Blood-based cognitive ability evaluation method and device
The blood-based cognitive ability assessment method addresses the limitations of traditional tests by classifying individuals using classification algorithms and identifying key blood factors, offering a more accessible and reliable evaluation of cognitive abilities in the elderly, particularly for 'Super-Agers', enhancing health management and reducing testing burdens.
Patent Information
- Application Number
- PCT/KR2024/007480
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2023-11-15
- Filing Date
- 2024-05-31
- Publication Date
- 2024-12-05
AI Technical Summary
Current cognitive function tests for the elderly are time-consuming, burdensome, and difficult for those with reading difficulties, and they do not allow for frequent assessments, while existing research focuses mainly on diseases rather than evaluating superior cognitive abilities in healthy elderly individuals known as 'Super-Agers'.
A blood-based cognitive ability assessment method that uses actual and augmented test data to classify individuals into 'Super-Agers' or 'Typical Agers' through classification algorithms, identifying major blood factors associated with superior cognitive functions like language, visual, and visuospatial abilities, and creating an evaluation model for cognitive ability evaluation.
This method provides a more accessible, cost-effective, and reliable way to evaluate cognitive abilities, enabling continuous health management and distinguishing superior cognitive performance in healthy elderly individuals, reducing the burden and time of traditional testing.
Smart Images

Figure KR2024007480_05122024_PF_FP_ABST
Abstract
Description
Blood-based cognitive ability assessment method and device
[0001] Embodiments of the present invention relate to a blood-based cognitive ability evaluation method and device.
[0002] This application is the result of research conducted with the support of the Korea Health Industry Development Institute (No. HU20C0271) and the National IT Industry Promotion Agency (No. RS-2022-00155966, Artificial Intelligence Convergence Innovation Talent Development (Ewha Womans University), Project Identification Number: 1711179344) with funding from the government (Ministry of Health and Welfare and Ministry of Science and ICT) in 2024.
[0003] With the rapidly growing elderly population, health care for the elderly is emerging as a significant social concern. Research is actively underway on cognitive changes in old age, particularly related to brain health. While cognitive decline or impairment in old age is common with increasing age, these changes are not fixed but rather individual characteristics that can vary depending on various factors.
[0004] Cognitive tests, which measure cognitive function, are primarily used to diagnose conditions like mild cognitive impairment and dementia, and are typically conducted using paper-based or document-based computer programs. However, these paper-based tests are time-consuming and burdensome for older adults, making them difficult to use for those with reading difficulties. Furthermore, the inability to administer the same test multiple times makes it impossible to uniformly monitor cognitive changes.
[0005] Meanwhile, existing research on cognitive function has primarily focused on patients with conditions such as Alzheimer's, Parkinson's, diabetes, and hypertension. However, recent research on "super-agers," elderly individuals with robust cognitive abilities beyond their age, has been gaining attention. Super-agers possess cognitive abilities far superior to their age, leading to healthier later life, both physically and mentally. Therefore, these super-agers can serve as a yardstick for assessing cognitive function in older adults.
[0006] Accordingly, there is a growing need for the development of cognitive ability measurement technology that can be used by healthy subjects rather than for the purpose of diagnosing diseases, and there is a need for the development of cognitive ability assessment technology that can assess whether a person has strong cognitive ability for his or her age.
[0007] Embodiments of the present invention are intended to solve the above-mentioned problems and / or limitations, and to provide a cognitive ability evaluation method capable of evaluating cognitive ability based on a blood test.
[0008] In addition, the purpose is to provide a computer-readable recording medium having recorded thereon a program for executing the above method on a computer.
[0009] However, these tasks are exemplary and the scope of the present invention is not limited thereby.
[0010] According to one embodiment of the present invention, the method may include: obtaining actual test data including actual blood test results and cognitive function test results of test subjects; determining, based on the actual test data, major cognitive function items of a cognitive function test related to whether the test subjects belong to a first group; generating augmented test data based on the obtained actual test data; predicting, using a classification algorithm, whether individual data of the augmented test data belong to the first group; classifying individual data of the actual test data and the augmented test data into one of the first group and the second group; and extracting one or more major blood factors related to the first group for each major cognitive function item based on the classified actual test data and the augmented test data.
[0011] In the present invention, the step of determining the main items of the cognitive function test may include: a step of classifying each of the subjects into a first group or a second group; a step of learning a correlation between each item of the cognitive function test and whether the subjects belong to the first group based on the classification of the subjects into the first or second group and the actual experimental data; and a step of determining the main cognitive function items based on the learning.
[0012] In the present invention, the first group may be a group of Super Agers who have cognitive abilities higher than the average cognitive abilities of the age group to which the test subject belongs, and the second group may be a group of Typical Agers who are not Super Agers.
[0013] In the present invention, the step of generating the augmented test data may generate augmented test data including virtual blood test results and cognitive function test results using a preset data generation algorithm.
[0014] In the present invention, the step of predicting whether or not it belongs to the first group may generate a first prediction result using at least two different classification algorithms for the augmented inspection data, and stack the first prediction results to generate a final prediction result.
[0015] In the present invention, before inputting the at least two different classification algorithms, the augmented inspection data can be preprocessed using a label encoder (LabelEncoder) and a simple imputer (SimpleImputer).
[0016] A blood-based cognitive ability evaluation method according to the present invention, wherein the determined main cognitive function items are language memory, visual memory, and visuospatial items.
[0017] In the present invention, a step of creating an evaluation model for evaluating cognitive ability based on the major blood factors extracted for each major cognitive function item may be additionally included.
[0018] A computer-readable recording medium having recorded thereon a program for executing the method of the present invention on a computer.
[0019] According to another embodiment of the present invention, there is provided a blood-based cognitive ability evaluation device, comprising: a processor; a communication unit; and a memory, wherein the processor obtains actual test data including actual blood test results and cognitive function test results of test subjects, determines, based on the actual test data, major cognitive function items of a cognitive function test related to whether the test subjects belong to a first group, generates augmented test data based on the obtained actual test data, predicts, using a classification algorithm, whether individual data of the augmented test data belong to the first group, classifies individual data of the actual test data and the augmented test data into one of a first group and a second group, and extracts one or more major blood factors related to the first group for each major cognitive function item based on the classified actual test data and the augmented test data.
[0020] According to the aforementioned means for solving the problem of the present invention, embodiments of the present invention can assess a subject's cognitive ability based on a blood test, thereby increasing accessibility to cognitive ability assessment. Furthermore, the financial burden on the subject can be reduced and the time required for cognitive ability testing can be saved.
[0021] In addition, embodiments of the present invention can provide highly reliable evaluation results by evaluating cognitive ability based on a blood test, and can assist in continuous health management by enabling follow-up observation over time.
[0022] Of course, the scope of the present invention is not limited by these effects.
[0023] FIG. 1 is a drawing for explaining a cognitive ability evaluation system according to one embodiment of the present invention.
[0024] Figure 2 is a drawing illustrating the inspection data of Figure 1.
[0025] FIG. 3 is a block diagram schematically illustrating a cognitive ability evaluation device according to one embodiment of the present invention.
[0026] Figure 4 is a flowchart illustrating a cognitive ability evaluation method according to one embodiment of the present invention.
[0027] Figure 5(a) is a graph showing the major blood factors in the attention area.
[0028] Figure 5(b) is a graph showing the major blood factors in the language area.
[0029] Figure 5(c) is a graph showing the major blood factors in the language memory area.
[0030] Figure 5(d) is a graph showing the major blood factors in the visual memory area.
[0031] Figure 5(e) is a graph showing major blood factors in the spatiotemporal domain.
[0032] Figure 5(f) is a graph showing major blood factors in the frontal lobe region.
[0033] Figure 6 is a flowchart for explaining a cognitive ability evaluation method according to another embodiment of the present invention.
[0034] Figure 7 is a graph showing the correlation between each item of the cognitive function test and the super-ager group according to one embodiment of the present invention.
[0035] Figure 8 is a graph showing the accuracy of a prediction model by a classification algorithm according to an embodiment of the present invention.
[0036] Figures 9(a) to 9(e) are graphs for explaining blood factors associated with the first group determined for each cognitive function test item.
[0037] The advantages and features of the present invention, and the methods for achieving them, will become clearer with reference to the embodiments described in detail with the accompanying drawings. However, the present invention is not limited to the embodiments presented below, but can be implemented in various different forms, and it should be understood that all modifications, equivalents, and alternatives fall within the spirit and technical scope of the present invention.
[0038] The terminology used in this application is only used to describe specific embodiments and is not intended to limit the present invention. The singular expression includes the plural expression unless the context clearly indicates otherwise. In this application, it should be understood that the terms "comprise" or "have" indicate the presence of a feature, number, step, operation, component, part, or combination thereof described in the specification, but do not exclude in advance the possibility of the presence or addition of one or more other features, numbers, steps, operations, components, parts, or combinations thereof.
[0039] Additionally, the connecting lines or connecting members between components depicted in the drawings are merely exemplary representations of functional connections and / or physical or circuit connections. In an actual device, connections between components may be represented by various functional connections, physical connections, or circuit connections that may be replaced or added.
[0040] Furthermore, some embodiments of the present disclosure are susceptible to various modifications. Specific embodiments are illustrated in the drawings and described in detail in the detailed description. The effects and features of the present embodiments, as well as the methods for achieving them, will become clearer with reference to the detailed descriptions below, along with the drawings. However, the present embodiments are not limited to the embodiments disclosed below and may be implemented in various forms.
[0041] 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 addition, in the following examples, singular expressions include plural expressions unless the context clearly indicates otherwise. In addition, 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 addition, in the following examples, when it is said that a part is “connected” to another part, this includes not only cases where it is “directly connected,” but also cases where it is “electrically connected” with another element in between.
[0042] Hereinafter, the following embodiments 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.
[0043] FIG. 1 is a drawing for explaining a cognitive ability evaluation system according to one embodiment of the present invention, and FIG. 2 is a drawing exemplifying the test data (20) of FIG. 1.
[0044] The cognitive ability evaluation server (100, see FIG. 3) can obtain test data (20) including blood test results and cognitive function test results of each subject from the subject group (10).
[0045] Here, a subject refers to someone who underwent blood tests and cognitive function tests and provided the test results. This group of subjects can be categorized into multiple age groups, and each subject can be categorized into those with cognitive abilities above average for their age group and those without.
[0046] That is, the group of subjects consisting of elderly people can be classified into a super-ager group and a typical-ager group. In this specification, "super-ager" refers to an elderly person who has the cognitive abilities of a middle-aged person. An elderly person is defined as someone aged 60 years or older. Of course, the age range defining elderly people can be set to a different age range. Middle age is defined as 45 years of age. Of course, the age range defining middle age can be set to a different age range. The cognitive abilities of the subjects can be determined through clinical assessment.
[0047] Super-agers are considered individuals with high cognitive reserve and / or brain resilience. Cognitive reserve is a concept based on brain plasticity, which refers to the ability to actively cope with the brain's changes associated with aging. While the specific factors that determine cognitive reserve have not been identified, environmental factors such as the subject's education level are known to influence cognitive reserve. Brain resilience refers to the ability to repair damaged nerve cells.
[0048] The test data (20) may include blood test data (21), cognitive test data (22), and additional data (23). The blood test data (21) may be obtained from a blood test performed during a general health checkup, which is a result of a blood test of a subject. For example, the blood test data (21) may include results of hemoglobin test, fasting blood sugar test, liver function test, serum creatinine test, cholesterol test, and glomerular filtration rate test, and may include values of factors such as HDL-Cholesterol, Cl, C-peptide, and PSK9. However, the present invention is not limited thereto, and the blood test data (21) may include results from various blood test methods.
[0049] Cognitive test data (22) is the result of a cognitive function test of the subject and may include results for multiple test items. For example, cognitive test data (22) may include test results for attention, language, verbal memory, visual memory, visuospatial, and frontal lobe.
[0050] In addition to blood test data (21) and cognitive test data (22), the test data (20) may include additional data (23) such as personal data such as age, gender, and education level.
[0051] The cognitive ability evaluation device (100) can extract major blood factors (30) from the test data (20) of a group of subjects (10). Specifically, the cognitive ability evaluation device (100) can perform a correlation analysis of blood test data (21) and cognitive test data (22) to extract major blood factors (30) that affect cognitive ability among a plurality of blood factors included in the blood test data (21).
[0052] The extracted major blood factors (30) can be used to create an evaluation model (40) for evaluating the cognitive ability of a subject (50). That is, the cognitive ability evaluation device (100) can build a cognitive ability evaluation model (40) that evaluates cognitive ability using the major blood factors (30) included in the blood test results.
[0053] The cognitive ability evaluation device (100) extracts blood information consisting of values corresponding to major blood factors (30) from blood sample data obtained from a blood sample (60) of a subject (50), and can evaluate the cognitive ability of the subject (50) using the extracted blood information and the evaluation model (40). The cognitive ability evaluation device (100) can evaluate whether the subject (50) has a cognitive ability higher than the average cognitive ability of the age group to which the subject belongs. In addition, the cognitive ability evaluation device (100) can evaluate the cognitive ability of the subject (50) for each item of a cognitive function test.
[0054] Hereinafter, an example of the operation of the cognitive ability evaluation device (100) will be described with reference to FIGS. 3 to 10.
[0055] FIG. 3 is a block diagram schematically illustrating a cognitive ability evaluation device (100) according to one embodiment of the present invention.
[0056] Referring to FIG. 3, a cognitive ability evaluation device (100) according to an embodiment of the present invention may include a communication unit (110), a processor (120), a memory (130), and an input / output interface (140). The cognitive ability evaluation device (100) illustrated in FIG. 3 only illustrates components related to the present embodiment in order to prevent the features of the present embodiment from being obscured. Therefore, it will be understood by those skilled in the art related to the present embodiment that other general components may be included in addition to the components illustrated in FIG. 3. In addition, it will be apparent to those skilled in the art related to the present embodiment that the communication unit (110), the processor (120), the memory (130), and the input / output interface (140) illustrated in FIG. 3 may be implemented as independent devices.
[0057] A cognitive ability evaluation device (100) according to one embodiment of the present invention can extract major blood factors used for cognitive ability evaluation from a subject's test data, create an evaluation model for evaluating cognitive ability based on the extracted blood factors, and evaluate the subject's cognitive ability using a blood sample of the subject.
[0058] A user can check the results of a cognitive ability assessment of a subject through a cognitive ability assessment device (100). Here, a user is someone who wishes to check the results of a cognitive ability assessment of a subject. For example, the user may be a doctor, a clinical scientist, the subject, or the subject's manager.
[0059] In one embodiment, the cognitive ability assessment device (100) may be mounted on a user terminal. The user terminal may be a fixed or mobile terminal implemented as a computer device, and may communicate with other user terminals and / or servers via a network using wireless or wired communication methods. For example, the user terminal may be, but is not limited to, a smartphone, a mobile phone, a navigation device, a computer, a laptop, a digital broadcasting terminal, a PDA (Personal Digital Assistant), a PMP (Portable Multimedia Player), a tablet PC, a wearable device, etc.
[0060] In another embodiment, the cognitive ability assessment device (100) may be installed on a server. In this case, the server may be implemented as a computer device or multiple computer devices that communicate with multiple user terminals via a network to provide commands, codes, files, content, services, etc. For example, the cognitive ability assessment device (100) may be installed on two servers, with one server learning an assessment model for evaluating cognitive ability, and the other server evaluating the subject's cognitive ability from blood sample data based on the learned algorithm.
[0061] When the cognitive ability assessment device (100) is installed on a server, the server can provide a file for installing an application to a user terminal connected via a network. In this case, the user terminal can install the application using the file provided by the server. In addition, the user terminal can access the server and receive services or content provided by the server under the control of the operating system (OS) included in the user terminal and at least one program (e.g., a browser or an installed application).
[0062] The communication unit (110) may provide a configuration or function for the cognitive ability evaluation device (100) to communicate with an external device (not shown) via a network. For example, control signals, commands, data, etc. provided under the control of the processor (120) may be transmitted to the external device via the communication unit (110) and the network.
[0063] The communication unit (110) can receive test data, including blood test results and cognitive function test results, from an external device. Furthermore, the communication unit (110) can receive blood sample data and additional information from the subject from the external device. Furthermore, the communication unit (110) can transmit cognitive ability assessment results to an external device, such as a user terminal.
[0064] The processor (120) may be configured to process commands of a computer program by performing basic arithmetic, logic, and input / output operations. Here, the commands may be provided to the processor (120) by the memory (130) or the communication unit (110). For example, the processor (120) may be configured to execute commands received according to program code stored in a recording device such as the memory (130). In addition, the processor (120) may generally control the operations of other components included in the cognitive ability assessment device (100).
[0065] The processor (120) may be implemented as an array of a plurality of logic gates, or may be implemented as a combination of a general-purpose microprocessor and a memory storing a program that can be executed on the microprocessor. For example, the processor (120) may include a general-purpose processor, a central processing unit (CPU), a microprocessor, a digital signal processor (DSP), a controller, a microcontroller, a state machine, etc. In some environments, the processor (120) may include an application-specific integrated circuit (ASIC), a programmable logic device (PLD), a field programmable gate array (FPGA), etc. For example, the processor (120) may refer to a combination of processing devices, such as a combination of a digital signal processor (DSP) and a microprocessor, a combination of a plurality of microprocessors, a combination of one or more microprocessors coupled with a digital signal processor (DSP) core, or any other such combination of configurations.
[0066] The memory (130) may include any non-transitory computer-readable recording medium. For example, the memory (130) may include a non-permanent mass storage device such as a random access memory (RAM), a read only memory (ROM), a disk drive, a solid state drive (SSD), a flash memory, etc. As another example, the non-permanent mass storage device such as a ROM, an SSD, a flash memory, a disk drive, etc. may be a separate permanent storage device distinct from the memory. In addition, the memory (130) may store an operating system (OS) and at least one program code (e.g., a code for the processor (120) to perform an operation to be described later with reference to FIG. 4).
[0067] The input / output interface (140) may be a means for interfacing with a device for input and / or output (e.g., a keyboard, a mouse, a touch screen, etc.) that may be connected to or included in the cognitive ability evaluation device (100). In FIG. 3, the input / output interface (140) is illustrated as an element configured separately from the processor (120), but is not limited thereto, and the input / output interface (140) may be configured to be included in the processor (120).
[0068] Figure 4 is a flowchart illustrating a cognitive ability evaluation method according to one embodiment of the present invention.
[0069] Referring to FIG. 4, a method for evaluating cognitive ability according to an embodiment of the present invention includes a step of obtaining test data of test subjects (S100), a step of extracting major blood factors used for evaluating cognitive ability from the test data (S200), a step of creating an evaluation model for evaluating cognitive ability based on the extracted major blood factors (S300), and a step of extracting information corresponding to major blood factors from a blood sample of the subject and evaluating the subject's cognitive ability using the extracted information and the evaluation model (S400). Hereinafter, the technical idea of the present invention will be explained by specifically exemplifying the contents of an actual experiment by a researcher.
[0070] First, a cognitive ability evaluation method according to one embodiment of the present invention obtains test data of test subjects (S100).
[0071] The cognitive ability evaluation device (100) can receive test data of subjects from an external device through a communication unit (110) and transmit the received test data to a processor (120).
[0072] The test data may include blood test data (21), cognitive test data (22), and additional data (23). The blood test data (21) may be obtained from a blood test performed during a general health checkup, representing the subject's blood test results. However, the present invention is not limited thereto, and the blood test data (21) may include results obtained from various blood test methods.
[0073] Cognitive test data (22) is the result of a cognitive function test of the subject and may include results for multiple test items. For example, cognitive test data (22) may include test results for attention, language, verbal memory, visual memory, visuospatial, and frontal lobe.
[0074] In addition to blood test data (21) and cognitive test data (22), the test data (20) may include additional data (23) such as personal data such as age, gender, and education level, and may also include physical examination data such as body mass index (BMI).
[0075] As a specific example, the group of subjects formed by the researcher consisted of 81 elderly people aged 60 or older, and blood tests and cognitive function tests were performed on all subjects.
[0076] Next, major blood factors used for cognitive ability evaluation are extracted from the test data (S200).
[0077] The processor (120) can perform a correlation analysis between blood test results and cognitive function test results included in the test data and extract key blood factors to be used in cognitive ability assessment. At this time, the processor (120) may also use additional data included in the test data for the correlation analysis.
[0078] Specifically, the processor (120) can classify subjects as having cognitive abilities higher than the average cognitive abilities of their age group using the cognitive function test results included in the test data. That is, the group of subjects is classified into multiple age groups, and each subject can be classified into those who have cognitive abilities higher than the average cognitive abilities of their age group and those who do not.
[0079] In a specific experimental example, the researcher classified the group of subjects into a group of 41 super-agers (those with cognitive abilities above the average for their age group) and a group of 40 typical-agers based on cognitive function test results. For convenience of explanation, the super-ager group will be referred to as Group 1, and the typical-ager group as Group 2.
[0080] The results of the statistical analysis of the cognitive function test results for the SuperAger group and the general elderly group are shown in Table 1 below. The cognitive function test results showed significant differences between the two groups in language, verbal memory, visual memory, and visuospatial, with p<0.05. Here, p represents the probability of significance (p-value). In addition, verbal memory, visual memory, and visuospatial showed significant differences between the two groups, with p<0.01.
[0081] [Table 1]
[0082]
[0083] The results of statistical analysis on the blood test results of the two groups are shown in Table 2 below. The blood test results showed significant differences between the two groups in HDL, Cl, C-peptide, and PCSK9 (p<0.05).
[0084] [Table 2]
[0085]
[0086] In addition, a correlation analysis was performed between the cognitive function test results and blood test results of the experimental group, and the following features were significantly derived for each item of the cognitive function test.
[0087] Attention showed a significant correlation with HDL Cholesterol (r=0.254, p=0.022) and Insulin (r=-0.223, p=0.046). Language showed a significant correlation with MCHC (r=0.261, p=0.019). Verbal memory showed a significant correlation with C-peptide (r=-0.281, p=0.011), EPO (r=0.227, p=0.042), and HDL Cholesterol (r=0.219, p=0.049). Visual memory showed a significant correlation with PCSK9 (r=0.281, p=0.011), HDL Cholesterol (r=0.264, p=0.017), and Free T4 (r=-0.234, p=0.036). Visuospatial showed a significant correlation with HDL Cholesterol (r=0.341, p=0.002), Triglyceride (r=-0.332, p=0.002), Free T4 (r=-0.326, p=0.003), ApoE (r=-0.287, p=0.009), Insuline (r=-0.231, p=0.038), and AST (r=-0.228, p=0.041). The frontal lobe showed a significant correlation with oxLDL (r=-0.356, p=0.001), Hb (r=-0.288, p=0.009), Hct (r=-0.262, p=0.018), Vimentin (r=-0.246, p=0.027), K (r=0.237, p=0.034), RBC (r=-0.227, p=0.041), and Creatinine (r=-0.226, p=0.042). Here, r represents the Pearson correlation coefficient.
[0088] The processor (120) performs a correlation analysis between each variable included in the test data using the test data of the super-aged group and the general elderly group classified from the test subject group, and extracts (i.e., selects variables (feature selection)) major blood factors to be used for cognitive ability evaluation based on a preset variable selection algorithm. At this time, the number of major blood factors extracted may be 10, but is not limited thereto.
[0089] When creating an evaluation model using machine learning, variables (features, or attributes) are used. However, using too many or inappropriate variables can lead to overfitting or reduced predictive accuracy. Therefore, for a machine learning model to achieve high predictive accuracy, it is necessary to use an appropriate combination of variables. In other words, by selecting the variables most highly correlated with the desired predictor, the complexity of the machine learning model can be reduced while using as few variables as possible.
[0090] The variable selection algorithm may be, but is not limited to, the selectKBest algorithm.
[0091] The major blood factors extracted from the test data using the preset variable selection algorithm may include at least one of HbA1c, Total calcium, Phosphorus, Glucose, BUN, Creatinine, eGFRMDRD, eGFRCKDEPI, Triglyceride, Total cholesterol, Total Protein, Albumin, AST, ALT, ALP, Total bilirubin, HDLCholesterol, LDLCholesterol, Na, K, Cl, CRP, TSH, Free T4, Insulin, C-peptide, Folic acid, Vit B12, WBC, RBC, Hb, Hct, MCV, MCH, MCHC, RDW(CV), Platelet, PCSK9, ApoE, CD36, EPO, AGE, RAGE, Vimentin, FABP-3, FABP-4, oxLDL, Adioponectin, and Leptin.
[0092] Meanwhile, the processor (120) can extract major blood factors for each item of the cognitive function test using a preset variable selection algorithm, in one embodiment, selectKBest. Hereinafter, examples of major blood factors selected for each item of the cognitive function test are described in FIGS. 5(a) to 5(e).
[0093] Figure 5(a) is a graph showing the main blood factors of the attention area, Figure 5(b) is a graph showing the main blood factors of the language area, Figure 5(c) is a graph showing the main blood factors of the verbal memory area, Figure 5(d) is a graph showing the main blood factors of the visual memory area, Figure 5(e) is a graph showing the main blood factors of the spatiotemporal area, and Figure 5(f) is a graph showing the main blood factors of the frontal lobe area. Each graph calculates the importance based on the degree to which each variable contributes to evaluating cognitive ability in each cognitive function test item, and presents the top 10 variables with high importance as a graph.
[0094] Referring to Figure 5(a), the major blood factors in the attention area were extracted as Total calcium, Triglyceride, Total cholesterol, Total bilirubin, HDL-cholesterol, Sodium (Na), Thyroid-Stimulating Hormone (TSH), CD36, Erythropoietin (EPO), and Leptin.
[0095] Referring to Figure 5(b), the major blood factors in the language region were extracted as Glucose, Insulin, C-peptide, Red Blood Cell (RBC), Hemoglobin (Hb), Mean Corpuscular Hemoglobin (MCH), Mean Corpuscular Hemoglobin Concentration (MCHC), Receptor for Advanced Glycation End Products (RAGE), Adiponectin, and Leptin.
[0096] Referring to Figure 5(c), the major blood factors in the language memory area were extracted as Hemoglobin A1c (HbA1c), Glucose, Alanine Aminotransferase (ALT), HDL-cholesterol, Free-T4, C-peptide, Mean Corpuscular Volume (MCV), MCH, Platelet, and EPO.
[0097] Referring to Fig. 5(d), the major blood factors in the visual memory area were extracted as HbA1c, Total cholesterol, HDL-cholesterol, LDL-cholesterol, Free T4, White Blood Cell (WBC), RBC, Hematocrit (Hct), PCSK9, and oxidized LDL (oxLDL).
[0098] Referring to Fig. 5(e), the major blood factors in the spatiotemporal domain are extracted as HbA1c, Blood Urea Nitrogen (BUN), Triglyceride, Total cholesterol, Aspartate Aminotransferase (AST), HDL-cholesterol, Free T4, Insulin, C-peptide, and Apolipoprotein E (ApoE).
[0099] Referring to Figure 5(f), the major blood factors in the frontal lobe region are extracted as Triglyceride, Total Cholesterol, Potassium (K), Hemoglobin (Hb), Hematocrit (Hct), MCH, MCHC, CD36, Advanced Glycation End Products (AGE), and oxLDL.
[0100] Figures 5(a) to 5(e) illustrate examples of major blood factors extracted for each cognitive function test item, and the present invention is not limited thereto.
[0101] Next, an evaluation model for evaluating cognitive ability is created based on the extracted major blood factors (S300).
[0102] The processor (120) can learn an evaluation model using major blood factors and generate an evaluation model for evaluating cognitive ability. The evaluation model can be trained to receive blood sample data from a subject and output the results of an evaluation of the subject's cognitive ability.
[0103] Meanwhile, in the step (S300) of generating an evaluation model, the processor (120) can generate augmented data based on the actual test data of the test subjects to learn the evaluation model in order to solve the quantitative problem of the test data (training data) acquired from the test subject group. In this specification, for the purpose of distinguishing the data, the test data acquired from the actual test subjects may be referred to as actual test data, and the generated augmented data may be referred to as augmented test data.
[0104] According to one embodiment, the processor (120) can generate augmented test data using a preset augmented data generation algorithm. As the augmented data generation algorithm, Generation of REAlistic Tabular Data (GReaT), which is a large language model-based tabular data augmentation, can be used, but is not limited thereto. As a specific example, the researcher generated 729 augmented data using an LLM-based algorithm, and can learn an evaluation model for each cognitive function item using a total of 810 data.
[0105] An evaluation model according to one embodiment of the present invention may be a statistical learning algorithm implemented based on the structure of a biological neural network in machine learning technology and cognitive science, or a machine learning model that executes the algorithm. For example, a machine learning model may represent a model with problem-solving capabilities by learning to reduce the error between the correct output corresponding to a specific input and the inferred output by repeatedly adjusting the weights of synapses, which are artificial neurons that form a network by combining synapses, as in a biological neural network. For example, a machine learning model may be implemented as a multilayer perceptron (MLP) composed of multiple layers of nodes and connections between them.
[0106] The evaluation model according to the present embodiment can be implemented using one of various artificial neural network model structures including MLP. For example, the evaluation model can be composed of an input layer that receives input signals or data from the outside, an output layer that outputs output signals or data corresponding to the input data, and at least one hidden layer located between the input layer and the output layer that receives signals from the input layer, extracts characteristics, and transmits them to the output layer. The output layer receives signals or data from the hidden layer and outputs them to the outside.
[0107] For example, the evaluation model may be generated using at least one of, but not limited to, linear regression, Lasso, decision tree, random forest, support vector machine, gradient boost, AdaBoost, and XgBoost.
[0108] Meanwhile, as described above, in the step (S300) of generating an evaluation model, the evaluation model can be verified using augmented data generated based on the test data. More specifically, the processor (120) can generate augmented data based on the test data obtained from a group of subjects. For example, the LLM-based GReaT model can be used as an augmented data generation algorithm.
[0109] As a specific example, the researcher generated 729 augmented data sets using the GreaT algorithm and performed 10-fold cross-validation on each learning model using a total of 810 sets of data. The cross-validation results showed that the 729 augmented data sets were used for training and validated against 81 sets of original data sets. Hyperparameter tuning was performed using Pycaret, a well-known automated machine learning (AutoML) tool. The evaluation functions used were the mean squared error (MSE), root mean squared error (RMSE), and mean absolute error (MAE).
[0110] [Table 3] below shows the error rates for 17 regression analysis algorithms (Linear, Least Angle, Lasso, Lasso Least Angle, Ridge, Bayesian Ridge, Orthogonal Matching Pursuit, Decision Tree, Extra Trees, Random Forest, Huber, Elastic Net, Passive Aggressive, K Neighbors, AdaBoost, Gradient Boost, Light Gradient Boost) for each cognitive domain. Referring to [Table 3], it can be seen that the augmented data shows an average error rate of 0.15 based on the MAE.
[0111] [Table 3]
[0112]
[0113] In addition, the processor (120) can measure the influence of blood factors in predicting scores for each item of cognitive ability in order to build an evaluation model. More specifically, the influence of 10 blood factors for each item of cognitive ability derived above can be evaluated using SHapley Additive explanation (SHAP). In a specific embodiment, the processor (120) can derive a blood factor that has a greater influence among the 10 factors for each cognitive ability item. In addition, the processor (120) can derive whether the level of each blood factor affects the increase in the cognitive ability score depending on the degree of high or low. For example, the processor (120) can determine that among the 10 blood factors, Leptin has the greatest influence on the attention item, and that the lower the Leptin level, the higher the attention score.
[0114] Next, blood information corresponding to major blood factors is extracted from blood sample data obtained from the subject, and the subject's cognitive ability is evaluated using the extracted blood information and the evaluation model (S400).
[0115] The processor (120) can use the assessment model to evaluate the cognitive ability of the subject. Furthermore, the processor (120) can also use the assessment model to evaluate the cognitive ability of the subject for each cognitive function test item. Furthermore, the processor (120) can use the assessment model to assess whether the subject possesses cognitive ability above the average cognitive ability of his or her age group.
[0116] Specifically, the processor (120) can obtain blood sample data generated from the blood test results of a blood sample of a subject. The processor (120) can extract values corresponding to major blood factors from the blood sample data and extract blood information comprised of major blood factor values.
[0117] The processor (120) can input the extracted blood information into an evaluation model to output the cognitive ability evaluation results of the subject. The cognitive ability evaluation results output from the processor (120) may be scores for each cognitive function test item, but are not limited thereto.
[0118] In another embodiment, the processor (120) may evaluate the cognitive ability of the subject by calculating a cognitive function score of the subject and comparing the calculated cognitive function score with the average cognitive function score of the subject for the corresponding age to determine whether the subject has cognitive ability above the average for the corresponding age. In this case, the processor (120) may also output the degree to which the subject's cognitive ability deviates from the average cognitive ability of the subject for the corresponding age as a probability value.
[0119] Figure 6 is a flowchart for explaining a cognitive ability evaluation method according to another embodiment of the present invention.
[0120] First, the cognitive ability assessment method of the present invention acquires actual test data including blood test results and cognitive function test results of the subjects (610). Step 610 may be similar to step S100 of FIG. 4 described above, and any duplicate description will be omitted. Similar to the experimental example described in step S100 of FIG. 4, the subject group consists of 81 elderly people aged 60 or older, and the actual test data may be blood test and cognitive function test data of a test conducted on the actual subject group.
[0121] With respect to the embodiment of FIG. 4 described above, steps 620 to 660 described below may replace steps S200 and S300 of FIG. 4. This will be described in detail below.
[0122] Next, the cognitive ability assessment method of the present invention determines the main cognitive function items of the cognitive function test related to whether the subjects belong to the first group based on actual test data (620). That is, according to one embodiment, the annual relationship between each item of the cognitive function test and whether the subjects are Super Ager can be learned based on the actual test results. In this regard, according to one embodiment of the present invention, all subjects can be classified into a Super Ager group (group 1) and a general elderly group (group 2). Therefore, based on the actual test results of the subjects classified into the two groups, the correlation between the scores for each item of the cognitive function test of the individual subjects and whether the subjects belong to the Super Ager group can be derived. To this end, according to one embodiment of the present invention, based on data classifying the subjects into the first or second group and actual test data, the correlation between each item of the cognitive function test and whether the subjects belong to the first group can be learned, and the main cognitive function items can be determined based on the learning results.
[0123] Figure 7 is a graph showing the correlation between each item of the cognitive function test and the super-ager group according to one embodiment of the present invention.
[0124] Referring to Figure 7, the horizontal axis of the graph represents the degree to which each item is related to whether or not it is a Super Ager (Variable Importance), and the vertical axis represents each item of the cognitive function test as an evaluation factor (feature). As shown in Figure 7, items with a high degree of correlation to whether or not it is a Super Ager, i.e., major cognitive function items, may be language memory, visual memory, and visuospatial items.
[0125] In relation to this, as described above in step S200 of the above-described FIG. 4, the processor (120) can perform a correlation analysis of the blood test results and cognitive function test results included in the test data, and extract major blood factors to be used for cognitive ability evaluation. As shown in [Table 1], in the cognitive function test, language memory, visual memory, and visuopatial showed a significantly significant difference between the two groups at p<0.01. It can be seen that this is a result consistent with the importance of each cognitive function test item according to the embodiment of FIG. 7.
[0126] Next, the cognitive function evaluation method of the present invention generates augmented test data that can supplement actual test data (630). More specifically, the cognitive function evaluation method of the present invention can generate augmented test data including virtual blood test results and cognitive function test results using a preset data generation algorithm. In this regard, as described above, the step of generating an evaluation model (S300) can generate augmented data based on actual test data using a data generation algorithm to generate (train) or verify (verify) the evaluation model. Augmented data generated from actual test data can be referred to as augmented test data, and according to one embodiment, 729 pieces of augmented test data can be generated from 81 pieces of actual test data.
[0127] Next, the cognitive function assessment method of the present invention predicts whether individual data of the augmented test data belong to the first group using a classification algorithm (640). That is, it is possible to predict the result of whether the generated augmented test data belongs to the Super Ager group. In the steps 610 and 620 described above, the results of classifying actual individual test subjects into the Super Ager group and the general elderly group could be used, but since the augmented test data is not about actual test subjects, there is no information on whether the individual augmented test data belong to the Super Ager group. Therefore, in order to classify the augmented test data into the Super Ager group and the general elderly group, that is, to predict whether the individual augmented test data corresponds to the Super Ager, a separate classification task is required, and then a step of labeling whether it is a Super Ager is required.
[0128] To this end, according to one embodiment of the present invention, a first prediction result may be generated using different classification algorithms to predict whether each augmented test data belongs to the super-age group, and then the first prediction results may be stacked to generate a final prediction result. According to a specific embodiment, the first prediction result may be generated using the classification algorithms of the ExtraTreesClassifier, the RandomForestClassifier, and the GradientBoosting Classifier, but the present invention is not limited thereto, and various classification algorithms may be used to generate the first prediction result. More specifically, the ExtraTreesClassifier is a model using Extremely Randomized Trees, and it is possible to make predictions by constructing trees for some randomly selected elements. In addition, the RandomForestClassifier is a model that ensembles multiple decision trees, and each tree independently learns and makes predictions, and the results are combined to make predictions. In addition, the GradientBoosting Classifier can improve prediction performance by creating new trees that compensate for the errors of the previous trees.
[0129] Furthermore, according to one embodiment of the present invention, after generating primary prediction results using two or more different classification algorithms, the prediction results are used as new elements to train a meta-model, and the meta-model can use the prediction results as input to generate a final prediction result. Thus, according to one embodiment of the present invention, a prediction model can be provided that provides a prediction result on whether augmented test data belongs to the super-age group, with improved overall prediction performance by complementing the weaknesses of each classifier.
[0130] Meanwhile, according to one embodiment of the present invention, after performing a preprocessing operation on the augmented inspection data, it can be input into the classification algorithms of the above-described prediction model. More specifically, a preprocessing operation can be performed on the augmented inspection data to convert categorical data into numerical data through LabelEncoder, and to replace missing values with other values such as the mean, median, and mode through SimpleImputer. That is, LabelEncoder handles encoding of categorical data, and SimpleImputer appropriately handles missing values to maintain the completeness of the data, thereby increasing the accuracy of the prediction model by the classification algorithm according to one embodiment of the present invention.
[0131] Figure 8 is a graph showing the accuracy of a prediction model by a classification algorithm according to an embodiment of the present invention.
[0132] The graph in FIG. 8 is a Receiver Operating Characteristic (ROC) curve for the prediction model of the present invention. The ROC curve is a graph used to visualize the performance of a binary classification model. In the ROC graph, the x-axis (False Positive Rate, FPR) represents the ratio of cases in which the prediction model of the present invention predicted a positive result when the prediction model predicted a positive result. Mathematically, FPR = FP / (FP + TN). Here, FP represents False Positive (the number of cases in which the prediction model predicted a positive result but was actually negative), and TN represents True Negative (the number of cases in which the prediction model predicted a negative result but was actually negative). In addition, the Y-axis (True Positive Rate, TPR) represents the ratio of cases in which the positive result was actually positive and the model accurately predicted a positive result. Mathematically, TPR = TP / (TP + FN). Here, TP represents True Positive (the number of cases in which the model accurately predicted a positive result), and FN represents False Negative (the number of cases in which the model predicted a negative result but was actually positive).
[0133] Referring to FIG. 8, the ROC curve according to the classification model of the present invention can evaluate the performance of the model at various thresholds. Each curve depicted in the graph of FIG. 8 can represent a combination of FPR and TPR obtained by the prediction model at a specific threshold. More specifically, the closer the curve is to the upper left, the better the performance of the prediction model can be judged, which means that the prediction model has a low FPR while maintaining a high TPR. Relatedly, the larger the AUC (Area Under the Curve), the better the performance of the model can be evaluated. In one embodiment of FIG. 8, since the AUC values of the ROC curves are all 0.9, the prediction model of the present invention can be evaluated as having high accuracy.
[0134] Additionally, in the additional embodiment of FIG. 8, a micro-average ROC curve may be used to evaluate the accuracy of the prediction model. A micro-average ROC curve refers to calculating a single ROC curve for all classes instead of drawing an ROC curve for each class in multi-class classification. In other words, a micro-average ROC curve can calculate an overall average ROC curve using the sum of FP (false positive), TP (true positive), and FN (false negative) of all classes. In addition, a macro-average ROC curve is a curve obtained by calculating ROC curves individually for each class and then averaging them. In the embodiment of FIG. 8, since the AUC values of the micro-average ROC curve and the macro-average ROC curve are both 0.9 or higher, it can be evaluated that the prediction model of the present invention has high accuracy. That is, according to one embodiment of the present invention, it is possible to predict whether individual data of the augmented test data belong to the super-age group with an accuracy of approximately 90%.
[0135] Next, the cognitive ability assessment method of the present invention classifies the actual test data and the augmented test data into a first group and a second group (650). In the following specification, for the convenience of explanation, the actual test data and the augmented test data may be referred to as the entire test data combined. As described above, in the step (620) of determining the main items of the cognitive function test, each subject was divided into a first group or a second group, so the actual test data can be classified into a first group or a second group. In addition, in the step (640) of predicting whether the augmented test data belongs to the first group, a classification algorithm is used to predict whether each piece of augmented test data belongs to the first group, so the augmented test data can also be classified into a first group or a second group based on the prediction result. In a specific example, if there are 81 actual experimental data and 791 augmented experimental data, 449 subjects can be classified into the SuperAger group and 361 subjects can be classified into the General Elderly group out of a total of 810 data.
[0136] Next, the cognitive ability assessment method of the present invention extracts at least one major blood factor associated with the first group for each major cognitive function item of the cognitive function test (660). More specifically, according to one embodiment of the present invention, SHAP (SHapley Additive exPlanations) values of blood factors for each cognitive function item are derived for the entire test data, and the SHAP values can be used to determine blood factors that affect whether a subject is classified as a Super Ager. SHAP is a technique used to explain the prediction of a model, and the SHAP value, also called the Shapley value, can measure the degree to which each blood factor feature contributes to the prediction of the model. In particular, SHAP can explain the prediction by considering various combinations while individually calculating the importance of each blood factor feature. In other words, by performing SHAP analysis for each cognitive ability item, the influence of a blood factor on classification as a Super Ager can be quantified.
[0137] Figures 9(a) to 9(e) are graphs for explaining blood factors associated with the first group determined for each cognitive function test item.
[0138] Referring to Figure 9(a), when comparing the Super Ager group with the general population group, the blood factors in the Attention cognitive function item are Leptin (↓), HDL (↑), Triglyceride (↓), Total bilirubin (↑), CD36 (↑), and Na (↓). Here, (↑) means that the higher the blood factor level, the higher the cognitive function score, and (↓) means that the lower the blood factor level, the higher the cognitive function score.
[0139] Referring to Figure 9(b), when comparing the Super Age group with the general population group, it can be confirmed that the blood factors in the language cognitive function item are RAGE (↑), Hb (↓), and MCHC (↓).
[0140] Referring to Figure 9(c), when comparing the Super Ager group with the general group, it can be confirmed that the blood factors in the Verbal Memory cognitive function item are Glucose (↓), C-peptide (↓), MCH (↑), HDL (↑), Platelet (↓), and Free T4 (↓).
[0141] Referring to Figure 9(d), when comparing the Super Ager group with the general group, it can be confirmed that the blood factors in the Visual Memory cognitive function item are WBC (↓), RBC (↑), and oxLDL (↑).
[0142] Referring to Fig. 9(e), when comparing the Superager group with the general group, it can be confirmed that the blood factors in the Visuospatial Superager cognitive function item are Score Visuospatial (↑): ApoE (↓), HbA1c (↓), AST (↓), Free T4 (↑), Triglyceride (↓).
[0143] Referring to Figure 9(f), when comparing the Super Ager group with the general elderly group, it can be confirmed that the blood factor in the frontal cognitive function item is CD36 (↑). That is, as examined in [Table 1], it can be confirmed that the frontal cognitive function item does not differ significantly between the general elderly and the Super Ager group.
[0144] Next, an evaluation model for evaluating cognitive ability is created based on the major blood factors determined for each major cognitive function item (670). The step of creating an evaluation model for evaluating cognitive ability based on the determined major blood factors is similar to step S300 of FIG. 4, and therefore, redundant descriptions will be omitted. Meanwhile, compared to the embodiment of FIG. 4, in the embodiment of FIG. 6, a step of predicting and labeling whether the augmented test data is a SuperAger using a classification algorithm is additionally performed, and whether the entire test data is a SuperAger can be classified based on the predicted result. Therefore, since the major blood factors extracted based on the embodiment of FIG. 6 show a regular distribution in the SuperAger group but a random distribution in the general elderly group, the major blood factors extracted in the embodiment of FIG. 6 can be said to be biomarkers that more effectively show the characteristics of a SuperAger.
[0145] As described above, in step S200 of FIG. 4, the processor (120) may extract major blood factors for each item of the cognitive function test using a preset variable selection algorithm, in one embodiment, selectKBest. Meanwhile, according to another embodiment of the present invention, major biomarkers for each cognitive domain may be selected using one or more feature selections instead of selectKBest. At this time, one or more feature selection methods may be feature selection with recursive feature elimination (RFE) combined with simulated annealing (SA), feature selection (cognitive ability) with RFE, or feature selection (cognitive ability) with BOURUTA.
[0146] In more detail, a processor (120) according to one embodiment of the present invention can extract the top 10 features obtained by performing RFE using a random forest, Adaboost, or lightGBM classifier depending on whether it is a super-age or not by using Feature Selection (SA) with RFE.
[0147] According to a specific embodiment, the processor (120) can perform RFE using a random forest to extract 10 top factors of EPO, FABP-4, Free T4, Adiponectin, Vit B12, Folic acid, HDLC holesterol, WBC, MCV, and Cl. In addition, the processor (120) can perform RFE using Adaboost to extract 10 top features of HDLC holesterol, FABP-4, EPO, WBC, oxLDL, CD36, RBC, Vit B12, Adiponectin, and MCV. In addition, the processor (120) can perform RFE using a lightGBM classifier to extract 10 top factors of Vit B12, HDLC holesterol, Platelet, Total cholesterol, EPO, Cl, AST, BUN, CRP, and RDW (CV). Additionally, the processor (120) can extract overlapping factors (EPO, VitB12, HDLCholesterol).
[0148] Meanwhile, according to the test data of the subjects acquired by the processor (120) in step S100 of FIG. 4, while each patient is binary in terms of whether or not he or she is a super-ager (belongs to group 1), the individual cognitive function scores are continuous. Therefore, the entire dataset of 81 subjects is quantitatively insufficient to detect biomarkers among the 56 blood factors through continuous variables.
[0149] As previously mentioned, whether or not someone is a Super Ager is determined by their cognitive ability scores across the cognitive domains of Attention, Language, Verbal Memory, Visual Memory, Visuospatial, and Frontal. Therefore, it is necessary to divide individuals into two groups: the top 25% of the most important test items that determine their cognitive ability scores in each cognitive domain, and the remaining groups, and identify the key factors that contribute to this top group's high cognitive ability scores in each cognitive domain.
[0150] To this end, the processor (120) may perform feature selection (cognitive ability) with RFE. Feature selection (cognitive ability) with RFE may be performed by utilizing random forest, adaboost, and lightGBM classifiers depending on whether the cognitive ability score group for each cognitive domain is high. The processor (120) may extract overlapping factors (HDLCholesterol, Leptin, FABP-3, Free T4, Glucose, RDW (CV), CD36, ApoE, EPO, Vit B12, RAGE, Total cholesterol, oxLDL, FABP-4, PCSK9, Triglyceride) from among the top 10 factors obtained through this.
[0151] In addition, the processor (120) can perform feature selection (cognitive ability) with BORUTA. Feature selection (cognitive ability) with BORUTA may be performing R-based BORUTA depending on whether the group has the highest cognitive ability score for each cognitive domain. Through this, the processor (120) can extract 10 factors (HDLCholesterol, BUN, ApoE, HbA1c, Insulin, EPO, Vimentin, FABP-4, Free T4, Total Cholesterol, RAGE, PCSK9).
[0152] Next, the processor (120) predicts whether or not it is a super-ager using factors extracted by the three methods of Feature Selection with Recursive Feature Elimination (RFE) combined with Simulated Annealing (SA), Feature Selection (recognition) with RFE, and Feature Selection (recognition) with BOURUTA, and determines the accuracy of the prediction result. According to one embodiment, since the accuracy is the highest when predicting SuperAger with factors extracted by Feature Selection (cognitive ability) with RFE and Feature Selection (cognitive ability) with BOURUTA, the processor (120) can select a total of 20 factors, including EPO, Vit B12, HDLCholesterol, Leptin, FABP-3, Free T4, Glucose, RDW (CV), CD36, ApoE, RAGE, Total cholesterol, oxLDL, FABP-4, PCSK9, Triglyceride, BUN, HbA1c, Insulin, and Vimentin, as SuperAger biomarkers.
[0153] Meanwhile, according to step 630 of FIG. 6 described above, augmented inspection data can be generated to increase the accuracy of the evaluation model. In this regard, according to another embodiment of the present invention, the augmented inspection data can be used to verify the evaluation model.
[0154] In more detail, the generated augmented test data was used to train the evaluation model, and the original data was used to validate the evaluation model using the large-language model-based 'GReaT' method. According to a specific example, out of a total of 81 data, 49 data were used for augmentation, and 70 augmented data were obtained. Through this, a correlation analysis between the actual data and the augmented data using a total of 20 biomarkers described above was performed, and the difference was very small.
[0155] Meanwhile, according to one embodiment of the present invention, some of the 20 biomarkers described above may be common to two or more cognitive domains. Such biomarkers that are common to multiple cognitive domains can be used as indicators for measuring cognitive reserve, thereby allowing for more accurate prediction of super-ager status.
[0156] As described above, the cognitive ability evaluation method and device according to embodiments of the present invention can evaluate cognitive ability based on a blood test, and can easily obtain cognitive ability evaluation results without a long-term cognitive function test or MRI scan, thereby significantly reducing the time and cost required for cognitive ability evaluation.
[0157] Furthermore, embodiments of the present invention may expand accessibility to cognitive ability assessment. For example, by assessing cognitive ability using the results of blood tests performed periodically in the elderly population, regular cognitive screenings may be conducted.
[0158] In addition, embodiments of the present invention can enable healthy people, not just those with a disease, to easily test and manage their cognitive abilities by evaluating whether the subject has cognitive abilities higher than the average cognitive abilities of the age group to which the subject belongs, and have the effect of tracking changes in cognitive abilities by providing a relatively objective cognitive ability evaluation method based on a blood test.
[0159] The present embodiments may also be implemented in the form of a recording medium containing computer-executable instructions, such as program modules, executed by a computer. Computer-readable media may be any available media that can be accessed by a computer, and includes both volatile and nonvolatile media, removable and non-removable media. Furthermore, computer-readable media may include both computer storage media and communication media. Computer storage media includes both volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information, such as computer-readable instructions, data structures, program modules, or other data. Communication media typically includes computer-readable instructions, data structures, other data, such as program modules, in a modulated data signal, or other transport mechanism, and includes any information delivery media.
[0160] Additionally, in this specification, a “part” may be a hardware component such as a processor or circuit, and / or a software component executed by a hardware component such as a processor.
[0161] While the present invention has been described with reference to the embodiments illustrated in the drawings, these are merely exemplary, and those skilled in the art will appreciate that various modifications and variations of the embodiments are possible. Therefore, the true scope of technical protection of the present invention should be determined by the technical spirit of the appended claims.
Claims
1. A step of obtaining actual test data including actual blood test results and cognitive function test results of the subjects; A step of determining the main cognitive function items of the cognitive function test related to whether the subjects belong to the first group based on the actual test data; A step of generating augmented inspection data based on the acquired actual inspection data; A step of predicting whether individual data of the above augmented inspection data belong to the first group using a classification algorithm; A step of classifying individual data of the actual inspection data and the augmented inspection data into one of the first group and the second group; and A step of extracting one or more major blood factors associated with the first group for each major cognitive function item based on the classified actual test data and augmented test data; A blood-based cognitive ability assessment method comprising:
2. In paragraph 1, The steps to determine the main items of the above cognitive function test are: A step of classifying each of the above subjects into a first group or a second group; A step of learning the correlation between each item of the cognitive function test and whether the subjects belong to the first group based on the actual experimental data and classifying the subjects into the first or second group; A step of determining the main cognitive function items based on the above learning; A blood-based cognitive ability assessment method comprising:
3. In paragraph 1, A blood-based cognitive ability assessment method, wherein the first group is a group of super-agers who have cognitive abilities higher than the average cognitive abilities of their age group, and the second group is a group of typical-agers who are not super-agers.
4. In paragraph 1, A blood-based cognitive ability assessment method, wherein the step of generating the above-mentioned augmented test data generates augmented test data including virtual blood test results and cognitive function test results using a preset data generation algorithm.
5. In paragraph 1, A blood-based cognitive ability assessment method, wherein the step of predicting whether the data belongs to the first group comprises generating a first prediction result using at least two different classification algorithms for the augmented test data, and stacking the first prediction results to generate a final prediction result.
6. In paragraph 5, A blood-based cognitive ability assessment method, wherein the augmented test data is preprocessed using a LabelEncoder and a SimpleImputer before being input into at least two different classification algorithms.
7. In paragraph 1, A blood-based cognitive ability assessment method, wherein the determined major cognitive function items are language memory, visual memory, and visuospatial items.
8. In paragraph 1, A blood-based cognitive ability evaluation method, further comprising a step of creating an evaluation model for evaluating cognitive ability based on the major blood factors extracted for each of the major cognitive function items.
9. A computer-readable recording medium recording a program for executing the method of Article 1 on a computer.
10. As a blood-based cognitive ability evaluation device, The device includes a processor; a communication unit; and a memory, The above processor, Obtain actual test data including actual blood test results and cognitive function test results of the subjects, Based on the above actual test data, the main cognitive function items of the cognitive function test related to whether the subjects belong to the first group are determined, Generate augmented inspection data based on the acquired actual inspection data, For each piece of data in the above augmented inspection data, a classification algorithm is used to predict whether it belongs to the first group, Classify the individual data of the above actual inspection data and the above augmented inspection data into one of the first group and the second group, A blood-based cognitive ability assessment method, which extracts one or more major blood factors associated with the first group for each major cognitive function item based on the classified actual test data and augmented test data.
Citation Information
Patent Citations
System and method for measuring biological ages
KR1020160036954A
A smart examination apparatus for dementia early diagnosis and the method by using the same
KR1020180002234A
Multiplexed assay and methods of use thereof
US20220120765A1
Real-time monitoring and control of physical and arousal status of individual organisms
WO2008003148A1
