the memory impairment type mild cognitive impairment group may refer to the degree of retrieval impairment, and the degree of severe cognitive impairment for a user included in the pre-dementia mild cognitive impairment group may refer to the degree of storage impairment. In addition, the degree of severe cognitive impairment for a user included in the severe cognitive impairment group may refer to the degree of progression of actual severe cognitive impairment. However, it is not limited thereto.

[0112] In various embodiments, the server 100 can use the scoring model to score points for users belonging to each of the normal range group, the latent cognitive impairment group, the non-Alzheimer's mild cognitive impairment group, the pre-dementia mild cognitive impairment group, and the severe cognitive impairment group. Here, the scoring model may be a model that inputs information about a user belonging to each group (e.g., information about the group to which the user belongs and the degree of severe cognitive impairment) and calculates the probability of dementia of the corresponding user as a score. For example, the scoring model may generate a scoring function for scoring a score for each user belonging to the latent cognitive impairment group, the non-Alzheimer's mild cognitive impairment group, the pre-dementia mild cognitive impairment group, and the severe cognitive impairment group, and may input the input values ​​to the generated scoring function to calculate a score for the user. Here, the scoring function may be implemented in the form of a linear function having a different slope for each group as shown in FIG. 10, but is not limited thereto.

[0113] In various embodiments, the server 100 may assign scores in the range of 0 to 50 to users in the normal range group based on the degree of severe cognitive impairment, and may assign scores in the range of 50 to 60 to users in the cognitive impairment latency group based on the scoring function shown in FIG. 10. In addition, based on the scoring function shown in FIG. 10, the server 100 may assign scores in the range of 60 to 70 to users in the non-Alzheimer's mild cognitive impairment group, scores in the range of 70 to 85 to users in the pre-dementia mild cognitive impairment group, and scores in the range of 85 to 100 to users in the severe cognitive impairment group, depending on the degree of severe cognitive impairment.

[0114] In various embodiments, when the server 100 receives an input of a dementia diagnosis request including brainwave data from a first user, it can perform an operation of generating quantified brainwave data for the first user using the first user's brainwave data in response to the dementia diagnosis request, an operation of analyzing the quantified brainwave data to classify the first user into one of the groups, and an operation of scoring based on the classified group and the degree of severe cognitive impairment, and can combine the results derived by performing each operation to provide a result report (e.g., Figures 11 and 12).

[0115] Through this, it is possible to determine whether or not a user has severe cognitive impairment with just the simple action of collecting and inputting the user's brainwave data. It also has the advantage of allowing users to more clearly understand their own condition in relation to severe cognitive impairment and take appropriate measures by providing a numerical value indicating the degree to which a user is likely to develop severe cognitive impairment, even if the user is not a patient with severe cognitive impairment.

[0116] The method for classifying patients with severe cognitive impairment through electroencephalogram data analysis has been described with reference to the flowchart shown in the drawings. For ease of explanation, the method for classifying patients with severe cognitive impairment through electroencephalogram data analysis has been illustrated and described as a series of blocks, but the present invention is not limited to the order of the blocks, and some blocks may be performed in a different order than those illustrated and described herein or simultaneously. In addition, new blocks not described in the present specification and drawings may be added, or some blocks may be deleted or modified.

[0117] Although the embodiments of the present invention have been described above with reference to the attached drawings, those skilled in the art to which the present invention pertains will understand that the present invention can be embodied in other specific forms without changing the technical spirit or essential features of the present invention. Therefore, the embodiments described above should be understood to be illustrative in all respects and not restrictive.

Claims

1. 1. A method performed by a computing device, comprising: an electroencephalogram collecting step of collecting electroencephalogram data for a plurality of users; a first classification step of classifying the plurality of users into a severe cognitive impairment group or a non-severe cognitive impairment group based on a calculated probability by analyzing the collected electroencephalogram data using a first classification model; A second classification step of classifying the users included in the non-severe cognitive impairment group into a normal group or an amnestic mild cognitive impairment (aMCI) group based on the derived presence or absence of memory impairment using a second classification model; and and a third classification step of classifying the users included in the normal group into a normal range group or a cognitive impairment latent group based on the presence or absence of a subjective cognitive decline (SCD) derived using the optimized classification model, and classifying the users included in the memory impairment type mild cognitive impairment group into a non-Alzheimer's mild cognitive impairment group or a pre-dementia mild cognitive impairment group based on the type of memory impairment the users have derived using the optimized classification model.

2. The electroencephalogram acquisition step includes: collecting a plurality of unit electroencephalogram data through a plurality of electroencephalogram measuring channels attached to different positions on the head of a first user; generating quantified electroencephalogram (QEEG) data by converting the collected plurality of unit electroencephalogram data into numerical values ​​and images; and The method for classifying patients with severe cognitive impairment through electroencephalogram data analysis according to claim 1 , further comprising a step of standardizing the generated quantified electroencephalogram data based on preset gender and age criteria.

3. The step of generating quantified electroencephalogram (QEEG) data comprises: calculating a frequency domain for the measured plurality of unit electroencephalogram data based on a Fast-Fourier Transform; rearranging frequency domains corresponding to a plurality of unit electroencephalogram data measured at the same time among the plurality of unit electroencephalogram data measured based on the calculated frequency domain according to positions of the plurality of electroencephalogram measurement channels; and 3. The method of claim 2, further comprising: accumulating the rearranged unit EEG data in a time series manner to generate quantified EEG data in a symmetry image form.

4. The first classification step comprises: calculating a probability of severe cognitive impairment and a probability of non-severe cognitive impairment for each of the plurality of users by performing image analysis of the generated quantified electroencephalogram data using a first classification model; and The method for classifying patients with severe cognitive impairment through electroencephalogram data analysis according to claim 2, further comprising a step of classifying the plurality of users into the severe cognitive impairment group or the non-severe cognitive impairment group using the calculated probability of severe cognitive impairment and the calculated probability of non-severe cognitive impairment.

5. The second classification step comprises:

2. The method of claim 1, further comprising: classifying users having a memory impairment among the users in the non-severe cognitive impairment group into the memory impairment type mild cognitive impairment group; and classifying users not classified into the memory impairment type mild cognitive impairment group into the normal group.

6. The third classification step comprises: generating the optimized classification models for a user classification, the optimized classification models including a third classification model and a fourth classification model; classifying users who do not have a subjective cognitive decline (SCD) among the users included in the normal group into the normal range group and classifying users who have the subjective cognitive decline into the cognitive impairment latency group using the third classification model; and 2. The method of claim 1, further comprising: classifying users having a retrieval failure among users included in the memory impairment type mild cognitive impairment group into the non-Alzheimer's mild cognitive impairment group and classifying users having an encoding failure into the pre-dementia mild cognitive impairment group using the fourth classification model.

7. The step of generating an optimized classification model comprises: generating a plurality of different feature value combinations by repeatedly selecting a predetermined number of feature values ​​from among a plurality of feature values ​​derived by analyzing a plurality of electroencephalogram data collected from each of the plurality of users, and generating a plurality of multidimensional models using each of the generated feature value combinations; 7. The method of claim 6, further comprising the steps of: calculating an accuracy for each of the generated multidimensional models; and selecting top N multidimensional models in order of the calculated accuracy, counting the number of feature values ​​included in each of the selected N multidimensional models, sequentially selecting M feature values ​​starting from the feature value having the highest count value, and generating the optimized classification model including the selected M feature values.

8. 2. The method of claim 1, further comprising the step of scoring a degree of severe cognitive impairment for each of the users included in the normal range group, the cognitive impairment latency group, the non-Alzheimer's mild cognitive impairment group, the pre-dementia mild cognitive impairment group, and the severe cognitive impairment group using the classification results of the first classification step, the second classification step, and the third classification step.

9. The scoring step comprises: assigning a score within a range of 0 to 50 to the users who are included in the normal range group according to the degree of the severe cognitive impairment; giving a score within a range of 50 to 60 to the users included in the cognitive impairment latency period group according to the severity of the severe cognitive impairment; assigning a score within a range of 60 to 70 to the users included in the non-Alzheimer's mild cognitive impairment group according to the severity of the severe cognitive impairment; assigning a score within a range of 70 to 85 to the users included in the pre-dementia mild cognitive impairment group according to the severity of the severe cognitive impairment; and 9. The method of claim 8, further comprising the step of giving a score within a range of 85 to 100 to the users included in the severe cognitive impairment group according to the degree of the severe cognitive impairment.

10. Processor; Network interface; memory; and A computer program that is loaded into the memory and executed by the processor, The computer program includes: electroencephalogram (EEG) collection instructions for collecting electroencephalogram data for a plurality of users; first classification instructions for classifying the plurality of users into a severe cognitive impairment group or a non-severe cognitive impairment group based on a calculated probability by analyzing the collected electroencephalogram data using a first classification model; second classification instructions for classifying the users included in the non-severe cognitive impairment group into a normal group or an amnestic mild cognitive impairment (aMCI) group based on the presence or absence of a derived memory impairment using a second classification model; and and a third classification instruction for classifying users included in the normal group into a normal range group or a cognitive impairment latent group based on the presence or absence of a subjective cognitive decline (SCD) derived using the optimized classification model, and classifying users included in the memory impairment type mild cognitive impairment group into a non-Alzheimer's mild cognitive impairment group or a pre-dementia mild cognitive impairment group based on the type of memory impairment the user has derived using the optimized classification model.

11. In combination with a computing device, an electroencephalogram collecting step of collecting electroencephalogram data for a plurality of users; a first classification step of classifying the plurality of users into a severe cognitive impairment group or a non-severe cognitive impairment group based on a calculated probability by analyzing the collected electroencephalogram data using a first classification model; A second classification step of classifying the users included in the non-severe cognitive impairment group into a normal group or an amnestic mild cognitive impairment (aMCI) group based on the derived presence or absence of memory impairment using a second classification model; and A computer program recorded on a computer-readable recording medium for executing a third classification step of classifying users included in the normal group into a normal range group or a cognitive impairment latent group based on the presence or absence of a subjective cognitive decline (SCD) derived using an optimized classification model, and classifying users included in the memory impairment type mild cognitive impairment group into a non-Alzheimer's mild cognitive impairment group or a pre-dementia mild cognitive impairment group based on the type of memory impairment possessed by the user derived using the optimized classification model.

Citation Information

Patent Citations

  • System and method for predicting decrease of cognitive ability

    JP2007125362A

  • Method for classifying and treating physiologic brain imbalance using quantitative eeg

    JP2011172942A

  • Field-deployable concussion detector

    JP2011514823A

  • Methods and devices for multi-style neurological assessment

    JP2015507516A

  • Brain disease diagnosis support system, brain disease diagnosis support method, and program

    JP2016106940A