Amyloid beta pathology risk stratification method using artificial intelligence model
An AI-based risk stratification method for amyloid beta pathology in Alzheimer's disease reduces the need for invasive tests by categorizing patients into risk groups, focusing additional testing on intermediates, thus enhancing accessibility and reducing costs.
Patent Information
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- BEAUBRAIN HEALTHCARE CO LTD
- Filing Date
- 2025-12-10
- Publication Date
- 2026-07-02
AI Technical Summary
Current methods for diagnosing amyloid beta pathology in Alzheimer's disease, such as PET scans and cerebrospinal fluid analysis, are expensive and invasive, limiting their accessibility and patient acceptance, necessitating a more affordable and minimally invasive approach.
A risk stratification method using an artificial intelligence model that analyzes brain atrophy data from multimodal imaging and determines risk groups, with additional testing only conducted on an intermediate group, employing a Random Forest model and blood biomarkers like plasma pTau217.
Reduces the need for additional testing in high and low-risk groups, allowing intensive analysis in the intermediate group, thereby decreasing physical and economic burden while maintaining accuracy.
Smart Images
Figure KR2025095776_02072026_PF_FP_ABST
Abstract
Description
Risk stratification method for amyloid beta pathology using an artificial intelligence model
[0001] The present invention relates to amyloid beta pathology, which is a cause of Alzheimer's disease, and more specifically, to a method for risk stratification of amyloid beta pathology using an artificial intelligence model.
[0002] The 'gold standard' method for confirming the pathology of Alzheimer's disease (AD) is pathological examination of the brain during autopsy (DeTure and Dickson, 2019). However, since the early 21st century, the ability to diagnose the pathology of Alzheimer's disease (AD) in living people has been made possible through the development of radioligands for amyloid beta (Aβ) positron emission tomography (PET) scans (Klunk et al., 2004; Schilling et al., 2016), tau PET scans (Marquie et al., 2015; Leuzy et al., 2019), magnetic resonance imaging (MRI) for neurodegeneration (Frisoni et al., 2010), and the analysis of amyloid beta (Aβ) and tau in cerebrospinal fluid (CSF) (Blennow, 2004; Holtzman, 2011).
[0003] These advancements gave rise to an in vivo biological system of biomarkers for Alzheimer's disease (AD) based on amyloid beta (Aβ), tau, and neurodegeneration, known as the A / T / N system (Jack et al., 2018). Indeed, the A / T / N system based on this technology clearly includes amyloid beta (Aβ)-positive (+) individuals in the Alzheimer's disease (AD) continuum, whereas individuals with amyloid beta-negative (-) profiles are considered to be in a normal state or possess non-Alzheimer's disease pathological changes (Jack et al., 2018).
[0004] In particular, many clinical trials recruiting early-stage patients use amyloid (Aβ) PET scans or cerebrospinal fluid (CSF) Aβ42 levels as an important step to improve the quality of the clinical trial cohort (Sperling et al., 2014; Honig et al., 2018). However, despite these advancements, PET scans are very expensive and not universally accessible.
[0005] In addition, although lumbar punctures are very safe (Peskind et al., 2009), there was still resistance among patients and medical professionals to the collection of cerebrospinal fluid (CSF) samples (Moulder et al., 2017).
[0006] Therefore, there is a significant increase in efforts to detect amyloid beta (Aβ) pathology in Alzheimer's disease using low-cost and minimally invasive methods compared to 'gold standards' such as PET scans or cerebrospinal fluid (CSF).
[0007] In particular, with the recent approval of anti-amyloid beta immunotherapies for Alzheimer's disease, there is an urgent need for an affordable system to identify amyloid beta positivity in patients with cognitive impairment.
[0008] Prior art literature
[0009] Patent documents
[0010] Korean Patent Publication No. 10-2019-0067477 (Method and apparatus for measuring the stage of Alzheimer's dementia using amyloid PET imaging.
[0011] Accordingly, the present invention has been devised to solve the above-mentioned problems, and the objective of the present invention is to provide a method for risk stratification of amyloid beta pathology using an artificial intelligence model.
[0012] Another objective of the present invention is to provide a risk stratification method for amyloid beta pathology using an artificial intelligence model that performs additional testing only on the intermediate risk group through risk stratification.
[0013] However, the technical problems to be solved by the present invention are not limited to those mentioned above, and other technical problems not mentioned will be clearly understood by those skilled in the art to which the present invention belongs from the description below.
[0014] To achieve the above technical task, the method comprises the steps of: extracting a dataset regarding brain atrophy from multimodal data (T1, T2 or CT); an artificial intelligence model (140) learning the dataset (100) regarding brain atrophy; the learned artificial intelligence model (140) outputting an amyloid beta-PET positive day prediction probability distribution and an amyloid beta-PET negative day prediction probability distribution (S100); the artificial intelligence model (140) determining an upper threshold and a lower threshold based on the amyloid beta-PET positive day prediction probability distribution and the amyloid beta-PET negative day prediction probability distribution (S100); and a control unit stratifying the risk of amyloid beta-PET positive into a high-risk group (151), an intermediate-risk group (153), and a low-risk group (155) based on the upper threshold and the lower threshold (S120). A method for stratifying the risk of amyloid beta pathology using an artificial intelligence model is provided, characterized by comprising: a step (S140) of performing a blood biomarker test on an intermediate risk group (153); and a step (S160) of determining a patient whose blood biomarker test result is positive as a patient positive for amyloid beta-PET, and determining a patient whose blood biomarker test result is negative as a patient negative for amyloid beta-PET.
[0015] Optionally, the artificial intelligence model (140) is a random forest model.
[0016] Optionally, the dataset (100) regarding brain atrophy includes brain segmentation images (110) and W-scores of brain volume, and
[0017] The W-score of brain volume is the following formula
[0018]
[0019] It is calculated from.
[0020] Optionally, the dataset (100) regarding brain atrophy further includes information regarding the APOE ε4 genotype.
[0021]
[0022] Optionally, the determining step (S100) determines an upper threshold and a lower threshold from a prediction probability distribution that combines the amyloid beta-PET positive day prediction probability distribution and the amyloid beta-PET negative day prediction probability distribution.
[0023] Optionally, the upper threshold is one of the probability that amyloid beta-PET is positive at 90% specificity (Sp) on the combined prediction probability distribution of 0.62, the probability that amyloid beta-PET is positive at 95% specificity (Sp) on the combined prediction probability distribution of 0.72, and the probability that amyloid beta-PET is positive at 97.5% specificity (Sp) on the combined prediction probability distribution of 0.77.
[0024] Optionally, the lower threshold is one of the probability that amyloid beta-PET is negative at 90% sensitivity (Se) on the combined prediction probability distribution of 0.38, the probability that amyloid beta-PET is negative at 95% sensitivity (Se) on the combined prediction probability distribution of 0.35, and the probability that amyloid beta-PET is negative at 97.5% sensitivity (Se) on the combined prediction probability distribution of 0.27.
[0025] Optionally, the upper threshold is the probability that amyloid beta-PET is positive at 95% specificity (Sp) on the combined prediction probability distribution of 0.72, and the lower threshold is the probability that amyloid beta-PET is negative at 95% sensitivity (Se) on the combined prediction probability distribution of 0.35.
[0026] Optionally, the blood biomarker test is the plasma pTau217 biomarker test.
[0027] According to one embodiment of the present invention, additional testing is not required for patients in high-risk and low-risk groups. This can reduce the physical and economic burden on patients.
[0028] In addition, by conducting additional tests (e.g., plasma pTau-217 biomarker test or PET scan) only on the intermediate risk group through risk stratification, it becomes possible to perform intensive and accurate analysis on patients in the intermediate risk group.
[0029] However, the effects obtainable from the present invention are not limited to those mentioned above, and other unmentioned effects will be clearly understood by those skilled in the art from the description below.
[0030] The following drawings attached to this specification illustrate preferred embodiments of the present invention and serve to further enhance understanding of the technical concept of the present invention together with the detailed description of the invention provided below; therefore, the present invention should not be interpreted as being limited only to the matters described in such drawings.
[0031] FIG. 1 is a block diagram of risk stratification of amyloid beta pathology using an artificial intelligence model according to an embodiment of the present invention,
[0032] FIG. 2 is a graph of the predicted probability distribution of the first patient group inferred by the artificial intelligence model of the present invention,
[0033] FIG. 3 is a graph of the predicted probability distribution of an external second patient group to verify the artificial intelligence model constructed in the present invention,
[0034] Figure 4 is a predicted probability distribution graph combining the graphs of Figures 2 and 3,
[0035] FIG. 5 is a flowchart of a method for risk stratification of amyloid beta pathology using an artificial intelligence model according to an embodiment of the present invention,
[0036] FIG. 6 is a graph showing the overall accuracy of the risk stratification method according to the present invention,
[0037] Figure 7 is a graph showing the rate reduction achieved in additional inspections by the risk stratification method according to the present invention.
[0038] Below, with reference to the attached drawings, embodiments of the present invention are described in detail so that those skilled in the art can easily implement the invention. However, since the description of the present invention is merely an example for structural or functional explanation, the scope of the present invention should not be interpreted as being limited by the embodiments described in the text. That is, since the embodiments are subject to various modifications and may take various forms, the scope of the present invention should be understood to include equivalents capable of realizing the technical concept. Furthermore, the objectives or effects presented in the present invention do not imply that a specific embodiment must include all of them or only such effects; therefore, the scope of the present invention should not be understood as being limited by them.
[0039] The meaning of the terms described in this invention should be understood as follows.
[0040] Terms such as "first" and "second" are intended to distinguish one component from another, and the scope of rights shall not be limited by these terms. For example, the first component may be named the second component, and similarly, the second component may be named the first component. When a component is referred to as being "connected" to another component, it should be understood that it may be directly connected to that other component, or that there may be other components in between. Conversely, when a component is referred to as being "directly connected" to another component, it should be understood that there are no other components in between. Meanwhile, other expressions describing the relationship between components, such as "between" and "exactly between," or "adjacent to" and "directly adjacent to," shall be interpreted in the same manner.
[0041] A singular expression should be understood to include a plural expression unless the context clearly indicates otherwise, and terms such as "include" or "have" are intended to specify the existence of the set-up features, numbers, steps, actions, components, parts, or combinations thereof, and should be understood not to preclude the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.
[0042] Unless otherwise defined, all terms used herein have the same meaning as generally understood by those skilled in the art to which this invention pertains. Terms defined in commonly used dictionaries should be interpreted as having meanings consistent with the context of the relevant technology and should not be interpreted as having an ideal or overly formal meaning unless explicitly defined in this invention.
[0043] Composition of the embodiment
[0044] Hereinafter, the configuration of a preferred embodiment will be described in detail with reference to the attached drawings. FIG. 1 is a block diagram of risk stratification of amyloid beta pathology using an artificial intelligence model according to an embodiment of the present invention. As shown in FIG. 1, a dataset (100) regarding brain atrophy includes brain segmentation images (110), W-scores of brain volume, demographics (120), and information regarding APOE ε4 genotype.
[0045] The brain segmentation image (110) is an image that visualizes the structure of the brain by analyzing and segmenting specific regions. The brain segmentation image (110) is a structural segmentation image of cerebral gray matter (GM), white matter (WM), cerebrospinal fluid (CSF), a functional segmentation image that analyzes the active regions of the brain using fMRI (functional MRI), and a lesion segmentation image that accurately segments lesion regions such as tumors, hemorrhages, and damaged areas. These brain segmentation images (110) are obtained from a first patient group consisting of multiple patients.
[0046] Demographics (120) include information such as gender, age, medical history, and underlying diseases of the first patient group.
[0047] The W-score of brain volume is the following formula
[0048]
[0049] It is calculated from.
[0050] And, the W-score includes the number of APOE ε4s.
[0051]
[0052] The APOE ε4 genotype is one of three major alleles of the Apolipoprotein E (APOE) gene and is associated with Alzheimer's disease and other neurodegenerative diseases. The APOE gene is located on chromosome 19 and is responsible for the production of apolipoprotein E, a protein that regulates lipid and cholesterol metabolism.
[0053] Among the major alleles of the APOE gene, APOE ε4 is an allele that increases the risk of Alzheimer's disease and cardiovascular disease, and is one of the major genetic risk factors for Alzheimer's disease. For example, it is known that possessing one ε4 allele increases the risk of developing Alzheimer's disease by about 2 to 3 times, and possessing two ε4 alleles (ε4 / ε4) increases the risk by about 8 to 12 times.
[0054] The artificial intelligence model (140) is a Random Forest model. The Random Forest model is one of the ensemble learning methods in machine learning and is a powerful model that improves prediction performance by combining multiple decision trees. It is mainly suitable for solving classification and regression problems.
[0055] The stratification section (150) stratifies the risk of testing positive for amyloid beta (Aβ) protein. For example, the stratification section (150) classifies into a high-risk group (151), a medium-risk group (153), and a low-risk group (155). These three classifications are merely examples, and the classification can be configured to be subdivided into five or more levels.
[0056] Additional tests (plasma pTau-217 biomarker test or PET scan) are performed only on patients in the intermediate risk group (153). And, patients in the intermediate risk group (153) who undergo additional tests receive either an amyloid beta (Aβ) positive result (170) or an amyloid beta (Aβ) negative result (180).
[0057] The control unit can be a computer running an operating system and application programs. Such a computer is equipped with a central control unit (CPU), storage devices, communication devices, a monitor, a mouse, a keyboard, USB ports, etc. Optionally, the control unit can be a server computer and is configured to enable bidirectional data communication with multiple client terminals (member computers, mobile phones, laptops, tablets, etc.) via a wired or wireless network.
[0058] Method of the example
[0059] Hereinafter, a method of a preferred embodiment will be described in detail with reference to the attached drawings. FIG. 5 is a flowchart of a method for risk stratification of amyloid beta pathology using an artificial intelligence model according to an embodiment of the present invention.
[0060] First, a dataset regarding brain atrophy is extracted from multimodal data (T1, T2, or CT). T1 and T2 are weighted MRI images, and CT is a computed tomography image.
[0061] Next, an artificial intelligence model (140) learns the extracted dataset (100) regarding brain atrophy.
[0062] Next, the trained artificial intelligence model (140) outputs the predicted probability distribution for amyloid beta-PET positive days and the predicted probability distribution for amyloid beta-PET negative days (S100). FIG. 2 is a graph of the predicted probability distribution of a first patient group inferred by the artificial intelligence model of the present invention, FIG. 3 is a graph of the predicted probability distribution of an external second patient group to verify the artificial intelligence model constructed in the present invention, and FIG. 4 is a graph of the predicted probability distribution combining the graphs of FIG. 2 and FIG. 4. As shown in FIG. 2 to FIG. 4, negative (blue dots) and positive (red dots) amyloid beta (Aβ)-PET scans appear on the horizontal axis, and predicted probabilities (0 to 1) appear on the vertical axis. In FIG. 2 to FIG. 4, one dot represents one patient. By comparing FIG. 2 and FIG. 3, it can be seen that the predicted probability distribution output by the artificial intelligence model (140) (Fig. 2) and the actual predicted probability distribution of the external patient group (Fig. 3) are similar, which indicates that the artificial intelligence model (140) was constructed correctly.
[0063] In this determining step (S100), the control unit (not shown) determines an upper threshold and a lower threshold from a prediction probability distribution that combines the amyloid beta-PET positive day prediction probability distribution and the amyloid beta-PET negative day prediction probability distribution. As shown in FIG. 4, the combined prediction probability distribution enables more accurate risk stratification.
[0064] Next, the artificial intelligence model (140) or the control unit determines an upper threshold and a lower threshold based on the amyloid beta-PET positive day prediction probability distribution and the amyloid beta-PET negative day prediction probability distribution (Fig. 2) or the combined prediction probability distribution (Fig. 4) (S100).
[0065] Risk Group Number of patients in each risk group (n) Aβ-PET Status Negative (%) Positive (%) 90% Se Lower Threshold (%), 90% Sp Upper Threshold (%) Low Risk (<38) 330 286 (86.7%) 44 (13.3%) Intermediate Risk (38-62) 372 178 (47.8%) 194 (52.2%) High Risk (>62) 593 71 (12.0%) 522 (88.0%) 95% Se Lower Threshold (%), 95% Sp Upper Threshold (%) Low Risk Group (<35) 29 2 2 5 8 (88.4%) 3 4 (11.6%) Medium Risk Group (35-72) 5 1 0 2 3 5 (46.1%) 2 7 5 (53.9%) High Risk Group (>72) 4 9 3 4 2 (8.5%) 4 5 1 (91.5%) 97.5% Se Lower Threshold (%) 97.5% Sp Upper Threshold (%) Low Risk Group (<27) 1 7 9 1 6 2 (90.5%) 1 7 (9.5%) Medium Risk Group (27-77) 6 9 2 3 3 8 (48.8%) 3 5 4 (51.2%) High Risk Group (>77) 4 2 4 3 5 (8.3%) 3 8 9 (91.7%)
[0066] Specifically, as shown in FIGS. 2 to 4 and [Table 1], the upper threshold is determined as one of the following: a probability of amyloid beta-PET being positive at 90% specificity (Sp) on the combined prediction probability distribution of 0.62, a probability of amyloid beta-PET being positive at 95% specificity (Sp) on the combined prediction probability distribution of 0.72, and a probability of amyloid beta-PET being positive at 97.5% specificity (Sp) on the combined prediction probability distribution of 0.77. Optionally, the lower threshold is determined as one of the following: a probability of amyloid beta-PET being negative at 90% sensitivity (Se) on the combined prediction probability distribution, 0.38, a probability of amyloid beta-PET being negative at 95% sensitivity (Se) on the combined prediction probability distribution, and a probability of amyloid beta-PET being negative at 97.5% sensitivity (Se) on the combined prediction probability distribution, as shown in FIGS. 2 to 4 and [Table 1].
[0067] In a specific embodiment of the present invention, the upper threshold is 0.72 and the lower threshold is 0.35.
[0068] Next, the control unit stratifies the risk of amyloid beta-PET being positive into a high-risk group (151), an intermediate-risk group (153), and a low-risk group (155) based on an upper threshold (0.72) and a lower threshold (0.32) (S120). As shown in FIG. 5, 38.1% of patients in the high-risk group (151) have a 91.5% probability of being amyloid beta-PET positive, so they do not need to undergo additional testing. And, 22.5% of patients in the low-risk group (155) have an 88.4% probability of being amyloid beta-PET negative, so they do not need to undergo additional testing.
[0069] However, 39.4% of patients in the intermediate risk group (153) may receive a positive or negative result on the amyloid beta-PET scan. Therefore, additional tests (detailed tests) are performed only on the intermediate risk group (153).
[0070] That is, a blood biomarker test is performed on the intermediate risk group (153) (S140). The blood biomarker test is a plasma pTau217 biomarker test or a PET scan.
[0071] Next, patients with positive blood biomarker test results are determined to be amyloid beta-PET positive (S160). This is because 82.7% of the 61.4% of patients who tested positive in the plasma pTau217 biomarker test are amyloid beta-PET positive. And, patients with negative blood biomarker test results are determined to be amyloid beta-PET negative (S160). This is because 91.9% of the 38.6% of patients who tested negative in the plasma pTau217 biomarker test are amyloid beta-PET negative.
[0072] FIG. 6 is a graph showing the overall accuracy of the risk stratification method according to the present invention, and FIG. 7 is a graph showing the reduction in the rate achieved in additional tests by the risk stratification method according to the present invention. As shown in FIG. 6, a two-step workflow was performed to reflect the accuracy of the classification rate in the low-risk group (155) and high-risk group (151), and the accuracy of plasma pTau217 classification for the intermediate-risk group (153). The overall accuracy of the risk stratification was calculated for each step by combining the first patient group and the external second patient group (n = 1,295). Error bars represent the 95% confidence interval (CI).
[0073] And, as shown in Figure 7, it can be seen that the additional testing rate decreased according to the threshold strategy of risk stratification (specificity / sensitivity 90%, n = 923; 95%, n = 785; 97.5%, n = 603). In particular, it was confirmed that the additional testing rate decreased when the specificity / sensitivity increased from 90% to 97.5%.
[0074] As described above, the detailed description of the preferred embodiments of the present invention disclosed is provided to enable those skilled in the art to implement and practice the present invention. Although the present invention has been described with reference to preferred embodiments, those skilled in the art will understand that various modifications and changes can be made to the present invention without departing from the scope of the invention. For example, those skilled in the art may utilize each configuration described in the embodiments described above in combination with one another. Accordingly, the present invention is not intended to be limited to the embodiments shown herein, but to be given the broadest scope consistent with the principles and novel features disclosed herein.
[0075] The present invention may be embodied in other specific forms without departing from the spirit and essential features of the invention. Accordingly, the above detailed description should not be interpreted restrictively in all respects but should be considered exemplary. The scope of the invention shall be determined by a reasonable interpretation of the appended claims, and all modifications within the equivalent scope of the invention are included within the scope of the invention. The invention is not intended to be limited to the embodiments shown herein, but to be given the broadest possible scope consistent with the principles and novel features disclosed herein. Furthermore, embodiments may be constructed by combining claims that are not explicitly related in the claims, or by including them as new claims through amendments made after filing.
[0076] Explanation of the symbols
[0077] 100 : dataset,
[0078] 110 : Brain segmentation image,
[0079] 120 : Demographics,
[0080] 130 : W-score calculation unit,
[0081] 140 : Artificial intelligence model,
[0082] 150 : Hierarchy section,
[0083] 151 : High-risk group,
[0084] 153 : Intermediate risk group,
[0085] 155 : Low-risk group,
[0086] 160: Plasma pTau-217 test or PET scan,
[0087] 170 : Aβ-PET positive test,
[0088] 180: Aβ-PET negative result.
Claims
1. A step of extracting a dataset regarding brain atrophy from multimodal data (T1, T2 or CT); A step in which an artificial intelligence model (140) learns a dataset (100) regarding brain atrophy; Step (S100) in which the learned artificial intelligence model (140) outputs the predicted probability distribution of amyloid beta-PET positive days and the predicted probability distribution of amyloid beta-PET negative days; Step (S100) in which the artificial intelligence model (140) determines an upper threshold and a lower threshold based on the amyloid beta-PET positive day prediction probability distribution and the amyloid beta-PET negative day prediction probability distribution; A step (S120) in which the control unit stratifies the risk of the amyloid beta-PET being positive into a high-risk group (151), an intermediate-risk group (153), and a low-risk group (155) based on the upper threshold and the lower threshold; Step (S140) of performing a blood biomarker test on the above intermediate risk group (153); and A method for risk stratification of amyloid beta pathology using an artificial intelligence model, characterized by including the step (S160) of determining a patient whose blood biomarker test result is positive as a patient who is positive for amyloid beta-PET, and determining a patient whose blood biomarker test result is negative as a patient who is negative for amyloid beta-PET.
2. In Paragraph 1, A method for stratifying the risk of amyloid beta pathology using an artificial intelligence model, characterized in that the above artificial intelligence model (140) is a random forest model.
3. In Paragraph 1, A method for stratifying the risk of amyloid beta pathology using an artificial intelligence model, characterized in that the above dataset (100) regarding brain atrophy includes brain segmentation images (110) and W-scores of brain volume.
4. In Paragraph 3, The W-score of brain volume is the following formula A method for risk stratification of amyloid beta pathology using an artificial intelligence model characterized by being derived from 5. In Paragraph 3, A method for risk stratification of amyloid beta pathology using an artificial intelligence model, characterized in that the above dataset (100) regarding brain atrophy further includes information regarding APOE ε4 genotype.
6. In Paragraph 1, A method for risk stratification of amyloid beta pathology using an artificial intelligence model, wherein the above-determining step (S100) is characterized by determining the upper threshold and the lower threshold from a prediction probability distribution that combines the amyloid beta-PET positive day prediction probability distribution and the amyloid beta-PET negative day prediction probability distribution.
7. In Paragraph 6, The above upper threshold is, On the combined prediction probability distribution above, the probability that the amyloid beta-PET is positive at 90% specificity (Sp) is 0.62, On the combined predicted probability distribution above, the probability that the amyloid beta-PET is positive at 95% specificity (Sp) is 0.72, and A method for risk stratification of amyloid beta pathology using an artificial intelligence model characterized by having one of the following: a probability of 0.77 that the amyloid beta-PET is positive at a specificity (Sp) of 97.5% on the combined prediction probability distribution.
8. In Paragraph 6, The above lower threshold is, On the combined prediction probability distribution above, the probability that the amyloid beta-PET is negative at 90% sensitivity (Se) is 0.38, On the combined prediction probability distribution above, the probability that the amyloid beta-PET is negative at 95% sensitivity (Se) is 0.35, and A method for risk stratification of amyloid beta pathology using an artificial intelligence model characterized by having one of the following: a probability of 0.27 that the amyloid beta-PET is negative at a sensitivity (Se) of 97.5% on the combined prediction probability distribution.
9. In Paragraph 6, The upper threshold is the probability that the amyloid beta-PET is positive at 95% specificity (Sp) on the combined prediction probability distribution, which is 0.72, and A method for risk stratification of amyloid beta pathology using an artificial intelligence model, characterized in that the lower threshold is a probability of 0.35 that the amyloid beta-PET is negative at a sensitivity (Se) of 95% on the combined prediction probability distribution.
10. In Paragraph 1, A method for risk stratification of amyloid beta pathology using an artificial intelligence model, characterized in that the above blood biomarker test is a plasma pTau217 biomarker test.