Intracerebral amyloid-β lesion estimation device, method, and program, and estimation model

A device using a machine learning-based estimation model with blood biomarkers addresses the invasive and costly issues of current amyloid-β lesion detection methods, providing a non-invasive and accurate diagnostic tool for Alzheimer's disease.

WO2024232246A1PCT designated stage expired Publication Date: 2024-11-14KEIO UNIV
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
PCT/JP2024/015659
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2023-05-09
Filing Date
2024-04-19
Publication Date
2024-11-14

AI Technical Summary

Technical Problem

Current methods for detecting amyloid-β lesions in the brain, such as amyloid PET and cerebrospinal fluid quantification, are invasive and expensive, making them unsuitable for widespread diagnosis and not covered by insurance, necessitating a non-invasive and cost-effective detection method for Alzheimer's disease diagnosis.

Method used

A device comprising an estimation model that uses blood-related information, specifically concentrations of phosphorylated tau, amyloid β40, amyloid β42, LYZL2, CAPN3, CEACAM3, and CEP152, to determine the presence or absence of amyloid-β lesions in the brain, eliminating the need for invasive tests by using a machine learning-based approach.

Benefits of technology

Enables non-invasive and cost-effective detection of amyloid-β lesions with high accuracy, as demonstrated by an AUC of 0.99 for the six-factor model, allowing for definitive Alzheimer's disease diagnosis.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure JP2024015659_14112024_PF_FP_ABST
    Figure JP2024015659_14112024_PF_FP_ABST
Patent Text Reader

Abstract

According to the present invention, intracerebral amyloid-β lesions are detected noninvasively and at low cost. An intracerebral amyloid-β lesion estimation device according to one embodiment of the present invention comprises an estimation model which outputs the presence or absence of an amyloid-β lesion in the brain when information pertaining to blood is input, and also comprises an information acquisition unit that acquires information pertaining to the blood of a subject, an estimation unit that inputs, to the estimation model, the acquired information pertaining to the blood of the subject, and estimates the presence or absence of an amyloid-β lesion in the brain of the subject, and an estimation result output unit that outputs information pertaining to the estimated presence or absence of an amyloid-β lesion in the brain of the subject.
Need to check novelty before this filing date? Find Prior Art

Description

Apparatus, method, program, and estimation model for estimating intracerebral amyloid-β lesions

[0001] The present invention relates to the estimation of amyloid-β pathology in the brain.

[0002] Detecting amyloid-β lesions in the brain has been essential for a definitive diagnosis of Alzheimer's disease, but no testing methods have been covered by insurance. Amyloid PET (Positron Emission Tomography) and quantification of amyloid-β in cerebrospinal fluid have been performed at the research level to determine the presence or absence of amyloid-β lesions in the brain, but these are invasive and expensive and have not yet been covered by insurance.

[0003] Benedet A. et al. Differences Between Plasma and Cerebrospinal Fluid Glial Fibrillary Acidic Protein Levels Across the Alzheimer Disease Continuum 2021 Dec 1;78(12):1471-1483.

[0004] When implementing anti-amyloid therapies, which are expected to make great advances in the future, it is necessary to be able to detect amyloid beta lesions in the brain non-invasively and at low cost in order to make a definitive diagnosis of Alzheimer's disease.

[0005] The present invention aims to detect amyloid-β lesions in the brain non-invasively and at low cost.

[0006] One embodiment of the present invention relates to an intracerebral amyloid beta lesion estimation device, which includes an estimation model that outputs the presence or absence of an amyloid beta lesion in the brain when information about the blood is input, and includes an information acquisition unit that acquires information about the subject's blood, an estimation unit that inputs the acquired information about the subject's blood into the estimation model and estimates the presence or absence of an amyloid beta lesion in the subject's brain, and an estimation result output unit that outputs the estimated information about the presence or absence of an amyloid beta lesion in the subject's brain.

[0007] According to the present invention, amyloid-β lesions in the brain can be detected non-invasively and at low cost.

[0008] FIG. 1 is a diagram showing the overall configuration of one embodiment of the present invention. FIG. 2 is a functional block diagram of an intracerebral amyloid β lesion estimation device according to one embodiment of the present invention. FIG. 3 is a functional block diagram of a learning device according to one embodiment of the present invention. FIG. 4 is a flowchart of a process for estimating the presence or absence of intracerebral amyloid β lesions according to one embodiment of the present invention. FIG. 5 is a flowchart of a learning process for generating an estimation model according to one embodiment of the present invention. FIG. 6 is data comparing a model of amyloid β42 / amyloid β40 and phosphorylated tau (p-tau181) with a six-factor model according to one embodiment of the present invention. FIG. 7 is a diagram showing the hardware configuration of an intracerebral amyloid β lesion estimation device and a learning device according to one embodiment of the present invention.

[0009] Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

[0010] <System Configuration> Fig. 1 is a diagram showing the overall configuration according to one embodiment of the present invention. The brain amyloid β lesion estimation system 1 can include an brain amyloid β lesion estimation device 10 and a learning device 20. Each of these will be described below.

[0011] <<Cerebral amyloid β lesion estimation device>> The cerebral amyloid β lesion estimation device (hereinafter also simply referred to as the "estimation device") 10 is a device that estimates the presence or absence of amyloid β lesions in the brain of a subject 30 (specifically, the presence or absence of amyloid β accumulation). Specifically, the estimation device 10 inputs information about the blood of the subject 30 into an estimation model to estimate the presence or absence of amyloid β lesions in the brain of the subject 30. The estimation device 10 is equipped with an estimation model generated by a learning device 20. The estimation device 10 is composed of one or more computers.

[0012] <<Learning Device>> The learning device 20 is a device that, when information about the blood of a subject 30 is input, generates an estimation model that outputs the presence or absence of an amyloid-β lesion in the brain of the subject 30. The learning device 20 is made up of one or more computers.

[0013] For example, the information about blood is the quantitative values ​​(specifically, the concentrations of each in the blood) of plasma phosphorylated tau, amyloid β40, amyloid β42, LYZL2, CAPN3, CEACAM3, and CEP152.

[0014] In addition, in FIG. 1, the intracerebral amyloid β lesion estimation device 10 and the learning device 20 are described as separate devices, but the intracerebral amyloid β lesion estimation device 10 and the learning device 20 may be implemented in the same device.

[0015] <Functional Block Diagram> Hereinafter, the functional configuration of the intracerebral amyloid β lesion estimation device 10 will be described with reference to FIG. 2, and the functional configuration of the learning device 20 will be described with reference to FIG.

[0016] <<Estimation Device>> Figure 2 is a functional block diagram of an intracerebral amyloid β lesion estimation device 10 according to one embodiment of the present invention. The intracerebral amyloid β lesion estimation device 10 can include an information acquisition unit 101, an estimation unit 102, an estimation model storage unit 103, and an estimation result output unit 104. Furthermore, the intracerebral amyloid β lesion estimation device 10 can function as the information acquisition unit 101, the estimation unit 102, and the estimation result output unit 104 by executing a program. Each of these components will be described below.

[0017] The information acquisition unit 101 acquires information about the blood of the subject 30.

[0018] For example, a doctor or the like obtains information about the subject's 30 blood by taking a blood sample and inputs the information about the subject's 30 blood into the estimation device 10.

[0019] [Information about Blood] Here, information about blood (specifically, information about blood molecules) will be described. For example, information about blood includes information about at least one of plasma and serum.

[0020] For example, the information about blood includes at least one of the following: a quantitative value of plasma phosphorylated tau (specifically, its concentration in the blood), a quantitative value of amyloid beta 40 (specifically, its concentration in the blood), a quantitative value of amyloid beta 42 (specifically, its concentration in the blood), a quantitative value of LYZL2 (Lysozyme-like protein 2) (specifically, its concentration in the blood), a quantitative value of CAPN3 (Calpain-3) (specifically, its concentration in the blood), a quantitative value of amyloid beta 42 / quantitative value (ratio) of amyloid beta 40, a quantitative value of CEACAM3 (Carcinoembryonic Antigen-Related Cell Adhesion Molecule 3) (specifically, its concentration in the blood), and a quantitative value of CEP152 (Centrosomal Protein 152) (specifically, its concentration in the blood).

[0021] The blood-related information is preferably a combination of four items: a quantitative value of plasma phosphorylated tau (specifically, its concentration in the blood), a quantitative value of amyloid beta 40 (specifically, its concentration in the blood), a quantitative value of LYZL2 (specifically, its concentration in the blood), and a quantitative value of CAPN3 (specifically, its concentration in the blood), and may be any measurement value including these four items. The blood-related information may further include at least one of a quantitative value of amyloid beta 42, a quantitative value of amyloid beta 42 / amyloid beta 40, a quantitative value of CEACAM3, and CEP152.

[0022] For example, information about blood can be obtained by measuring a sample (blood) collected from the subject 30 using a highly sensitive immunodetection method.

[0023] The estimation unit 102 inputs information about the blood of the subject 30 acquired by the information acquisition unit 101 into an estimation model stored in the estimation model memory unit 103, and estimates the presence or absence of amyloid beta lesions in the brain of the subject 30.

[0024] Specifically, the estimation unit 102 inputs information about the blood of the subject 30 acquired by the information acquisition unit 101 into an estimation model stored in the estimation model memory unit 103, and outputs the presence or absence of amyloid beta lesions in the brain of the subject 30.

[0025] The estimation model storage unit 103 stores an estimation model generated by the learning device 20. The estimation model is a trained model that outputs the presence or absence of amyloid-β lesions in the brain of the subject 30 when information about the blood of the subject 30 is input.

[0026] The estimation result output unit 104 outputs information on the presence or absence of amyloid-β lesions in the brain of the subject 30 estimated by the estimation unit 102 (for example, displays the estimation result on a display or the like of the estimation device 10).

[0027] <<Learning Device>> Figure 3 is a functional block diagram of a learning device 20 according to one embodiment of the present invention. The learning device 20 can include a learning data acquisition unit 201, a machine learning unit 202, and an estimation model storage unit 203. Furthermore, the learning device 20 can function as the learning data acquisition unit 201 and the machine learning unit 202 by executing a program. Each of these units will be described below.

[0028] The learning data acquisition unit 201 acquires learning data.

[0029] Specifically, the training data is information about the blood of multiple individuals and information about the presence or absence of amyloid-β lesions in the brain. For example, a doctor or the like can look at images from an amyloid PET test, determine whether the result is positive or negative, and create training data (i.e., if the result is positive, a label indicating that amyloid-β lesions are present in the brain is assigned, and if the result is negative, training data indicating that amyloid-β lesions are absent in the brain is created). Note that the information used to create the training data is not limited to images from an amyloid PET test, and may be measurement values ​​from a cerebrospinal fluid test, etc.

[0030] The machine learning unit 202 uses any machine learning method to generate an estimation model that outputs the presence or absence of amyloid-β lesions in the brain when blood information is input (i.e., determines the parameters of the estimation model). For example, the learning device 20 can use the LightGBM (Light Gradient Boosting Machine) method. The machine learning unit 202 stores the generated estimation model in the estimation model storage unit 203.

[0031] The estimation model storage unit 203 stores the estimation model generated by the machine learning unit 202. The estimation model is a trained model that outputs the presence or absence of amyloid-β lesions in the brain of the subject 30 when information about the blood of the subject 30 is input.

[0032] <Processing Method> The estimation process will be described below with reference to FIG. 4, and the learning process will be described with reference to FIG.

[0033] <<Estimation Method>> FIG. 4 is a flowchart of a process for estimating the presence or absence of amyloid-β lesions in the brain according to one embodiment of the present invention.

[0034] In step 101 (S101), the estimation device 10 acquires information about the blood of the subject 30.

[0035] In step 102 (S102), the estimation device 10 inputs the information about the blood of the subject 30 acquired in S101 into an estimation model and estimates the presence or absence of amyloid β lesions in the brain of the subject 30.

[0036] In step 103 (S103), the estimation device 10 outputs information on the presence or absence of amyloid-β lesions in the brain of the subject 30 estimated in S102 (for example, displays the estimation result on a display or the like of the estimation device 10).

[0037] Thereafter, a doctor or the like can make a definitive diagnosis of Alzheimer's disease in the subject 30 based on the estimation results of the estimation device 10.

[0038] <<Learning Method>> FIG. 5 is a flowchart of a learning process for generating an estimation model according to one embodiment of the present invention.

[0039] In step 201 (S201), the learning device 20 acquires learning data.

[0040] In step 202 (S202), the learning device 20 generates an estimation model using any machine learning method that outputs the presence or absence of amyloid β lesions in the brain when blood-related information is input.

[0041] <Accuracy of Estimation Model> The accuracy of the estimation model will be explained. When the quantitative values ​​of plasma phosphorylated tau, amyloid beta 40, amyloid beta 42, LYZL2, CAPN3, CEACAM3, and CEP152 are input, the AUC (area under the ROC (Receiver Operating Characteristic) curve)) of the estimation model that outputs the presence or absence of amyloid beta lesions in the brain is approximately 1.00 (0.99). This shows that a positive amyloid PET test can be estimated with extremely high accuracy.

[0042] 6 shows data comparing a model using amyloid-β42 / amyloid-β40 and phosphorylated tau (p-tau181) as inputs according to one embodiment of the present invention with a six-factor model (a model using six factors: amyloid-β42 / amyloid-β40, phosphorylated tau (p-tau181), CAPN3, LYZL2, CEACAM3, and CEP152 as inputs). The AUC for amyloid-β42 / amyloid-β40 and phosphorylated tau (p-tau181) alone was 0.94, while the AUC for the six-factor model was 0.99, indicating that the six-factor model had a significantly higher accuracy rate.

[0043] <Effects> Instead of conducting conventional invasive and costly amyloid PET tests or cerebrospinal fluid tests, the present invention requires only a blood sample, making it possible to detect intracerebral amyloid-β lesions found in the brains of patients with Alzheimer's disease non-invasively and at low cost.

[0044] 7 is a diagram showing the hardware configuration of the intracerebral amyloid β lesion estimation device 10 and the learning device 20 according to one embodiment of the present invention. The intracerebral amyloid β lesion estimation device 10 and the learning device 20 can include a control unit 1001, a main memory unit 1002, an auxiliary memory unit 1003, an input unit 1004, an output unit 1005, and an interface unit 1006. Each of these will be described below.

[0045] The control unit 1001 is a processor (for example, a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), etc.) that executes various programs installed in the auxiliary storage unit 1003 .

[0046] The main memory unit 1002 includes a non-volatile memory (Read Only Memory (ROM)) and a volatile memory (Random Access Memory (RAM)). The ROM stores various programs, data, etc. required for the control unit 1001 to execute various programs installed in the auxiliary memory unit 1003. The RAM provides a working area into which the various programs installed in the auxiliary memory unit 1003 are expanded when executed by the control unit 1001.

[0047] The auxiliary storage unit 1003 is an auxiliary storage device that stores various programs and information used when the various programs are executed.

[0048] The input unit 1004 is an input device through which an operator of the intracerebral amyloid β lesion estimation device 10 and the learning device 20 inputs various instructions to the intracerebral amyloid β lesion estimation device 10 and the learning device 20 .

[0049] The output unit 1005 is an output device that outputs the internal states of the intracerebral amyloid β lesion estimation device 10 and the learning device 20, etc.

[0050] The interface unit 1006 is a communication device for connecting to a network and communicating with other devices.

[0051] Although the examples of the present invention have been described in detail above, the present invention is not limited to the specific embodiments described above, and various modifications and changes are possible within the scope of the gist of the present invention as set forth in the claims.

[0052] This international application claims priority based on Japanese Patent Application No. 2023-077208, filed on May 9, 2023, the entire contents of which are incorporated herein by reference.

[0053] REFERENCE SIGNS LIST 1 Cerebral amyloid β lesion estimation system 10 Cerebral amyloid β lesion estimation device 20 Learning device 30 Subject 101 Information acquisition unit 102 Estimation unit 103 Estimation model storage unit 104 Estimation result output unit 201 Learning data acquisition unit 202 Machine learning unit 203 Estimation model storage unit 1001 Control unit 1002 Main storage unit 1003 Auxiliary storage unit 1004 Input unit 1005 Output unit 1006 Interface unit

Claims

1. An apparatus for estimating amyloid beta lesions in the brain, comprising: an information acquisition unit that acquires information about a subject's blood; an estimation unit that inputs the acquired information about the subject's blood into the estimation model and estimates the presence or absence of amyloid beta lesions in the subject's brain; and an estimation result output unit that outputs information about the presence or absence of amyloid beta lesions in the subject's brain.

2. The device for estimating intracerebral amyloid-β lesions as described in claim 1, wherein the information regarding blood includes information regarding at least one of plasma and serum.

3. The device for estimating amyloid beta lesions in the brain as described in claim 1 or 2, wherein the information regarding the blood includes a quantitative value of plasma phosphorylated tau, a quantitative value of amyloid beta 40, a quantitative value of LYZL2, and a quantitative value of CAPN3.

4. The intracerebral amyloid beta lesion estimation device as described in claim 3, wherein the information regarding the blood further includes at least one of a quantitative value of amyloid beta 42, a quantitative value of amyloid beta 42 / quantitative value of amyloid beta 40, a quantitative value of CEACAM3, and CEP152.

5. The intracerebral amyloid β lesion estimation device of claim 1, wherein the estimation model is generated by machine learning using the results of an amyloid PET test as learning data.

6. A method executed by an intracerebral amyloid beta lesion estimation device having an estimation model that outputs the presence or absence of amyloid beta lesions in the brain when information regarding blood is input, the method comprising the steps of: acquiring information regarding a subject's blood; inputting the acquired information regarding the subject's blood into the estimation model to estimate the presence or absence of amyloid beta lesions in the subject's brain; and outputting the estimated information regarding the presence or absence of amyloid beta lesions in the subject's brain.

7. A program for causing an intracerebral amyloid beta lesion estimation device having an estimation model that outputs the presence or absence of amyloid beta lesions in the brain when information regarding blood is input, to execute the following steps: acquiring information regarding a subject's blood; inputting the acquired information regarding the subject's blood into the estimation model to estimate the presence or absence of amyloid beta lesions in the subject's brain; and outputting information regarding the estimated presence or absence of amyloid beta lesions in the subject's brain.

8. An estimation model that outputs the presence or absence of amyloid beta lesions in a subject's brain when information about the subject's blood is input.

Citation Information

Patent Citations

  • Biomarker and device for diagnosing Alzheimer's disease

    CN115201495A

  • Biomarkers for diagnosing the asymptomatic stage of Alzheimer's disease and their use

    JP2023505008A

  • Malware detection system using LSTM method to provide a service vaccine platform with high detction rate

    KR102495329B1

  • Non-invasive assessment of alzheimer's disease

    WO2022115705A2

  • Information processing device, medical assistance device, and medical assistance method using multi-modal type machine learning model

    WO2022265022A1