Brain age prediction and brain disease risk assessment method and related system thereof

Abstract

The invention belongs to the technical field of brain science disease early screening, prevention and mechanism analysis, and particularly relates to a brain age prediction and brain disease risk assessment method and a related system thereof. According to preprocessing of human tissue transcriptome gene expression data and crowd plasma proteome data, it is guaranteed that input data is coordinated and consistent; establishing brain age prediction models of different brain tissue areas on the basis of LASSO regression of bootstrap sampling; and taking the selected brain region model with high correlation between the predicted age and the actual age as input, and evaluating the risk of poor brain age of the brain region model on brain diseases. According to the method, transcriptome information and plasma proteome information are combined, and the onset risk of the brain diseases is predicted through Cox risk regression, so that the accuracy of model prediction is improved, and a potential target is provided for subsequent early intervention of the brain diseases.

Description

Technical Field

[0001] The present invention belongs to the technical field of early screening, prevention, and mechanism analysis of brain science diseases, and particularly relates to a brain age prediction and brain disease risk assessment method and its related system. Background Art

[0002] Currently, the problem of population aging is gradually emerging. According to the estimation of the World Health Organization (WHO), it is expected that by 2030, one in every six people (i.e., a total of 210 million people) will be over 60 years old. In low-income, middle-income, and high-income countries, the incidence, mortality, and medical costs of most diseases are related to the actual age of the population. Throughout a person's life cycle, aging occurs at the sites of the origin of various acute and chronic diseases, and in fact, the basic aging process even begins before conception, such as the aging trend shown by oocytes related to Down syndrome.

[0003] Brain tissue age, that is, brain age (Brain Age), refers to the biological age of the brain predicted by a machine learning model established through characteristic data. The difference between it and the individual's actual age, namely the brain age gap (Brain Age Gap, BAG), can reflect the health status of the individual's brain tissue and predict the aging progress of brain tissue regions. Brain age research not only has important scientific significance but also has extensive clinical value. Research shows that the brain age gap can quantify the individual's brain aging speed. A positive deviation (predicted age > actual age, brain age gap > 0) indicates accelerated aging and is related to cognitive decline and disease risk. And the brain age difference can distinguish normal aging from pathological aging (such as Alzheimer's disease) and provide a biological marker for the individual aging trajectory. The greater the BAG gap of aging individuals, the higher the risk of mental or physical problems and the easier it is to die prematurely.

[0004] Among the issues of population aging, brain aging may lead to the occurrence of a variety of complex brain diseases, which will impose a serious burden on medical resources. Existing brain aging studies based on magnetic resonance imaging (MRI) or clinical indicators have a small sample size and high sample collection costs, making it difficult to promote the sinking of medical technology resources for brain disease prediction and early screening. In recent years, population cohort aging studies based on blood biomarkers can provide insights at the protein level, but they all focus on the aging of organs or whole brain tissues, resulting in the current lack of clarity on regional brain aging and its association with brain aging diseases. Different brain tissue regions have different morphological characteristics, gene expression, and disease associations. Studies have shown that the degree of brain softening is regionally specific with age, there are regional differences in gene expression of brain cells, and there are regional differences in metabolism and oxidative stress. While the overall decrease in the level of DNA methylation in the whole genome, abnormal high methylation of certain CpG-rich regions (such as the PRNP gene) may be associated with degenerative lesions in specific brain regions. These differences in brain regions make it challenging to explore the association between brain age and brain diseases in different brain regions in highly complex and dynamically changing brain tissues. In addition, the current large-scale organ age research on plasma proteins in the population has not gone deep into specific areas of the brain. There is still a lack of in-depth research on the association between brain age prediction models constructed with brain region-specific proteins and brain disease risks, which brings obstacles to clinical brain region age prediction and disease risk research. Summary of the invention

[0005] The present invention provides a method for predicting brain age and assessing brain disease risk and a related system thereof, so as to solve the technical problems existing in the prior art, such as limited application of brain age research, small sample size and high sample collection cost in brain aging research, and lack of research on the association between brain age prediction and brain disease risk.

[0006] In order to achieve the above object, the present invention adopts the following technical solution: A method for predicting brain age and assessing brain disease risk comprises the following steps: Obtain human tissue transcriptome data, plasma proteome data of healthy people, and plasma proteome data of diseased people, and perform data preprocessing; Obtain brain region marker gene data based on human tissue transcriptome data; Mapping brain region marker gene data with human plasma proteome data to obtain brain region marker protein data; The brain region marker protein data were split into a training set and an independent test set. Brain age prediction models for different brain regions were established based on the training set. The brain age prediction models were evaluated based on the independent test set, and the optimal brain age prediction models for different brain regions were retained. Based on the plasma proteome data of the disease population, the optimal actual age prediction model of different brain regions is used to calculate the brain age difference. According to the brain age difference, the correlation between the brain age difference and the brain disease risk is obtained. Based on the correlation between the brain age difference and the brain disease risk, brain age prediction and brain disease risk assessment are realized.

[0007] The preprocessing of the human tissue transcriptome data is specifically as follows: the expression matrix of the human tissue transcriptome data is subjected to CPM normalization and z-score standardization; the preprocessing of the plasma proteome data of the healthy population and the plasma proteome data of the disease population is specifically as follows: the data with a NaN ratio greater than 30% in the human plasma proteome data is excluded, and for the data with NaN and a NaN ratio less than or equal to 30%, KNN-impute imputation is performed to fill in the missing values.

[0008] The obtaining of the brain region marker gene data is specifically as follows: based on the gene expression matrix data in the human tissue transcriptome data, the average gene expression in different tissues and organs is calculated. If the average gene expression of a gene in a certain tissue or organ is 2 times or more than its average gene expression in other tissues or organs, this gene is used as the marker gene of this tissue or organ, and the marker gene data of different tissues and organs is obtained; from the marker gene data of different tissues and organs, the brain tissue marker gene data is screened; the gene expression matrix of the human tissue transcriptome data is filtered to obtain a gene expression matrix containing only brain tissues; the brain tissue marker gene data and the gene expression matrix containing only brain tissues are combined into a new matrix. For each gene in the new matrix, according to its gene expression in different brain regions, if the expression of a gene in a certain brain region is 1.5 times or more than its expression in other brain regions, then this gene is the marker gene of this brain region, and finally the brain region marker gene data is obtained.

[0009] The method for mapping the brain region marker gene data and the human plasma proteome data is specifically as follows: the SYMBOL names of all marker genes in the brain region marker gene data are extracted, and the SYMBOL names of the marker genes are intersected with the protein names of the plasma proteome data of the healthy population. The plasma proteome data of the intersection part is the brain region marker protein data.

[0010] The method for establishing the actual age prediction model is specifically as follows: Establish a brain age prediction model based on the machine learning method of LASSO regression. Split the brain region biomarker data into a training set and an independent test set according to a split ratio of 8:2. Perform 100 resamplings on the training set using the bagging method. Each sampling obtains a Bootstrap sample with the same dimension as the original training set to form a new training set. Using gender as a covariate and the new training set as input features, perform Bootstrap training on the brain age prediction model to obtain brain age prediction models for different brain regions; during the Bootstrap training of the brain age prediction model, use grid search to perform hyperparameter tuning on the L1 regularization parameter of the brain age prediction model, and perform five-fold cross-validation. Use as the model evaluation criterion for five-fold cross-validation, and save the optimized brain age prediction models for different brain regions obtained from each training.

[0011] Use the independent test set to test the optimized brain age prediction models for different brain regions saved after training, obtain the predicted ages of the independent test set, perform pearson calculation on the actual ages and predicted ages of the population in the independent test set, and use the correlation coefficient R and P value as the model evaluation criteria. Save the optimized brain age prediction models for different brain regions with a correlation coefficient R greater than 0.5 to obtain the optimal brain age prediction models for different brain regions.

[0012] The method for obtaining the correlation between brain age difference and brain disease risk is specifically as follows: According to the optimal brain age prediction models for different brain regions, perform prediction on the plasma proteome data of the disease population to output different predicted brain ages. Based on the predicted brain ages and the actual ages of the population in the plasma proteome data of the disease population, obtain the brain age difference. Use the brain age difference obtained from the optimal brain age prediction models for different brain regions as the independent variable, use the time difference between the observation start date and the observation end date of the healthy population and the diseased population as the dependent variable, and use gender and age as covariates to perform Cox proportional hazards regression analysis to calculate the correlation between the brain age difference and the risk of developing new diseases.

[0013] A brain age prediction and brain disease risk assessment system, including a data acquisition module, a data processing module, a model establishment module, and a risk assessment module; The data acquisition module is used to acquire human tissue transcriptome data, healthy population plasma proteome data, and disease population plasma proteome data; The data processing module is used to preprocess the human tissue transcriptome data, healthy population plasma proteome data, and disease population plasma proteome data; according to the human tissue transcriptome data, obtain brain region biomarker gene data; map the brain region biomarker gene data and the human plasma proteome data; split the brain region biomarker data; A model establishment module, configured to establish a brain age prediction model for different brain regions according to a training set, evaluate the brain age prediction model according to an independent test set, and retain the optimal brain age prediction model for different brain regions; A risk assessment module, configured to calculate a brain age difference according to the plasma proteome data of a disease population by using the optimal actual age prediction model of different brain regions, obtain the correlation between the brain age difference and the brain disease risk according to the brain age difference, and implement brain age prediction and brain disease risk assessment based on the correlation between the brain age difference and the brain disease risk.

[0014] An electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the steps of a method for brain age prediction and brain disease risk assessment.

[0015] A computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, implements the steps of a method for brain age prediction and brain disease risk assessment.

[0016] Compared with the prior art, the present invention has the following beneficial effects: The core of the method for brain age prediction and brain disease risk assessment based on plasma proteome of the present invention is to utilize the combination of the information of the human plasma proteome and the human organ transcriptome, that is, the differential expression data of the plasma proteome in the human brain tissue region with age, and use it as the eigenvalue of the model, so as to perform more accurate and rigorous prediction of the onset risk of brain diseases. The present invention obtains brain region-specific marker proteins by mapping marker genes to the brain tissue-brain region, thereby improving the resolution of model prediction and providing potential targets for the early intervention of subsequent brain diseases. At the same time, the method of the present invention makes up for the deficiencies of brain medical imaging examinations such as magnetic resonance to a certain extent, and supplements the problems such as the difficulty in deeply studying the molecular mechanism of chronic diseases in the brain region in clinical practice. Therefore, the present invention maps the plasma proteins to the marker genes of the human organ transcriptome, deeply explores the relationship between the expression of brain tissue-brain region-specific marker proteins and brain age and brain diseases, makes a more rigorous prediction of the occurrence risk of brain diseases, and further achieves prevention, control and treatment.

[0017] Furthermore, according to the prediction model method of the present invention, gene expression data from transcriptome sequencing of 8,388 donors, 17,382 samples, and 52 tissue sources in the GTEx database were collected to identify brain tissue-brain region specific marker genes, which were mapped to the processed plasma proteome data of 7,082 healthy individuals in the UKB, including 2,920 protein expression information, and a brain age prediction model was trained; model results with a Pearson R greater than 0.5 in the model were screened to obtain three brain region models, and brain age prediction and brain age difference calculation were performed on 45 brain disease populations screened from the UKB plasma proteome data; and the association between the brain age difference of different brain region models and the risk of brain diseases was calculated based on the Cox proportional hazards regression model; finally, 69 specific brain region models were found to have a significant impact on brain diseases (P<0.05); among them, for example, according to the cerebellum model, hemiplegia can be predicted, and an increase in the cerebellum brain age difference (accelerated aging) will cause an increase in the risk of this disease. Furthermore, this method is applicable to studying the impact of brain age differences in multiple brain tissue regions on multiple brain diseases, effectively solving the problem of risk prediction for complex brain diseases or dysfunctions. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 : Flowchart of the brain age prediction and brain disease risk assessment method; Figure 2 : Schematic diagram of brain region marker protein data; Figure 3 : Schematic diagram of plasma proteome data of healthy individuals; Figure 4 : Visual scatter plot of Pearson correlation coefficient calculation between actual age and predicted age; Figure 5 : Schematic diagram of the output of the degree of association between prediction results and disease risk; Figure 6 : Schematic diagram of the brain age prediction and brain disease risk assessment method; Figure 7 : Schematic diagram of the system module of the brain age prediction and brain disease risk assessment; Figure 8 : Schematic diagram of the electronic device for brain age prediction and brain disease risk assessment.

[0019] Label description: 100, electronic device; 101, memory; 102, processor; 103, computer program; 104, communication bus. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0020] To further understand the content of the present invention, the following describes the present invention in detail with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments are only for explaining the present invention and not for limiting it.

[0021] The embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0022] Embodiment 1 Refer to Figure 6 , a method for predicting brain age and assessing the risk of brain diseases, comprising the following steps: Obtain human tissue transcriptome data, healthy population plasma proteome data, and diseased population plasma proteome data, and perform data preprocessing; According to the human tissue transcriptome data, obtain brain region marker gene data; Map the brain region marker gene data to the human plasma proteome data to obtain brain region marker protein data; Split the brain region marker protein data into a training set and an independent test set, establish a brain age prediction model for different brain regions according to the training set; evaluate the brain age prediction model according to the independent test set, and retain the optimal brain age prediction model for different brain regions; According to the diseased population plasma proteome data, calculate the brain age difference using the optimal actual age prediction model for different brain regions, obtain the correlation between the brain age difference and the risk of brain diseases according to the brain age difference, and realize brain age prediction and brain disease risk assessment based on the correlation between the brain age difference and the risk of brain diseases.

[0023] Embodiment 2 Refer to Figure 7 , a system for predicting brain age and assessing the risk of brain diseases, comprising a data acquisition module, a data processing module, a model establishment module, and a risk assessment module; The data acquisition module is used to obtain human tissue transcriptome data, healthy population plasma proteome data, and diseased population plasma proteome data; The data processing module is used to perform preprocessing on the human tissue transcriptome data, healthy population plasma proteome data, and diseased population plasma proteome data; obtain brain region marker gene data according to the human tissue transcriptome data; map the brain region marker gene data to the human plasma proteome data; split the brain region marker protein data; The model establishment module is used to establish a brain age prediction model for different brain regions according to the training set, and evaluate the brain age prediction model according to the independent test set, and retain the optimal brain age prediction model for different brain regions; The risk assessment module is used to calculate the brain age difference using the optimal actual age prediction model for different brain regions according to the diseased population plasma proteome data, obtain the correlation between the brain age difference and the risk of brain diseases according to the brain age difference, and realize brain age prediction and brain disease risk assessment based on the correlation between the brain age difference and the risk of brain diseases.

[0024] Embodiment 3 Based on Figure 1In the method flow of brain age prediction and brain disease risk assessment, the present invention proposes a method for brain age prediction and brain disease risk assessment. The specific implementation method includes the following steps: Download human tissue transcriptome data from the GTEx database server. The human tissue transcriptome data includes the original count matrix and human tissue source information. Download healthy population plasma proteome data and disease population plasma proteome data from the UKB database server. Both the healthy population plasma proteome data and the disease population plasma proteome data contain clinical information such as population gender and age. Perform data preprocessing on the human tissue transcriptome data, healthy population plasma proteome data, and disease population plasma proteome data. The data preprocessing is specifically as follows: For the expression matrix of the human tissue transcriptome data, use the pydeseq2 package for normalization to facilitate subsequent calculation of marker genes. For the healthy population plasma proteome data and the disease population plasma proteome data, exclude the data with a NaN ratio greater than 30% in the human plasma proteome data. For the data with NaN and a NaN ratio less than or equal to 30%, perform KNN-impute interpolation to fill in the missing values and avoid affecting subsequent modeling. The above data preprocessing solves the problems of data scale difference and data missing.

[0025] The gene expression matrix data of the human tissue transcriptome data is divided into organs and brain regions according to tissue organs, mainly including 19 organs and 7 brain regions. Among them, the organ types include fat, adrenal gland, artery, bladder, brain, intestine, esophagus, heart, kidney, liver, lung, salivary gland, muscle, pancreas, pituitary gland, skin, immune tissue, stomach, and thyroid gland. The brain tissue types include cerebellum, cerebral cortex, cerebral nucleus, hippocampus, hypothalamus, substantia nigra, and spinal cord, covering the main regions of the entire brain nervous system.

[0026] According to the gene expression matrix data in the human tissue transcriptome data after data preprocessing, calculate the average gene expression in different tissue organs. If the average gene expression of a gene in a certain tissue organ is 2 times or more than its average gene expression in other tissue organs, regard this gene as the marker gene of this tissue organ, and obtain the marker gene data of different tissue organs. Screen the brain tissue marker gene data from the marker gene data of different tissue organs. Filter the gene expression matrix of the human tissue transcriptome data to obtain a gene expression matrix that only contains brain tissue. Combine the brain tissue marker gene data and the gene expression matrix that only contains brain tissue into a new matrix. For each gene in the new matrix, according to its gene expression in different brain regions, if the expression of a gene in a certain brain region is 1.5 times or more than its expression in other brain regions, then this gene is the marker gene of this brain region, and finally obtain the brain region marker gene data.

[0027] Extract the SYMBOL names of all marker genes in the brain region marker gene data. Take the intersection of the SYMBOL names of the marker genes and the protein names in the plasma proteome data of healthy individuals. Map the brain region marker gene data to the human plasma proteome data to identify brain region marker proteins. Among them, the plasma proteome data in the intersection part is the brain region marker protein data.

[0028] Establish a brain age prediction model based on the machine learning method of LASSO regression. Split the brain region marker protein data into a training set and an independent test set according to a split ratio of 8:2. Perform 100 resamplings on the training set using the bagging method. Each sampling obtains a Bootstrap sample with the same dimension as the original training set to form a new training set. Use gender (F = 1, M = 0) as a covariate and the new training set as input features to perform Bootstrap training on the brain age prediction model to obtain brain age prediction models for different brain regions; during the Bootstrap training of the brain age prediction model, use grid search to perform hyperparameter tuning and conduct five-fold cross-validation. Use as the model evaluation criterion for five-fold cross-validation. is the regression coefficient. , When the value is the largest, the evaluation result is better. Save the brain age prediction model when the value is the largest to obtain optimized brain age prediction models for different brain regions. Use the independent test set to test the optimized brain age prediction models for different brain regions saved after training to obtain the predicted ages of the independent test set. Calculate the pearson correlation coefficient between the actual ages and the predicted ages of the population in the independent test set. Use the correlation coefficient R and the P value as the model evaluation criteria and perform visualization of the scatter plot. The scatter plot is as Figure 4 shown; Save the optimized brain age prediction models for different brain regions with a correlation coefficient R greater than 0.5 to obtain the optimal brain age prediction models for different brain regions for subsequent association with disease risks.

[0029] According to the optimal brain age prediction models for different brain regions, different predicted brain ages are output by predicting the plasma proteome data of the disease population. Based on the predicted brain ages and the actual ages of the population in the plasma proteome data of the disease population, BAG is calculated to obtain the brain age difference. Using the brain age difference obtained from the optimal brain age prediction models of different brain regions as the independent variable, and using the time difference (in years) between the observation start date (Time) and the observation end date (Event) of the healthy population and the diseased population as the dependent variable (event time), and using gender (F = 0, M = 1) and age (Age) as covariates, Cox proportional hazards regression analysis is carried out to associate the brain age difference, the previous disease structure, and the new disease outcome, and calculate the correlation between the brain age difference and the risk of new diseases, so as to achieve brain age prediction and brain disease risk assessment. Among them, the observation start date (Time) of the healthy population is the blood collection date of the population, and the observation end date (Event) of the healthy population is April 30, 2024; the observation start date (Time) of the new disease population is the blood collection date of the population, and the observation end date (Event) of the new disease population is the date of the first illness of the population.

[0030] The above data preprocessing, model establishment, model training, and model prediction and other program operations are all based on the Linux operating system and are compiled using the python program. The python program is pre-installed with pandas, numpy, scikit-learn, and pydeseq2 packages, and the output results are used for the next analysis.

[0031] Example 4 The method for predicting brain age and assessing brain disease risk based on plasma proteome according to the present invention mainly considers the following two advantages: (1) Selection of input data Proteomics research can locate key genes that directly regulate basic life activities from the root, and at the same time, construct the gene regulatory network of the disease by studying the interactions between regulatory genes, so as to reveal the molecular mechanism in the process of complex diseases and provide basic theoretical support for the clinical intervention, later treatment and prognosis of the disease. Among them, plasma proteome has become an important data source for proteome research in recent years due to its advantages such as easy access and low cost in clinical practice. Studies have shown that the blood of young mice can reverse aging and diseases of various tissues. This supports the hypothesis that age-related changes in protein molecules in the blood can provide new biological insights into aging diseases. At the same time, the results of clinical practice show that it is feasible to use organ-specific plasma proteins to non-invasively assess organ health and disease status, which provides theoretical support for the development of brain region-specific plasma protein-specific aging and brain disease risk assessment. It has high clinical research value to reflect the age of specific brain regions and link its protein level with brain aging diseases by measuring the human plasma proteome. Therefore, the present invention can use plasma protein data as a molecular basis and characterization for predicting brain tissue-brain region aging, and make more rigorous predictions on the occurrence, prevention, control and diagnosis and treatment of brain diseases.

[0032] (2) Program automation and performance optimization Simplify the analysis process as much as possible, define the input once, and get the final prediction result directly. In addition, because it involves large-scale prediction features, it is necessary to optimize the internal algorithm design to minimize the program operation time.

[0033] The method for predicting brain age and assessing brain disease risk described in the present invention mainly adopts the following technical solutions: Preliminary preparation Considering that the present invention is implemented using shell scripts, it is recommended to use a server-side Linux operating system, install Python software (https: / / www.python.org / ), and load the required Python packages (https: / / pypi.org / ) to greatly speed up the program. The public data required for the model can be unzipped and downloaded by yourself. In addition, users can also apply to download the same functional data as the public data in the program and replace it. The public data connections required in the model are as follows: https: / / www.ukbiobank.ac.uk / Identifying brain region marker genes is divided into the first module and the second module, namely the organ marker gene module and the brain region marker gene module. The first module will output the results of different organ marker genes. The first column is the organ name Group, and the second column is the gene name Gene of the corresponding marker gene. When running the second module program, the user needs to use the output data of the first module as the input data of the second module and run the calculation again. The output result is the result of the corresponding marker genes in different brain regions, which has the same structure as the result of the first module. Using the final results of the above two modules as input, map them to the protein names in the plasma proteome data of healthy people, and obtain the brain region marker protein data of different brain regions after taking the intersection as the final result. The first column is Group, and the second column is Protein. A total of 124 brain region-specific marker proteins were identified, specifically including: 56 in the cerebellum region; 27 in the spinal cord region; 22 in the cerebral cortex region; 12 in the cerebral nucleus region; 7 in the hypothalamus. Among them, 2 brain regions in the original marker gene list were discarded because their genes do not exist in the plasma protein sequence. The output result is as Figure 2 shown.

[0034] Actually conduct the training of the brain age prediction model for brain regions, specifically including the following steps: All the above files are saved in the current working directory. Taking the cerebellum marker protein data list as an example, the process of a brain age prediction and brain disease risk assessment method is carried out under the linux operating system.

[0035] In the current path, on the premise that all files are ready. Among them, the file of brain region marker protein data is saved as region_markers.tsv, and the file of plasma proteome data of healthy people is saved as proteins.tsv. The specific content of the plasma proteome data of healthy people is as Figure 3 shown.

[0036] Running the program for model training will output 1 folder, which is named after the current brain region name Cerebellum. It contains 100 result files for model training, such as the model binary file like ukb_hc_bs1.pkl. The number after bs represents the result of the nth Bootstrap training. At the same time, for the brain age prediction models of different brain regions, use the models of 100 trainings to predict the age of the independent test set, take the average prediction as the predicted age of each individual, calculate the pearson R and p values as the model evaluation criteria, and conduct the visualization of the scatter plot. The scatter plot is as Figure 4 shown; The user selects the brain age prediction models of the brain regions with a Pearson R value greater than or equal to the threshold of Pearson R (the default threshold of Pearson R is 0.05) from the model prediction results generated by the model training program, and uses the model file output by the model training program as the input file for the subsequent module program.

[0037] For the specific embodiments described above, the experimental results are interpreted. For the selected cerebellar brain region, the model prediction result shows that the Pearson R value is 0.57 and the p value is 4.4×10-121, indicating that the correlation between the predicted age result of the brain age prediction model of the cerebellar brain region and its actual age for the independent test set is 57%, which is above 50% and meets the default threshold. The brain age prediction model of this brain region can be retained for use in the subsequent brain disease risk prediction module.

[0038] The actual evaluation of the association between the brain region model and the brain disease risk specifically includes the following steps: All the above files are saved in the current working directory. Taking the brain age prediction model of the cerebellum and the disease of Hemiplegia as an example, under the linux operating system, when all files are ready in the current path. The brain age prediction model file of the cerebellum is the.pkl suffix file set under the file path output by the model training program, and the input file of the plasma proteome of the disease population is Hemiplegia_pro.tsv, which has the same format as the plasma proteome dataset of the healthy population; run the brain age prediction model of the cerebellum, calculate the degree of association between the prediction result and the disease risk, and output 1 file result, which is a result file named disease_associations.tsv, as Figure 5 shown, where Disease_Name represents the disease name, Model represents the model name, HR represents the Hazard Ratio (i.e., the risk ratio of the model for the disease), and p represents the significance of the risk ratio; for the specific embodiments described above, the experimental results are interpreted. The selected cerebellar brain region is associated with Hemiplegia, and the association result shows that the HR value is 1.069 (±0.03) and the p value is 8.1×10 -8 , indicating that the correlation between the predicted age result of the brain age difference of the cerebellar brain region model and its actual age for the hemiplegic population is about 6.9%, and it is extremely significant in the population effect; Example 5 Please refer to Figure 8 shown, the present invention also provides an electronic device 100 for implementing a method for brain age prediction and brain disease risk assessment; the electronic device 100 includes a memory 101, at least one processor 102, a computer program 103 stored in the memory 101 and executable on the at least one processor 102, and at least one communication bus 104.

[0039] The memory 101 can be used to store the computer program 103. The processor 102 realizes the steps of the method for predicting brain age and assessing brain disease risk described in Embodiment 1 by running or executing the computer program stored in the memory 101 and calling the data stored in the memory 101. The memory 101 may mainly include a program storage area and a data storage area. Among them, the program storage area can store an operating system, application programs required for at least one function (such as a sound playback function, an image playback function, etc.); the data storage area can store data created according to the use of the electronic device 100 (such as audio data, etc.). In addition, the memory 101 may include non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) card, a Flash Card, at least one magnetic disk storage device, a flash memory device, or other non-volatile solid-state storage devices.

[0040] The at least one processor 102 can be a Central Processing Unit (CPU), or can also be other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The processor 102 can be a microprocessor or the processor 102 can also be any conventional processor, etc. The processor 102 is the control center of the electronic device 100 and connects various parts of the entire electronic device 100 through various interfaces and lines.

[0041] The memory 101 in the electronic device 100 stores multiple instructions to implement a method for predicting brain age and assessing brain disease risk. The processor 102 can execute the multiple instructions to thereby implement: Obtain human tissue transcriptome data, healthy population plasma proteome data, and diseased population plasma proteome data, and perform data preprocessing; Obtain brain region marker gene data according to the human tissue transcriptome data; Map the brain region marker gene data to the human plasma proteome data to obtain brain region marker protein data; The brain region marker protein data is split into a training set and an independent test set, and a brain age prediction model for different brain regions is established based on the training set; the brain age prediction model is evaluated according to the independent test set, and the optimal brain age prediction model for different brain regions is retained; According to the plasma proteome data of the disease population, the optimal actual age prediction model of different brain regions is used to calculate the brain age difference. Based on the brain age difference, the correlation between the brain age difference and the brain disease risk is obtained, and the brain age prediction and the brain disease risk assessment are realized based on the correlation between the brain age difference and the brain disease risk.

[0042] Example 6 If the modules / units integrated in the electronic device 100 are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, to implement all or part of the processes in the above method embodiments of the present invention, it can also be completed by a computer program instructing relevant hardware. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by a processor, the steps of the above method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file or some intermediate form, etc. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory and read-only memory (ROM, Read-Only Memory).

[0043] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0044] The present invention is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram can be realized by computer program instructions, and the combination of the processes and / or blocks in the flowchart and / or block diagram can also be realized by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate for realizing the processes in Figure 1 one process or multiple processes and / or blocks Figure 1a device for the functions specified in one or more boxes.

[0045] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory produce a manufacture including an instruction device that implements the functions specified in one Figure 1 process or more processes and / or boxes Figure 1 a box or more boxes.

[0046] These computer program instructions may also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to produce a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one Figure 1 process or more processes and / or boxes Figure 1 a box or more boxes.

[0047] In addition, it should be understood that although this specification is described in terms of embodiments, not every embodiment contains only an independent technical solution. This narrative manner of the specification is only for clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment may also be appropriately combined to form other embodiments that can be understood by those skilled in the art. The above content is only to illustrate the technical idea of the present invention and cannot be used to limit the protection scope of the present invention. Any modification made on the basis of the technical solution according to the technical idea proposed by the present invention falls within the protection scope of the claims of the present invention.

Claims

1. A method for predicting brain age and assessing brain disease risk, characterized in that, It includes the following steps: Obtain human tissue transcriptome data, healthy population plasma proteome data, and diseased population plasma proteome data, and perform data preprocessing; According to the human tissue transcriptome data, obtain brain region marker gene data; Map the brain region marker gene data to the human plasma proteome data to obtain brain region marker protein data; Split the brain region marker protein data into a training set and an independent test set, and establish brain age prediction models for different brain regions based on the training set; evaluate the brain age prediction models according to the independent test set, and retain the optimal brain age prediction models for different brain regions; According to the diseased population plasma proteome data, calculate the brain age difference using the optimal actual age prediction models for different brain regions. Based on the brain age difference, obtain the correlation between the brain age difference and the brain disease risk, and realize brain age prediction and brain disease risk assessment based on the correlation between the brain age difference and the brain disease risk.

2. The method for predicting brain age and assessing brain disease risk according to claim 1, wherein The preprocessing of the human tissue transcriptome data is specifically as follows: perform CPM normalization and z-score standardization on the expression matrix of the human tissue transcriptome data; the preprocessing of the healthy population plasma proteome data and the diseased population plasma proteome data is specifically as follows: exclude the data with a NaN ratio greater than 30% in the human plasma proteome data, and perform KNN-impute interpolation on the data with NaN data and a NaN ratio less than or equal to 30% to fill in the missing values.

3. A method for predicting brain age and assessing brain disease risk according to claim 1, characterized in that, The specific method for obtaining the brain region marker gene data is as follows: based on the gene expression matrix data in the human tissue transcriptome data, calculate the average gene expression in different tissues and organs. If the average gene expression of a gene in a certain tissue or organ is 2 times or more greater than its average gene expression in other tissues and organs, regard this gene as the marker gene of this tissue or organ to obtain the marker gene data of different tissues and organs; screen the brain tissue marker gene data from the marker gene data of different tissues and organs; Filter the gene expression matrix of the human tissue transcriptome data to obtain a gene expression matrix containing only the brain tissue; combine the brain tissue marker gene data and the gene expression matrix containing only the brain tissue into a new matrix. For each gene in the new matrix, according to its gene expression in different brain regions, if the expression of the gene in a certain brain region is 1.5 times or more greater than its expression in other brain regions, then this gene is the marker gene of this brain region, and finally obtain the brain region marker gene data.

4. A method for predicting brain age and assessing brain disease risk according to claim 1, characterized in that, The method for mapping the brain region marker gene data to the human plasma proteome data is specifically as follows: extract the SYMBOL names of all marker genes in the brain region marker gene data, and take the intersection of the SYMBOL names of the marker genes and the protein names of the healthy population plasma proteome data. The plasma proteome data of the intersection part is the brain region marker protein data.

5. A brain age prediction and brain disease risk assessment method according to claim 1, characterized in that The method for establishing the actual age prediction model is specifically as follows: A brain age prediction model is established based on the machine learning method of LASSO regression. The brain region biomarker data is split into a training set and an independent test set according to a split ratio of 8:

2. The training set is resampled 100 times using the bagging method. Each sampling obtains a Bootstrap sample with the same dimension as the original training set to form a new training set. Using gender as a covariate and the new training set as input features, the brain age prediction model is trained by Bootstrap to obtain the brain age prediction models for different brain regions. During the Bootstrap training of the brain age prediction model, grid search is used to perform hyperparameter tuning on the L1 regularization parameter of the brain age prediction model, and five-fold cross-validation is performed. is used as the model evaluation criterion for five-fold cross-validation, and the optimized brain age prediction models for different brain regions obtained from each training are saved.

6. The method for predicting brain age and assessing brain disease risk according to claim 7, characterized in that, Use an independent test set to test the optimized brain age prediction models of different brain regions saved after training, obtain the predicted ages of the independent test set, calculate the Pearson correlation between the actual ages and predicted ages of the population in the independent test set, and use the correlation coefficient R and P-value as the model evaluation criteria. Save the optimized brain age prediction models of different brain regions with a correlation coefficient R greater than 0.5 to obtain the optimal brain age prediction models of different brain regions.

7. A method for predicting brain age and assessing brain disease risk according to claim 1, characterized in that, The method for obtaining the correlation between brain age difference and brain disease risk is as follows: According to the optimal brain age prediction models of different brain regions, obtain the different predicted brain ages output by predicting the plasma proteome data of the disease population. Based on the predicted brain ages and the actual ages of the population in the plasma proteome data of the disease population, obtain the brain age difference. Use the brain age difference obtained from the optimal brain age prediction models of different brain regions as the independent variable, use the time difference between the observation start date and the observation end date of the healthy population and the diseased population as the dependent variable, and use gender and age as covariates to perform Cox proportional hazards regression analysis to calculate the correlation between the brain age difference and the risk of developing new diseases.

8. A brain age prediction and brain disease risk assessment system, based on the brain age prediction and brain disease risk assessment method according to any one of claims 1 to 7, characterized in that, It includes a data acquisition module, a data processing module, a model establishment module, and a risk assessment module; The data acquisition module is used to acquire human tissue transcriptome data, healthy population plasma proteome data, and disease population plasma proteome data; The data processing module is used to preprocess the human tissue transcriptome data, healthy population plasma proteome data, and disease population plasma proteome data; According to the human tissue transcriptome data, obtain brain region marker gene data; Map the brain region marker gene data to the human plasma proteome data; split the brain region marker protein data; The model establishment module is used to establish brain age prediction models of different brain regions based on the training set, evaluate the brain age prediction models according to the independent test set, and retain the optimal brain age prediction models of different brain regions; The risk assessment module is used to calculate the brain age difference according to the plasma proteome data of the disease population using the optimal actual age prediction models of different brain regions, obtain the correlation between the brain age difference and the brain disease risk based on the brain age difference, and realize brain age prediction and brain disease risk assessment based on the correlation between the brain age difference and the brain disease risk.

9. An electronic device, including a memory (101), a processor (102), and a computer program (103) stored in the memory (101) and executable on the processor (102). When the processor (102) executes the computer program (103), it implements the steps of the method for predicting brain age and assessing brain disease risk according to any one of claims 1-7.

10. A computer-readable storage medium stores a computer program (103). When the computer program (103) is executed by a processor (102), it implements the steps of the method for predicting brain age and assessing brain disease risk according to any one of claims 1-7.

Images