Organ aging assessment method, device and equipment and storage medium

Abstract

The invention relates to an organ aging assessment method and device, equipment and a storage medium. According to the main technical scheme, the method comprises the steps of obtaining proteomics data of a target population and protein specific data of human organs; performing matching operation on the proteomics data and the protein specific data to obtain protein matching data of organs of the corresponding target population; performing correlation analysis on the protein matching data and real age data of the corresponding target population to obtain age-related protein data of organs of the corresponding target population; performing model training according to the age-related protein data and the real age data to obtain an organ aging evaluation model and a predicted age value of an organ of any person in the target population; and obtaining an aging state evaluation result corresponding to the human organ according to the predicted age value and the corresponding actual age value. The organ aging state evaluation method can achieve the effects of being beneficial to comprehensively and accurately evaluating the organ aging state, and further providing guarantee for organ aging and anti-aging intervention and the like.

Description

Technical Field

[0001] This application relates to the technical field of aging assessment, and particularly to an organ aging assessment method, device, equipment, and storage medium. Background Art

[0002] As is well known, the human body is composed of multiple organs, each of which plays an important role. Different organs naturally play different roles, and as people age, organs will gradually age. However, the degree of organ aging may vary among different individuals. Therefore, it is possible to evaluate the aging state of human organs to facilitate the prevention and diagnosis of related diseases.

[0003] Specifically, taking the liver organ as an example, the liver is one of the largest glandular organs in the human body, responsible for functions such as nutrient metabolism, synthesis and secretion of various proteins, decomposition of toxic substances, digestion, immune defense, storage of vitamins and minerals, and regulation of hormone levels. As people age, a series of degenerative changes occur in the morphology and function of the liver, including a decrease in liver blood flow, a weakened regenerative ability of hepatocytes, an increase in polyploid hepatocytes, an increase in immune inflammation and fibrosis deposition, and a decrease in liver metabolic capacity. At the level of liver tissue cells, liver aging causes a relative decrease in the number of hepatocytes, changes in cell structure, impaired autophagy activity, loss of protein stability, and release of inflammation-related cytokines; senescent hepatic sinusoidal endothelial cells dedifferentiate, form pseudo-capillarization, vasodilation, reduced vascular secretion ability, and a moderately pro-inflammatory state; the number of hepatic stellate cells in the elderly liver increases, showing a moderately activated state and increased expression of activation markers; in addition, liver aging also causes changes in the number and phenotype of immune cells. Liver aging increases the risk of developing various liver diseases such as fatty liver disease, cirrhosis, and liver cancer. Therefore, constructing an aging assessment model for corresponding organs such as the liver to achieve the assessment of corresponding organs can prevent and diagnose organ aging-related diseases early and improve the healthy lifespan of the elderly population.

[0004] In traditional implementation methods, clinically, blood biochemical tests are usually used to evaluate organs such as liver function, B-ultrasound is used to evaluate fatty liver, and elastography is used to evaluate the degree of liver fibrosis; in the academic research field, organ aging markers are generally determined through three dimensions: corresponding organ function, imaging, and body fluids, and compared and analyzed with the general population to determine the degree of aging of the corresponding organs; there are also proposals to evaluate the organ aging index by detecting the telomere length (T), NAD+ concentration (N), and serum albumin (C) in the blood, or to analyze the aging level of the corresponding organs using lipid metabolism markers, glucose metabolism markers, and liver enzymes.

[0005] However, the above methods all have certain limitations and are greatly affected by manual intervention, resulting in low evaluation accuracy and being relatively one-sided. Summary of the Invention

[0006] Based on this, the present application provides a method, device, equipment, and storage medium for organ aging assessment. By matching, analyzing, and model training on the proteomics data of the target population and the protein-specific data of the human organs, the assessment result of the aging state of the corresponding human organs is obtained, achieving the effects of helping to characterize the aging state of the human organs, dynamically monitoring its change process, and helping to comprehensively evaluate the occurrence and development trends of organ aging-related diseases, providing further guarantee for carrying out organ aging and anti-aging intervention measures, etc.

[0007] In the first aspect, a method for organ aging assessment is provided. The method includes: Obtain the proteomics data of the target population and the protein-specific data of the human organs; Perform a matching operation on the proteomics data and the protein-specific data to obtain the protein matching data of the organs of the corresponding target population; Perform a correlation analysis on the protein matching data and the real age data of the corresponding target population to obtain the age-related protein data of the organs of the corresponding target population; Perform model training based on the age-related protein data and the real age data to obtain an organ aging assessment model and the predicted age value of any person's organ in the target population; Based on the predicted age value and the corresponding actual age value, obtain the assessment result of the aging state of the corresponding person's organ.

[0008] According to an achievable manner in the embodiments of the present application, obtaining the proteomics data of the target population includes: Obtain the physical examination information data of multiple people; the physical examination information data includes historical medical history information data, measurement information data, and detection information data; According to the physical examination information data and the preset screening rules, obtain the plasma sample data of the target population; Perform detection and analysis on the plasma sample data to obtain the protein detection data of the target population; Perform preprocessing on the protein detection data to obtain the proteomics data of the target population.

[0009] According to an achievable manner in the embodiments of the present application, obtaining the protein-specific data of the human organs includes: According to the Human Protein Atlas, obtain the normalized gene expression level data of the same gene in different organ tissues of the human body; Sort the normalized gene expression level data of the gene in different organ tissues from largest to smallest to obtain the first organ normalized gene expression level data and the second organ normalized gene expression level data; Based on the normalized gene expression data of the first organ and the normalized gene expression data of the second organ, the ratio data of the corresponding genes is obtained; Based on the ratio data and a preset ratio threshold, the enriched gene data of the corresponding human organ is obtained; Based on the enriched gene data and the corresponding protein identification data, the protein-specific data of the corresponding human organ is obtained.

[0010] According to an implementable manner in the embodiments of the present application, a correlation analysis is performed on the protein matching data and the real age data of the corresponding target population to obtain the age-related protein data of the corresponding target population's organ, including: The protein matching data and the real age data are input into a correlation analysis model to obtain the correlation coefficient data of the corresponding target population's organ; Based on the correlation coefficient data and a preset distribution table, the significant index data of the corresponding target population's organ is obtained; The significant index values in the significant index data are sorted to obtain the probability significant data of the corresponding target population's organ; Based on the probability significant data, the age-related protein data of the corresponding target population's organ is obtained.

[0011] According to an implementable manner in the embodiments of the present application, model training is performed based on the age-related protein data and the real age data to obtain an organ aging assessment model and the predicted age value of any person's organ in the target population, including: The age-related protein data is partitioned into a data set to obtain the age-related protein partition data of the corresponding target population's organ; A normalization operation is performed on the age-related protein partition data to obtain model training data; An initial organ training model is constructed, and an optimization objective function and a loss function are determined; Based on the model training data, the optimization objective function, and the loss function, the initial organ training model is trained to obtain an organ aging assessment model; The age-related protein information of any person in the age-related protein data is input into the organ aging assessment model to obtain the predicted age value of the corresponding person's organ.

[0012] According to an implementable manner in the embodiments of the present application, based on the model training data, the optimization objective function, and the loss function, the initial organ training model is trained to obtain an organ aging assessment model, including: The model parameters of the initial organ training model are trained and screened through a grid search algorithm and a cross-validation algorithm to obtain optimal training parameters; The initial organ training model is repeatedly trained according to the model training data, the optimization objective function, the loss function, and the optimal training parameters to obtain an organ aging assessment model.

[0013] According to an implementable manner in the embodiments of the present application, based on the predicted age value and the corresponding actual age value, an aging state assessment result corresponding to the person's organ is obtained, including: The predicted age value and the corresponding actual age value are input into the organ aging index model to obtain the organ aging index value corresponding to the person; It is determined whether the organ aging index value is within the range of the preset standard state value; If the organ aging index value is within the range of the preset standard state value, it is determined that the person's organ is in a normal aging state; If the organ aging index value is not within the range of the preset standard state value, it is determined that the person's organ is in an abnormal aging state.

[0014] In a second aspect, an organ aging assessment device is provided, and the device includes: An acquisition unit that acquires the proteomics data of the target population and the protein-specific data of the human organ; A matching unit that performs a matching operation on the proteomics data and the protein-specific data to obtain the protein matching data corresponding to the target population's organ; An analysis unit that performs a correlation analysis on the protein matching data and the real age data of the corresponding target population to obtain the age-related protein data of the target population's organ; A prediction unit that performs model training based on the age-related protein data and the real age data to obtain an organ aging assessment model and the predicted age value of any person's organ in the target population; An evaluation unit that obtains an aging state assessment result corresponding to the person's organ based on the predicted age value and the corresponding actual age value.

[0015] In a third aspect, a computer device is provided, including: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores computer instructions executable by the at least one processor, and the computer instructions are executed by the at least one processor so that the at least one processor can execute the method involved in the first aspect above.

[0016] In a fourth aspect, a computer-readable storage medium is provided, on which computer instructions are stored, and the computer instructions are characterized in that the computer instructions are used to cause a computer to execute the method involved in the first aspect above.

[0017] According to the technical content provided by the embodiments of the present application, proteomics data of the target population and protein-specific data of human organs are obtained; the proteomics data and the protein-specific data are subjected to a matching operation to obtain protein matching data of the corresponding organs of the target population; the protein matching data is subjected to a correlation analysis with the true age data of the corresponding target population to obtain age-related protein data of the corresponding organs of the target population; model training is performed according to the age-related protein data and the true age data to obtain an organ aging assessment model and a predicted age value of any organ in the target population; according to the predicted age value and the corresponding actual age value, an aging state assessment result of the corresponding person's organ is obtained. Through the above operations, by matching, analyzing, and performing model training on the proteomics data of the target population and the protein-specific data of human organs, an aging state assessment result of the corresponding human organs is obtained, which helps to characterize the aging state of human organs, dynamically monitor its change process, and helps to comprehensively evaluate the occurrence and development trends of organ aging-related diseases, providing further guarantee for carrying out organ aging and anti-aging intervention measures and other aspects. Description of the Drawings

[0018] Figure 1 It is an application environment diagram of an organ aging assessment method in an embodiment; Figure 2 It is a schematic flowchart of an organ aging assessment method in an embodiment; Figure 3 It is a structural block diagram of an organ aging assessment device in an embodiment; Figure 4 It is a schematic structural diagram of a computer device in an embodiment. Detailed Embodiments

[0019] In order to make the objectives, technical solutions, and advantages of the present application clearer, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0020] An organ aging assessment method provided by the present application can be applied to, for example Figure 1In the application environment shown. Among them, the terminal 101 communicates with the server 103 through the network. Specifically, the server 103 obtains the proteomics data of the target population uploaded by the terminal 101, as well as the protein-specific data of the human organs; performs a matching operation on the proteomics data and the protein-specific data to obtain the protein matching data corresponding to the organs of the target population; performs a correlation analysis on the protein matching data and the real age data of the corresponding target population to obtain the age-related protein data of the organs of the corresponding target population; performs model training based on the age-related protein data and the real age data to obtain an organ aging assessment model and the predicted age value of any person's organ in the target population; and obtains the aging state assessment result of the corresponding person's organ according to the predicted age value and the corresponding actual age value. Among them, the terminal 101 can be, but is not limited to, various personal computers, laptop computers, smart phones, tablet computers, etc., and the server 103 can be implemented by an independent server or a server cluster composed of multiple servers.

[0021] In one embodiment, as Figure 2 shown, a method for assessing organ aging is provided, and this method can be executed by the server 103 in the application environment as Figure 1 shown. As Figure 2 shown, this method may include the following steps: Step S201: Obtain the proteomics data of the target population, as well as the protein-specific data of the human organs.

[0022] Among them, the proteomics data refers to the data obtained by detecting the plasma samples of the collected target population; the protein-specific data refers to the data obtained by screening based on the Human Protein Atlas according to a preset rule; the target population refers to a healthy population within a certain age range.

[0023] Specifically, existing research has shown that dozens of proteins directly regulate human lifespan have been found in the plasma proteome aging clock, and hundreds of proteins are biologically related to the health status of different organs. Therefore, plasma samples of the target population can be collected, and the collected plasma samples can be further detected to obtain proteomics data. At the same time, protein-specific data corresponding to the organs can also be obtained by screening based on the Human Protein Atlas according to a preset rule; here, the Human Protein Atlas refers to an expression map of human proteins in cells, tissues, and organs drawn by integrating genomics, transcriptomics, and proteomics data, and thus protein-specific data of human organs can be obtained based on this map. It should be noted that the proteomics data of the target population can be uploaded to the server by staff using a PC through the web side; the protein-specific data of human organs can be automatically screened on the server.

[0024] Step S203: Perform a matching operation on the proteomics data and the protein-specific data to obtain the protein matching data for the organs of the corresponding target population.

[0025] Here, since the proteomics data contains the protein information of the overall organs of each person in the target population, and the protein-specific data is the specific protein information of a certain organ of the human body, therefore, the protein information of the overall organs of each person in the population can be matched with the specific protein information of the human organs, and then the protein matching data for the organs of the corresponding target population can be obtained, that is, the protein matching information of each person's organs in the target population.

[0026] Step S205: Perform a correlation analysis on the protein matching data and the real age data of the corresponding target population to obtain the age-related protein data for the organs of the corresponding target population.

[0027] Here, correlation analysis refers to analyzing two or more variable elements with correlation to measure the degree of correlation between the two variable factors. Since the composition and function of human proteins change significantly with age, and these changes are closely related to the aging process and various age-related diseases, therefore, the protein matching data of the organs of the target population can be correlated with the real age data of the corresponding target population, and then the age-related protein data for the organs of the corresponding target population can be obtained, that is, the age-related protein information of each person's organs in the target population.

[0028] Step S207: Perform model training based on the age-related protein data and the real age data to obtain an organ aging assessment model and the predicted age value of any person's organ in the target population.

[0029] Among them, the organ aging assessment model can be trained based on the SVR (Support Vector Regression) model. SVR is a supervised learning algorithm that fits the data by finding an optimal hyperplane in a high-dimensional space that can tolerate a certain error range. This hyperplane can be regarded as a prediction model for the target value, which can make most data points lie within the ε-insensitive band of the plane while maximizing the width of the band.

[0030] Here, based on the SVR model, model training is performed according to the age-related protein data and the real age data to obtain the hyperparameter data suitable for the current evaluation scenario, and then the organ aging assessment model is obtained. Inputting the age-related protein information of any person in the target population into the organ aging assessment model can obtain the predicted age value of any person's organ in the target population.

[0031] Step S209: Obtain the evaluation result of the aging state of the organ corresponding to the person according to the predicted age value and the corresponding actual age value.

[0032] Here, through the above organ aging evaluation model, the predicted age value of any person's organ in the target population can be obtained. The predicted age value and the corresponding actual age value are further compared to obtain the evaluation result of the aging state of the organ corresponding to the person.

[0033] It can be seen that in the embodiment of the present application, by obtaining the proteomics data of the target population and the protein-specific data of the human organs; performing a matching operation on the proteomics data and the protein-specific data to obtain the protein matching data corresponding to the organs of the target population; performing a correlation analysis on the protein matching data and the real age data of the corresponding target population to obtain the age-related protein data of the organs of the target population; performing model training according to the age-related protein data and the real age data to obtain an organ aging evaluation model and the predicted age value of any person's organ in the target population; obtaining the evaluation result of the aging state of the organ corresponding to the person according to the predicted age value and the corresponding actual age value. Through the above operations, by matching, analyzing, and performing model training on the proteomics data of the target population and the protein-specific data of the human organs, the evaluation result of the aging state of the corresponding human organs is obtained, which helps to characterize the aging state of the human organs, dynamically monitor its change process, and helps to comprehensively evaluate the occurrence and development trend of organ aging-related diseases, providing further guarantee for carrying out organ aging and anti-aging intervention measures and other aspects.

[0034] The following will describe the different steps in the above method process in detail. First, in combination with the embodiments, the above step 201, that is, "obtain the proteomics data of the target population and the protein-specific data of the human organs", will be described in detail.

[0035] In a feasible manner, obtain the physical examination information data of multiple people; the physical examination information data includes historical medical history information data, measurement information data, and detection information data; according to the physical examination information data and the preset screening rules, obtain the plasma sample data of the target population; perform detection and analysis on the plasma sample data to obtain the protein detection data of the target population; perform preprocessing on the protein detection data to obtain the proteomics data of the target population.

[0036] Among them, multiple people include, but are not limited to, people in the age range of 19 to 97 years old; the physical examination information data includes historical medical history information data, measurement information data, and detection information data; the preset screening rules refer to the rules for eliminating abnormal items.

[0037] Here, obtain the physical examination information data of multiple people, that is, the historical medical history information, measurement information, and test information of each person among multiple people. According to the physical examination information data and preset screening rules, obtain the plasma sample data of the target population. Specifically, according to the historical medical history information reported by each person among multiple people, that is, the disease history and surgical history information, eliminate the personnel with any of the following reported disease histories or surgical histories. Among them, the disease history information includes but is not limited to: hypertension, diabetes, hepatitis, coronary heart disease, chronic bronchitis, bronchial asthma, pulmonary tuberculosis, hyperthyroidism, cerebral hemorrhage, cerebral infarction, fibroma, and chronic nephritis, etc.; the surgical history information includes but is not limited to: cholecystectomy, total gastrectomy, lobectomy, hysterectomy, and artificial pacemaker, etc.

[0038] The measurement information includes but is not limited to BMI information, blood pressure information, etc., and eliminate the personnel with abnormal BMI information and abnormal blood pressure information. Among them, abnormal BMI information refers to underweight or overweight personnel; abnormal blood pressure information refers to personnel whose SBP or DBP reaches the diagnostic criteria for hypertension or hypotension.

[0039] The test information includes but is not limited to blood glucose information, blood lipid information, blood routine information, liver function information, and kidney function information, etc. Eliminate the personnel with abnormal blood glucose information, abnormal blood lipid information, abnormal blood routine information, abnormal liver function information, and abnormal kidney function information. Abnormal blood glucose information refers to personnel whose fasting blood glucose value reaches the diagnostic criteria for diabetes or hypoglycemia; abnormal blood lipid information refers to personnel whose fasting four blood lipid items (total cholesterol, low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, triglyceride) reach the diagnostic criteria for dyslipidemia, or the values of total cholesterol and triglyceride are lower than the normal reference value range of the test; abnormal blood routine information refers to personnel whose value of any one of white blood cell count, neutrophil count, neutrophil percentage, lymphocyte count, lymphocyte percentage, monocyte count, monocyte percentage, basophil count, basophil percentage, eosinophil count, eosinophil percentage, red blood cell count, hemoglobin, mean corpuscular volume, mean corpuscular hemoglobin content, hematocrit, mean corpuscular hemoglobin concentration, red blood cell distribution width, platelet count, mean platelet volume, plateletcrit, platelet distribution width, and large platelet ratio exceeds the normal reference value range of the test; abnormal liver function information refers to personnel whose value of any one of total protein, albumin, globulin, albumin / globulin ratio, alanine aminotransferase, aspartate aminotransferase, and total bilirubin exceeds the normal reference value range of the test; abnormal kidney function information refers to personnel whose value of any one of creatinine, uric acid, and urea exceeds the normal reference value range of the test.

[0040] Through the above screening, the selected personnel are matched with the plasma information in the biobank to screen out phenotypically healthy people with plasma samples and good plasma quality; further, according to the principle of 2 people per age, phenotypically healthy elderly people, that is, the target population, can be randomly selected. It should be noted that since the number of people over 70 years old does not meet the requirements, considering that it is extremely rare for people in this age group to have all normal test indicators, the screening criteria for the measurement information data and test information data of people in this age group are slightly relaxed, that is, the ratio of the value exceeding the upper limit of the normal reference value range to the upper limit of the normal reference value range is less than 15%, and then the plasma sample data of the target number of people can be obtained. In this application, the target population can be 490 people, whose age range is between 19 years old and 97 years old, including a part of men and a part of women.

[0041] The plasma sample data is detected and analyzed to obtain the protein detection data of the target population. Specifically, the plasma sample data can be preliminarily processed, that is, the plasma sample data is collected using an ethylenediaminetetraacetic acid anticoagulant tube. After inverting and mixing evenly, it is centrifuged at 1600 revolutions for 10 minutes using a low-temperature centrifuge. After centrifugation, the plasma is collected into a microcentrifuge tube and stored at -80 °C for later use. After thawing on ice, 30 μl is taken from each sample, 270 μl of phosphate buffer is added, and then phenylmethylsulfonyl fluoride with a final concentration of 0.1 M is added; after shaking and mixing evenly, it is the protein solution, and then the protein concentration is measured. According to the protein concentration, an equal amount of protein solution is taken into a centrifuge tube, and then incubated at 37 °C with 10 mM dithiothreitol (DTT) for 1 h, alkylated with 50 mM iodoacetamide in the dark at room temperature for 1 h, DDT is added again to a final concentration of 5 mM, left standing at room temperature for 10 min, and the urea and ammonium bicarbonate are repeatedly filtered through an ultrafiltration tube, and then trypsin is added to digest overnight at 37 °C. After digestion, the peptides of each sample are desalted on a C18 column material, concentrated by vacuum centrifugation and redissolved in 0.1% formic acid for peptide quantification.

[0042] Here, a high-resolution mass spectrometry system can be coupled with a high-performance liquid chromatography system for detection to obtain the mass spectrometry data of the whole protein corresponding to the peptide sample. Specifically, each of the above-obtained samples is separated using a chromatography system. Mobile phase A is an aqueous solution containing 0.1% formic acid and 2% acetonitrile, and mobile phase B is an aqueous solution containing 0.1% formic acid and 80% acetonitrile. The C18 analytical column (monolithic column) is equilibrated with 100% of solution A. The sample is loaded by an autosampler and separated by the analytical column. The separated sample is subjected to mass spectrometry analysis using a timsTOF ProII mass spectrometer with data-independent acquisition (DIA) mass spectrometry detection means. The mass spectrometry detection parameters can be set as follows: ion mode, positive ion; capillary, 1800 V; drying temperature: 180 °C; drying gas: 3.0 L / min; mass range: 349.2 - 1229.2 Da; ion mobility: 0.71 - 1.26 V·S / cm²; collision energy: 20 eV - 59 eV. The data obtained from mass spectrometry is further identified and analyzed using DIA-NN software and the human Uniprot database; among them, the DIA-NN software generally uses the default settings, specifically: trypsin as the specific protease (allowing 1 missed cleavage site); fixed modification, carbamidomethylation on cysteine residues; variable modifications, oxidation on methionine residues and acetylation at the N-terminus of proteins; peptide mass tolerance is 15 ppm; fragment mass tolerance is 15 ppm; precursor q value < 0.01; protein grouping q value < 0.01. The generated files of DIA-NN are analyzed and filtered using Perseus software. Since the protein quantification list defines the groups, proteins missing in more than 20% of the samples can be removed, and thus 359 proteins for each individual in the target population can be identified.

[0043] Preprocess the protein detection data to obtain the proteomics data of the target population. Specifically, for the protein information of each person in the identified target population, further preprocessing operations can be performed, including but not limited to log2 transformation, median normalization, and left-shift Gaussian distribution to fill in missing values, etc., so as to obtain the proteomics data of the target population. It should be noted that in addition to the mass spectrometry identification platform, this application can also use the Olink proteomics detection platform, the SomaScan proteomics detection platform, etc., all of which can obtain the proteomics data of the target population. It should also be emphasized that for the proteomics data of the target population screened by the mass spectrometry detection platform, detection and fixation can be further performed using detection methods such as enzyme-linked immunosorbent assay, targeted mass spectrometry method, fluorescence-based protein array, etc. Since the aforementioned detection methods are all existing technologies, no detailed description will be given here; for the proteomics data obtained by the Olink proteomics detection platform and the SomaScan proteomics detection platform, further detection and fixation are no longer required.

[0044] Through the above operations, by obtaining the physical examination information data of multiple people; according to the physical examination information data and the preset screening rules, the plasma sample data of the target population is obtained, so that the obtained plasma sample data has a wider coverage range and stronger pertinence; at the same time, the mass spectrometry detection platform is used to identify plasma proteins, which has the advantages of lower cost, more comprehensive data collection, higher repeatability and stability, etc.

[0045] In another feasible way, according to the Human Protein Atlas, obtain the standardized gene expression level data of the same gene in different organ tissues of an individual; sort the standardized gene expression level data of the gene in different organ tissues from largest to smallest to obtain the first organ standardized gene expression level data and the second organ standardized gene expression level data; according to the first organ standardized gene expression level data and the second organ standardized gene expression level data, obtain the ratio data of the corresponding gene; according to the ratio data and the preset ratio threshold, obtain the enriched gene data of the corresponding human organ; according to the enriched gene data and the corresponding protein identification data, obtain the protein-specific data of the corresponding human organ.

[0046] Here, based on the RNA-tissue expression data in the Human Protein Atlas public database, proteins specifically expressed in corresponding human organs can be screened according to the data differences of normalized gene expression levels in different organ tissues, that is, the data of normalized expression of transcripts per million (nTPM). Specifically, since the same gene in the same organ or system may be derived from different tissues of that organ or system, the nTPM of the tissue with the largest expression level can be retained to represent the expression level of the gene in that organ or system. For example, the nTPM of gene TSPAN6 in the stomach tissue is 14.4, in the small intestine tissue is 18, in the colon tissue is 38.9, and in the rectal tissue is 32. Then, it can be considered that the nTPM of gene TSPAN6 in the gastrointestinal system is 38.9. Therefore, according to the Human Protein Atlas, the nTPM data of the same gene in different organ tissues of the human body can be obtained.

[0047] Sorting the nTPM data of genes in different organ tissues from largest to smallest, the first organ nTPM data and the second organ nTPM data can be obtained, that is, for the same gene, the top two organs with the largest nTPM expression levels are retained. For example, for gene MASP2, after sorting from largest to smallest, it is found that the nTPM in the liver organ and the brain organ are the top two with the largest values among all organs or systems. Then, for gene MASP2, we only retain the nTPM in the liver organ and the brain organ, and the nTPM of gene MASP2 in other organs or systems can be excluded.

[0048] Based on the first organ nTPM data and the second organ nTPM data, the ratio data of the corresponding gene can be obtained, and then the ratio, that is, FC, of the expression levels of the top two organs of the corresponding gene can be calculated. For example, for gene MASP2, the nTPM in the liver organ is 464.7, which is the organ with the highest expression level, and the organ with the second highest expression level is the brain organ, with an nTPM of 8.2. Then, the FC of gene MASP2 in these two organs is 56.67.

[0049] According to the ratio quantity data and the preset ratio quantity threshold, the enriched gene data of the corresponding organ is obtained. Here, since each gene has the FC values of the top two organs with the largest expression levels, therefore, the FC of each gene is compared with the preset ratio quantity threshold of 4. When FC > 4 and the maximum value of the expression level of this gene is greater than or equal to 1, that is, nTPM >= 1, it indicates that this gene is enriched in this organ. Furthermore, the number of genes enriched in the corresponding organ can be obtained, that is, the enriched gene data of the corresponding organ. Here, since the liver is the largest gland in the human body and the proteins secreted by it account for a large proportion in the plasma, the liver organ can be taken as an example. Through the above method, 266 genes enriched in the liver organ can be obtained.

[0050] According to the enriched gene data and the corresponding protein identification data, the protein specific data of the corresponding human organ can be obtained. Here, since each protein has a corresponding Uniprot number, therefore, the enriched gene data and the corresponding protein identification data can be matched correspondingly, and then the protein specific data of the corresponding organ can be obtained. Still taking the liver organ as an example, 258 proteins specifically expressed in the liver organ can be obtained.

[0051] Through the above operations, according to the Human Protein Atlas, the nTPM data of the same gene in different human organs and tissues are obtained. Furthermore, based on the nTPM data of each gene, the enriched gene data of the corresponding organ are obtained. The enriched gene data and the corresponding protein identification data are matched to obtain the protein specific data of the corresponding human organ, which can improve the convenience and accuracy of the selection of protein specific data and provide further guarantee for the subsequent organ aging assessment.

[0052] Next, the above step S205, that is, "Perform a correlation analysis on the protein matching data and the real age data of the corresponding target population to obtain the age-related protein data of the corresponding target population's organs", will be described in detail in combination with the embodiments.

[0053] Input the protein matching data and the real age data into the correlation analysis model to obtain the correlation coefficient data of the corresponding target population's organs; according to the correlation coefficient data and the preset distribution table, obtain the significant index data of the corresponding target population's organs; sort the significant index values in the significant index data to obtain the probability significant data of the corresponding target population's organs; according to the probability significant data, obtain the age-related protein data of the corresponding target population's organs.

[0054] Here, still taking the liver organ as an example, 359 proteins of each person in the target population obtained by mass spectrometry identification can be matched with 258 proteins specifically expressed in the human liver organ, and 98 liver organ protein matching information of each person in the target population can be obtained, that is, the protein matching data.

[0055] The protein matching data can be subjected to correlation analysis with the true age data of the corresponding target population to screen for age-related proteins. In addition, the protein matching data of the target population's organs can be divided into a male protein data set and a female protein data set according to gender, and then correlation analysis can be performed with the corresponding true age data respectively to screen for age-related protein data of male organs and age-related protein data of female organs.

[0056] Specifically, the protein matching data and the true age data of the corresponding target population are input into a correlation analysis model to obtain the correlation coefficient data of the corresponding target population's organs. Among them, the specific expression of the correlation analysis model can be expressed as follows: ; In the formula, N represents the total number of people in the target population; represents the expression value of a certain protein corresponding to the th person in the target population; represents the average expression value of a certain protein corresponding to all people in the target population; represents the true age value of the th person in the target population; represents the average value of the true ages of all people in the target population; r represents the strength of the linear association between the true age and a certain protein of the corresponding target population's organs. Therefore, through the above expression, the correlation coefficient data of the corresponding target population's organs can be obtained.

[0057] According to the correlation coefficient data and the preset distribution table, the significant index data of the corresponding target population's organs can be obtained. Here, in the correlation analysis model, multiple hypothesis tests are involved. Generally, the false discovery rate (FDR) also needs to be controlled. Therefore, the Benjamini-Hochberg (BH) correction method can be used for the correction operation. As is well known, after obtaining the correlation coefficient data, that is, r, the correlation coefficient can be converted into a t value, and then the significant index data can be obtained by looking up the t preset distribution table based on the degrees of freedom, and then the correction operation can be performed on it.

[0058] Sort the significant index values in the significant index data to obtain the probability significant data of the corresponding target population's organs. Specifically, sort the significant index values, that is, the P values, in the significant index data from small to large, denoted as , , …, , where m represents the total number of types of proteins in the protein matching data; based on the sorting result, determine the maximum rank k, and the specific expression can be expressed as follows: ; Among them, FDR can generally be set to 0.05, k represents the k-th protein. Find the largest P-value of the k-th that satisfies the above inequality, that is, consider these k P-values to be significant, and then obtain the probability-significant data.

[0059] Based on the probability-significant data, obtain the age-related protein data of the corresponding target population's organs. According to the probability-significant data, remove those with non-significant P-values. Then the first k proteins are the selected age-related proteins, which are the age-related protein data of the corresponding target population's organs. Still taking the liver organ as an example, 62 age-related proteins can be obtained. Specifically, the number of types of age-related proteins in each person's organ in the target population is as shown in the following table:

[0060] Through the above operations, through correlation analysis, age-related protein data can be obtained, providing further guarantee for the next prediction of the aging model, so as to obtain a more accurate analysis result of the aging state.

[0061] Next, in combination with the embodiments, the step "perform model training based on the age-related protein data and the real age data to obtain an organ aging assessment model and the predicted age value of any person's organ in the target population" in step S207 is described in detail.

[0062] Divide the age-related protein data into data sets to obtain the age-related protein division data of the corresponding target population's organs; perform normalization operations on the age-related protein division data to obtain model training data; construct an initial organ training model, and determine the optimization objective function and loss function; perform model training on the initial organ training model according to the model training data, the optimization objective function, and the loss function to obtain an organ aging assessment model; input the age-related protein information of any person in the age-related protein data into the organ aging assessment model to obtain the predicted age value of the corresponding person's organ.

[0063] Specifically, there are two construction strategies for constructing the organ aging assessment model, that is, constructing the organ aging assessment model without considering gender and constructing the organ aging assessment model considering gender. The difference between these two model constructions lies only in the input data sets. When constructing the organ aging assessment model without considering gender, there is no need to distinguish the gender of the people in the target population, while when constructing the organ aging assessment model considering gender, it is necessary to distinguish the gender of the people in the target population. Here, taking the construction of the organ aging assessment model without considering gender as an example for further explanation.

[0064] Partition the dataset of age-related protein data to obtain the age-related protein partition data corresponding to the organs of the target population. Here, the age-related protein data can be partitioned into a training set and a test set in a ratio of 7:3. Taking the target population of 490 people as an example, the number of samples in the training set is 343 people, and the test set is 147 people, thus obtaining the age-related protein partition data corresponding to the organs of the target population.

[0065] Perform a normalization operation on the age-related protein partition data to obtain the model training data. Here, the MinMaxScaler preprocessing method can be used to normalize the expression values of age-related proteins in each dataset to between 0 and 1. The specific expression is as follows: ; Among them, represents the expression value of each age-related protein in the partitioned dataset; represents the minimum expression value of age-related proteins in the partitioned dataset; represents the maximum expression value of age-related proteins in the partitioned dataset; represents the normalized expression value of each age-related protein in the partitioned dataset. After normalization, the expression values of age-related proteins in the partitioned dataset will all fall within the interval [0, 1], thus obtaining the model training data.

[0066] Construct an initial organ training model, and determine the optimization objective function and loss function. Here, an initial SVR organ training model including a single kernel function and a mixed kernel function can be constructed. The single kernel function includes, but is not limited to, linear kernel function, polynomial kernel function, radial basis kernel function, and Sigmoid kernel function, etc.; the mixed kernel function is a function formed by combining any two of the aforementioned single kernel functions according to a certain weight. Among them, the specific expression of the mixed kernel function can be expressed as follows: ; Among them, represents the expression value of age-related proteins in the input training data, is the mixed kernel function, which is a function that can map the input space to a high-dimensional feature space; is the first kernel function of the mixed kernel function, which can be any one of the above four single kernel functions; is the second kernel function of the mixed kernel function, and this kernel function is different from the first kernel function; represents the weight of the first kernel function, represents the weight of the second kernel function; represents the regularization parameter, I is the identity matrix, and I are both preset fixed values, and and They are all hyperparameters that need to be optimized through model training. It should be noted that the reason for adding a regularization term to the kernel function is to improve the numerical stability of the kernel matrix, help limit the complexity of the training model, and prevent overfitting, that is, by reducing the dependence of the training model on the training data, so as to achieve the effect of improving the numerical stability of the kernel matrix.

[0067] For the selected kernel function, whether it is a single kernel function or a mixed kernel function, the corresponding optimization objective function and loss function can be determined. By training the model through the loss function, an organ aging assessment model with the best model effect can be obtained.

[0068] Specifically, the expression of the optimization objective function can be represented as follows: ; This optimization objective function is restricted by the following conditions: ; Among them, w represents the weight vector; b represents the bias term; represents the regularization term, which can control the complexity of the training model; C represents the penalty parameter, which can balance the relationship between the complexity of the training model and the training error; and represent slack variables, corresponding to the cases where the actual age value is higher and lower than the predicted age value respectively. n represents the number of training samples, which can be 343 people in this application document; represents the true age value of the i th person in the target population, represents the predicted age value of the i th person in the target population, represents the function that maps the input space to the high-dimensional feature space; represents the loss function. Here, the loss function can adopt any one of the minimized epsilon-insensitive loss function, Huber loss function, or Log-Cosh loss function. Epsilon represents the insensitive loss, which can be used to define the acceptable error range of the training model, and no penalty is imposed within this range. It should be noted that w, b , , and values are all given initial values at the beginning of model training and will change automatically during the later model training process.

[0069] Here, the expression of the minimized epsilon-insensitive loss function can be represented as follows: ; Among them, ϵ represents the insensitive loss; represents the true age value of the i th individual in the target population, represents the i th individual in the target population; minimizing the insensitive loss function can define the error range ϵ that the training model can accept, and no penalty is imposed within this range.

[0070] The expression of the Huber loss function can be represented as follows: ; Among them, ϵ represents the insensitive loss; represents the i th individual in the target population, represents the i th individual in the target population. The Huber loss function can use the squared loss for small errors and the absolute loss for large errors, thus achieving a balance between robustness to outliers and sensitivity to small errors.

[0071] The expression of the Log-Cosh loss function can be represented as follows: ; Among them, represents the i th individual in the target population, represents the i th individual in the target population. The penalty of this loss function increases with the increase of the error. Using this loss function can avoid introducing insensitive loss ϵ, Therefore, it is not necessary to train it during the process of model training and optimization.

[0072] According to the model training data, the optimization objective function, and the loss function, the initial organ training model is trained to obtain an organ aging assessment model.

[0073] In an implementable manner, the model parameters of the initial organ training model are trained and screened through a grid search algorithm and a cross-validation algorithm to obtain the optimal training parameters; according to the model training data, the optimization objective function, the loss function, and the optimal training parameters, the initial organ training model is repeatedly trained to obtain an organ aging assessment model.

[0074] Among them, the trainable parameters that can be optimized include but are not limited to the weight value of the mixed kernel function, the penalty parameter C, and the insensitive loss ϵ, etc.

[0075] Here, the grid search algorithm is a simple and systematic method for optimizing hyperparameters, i.e., training parameters, which is widely used in the training and tuning of machine learning models. Its core idea is to define a set of candidate values for hyperparameters, namely the "grid", train and validate the model for each combination of hyperparameters, and finally select the combination of hyperparameters with the best performance on the validation set as the optimal combination of hyperparameters. Specifically, the specific process of the grid search algorithm is as follows: First, a parameter grid can be defined, defining the possible value ranges of the above training parameters. Among them, the value range of the kernel function kernel can be one or two of the linear kernel function, polynomial kernel function, radial basis kernel function, and sigmoid kernel function; the value range of the regularization parameter C can be from 1 to 51, the value range of the insensitive loss ϵ can be from 0.1 to 1.0, and the value range of the mixed kernel weight can be from 0 to 1. Then, a constructed parameter grid can be obtained; traverse each combination in the parameter grid, perform model training and validation, and finally select the combination of hyperparameters with the best performance on the validation set as the optimal combination of hyperparameters.

[0076] The machine is divided into K subsets of equal or nearly equal size (referred to as "folds"). Each time, one of the subsets is taken as the validation set, and the remaining K - 1 subsets are taken as the training set. The model is trained and the performance metrics on the validation set are recorded. After each selection, there will be corresponding validation results. Repeat K times so that each subset serves as a validation set once. Then, K validation results can be obtained, that is, one combination corresponds to K validation results. The average value of the K validation results is calculated to obtain the average validation result value corresponding to each combination. Select the set of hyperparameters, i.e., training parameters, with the best average validation result value, and retrain the model to obtain the optimal training model, i.e., the organ aging assessment model.

[0077] It should be noted that the performance metrics of the training model include but are not limited to the Pearson correlation coefficient, mean absolute error, and coefficient of determination, etc. Specifically, the Pearson correlation coefficient can be represented by r, and its expression can be as follows: ; Among them, represents the true age value corresponding to the i th person in the target population; represents the average value of; represents the predicted organ age value corresponding to the i th person in the target population; represents the average value of; N represents the number of people in the target population; r represents the strength of the linear association between the predicted organ age and the actual age. The larger this value, the better the performance of the training model.

[0078] The mean absolute error can be expressed as MAE, and its specific expression can be shown as follows: ; where, represents the predicted organ age value of the i -th person in the target population; represents the actual age value of the i -th person in the target population; N represents the number of the target population; MAE represents the average error between the actual age value and the predicted organ age. The smaller this value is, the better the performance of the trained model.

[0079] The coefficient of determination can be expressed as and its specific expression can be shown as follows: ; where, represents the predicted organ age value of the i -th person in the target population; represents the actual age of the i -th person in the target population; represents the average value; N represents the number of the target population; represents the percentage explained by the regression relationship between the predicted organ age value and the actual age value. The larger this value is, the better the performance of the trained model.

[0080] Here, a set of hyperparameters with the optimal validation average result value is selected, that is, the hyperparameter combination corresponding to the largest average value of r, the smallest average value of MAE, and the largest average value is the optimal set of hyperparameter combinations. For each loss function, there can be a corresponding optimal set of hyperparameters. That is, using any of the above loss functions can obtain the corresponding optimal trained model. However, if a better one is desired, the optimal trained models corresponding to the above three loss functions can be further screened, that is, the performance indicators corresponding to the optimal hyperparameter combinations of each loss function are compared again, and still the hyperparameter combination corresponding to the largest average value of r, the smallest average value of MAE, and the largest average value is selected. The trained model under the loss function corresponding to this combination is the better model, and this model is used as the organ aging assessment model. It should be noted that different organs correspond to different organ aging assessment models. For example, if the protein-specific data of the liver organ is obtained above, the trained model corresponds to the liver organ aging assessment model; if the protein-specific data of the kidney organ is obtained, the trained model corresponds to the kidney organ aging assessment model.

[0081] Input the age-related protein information of any person in the age-related protein data into the organ aging assessment model to obtain the predicted age value of the corresponding person's organ. Here, after obtaining the organ aging assessment model, the age-related protein information of any person in the age-related protein data can be input into the organ aging assessment model to obtain the predicted age value of the corresponding person's organ.

[0082] It should also be emphasized that when using the organ aging assessment model to predict the organ-specific biological ages of different populations, the datasets obtained may be different. If the expression values of some proteins are in a missing state, the model training set can be used to fill in the missing protein expression values. The filling methods include, but are not limited to, mean filling method, median filling method, fixed value 0 filling method, K-nearest neighbor filling method, and regression filling method, etc. After that, the organ aging assessment model is used to predict the organ-specific biological age.

[0083] For the above operations, the support vector regression model is used as the initial model, and the grid search algorithm is combined with the cross-validation algorithm to specify the range and step size of the hyperparameters for training the model, traverse each parameter combination, and determine the optimal model based on the performance evaluation index, which is the organ aging assessment model, achieving the effect of fully exploring the complex relationship between biomarkers and organ aging and improving the accuracy and adaptability of organ aging assessment.

[0084] Finally, in combination with the embodiments, the "obtaining the aging state assessment result of the corresponding person's organ according to the predicted age value and the corresponding actual age value" in the above step S209 is described in detail.

[0085] Input the predicted age value and the corresponding actual age value into the organ aging index model to obtain the organ aging index value of the corresponding person; determine whether the organ aging index value is within the range of the preset standard state value; if the organ aging index value is within the range of the preset standard state value, it is determined that the person's organ is in a normal aging state; if the organ aging index value is not within the range of the preset standard state value, it is determined that the person's organ is in an abnormal aging state.

[0086] Among them, the range of the preset standard state value can be the interquartile range of the organ aging index values of the target population.

[0087] Here, after obtaining the predicted age value, the predicted age value and the corresponding actual age value can be input into the organ aging index model to obtain the organ aging index value of the corresponding person. The specific expression of the organ aging index model can be shown as follows: ; Among them, represents the predicted age value, that is, the biological age of any person in the target population output by the organ aging assessment model, represents the corresponding actual age value.

[0088] Determine whether the organ aging index value is within the range of the preset standard state value; if the organ aging index value is within the range of the preset standard state value, it is determined that the organs of this person are in a normal aging state; if the organ aging index value is not within the range of the preset standard state value, it is determined that the organs of this person are in an abnormal aging state. Specifically, since the range of the preset standard state value can include an upper limit value and a lower limit value, when the organ aging index value is lower than the lower limit value of the range of the preset standard state value, it is determined that the organs of this person are in a decelerated aging state; when the organ aging index value is higher than the upper limit value of the range of the preset standard state value, it is determined that the organs of this person are in an accelerated aging state.

[0089] Through the above operations, by determining whether the organ aging index value is within the range of the preset standard state value, the aging state of the corresponding human organs is obtained, which is convenient for dynamically monitoring its change process, evaluating the occurrence and development trend of aging-related diseases, and providing a basis and support for carrying out anti-aging intervention measures and other aspects.

[0090] Here, in order to determine the feasibility of the method of this application, further verification operations were carried out. Specifically, taking the liver organ as an example, in the mass spectrometry public dataset on the ProteomeXchange Consortium platform, a mass spectrometry detection dataset related to the liver was screened for further verification. The specific process is as follows: (1) Data description The protein data used for model verification: sourced from a publicly available literature library.

[0091] The proteomics detection platform is: using a high-performance liquid chromatography system coupled with a Q Exactive HF Orbitrap high-resolution mass spectrometry system and a nanoelectrospray ion source for detection. The sample type for detection is plasma protein data, and 520 proteins can be obtained. Sample size and data grouping: A total of 48 people. Obese, normal glucose tolerance, and no liver disease group (Group1): 10 people; Obese, normal glucose tolerance, with non-alcoholic fatty liver disease group and non-alcoholic fatty liver disease group with diabetes (Group2): 20 people; Type 2 diabetes and no liver disease group (Group3): 8 people; Liver cirrhosis group (Group4): 10 people.

[0092] (2) Construction of the liver aging assessment model Match the 520 proteomics data identified above, i.e., the 258 liver organ-specific expressed proteins which are protein-specific data, and then perform correlation analysis to obtain 57 age-related protein data. Based on the age-related protein data and the corresponding true age data, perform model training to obtain an organ aging assessment model and the predicted age values of each individual organ in the above population. According to the predicted age values and the corresponding actual age values, obtain the aging index values of each individual organ.

[0093] (3) Model verification results According to the actual age values, predicted age values and aging index values of each individual, compare the differences in liver aging indices between different disease groups. The analysis results are shown in the following table. The results show that the liver aging rate of patients with liver cirrhosis is significantly higher than that of patients with non-alcoholic fatty liver, obesity, and diabetes; the liver aging rates of patients with non-alcoholic fatty liver, obesity, and diabetes are comparable. It should also be noted that variance test and HSD test can be further performed to further verify the feasibility of the method of this application. In summary, although non-alcoholic fatty liver, obesity, and diabetes are all related to liver health problems, their impacts on liver aging are lower than that of liver cirrhosis. The model verification results are in line with the expected hypothesis.

[0094]

[0095] Through further verification, it shows that the organ aging assessment model of this application has high effectiveness and stability. It not only helps to characterize the aging state of the human liver organ, but also can dynamically monitor its change process, helps to evaluate the occurrence and development trends of liver aging-related diseases, and provides a scientific basis and technical support for scientific research in aspects such as liver aging and anti-aging intervention measures.

[0096] It should be understood that although Figure 1 the steps in the flowchart of Figure 1 are shown in sequence according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless there is a clear description in this application, the execution of these steps has no strict order limit, and these steps can be executed in other orders. Moreover,

[0097] Figure 3 is a schematic structural diagram of an organ aging assessment device provided by an embodiment of this application. This device can be set in Figure 1A server in the application environment shown is used to execute the method flow as shown in Figure 1 As shown in. As shown in Figure 3 The device may include: an acquisition unit 301, a matching unit 303, an analysis unit 305, a prediction unit 307, and an evaluation unit 309. The main functions of each component module are as follows: The acquisition unit 301 is used to acquire proteomic data of the target population and protein-specific data of human organs; The matching unit 303 performs a matching operation on the proteomic data and the protein-specific data to obtain protein matching data of the corresponding target population organs; The analysis unit 305 performs a correlation analysis on the protein matching data and the true age data of the corresponding target population to obtain age-related protein data of the corresponding target population organs; The prediction unit 307 performs model training based on the age-related protein data and the true age data to obtain an organ aging assessment model and a predicted age value of any one organ in the target population; The evaluation unit 309 obtains an aging state assessment result of the corresponding person's organ according to the predicted age value and the corresponding actual age value.

[0098] In one embodiment, the acquisition unit 301 is further used to: Acquire physical examination information data of multiple people; the physical examination information data includes historical medical history information data, measurement information data, and detection information data; According to the physical examination information data and a preset screening rule, obtain plasma sample data of the target population; Perform detection and analysis on the plasma sample data to obtain protein detection data of the target population; Perform preprocessing on the protein detection data to obtain proteomic data of the target population.

[0099] In one embodiment, the acquisition unit 301 is further used to: According to the Human Protein Atlas, obtain standardized gene expression level data of the same gene in different organ tissues of the human body; Sort the standardized gene expression level data of the gene in different organ tissues in descending order to obtain first-organ standardized gene expression level data and second-organ standardized gene expression level data; According to the first-organ standardized gene expression level data and the second-organ standardized gene expression level data, obtain ratio data of the corresponding gene; According to the ratio data and a preset ratio threshold, obtain enriched gene data of the corresponding human organ; According to the enriched gene data and the corresponding protein identification data, obtain protein-specific data of the corresponding human organ.

[0100] In one embodiment, the analysis unit 305 is further configured to: Input the protein matching data and the true age data into a correlation analysis model to obtain correlation coefficient data corresponding to the organs of the target population; According to the correlation coefficient data and a preset distribution table, obtain significant index data corresponding to the organs of the target population; Sort the significant index values in the significant index data to obtain probability significant data corresponding to the organs of the target population; According to the probability significant data, obtain age-related protein data corresponding to the organs of the target population.

[0101] In one embodiment, the prediction unit 307 is further configured to: Partition the age-related protein data to obtain age-related protein partition data corresponding to the organs of the target population; Perform a normalization operation on the age-related protein partition data to obtain model training data; Construct an initial organ training model, and determine an optimization objective function and a loss function; Train the initial organ training model according to the model training data, the optimization objective function, and the loss function to obtain an organ aging assessment model; Input the age-related protein information of any person in the age-related protein data into the organ aging assessment model to obtain a predicted age value corresponding to the organs of this person.

[0102] In one embodiment, the prediction unit 307 is further configured to: Train and screen the model parameters of the initial organ training model through a grid search algorithm and a cross-validation algorithm to obtain optimal training parameters; According to the model training data, the optimization objective function, the loss function, and the optimal training parameters, repeatedly train the initial organ training model to obtain an organ aging assessment model.

[0103] In one embodiment, the evaluation unit 309 is further configured to: Input the predicted age value and the corresponding actual age value into an organ aging index model to obtain an organ aging index value corresponding to this person; Determine whether the organ aging index value is within a preset standard state value range; If the organ aging index value is within the preset standard state value range, determine that the organs of this person are in a normal aging state; if the organ aging index value is not within the preset standard state value range, determine that the organs of this person are in an abnormal aging state.

[0104] For the same or similar parts among the above embodiments, reference can be made to each other. Each embodiment focuses on the differences from other embodiments. In particular, for the apparatus embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and for the relevant parts, reference can be made to the description in the method embodiments.

[0105] It should be noted that the embodiments of this application may involve the use of user data. In actual applications, user-specific personal data can be used in the solutions described in this article within the scope permitted by applicable laws and regulations in compliance with the requirements of applicable laws and regulations in the country where it is located (such as when the user gives explicit consent, gives actual notice to the user, and the user gives explicit authorization, etc.).

[0106] According to the embodiments of this application, this application also provides a computer device and a computer-readable storage medium.

[0107] As Figure 3 shown, it is a block diagram of a computer device according to an embodiment of this application. The computer device is intended to represent various forms of digital computers or mobile devices. Among them, digital computers can include desktop computers, portable computers, workstations, personal digital assistants, servers, mainframe computers, and other suitable computers. Mobile devices can include tablet computers, smart phones, wearable devices, etc.

[0108] As Figure 4 shown, the device 400 includes a computing unit 401, a ROM 402, a RAM 403, a bus 404, and an input / output (I / O) interface 405. The computing unit 401, the ROM 402, and the RAM 403 are connected to each other through the bus 404. The input / output (I / O) interface 405 is also connected to the bus 404.

[0109] The computing unit 401 can execute various processes in the method embodiments of this application according to the computer instructions stored in the read-only memory (ROM) 402 or the computer instructions loaded from the storage unit 408 into the random access memory (RAM) 403. The computing unit 401 can be various general and / or special processing components with processing and computing capabilities. The computing unit 401 can include, but is not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. In some embodiments, the method provided by the embodiments of this application can be implemented as a computer software program, which is tangibly included in a computer-readable storage medium, such as the storage unit 408.

[0110] The RAM 403 can also store various programs and data required for the operation of the device 400. Part or all of the computer programs can be loaded and / or installed onto the device 400 via the ROM 402 and / or the communication unit 409.

[0111] The input unit 406, output unit 407, storage unit 408, and communication unit 409 in the device 400 can be connected to the I / O interface 405. Among them, the input unit 406 can be, for example, a keyboard, a mouse, a touch screen, a microphone, etc.; the output unit 407 can be, for example, a display, a speaker, an indicator light, etc. The device 400 can exchange information, data, etc. with other devices through the communication unit 409.

[0112] It should be noted that the device may also include other components necessary for normal operation. It may also only include the components necessary to implement the solution of this application, and does not necessarily include all the components shown in the figure.

[0113] The various embodiments of the systems and technologies described herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on a chip (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof.

[0114] The computer instructions for implementing the methods of this application can be written in any combination of one or more programming languages. These computer instructions can be provided to the computing unit 401, such that when the computer instructions are executed by a computing unit 401 such as a processor, the various steps involved in the embodiments of the methods of this application are executed.

[0115] The computer-readable storage medium provided by this application can be a tangible medium that can contain or store computer instructions for executing the various steps involved in the embodiments of the methods of this application. The computer-readable storage medium can include, but is not limited to, storage media in the forms of electronic, magnetic, optical, electromagnetic, etc.

[0116] The above specific embodiments do not constitute a limitation on the protection scope of this application. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application shall be included within the protection scope of this application.

Claims

1. A method for evaluating organ aging, characterized in that, The method includes: Obtaining the proteomics data of the target population and the protein-specific data of the human organs; Performing a matching operation on the proteomics data and the protein-specific data to obtain protein matching data for the organs of the corresponding target population; Performing a correlation analysis on the protein matching data and the true age data of the corresponding target population to obtain age-related protein data for the organs of the corresponding target population; Performing model training based on the age-related protein data and the true age data to obtain an organ aging assessment model and the predicted age value of any one person's organ in the target population; Obtaining an aging status assessment result for the corresponding person's organ according to the predicted age value and the corresponding actual age value.

2. The organ aging assessment method according to claim 1, wherein The obtaining of the proteomics data of the target population includes: Obtaining the physical examination information data of multiple people; the physical examination information data includes historical medical history information data, measurement information data, and detection information data; Obtaining the plasma sample data of the target population according to the physical examination information data and a preset screening rule; Performing detection and analysis on the plasma sample data to obtain protein detection data of the target population; Performing preprocessing on the protein detection data to obtain the proteomics data of the target population.

3. The organ aging assessment method according to claim 1, wherein The obtaining of the protein-specific data of the human organs includes: Obtaining the normalized gene expression level data of the same gene in different organ tissues of the human body according to the Human Protein Atlas; Sorting the normalized gene expression level data of the gene in different organ tissues from largest to smallest to obtain first organ normalized gene expression level data and second organ normalized gene expression level data; Obtaining ratio data corresponding to the gene according to the first organ normalized gene expression level data and the second organ normalized gene expression level data; Obtaining enriched gene data corresponding to the human organs according to the ratio data and a preset ratio threshold; Obtaining the protein-specific data of the corresponding human organs according to the enriched gene data and the corresponding protein identification data.

4. The organ aging assessment method according to claim 1, wherein, The performing of the correlation analysis on the protein matching data and the true age data of the corresponding target population to obtain age-related protein data for the organs of the corresponding target population includes: Inputting the protein matching data and the true age data into a correlation analysis model to obtain correlation coefficient data for the organs of the corresponding target population; Obtaining significant index data for the organs of the corresponding target population according to the correlation coefficient data and a preset distribution table; Sorting the significant index values in the significant index data to obtain probability significant data for the organs of the corresponding target population; Obtaining age-related protein data for the organs of the corresponding target population according to the probability significant data.

5. The organ aging assessment method according to claim 4, wherein The performing of model training based on the age-related protein data and the true age data to obtain an organ aging assessment model and the predicted age value of any one person's organ in the target population includes: Performing dataset partitioning on the age-related protein data to obtain age-related protein partition data for the organs of the corresponding target population; Performing a normalization operation on the age-related protein partition data to obtain model training data; Construct an initial organ training model, and determine an optimization objective function and a loss function; Perform model training on the initial organ training model according to the model training data, the optimization objective function, and the loss function to obtain an organ aging assessment model; Input the age-related protein information of any person in the age-related protein data into the organ aging assessment model to obtain the predicted age value of the corresponding person's organ.

6. The organ aging assessment method according to claim 5, wherein The performing model training on the initial organ training model according to the model training data, the optimization objective function, and the loss function to obtain an organ aging assessment model includes: Train and screen the model parameters of the initial organ training model through a grid search algorithm and a cross-validation algorithm to obtain optimal training parameters; According to the model training data, the optimization objective function, the loss function, and the optimal training parameters, repeatedly train the initial organ training model to obtain an organ aging assessment model.

7. The organ aging assessment method according to any one of claims 1-6, characterized in that The obtaining the aging state assessment result of the corresponding person's organ according to the predicted age value and the corresponding actual age value includes: Input the predicted age value and the corresponding actual age value into an organ aging index model to obtain the organ aging index value corresponding to the person; Judge whether the organ aging index value is within the range of a preset standard state value; If the organ aging index value is within the range of the preset standard state value, determine that the person's organ is in a normal aging state; If the organ aging index value is not within the range of the preset standard state value, determine that the person's organ is in an abnormal aging state.

8. An organ aging assessment device, characterized in that, The device includes: An acquisition unit that acquires proteomics data of a target population and protein-specific data of a human organ; A matching unit that performs a matching operation on the proteomics data and the protein-specific data to obtain protein matching data corresponding to the organs of the target population; An analysis unit that performs a correlation analysis on the protein matching data and the real age data of the corresponding target population to obtain age-related protein data of the organs of the corresponding target population; A prediction unit that performs model training according to the age-related protein data and the real age data to obtain an organ aging assessment model and the predicted age value of an organ of any person in the target population; An evaluation unit that obtains the aging state assessment result of the corresponding person's organ according to the predicted age value and the corresponding actual age value.

9. A computer device, comprising: At least one processor; And A memory communicatively connected to the at least one processor; wherein, The memory stores computer instructions executable by the at least one processor, and the computer instructions are executed by the at least one processor so that the at least one processor can execute the method according to any one of claims 1-7.

10. A computer-readable storage medium having computer instructions stored thereon, characterized in that, The computer instructions are used to cause a computer to execute the method according to any one of claims 1 to 7.

Images