Application of biomarkers in preparation of kit for predicting aging level of human body, and method and electronic device for constructing prediction model of aging level of human body

Abstract

The application provides an application of a biological index in preparation of a kit for predicting a human aging level, and a construction method and an electronic device of a prediction model of the human aging level. The biological index comprises the following indexes: mean corpuscular volume, platelet hematocrit, red blood cell hematocrit, red blood cell distribution width, monocyte percentage, mean corpuscular hemoglobin, neutrophil count, monocyte count, red blood cell count, platelet count, lymphocyte count, total cholesterol content, triglyceride content, low-density lipoprotein cholesterol content, total protein content, uric acid content, urea content, hemoglobin content, alkaline phosphatase content, albumin content, alanine aminotransferase content, glucose content, aspartate aminotransferase content, apolipoprotein A content, or creatinine content. The application can solve the problem that the prediction method in the prior art is difficult to consider cost and effect, and is suitable for the field of aging prediction.

Description

[0001] Cross-references

[0002] This application claims priority to Chinese Patent Application No. 202610165120.X, filed on February 4, 2026, entitled "Method for constructing a predictive model of human aging level and a predictive model, device and method of human aging level", the entire contents of which are incorporated herein by reference. Technical Field

[0003] This invention relates to the field of aging prediction, and more specifically, to the application of a biomarker in the preparation of a reagent kit for predicting human aging levels, as well as a method and electronic device for constructing a predictive model of human aging levels. Background Technology

[0004] Two mainstream existing technologies for predicting biological age include gene methylation clocks and aging assessment based on blood test indicators. Gene methylation clocks are considered the gold standard for biological age prediction due to their accuracy. However, the high cost, cumbersome sample processing, and lengthy analysis cycle of this approach limit its practical application, especially in resource-scarce medical environments. In contrast, aging assessment based on blood test indicators is widely popular due to its convenience and cost-effectiveness. The ease of obtaining blood samples and the routine testing of most indicators in clinical laboratories make this assessment method indispensable in health monitoring. However, existing models often become complex due to the need for large amounts of data, and the detection methods for some indicators are complex, increasing costs and time, while also resulting in relatively low accuracy and predictive value in describing an individual's aging state.

[0005] While biomedical research has utilized multi-omics data such as DNA methylation, transcriptomics, and lipidomics to construct highly accurate biological clocks, the specialized nature and cost of these models limit their widespread application in routine health management. Furthermore, many existing models focus on aging analysis of local tissues or organs, neglecting a holistic view of individual aging, thus weakening their comprehensiveness and guiding value. Therefore, current technology lacks a method for predicting human aging levels that balances cost and effectiveness. Summary of the Invention

[0006] The main objective of this invention is to provide an application of bioindicators in the preparation of reagent kits for predicting human aging levels, as well as a method and electronic device for constructing a predictive model of human aging levels, in order to solve the problem that existing methods for predicting human aging levels are difficult to balance cost and effectiveness.

[0007] To achieve the above objectives, according to a first aspect of the present invention, a method for constructing a predictive model for human aging levels is provided. The method includes: a) acquiring the calendar age and indicators of a sample population to establish a dataset of the sample population; wherein the indicators include any one or more of the following: mean corpuscular volume, platelet-to-volume ratio, hematocrit, red blood cell distribution width, monocyte percentage, mean corpuscular hemoglobin, neutrophil count, monocyte count, red blood cell count, platelet count or lymphocyte count, total cholesterol content, triglyceride content, low-density lipoprotein cholesterol content, total protein content, uric acid content, urea content, hemoglobin content, alkaline phosphatase content, albumin content, alanine aminotransferase content, glucose content, aspartate aminotransferase content, apolipoprotein A content, or creatinine content, and the sample population's diastolic blood pressure, systolic blood pressure, height, weight, or body mass index; b) constructing the model using the above dataset through machine learning training.

[0008] Optionally, the above indicators include 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29 or 30 of the above a).

[0009] Optionally, the above indicators include any one of the following sets:

[0010] 1) Systolic blood pressure;

[0011] 2) Systolic blood pressure + urea content;

[0012] 3) Systolic blood pressure + urea content + alkaline phosphatase content;

[0013] 4) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content;

[0014] 5) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content;

[0015] 6) Height + weight + apolipoprotein A level + platelet count + diastolic blood pressure;

[0016] 7) Alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight;

[0017] 8) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height;

[0018] 9) Height + weight + apolipoprotein A content + platelet count + diastolic blood pressure + alanine aminotransferase content;

[0019] 10) Alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content;

[0020] 11) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight.

[0021] 12) Height + weight + apolipoprotein A content + platelet count + diastolic blood pressure + alanine aminotransferase content + albumin content;

[0022] 13) Alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + urea content;

[0023] 14) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content;

[0024] 15) Height + weight + apolipoprotein A content + platelet count + diastolic blood pressure + alanine aminotransferase content + alkaline phosphatase content + albumin content;

[0025] 16) Alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + urea content + platelet-to-volume ratio.

[0026] 17) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio.

[0027] 18) Alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + urea content + platelet count + diastolic blood pressure.

[0028] 19) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-weight ratio + diastolic blood pressure.

[0029] 20) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content.

[0030] 21) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume.

[0031] 22) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content.

[0032] 23) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase.

[0033] 24) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content.

[0034] 25) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content.

[0035] 26) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage.

[0036] 27) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content;

[0037] 28) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width;

[0038] 29) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index;

[0039] 30) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit;

[0040] 31) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit + mean corpuscular hemoglobin;

[0041] 32) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit + mean corpuscular hemoglobin + triglyceride content;

[0042] 33) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit + mean corpuscular hemoglobin + triglyceride content + neutrophil count;

[0043] 34) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit + mean corpuscular hemoglobin + triglyceride content + neutrophil count + uric acid content;

[0044] 35) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit + mean corpuscular hemoglobin + triglyceride content + neutrophil count + uric acid content + monocyte count;

[0045] 36) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit + mean corpuscular hemoglobin + triglyceride content + neutrophil count + uric acid content + monocyte count + hemoglobin content and concentration;

[0046] 37) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit + mean corpuscular hemoglobin + triglyceride content + neutrophil count + uric acid content + monocyte count + hemoglobin concentration + red blood cell count;

[0047] 38) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit + mean corpuscular hemoglobin + triglyceride content + neutrophil count + uric acid content + monocyte count + hemoglobin concentration + red blood cell count + platelet count;

[0048] 39) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit + mean corpuscular hemoglobin + triglyceride content + neutrophil count + uric acid content + monocyte count + hemoglobin concentration + red blood cell count + platelet count + lymphocyte count.

[0049] Furthermore, the above indicators include any one, two, three, four, five, six, seven, eight, nine, or ten of the following indicators: the content of urea, alkaline phosphatase, albumin, alanine aminotransferase, apolipoprotein A, platelet-to-volume ratio, as well as systolic blood pressure, height, weight, and diastolic blood pressure in the blood of the sample population.

[0050] Optionally, the above indicators may include any one of the following sets:

[0051] 1) Systolic blood pressure;

[0052] 2) Systolic blood pressure + urea content;

[0053] 3) Systolic blood pressure + urea content + alkaline phosphatase content;

[0054] 4) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content;

[0055] 5) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content;

[0056] 6) Height + weight + apolipoprotein A level + platelet count + diastolic blood pressure;

[0057] 7) Alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight;

[0058] 8) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height;

[0059] 9) Height + weight + apolipoprotein A content + platelet count + diastolic blood pressure + alanine aminotransferase content;

[0060] 10) Alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content;

[0061] 11) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight.

[0062] 12) Height + weight + apolipoprotein A content + platelet count + diastolic blood pressure + alanine aminotransferase content + albumin content;

[0063] 13) Alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + urea content;

[0064] 14) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content;

[0065] 15) Height + weight + apolipoprotein A content + platelet count + diastolic blood pressure + alanine aminotransferase content + alkaline phosphatase content + albumin content;

[0066] 16) Alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + urea content + platelet-to-volume ratio.

[0067] 17) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio.

[0068] 18) Alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + urea content + platelet count + diastolic blood pressure.

[0069] 19) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure.

[0070] Furthermore, the indicators must contain at least total cholesterol; optionally, they may also include: diastolic blood pressure, systolic blood pressure, height, weight, or body mass index of the sample population, and / or any one or more of the following indicators in the blood of the sample population: mean corpuscular volume, platelet count, hematocrit, red blood cell distribution width, monocyte percentage, mean corpuscular hemoglobin, neutrophil count, monocyte count, red blood cell count, platelet count, lymphocyte count, triglyceride level, low-density lipoprotein cholesterol level, total protein level, uric acid level, urea level, hemoglobin level, etc. The levels of albumin, alkaline phosphatase, albumin, alanine aminotransferase, glucose, aspartate aminotransferase, apolipoprotein A, or creatinine in the blood of the sample population; optionally, the indicators include total cholesterol, mean corpuscular volume, urea, alkaline phosphatase, albumin, alanine aminotransferase, apolipoprotein A, platelet-to-volume ratio, glucose, aspartate aminotransferase, or creatinine, as well as systolic blood pressure, height, weight, and diastolic blood pressure, or a combination thereof.

[0071] Optionally, the above indicators include any one of the following sets:

[0072] 24) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content.

[0073] 25) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content.

[0074] 26) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage.

[0075] 27) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content;

[0076] 28) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width;

[0077] 29) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index;

[0078] 30) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit;

[0079] 31) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit + mean corpuscular hemoglobin;

[0080] 32) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit + mean corpuscular hemoglobin + triglyceride content;

[0081] 33) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit + mean corpuscular hemoglobin + triglyceride content + neutrophil count;

[0082] 34) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit + mean corpuscular hemoglobin + triglyceride content + neutrophil count + uric acid content;

[0083] 35) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit + mean corpuscular hemoglobin + triglyceride content + neutrophil count + uric acid content + monocyte count;

[0084] 36) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit + mean corpuscular hemoglobin + triglyceride content + neutrophil count + uric acid content + monocyte count + hemoglobin content and concentration;

[0085] 37) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit + mean corpuscular hemoglobin + triglyceride content + neutrophil count + uric acid content + monocyte count + hemoglobin concentration + red blood cell count;

[0086] 38) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit + mean corpuscular hemoglobin + triglyceride content + neutrophil count + uric acid content + monocyte count + hemoglobin concentration + red blood cell count + platelet count;

[0087] 39) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit + mean corpuscular hemoglobin + triglyceride content + neutrophil count + uric acid content + monocyte count + hemoglobin concentration + red blood cell count + platelet count + lymphocyte count.

[0088] Furthermore, the aforementioned human aging level includes biological age; optionally, before using the aforementioned machine learning training, the aforementioned dataset is divided into a training set and a test set, the aforementioned training set is used to build the model, and the aforementioned test set is used to verify the model effect; optionally, the aforementioned machine learning training method includes any one or more of the following: XGBoost regression model, Ridge regression model, Lasso regression model, or Ealstic Net regression model; optionally, the objective function of the aforementioned XGBoost regression model is reg:squarederror.

[0089] Further, the method for dividing the above dataset into training and test sets includes: randomly splitting the above sample population into the above training set or the above test set; optionally, before the above machine learning training, the above dataset is subjected to any one or more of the following processing: i) missing value processing, including: calculating the average value of each of the above indicators in the above dataset, and using the above average value to fill the missing indicators in the above dataset; ii) standardization, including: converting all data of each of the above biomarkers in the above dataset into relative values ​​respectively; optionally, the above standardization includes: performing Z-score standardization processing on all data of each of the above biomarkers in the above dataset respectively.

[0090] Further, step b) above also includes: dividing the dataset into three overlapping age groups according to the calendar age, and independently training a corresponding age segmentation sub-model for each age group. The predicted values ​​of samples in the test set that fall into the overlapping age groups are weighted and summed by two adjacent age segmentation sub-models according to the linear weight of the distance from the boundary. Optionally, the three overlapping age groups include 0-A years, BC years, and D-100 years, where A≥B, C≥D; A includes 47-50, B includes 42-50, C includes 60-63, and D includes 57-60. Optionally, the three overlapping age groups include any one of the following groups: 0-48 years, 42-63 years and 57-100 years; 0-50 years, 50-60 years and 60-100 years; or 0-47 years, 43-62 years and 58-100 years. Further, step b) above includes: dividing the above dataset into a male training set and a female training set according to gender, and using the above machine learning training to train male sub-models and female sub-models respectively.

[0091] Furthermore, before constructing the model using the training set, the samples are weighted based on the distribution of the calendar ages in the training set. Optionally, the weight of the weighting is A ÷ (B × C), where A is the total number of samples in the training set, B is the total number of unique calendar ages in the training set, and C is the number of times the calendar age appears in the training set.

[0092] Further, step b) above includes: dividing the sample population into a training set and a test set; establishing a mapping relationship between the above-mentioned indicators and the above-mentioned calendar age in the training set or the test set, respectively obtaining a training dataset and a test dataset; constructing a model on the training dataset using the above-mentioned machine learning method, including determining the optimal hyperparameters in the training dataset using five-fold cross-validation combined with random search, to obtain a preliminary model; validating the preliminary model using the above-mentioned test dataset, calculating and evaluating any one or more of the following parameters of the preliminary model: mean absolute error, coefficient of determination, or Pearson correlation coefficient, to obtain the above-mentioned prediction model.

[0093] To achieve the above objectives, according to a second aspect of the present invention, a predictive model for human aging level is provided, wherein the model is obtained using the method for constructing the predictive model for human aging level.

[0094] To achieve the above objectives, according to a third aspect of the present invention, a device for predicting human aging level is provided. The device includes an input module and a prediction module. The input module is used to acquire an index of a sample to be tested and input the index into the prediction module. The index is the aforementioned indicator. The prediction module is used to calculate the human aging level based on the index. The prediction module contains a prediction model, which is a model obtained using a method for constructing a prediction model for human aging level.

[0095] To achieve the above objective, according to a fourth aspect of the present invention, a method for predicting the level of human aging is provided, the method comprising: inputting an index of a sample to be tested into a prediction model, and obtaining the level of human aging output by the prediction model; wherein the prediction model is a model obtained using the method for constructing the prediction model of the level of human aging.

[0096] To achieve the above objectives, according to a fifth aspect of the present invention, an application of a biomarker in predicting human aging levels is provided, wherein the biomarker includes the aforementioned indicator.

[0097] To achieve the above objectives, according to a sixth aspect of the present invention, a method for establishing a dataset is provided, the method comprising: obtaining the calendar age and / or indicators of a target sample from a dataset to obtain the dataset; wherein the indicators used include the aforementioned indicators.

[0098] Furthermore, the above method also includes: i) missing value handling, including: calculating the average value of each of the above indicators in the above dataset, and using the average value to fill the missing indicators in the above dataset; and / or ii) standardization, including: converting all data of each of the above biomarkers in the above dataset into relative values ​​respectively; optionally, the above standardization includes: performing Z-score standardization on all data of each of the above biomarkers in the above dataset respectively.

[0099] To achieve the above objectives, according to a seventh aspect of the present invention, a training method for a predictive model of human aging level is provided, the training method comprising: obtaining the calendar age and indicators of a sample population, establishing a dataset of the sample population, wherein the indicators used include the aforementioned indicators; and constructing the model on the dataset using the aforementioned machine learning method.

[0100] Furthermore, the aforementioned machine learning method includes determining the optimal hyperparameters using five-fold cross-validation combined with random search within the aforementioned training dataset to obtain the aforementioned model; optionally, before the aforementioned machine learning training, the aforementioned dataset is subjected to any one or more of the following processing: i) missing value processing, including: calculating the average value of each of the aforementioned indicators in the aforementioned dataset, and using the aforementioned average value to fill in the missing of the aforementioned indicators in the aforementioned dataset; ii) standardization, including: converting all data of each of the aforementioned biomarkers in the aforementioned dataset into relative values ​​respectively;

[0101] Optionally, the above standardization includes: performing Z-score standardization on all data for each of the above biomarkers in the above dataset; Optionally, the above training method further includes: dividing the above dataset into three overlapping age groups according to the above calendar age, and independently training the corresponding age segment sub-model for each age group; Optionally, the above three overlapping age groups include 0-A years, BC years, and D-100 years; A includes 47-50, B includes 42-50, C includes 60-63, and D includes 57-60, where A≥B and C≥D; Optionally, the above three overlapping age groups include any one of the following groups: 0-48 years, 42-63 years and 57-100 years; 0-50 years, 50-60 years and 60-100 years; or 0-47 years, 43-62 years and 58-100 years;

[0102] Optionally, the above training method further includes: dividing the above dataset into a male training set and a female training set according to gender, and using the above machine learning training to train male sub-models and female sub-models respectively; Optionally, the above training method further includes: before building the model using the above training set, weighting the above samples based on the distribution of the above calendar ages in the above training set; Optionally, the weight of the above weighting = A ÷ (B × C), where A is the total number of samples in the above training set, B is the total number of unique above calendar ages in the above training set, and C is the number of times the above calendar age appears in the above training set.

[0103] To achieve the above objectives, according to an eighth aspect of the present invention, a computer device is provided, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method for constructing a prediction model of human aging level, the method for predicting human aging level, the method for establishing a dataset, or the method for training a model for predicting human aging level.

[0104] To achieve the above objectives, according to a ninth aspect of the present invention, a computer-readable storage medium is provided, on which a computer program is stored, wherein when the computer program is executed by a processor, the steps of the method for constructing the prediction model of human aging level, the method for predicting human aging level, the method for establishing a dataset, or the method for training the model for predicting human aging level are implemented.

[0105] To achieve the above objectives, according to a tenth aspect of the present invention, a computer program product is provided, comprising a computer program that, when executed by a processor, implements the steps of a method for constructing a predictive model of human aging level, a method for predicting human aging level, a method for establishing a dataset, or a method for training a model for predicting human aging level.

[0106] By applying the technical solution of this invention and utilizing the aforementioned model construction method for predicting human aging levels, a dataset of the sample population is established by collecting chronological ages and indicators. This dataset is then trained using machine learning techniques to construct a model capable of predicting human aging levels. The model constructed in this invention can capture the complex relationship between indicators that can be detected through routine physical examinations and aging levels, thus exhibiting high sensitivity and specificity in predicting human aging levels. It can accurately judge and predict aging levels, and compared to existing technologies, the method for obtaining these indicators is simpler, thereby reducing the overall cost of aging detection. Attached Figure Description

[0107] The accompanying drawings, which form part of this application, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings:

[0108] Figure 1 A scatter plot of the model prediction results in Embodiment 3 of the present invention is shown. Detailed Implementation

[0109] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. The present invention will now be described in detail with reference to the embodiments.

[0110] As mentioned in the background section, existing methods for predicting human aging levels require high-quality data, have high prediction costs, and poor performance. Therefore, in this application, the inventors attempt to develop a low-cost method for constructing a predictive model of human aging levels, and based on this, propose a series of protection schemes for this application.

[0111] In a first typical embodiment of this application, a method for constructing a predictive model for human aging levels is provided. The method includes: a) obtaining the chronological age and indicators of a sample population, and establishing a dataset for the sample population; wherein the indicators include one or more of the following: mean corpuscular volume, platelet criterion, hematocrit, red cell distribution width, monocytes percentage, mean corpuscular hemoglobin, neutrophil count, monocyte count, red blood cell count, platelet count, or lymphocyte count, total cholesterol, triglycerides, low-density lipoprotein cholesterol, and total protein. The data includes the following: protein, uric acid, urea, hemoglobin, alkaline phosphatase, albumin, alanine aminotransferase, glucose, aspartate aminotransferase, apolipoprotein A, or creatinine levels; and the sample population's diastolic blood pressure, systolic blood pressure, height, weight, and body mass index (BMI, calculated from the relationship between an individual's weight and height); or b) using the above datasets and machine learning training methods to construct the above model.

[0112] In this application, "calendar age" represents an individual's actual age. Optionally, in the above construction method, calendar age is taken as an integer.

[0113] The above-described method of this application constructs a detailed dataset by collecting the chronological age and various indicators of the sample population. These indicators cover, but are not limited to, components of the blood in the sample population, such as mean corpuscular volume, platelet-to-volume hematocrit, hematocrit, red blood cell distribution width, monocyte percentage, mean corpuscular hemoglobin, neutrophil count, monocyte count, red blood cell count, platelet count, or lymphocyte count; total cholesterol content, triglyceride content, low-density lipoprotein cholesterol content, total protein content, uric acid content, urea content, hemoglobin content, alkaline phosphatase content, albumin content, alanine aminotransferase content, glucose content, aspartate aminotransferase content, apolipoprotein A content, or creatinine content; and the diastolic blood pressure, systolic blood pressure, height, weight, or body mass index of the sample population.

[0114] Optionally, the above indicators include 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29 or 30 of the above a).

[0115] The indicators mentioned above in this application are all routine blood indicators and general body measurement indicators used in routine physical examinations of human health in the prior art. In this application, the use of the above indicators can enable the construction of a more accurate model for predicting the level of cognitive aging. Moreover, the above indicators mentioned above in this application are easier to obtain than those in the prior art, reducing the pressure and cost of data collection, and balancing cost and effectiveness.

[0116] The aforementioned indicators include any one or more of the indicators, including but not limited to any one or more components in the blood, and / or one or more anthropometric data, and / or the relevant ratios or proportions of one or more compounds in the blood. It should be noted that the multiple compounds in the blood mentioned above should be understood as corresponding multiple indicators, not a single indicator.

[0117] In step b) of the above-described construction method, after the dataset is ready, the method utilizes existing machine learning techniques to perform in-depth analysis of the collected data. The aim is to uncover the intrinsic relationship between biomarkers and aging levels, forming a reliable predictive model. The machine learning training process includes algorithm selection, model construction, and validation to ensure the model achieves the required performance. This method enables the construction of highly predictive models for accurately assessing individual aging levels, while reducing the need for a large number of indicators and lowering the difficulty of obtaining them, thereby reducing application costs.

[0118] Optionally, the indicators in a) above may consist of any one or more of the following indicators: mean corpuscular volume, platelet-to-volume ratio, hematocrit, red blood cell distribution width, monocyte percentage, mean corpuscular hemoglobin, neutrophil count, monocyte count, red blood cell count, platelet count or lymphocyte count, total cholesterol content, triglyceride content, low-density lipoprotein cholesterol content, total protein content, uric acid content, urea content, hemoglobin content, alkaline phosphatase content, albumin content, alanine aminotransferase content, glucose content, aspartate aminotransferase content, apolipoprotein A content, or creatinine content, as well as diastolic blood pressure, systolic blood pressure, height, weight or body mass index of the sample population.

[0119] Optionally, the above indicators may include any one of the following sets, or may consist of any one of the following sets:

[0120] 1) Systolic blood pressure;

[0121] 2) Systolic blood pressure + urea content;

[0122] 3) Systolic blood pressure + urea content + alkaline phosphatase content;

[0123] 4) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content;

[0124] 5) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content;

[0125] 6) Height + weight + apolipoprotein A level + platelet count + diastolic blood pressure;

[0126] 7) Alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight;

[0127] 8) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height;

[0128] 9) Height + weight + apolipoprotein A content + platelet count + diastolic blood pressure + alanine aminotransferase content;

[0129] 10) Alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content;

[0130] 11) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight.

[0131] 12) Height + weight + apolipoprotein A content + platelet count + diastolic blood pressure + alanine aminotransferase content + albumin content;

[0132] 13) Alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + urea content;

[0133] 14) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content;

[0134] 15) Height + weight + apolipoprotein A content + platelet count + diastolic blood pressure + alanine aminotransferase content + alkaline phosphatase content + albumin content;

[0135] 16) Alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + urea content + platelet-to-volume ratio.

[0136] 17) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio.

[0137] 18) Alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + urea content + platelet count + diastolic blood pressure.

[0138] 19) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-weight ratio + diastolic blood pressure.

[0139] 20) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content.

[0140] 21) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume.

[0141] 22) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content.

[0142] 23) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase.

[0143] 24) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content.

[0144] 25) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content.

[0145] 26) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage.

[0146] 27) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content;

[0147] 28) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width;

[0148] 29) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index;

[0149] 30) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit;

[0150] 31) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit + mean corpuscular hemoglobin;

[0151] 32) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit + mean corpuscular hemoglobin + triglyceride content;

[0152] 33) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit + mean corpuscular hemoglobin + triglyceride content + neutrophil count;

[0153] 34) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit + mean corpuscular hemoglobin + triglyceride content + neutrophil count + uric acid content;

[0154] 35) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit + mean corpuscular hemoglobin + triglyceride content + neutrophil count + uric acid content + monocyte count;

[0155] 36) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit + mean corpuscular hemoglobin + triglyceride content + neutrophil count + uric acid content + monocyte count + hemoglobin content and concentration;

[0156] 37) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit + mean corpuscular hemoglobin + triglyceride content + neutrophil count + uric acid content + monocyte count + hemoglobin concentration + red blood cell count;

[0157] 38) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit + mean corpuscular hemoglobin + triglyceride content + neutrophil count + uric acid content + monocyte count + hemoglobin concentration + red blood cell count + platelet count;

[0158] 39) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit + mean corpuscular hemoglobin + triglyceride content + neutrophil count + uric acid content + monocyte count + hemoglobin concentration + red blood cell count + platelet count + lymphocyte count.

[0159] Note: The "content" in 1)-39) above refers to the content in blood. For example, the molar concentration of the corresponding substance in blood, in units such as mmol / L.

[0160] In one optional embodiment, the above indicators include any one, two, three, four, five, six, seven, eight, nine, or ten of the following indicators: the content of urea, alkaline phosphatase, albumin, alanine aminotransferase, apolipoprotein A, platelet-to-volume ratio, systolic blood pressure, height, weight, and diastolic blood pressure in the blood of the sample population.

[0161] Optionally, the above indicators may include any one, two, three, four, five, six, seven, eight, nine, or ten of the following indicators: the urea content, alkaline phosphatase content, albumin content, alanine aminotransferase content, apolipoprotein A content, platelet-to-volume ratio, as well as systolic blood pressure, height, weight, and diastolic blood pressure in the blood of the sample population.

[0162] Optionally, the above indicators may include any one of the following sets, or may consist of any one of the following sets:

[0163] 1) Systolic blood pressure;

[0164] 2) Systolic blood pressure + urea content;

[0165] 3) Systolic blood pressure + urea content + alkaline phosphatase content;

[0166] 4) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content;

[0167] 5) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content;

[0168] 6) Height + weight + apolipoprotein A level + platelet count + diastolic blood pressure;

[0169] 7) Alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight;

[0170] 8) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height;

[0171] 9) Height + weight + apolipoprotein A content + platelet count + diastolic blood pressure + alanine aminotransferase content;

[0172] 10) Alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content;

[0173] 11) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight.

[0174] 12) Height + weight + apolipoprotein A content + platelet count + diastolic blood pressure + alanine aminotransferase content + albumin content;

[0175] 13) Alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + urea content;

[0176] 14) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content;

[0177] 15) Height + weight + apolipoprotein A content + platelet count + diastolic blood pressure + alanine aminotransferase content + alkaline phosphatase content + albumin content;

[0178] 16) Alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + urea content + platelet-to-volume ratio.

[0179] 17) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio.

[0180] 18) Alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + urea content + platelet count + diastolic blood pressure.

[0181] 19) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure.

[0182] In a preferred embodiment, the indicator contains at least total cholesterol; the indicator may also optionally contain: diastolic blood pressure, systolic blood pressure, height, weight, or body mass index of the sample population, and / or any one or more of the following indicators in the blood of the sample population: mean corpuscular volume, platelet-to-volume ratio, hematocrit, red blood cell distribution width, monocyte percentage, mean corpuscular hemoglobin, neutrophil count, monocyte count, red blood cell count, platelet count, lymphocyte count, triglyceride content, low-density lipoprotein cholesterol content, total protein content, uric acid content, urea content, hemoglobin content, alkaline phosphatase content, albumin content, alanine aminotransferase content, glucose content, aspartate aminotransferase content, apolipoprotein A content, or creatinine content;

[0183] Optionally, the indicators may include total cholesterol, mean corpuscular volume, urea levels, alkaline phosphatase levels, albumin levels, alanine aminotransferase levels, apolipoprotein A levels, platelet-to-volume ratio, glucose levels, aspartate aminotransferase levels, or creatinine levels in the blood of the sample population, as well as systolic blood pressure, height, weight, and diastolic blood pressure, or a combination thereof.

[0184] Optionally, the above indicators may include any one of the following sets, or may consist of any one of the following sets:

[0185] 24) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content.

[0186] 25) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content.

[0187] 26) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage.

[0188] 27) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content;

[0189] 28) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width;

[0190] 29) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index;

[0191] 30) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit;

[0192] 31) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit + mean corpuscular hemoglobin;

[0193] 32) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit + mean corpuscular hemoglobin + triglyceride content;

[0194] 33) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit + mean corpuscular hemoglobin + triglyceride content + neutrophil count;

[0195] 34) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit + mean corpuscular hemoglobin + triglyceride content + neutrophil count + uric acid content;

[0196] 35) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit + mean corpuscular hemoglobin + triglyceride content + neutrophil count + uric acid content + monocyte count;

[0197] 36) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit + mean corpuscular hemoglobin + triglyceride content + neutrophil count + uric acid content + monocyte count + hemoglobin content and concentration;

[0198] 37) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit + mean corpuscular hemoglobin + triglyceride content + neutrophil count + uric acid content + monocyte count + hemoglobin concentration + red blood cell count;

[0199] 38) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit + mean corpuscular hemoglobin + triglyceride content + neutrophil count + uric acid content + monocyte count + hemoglobin concentration + red blood cell count + platelet count;

[0200] 39) Systolic blood pressure + urea content + alkaline phosphatase content + albumin content + alanine aminotransferase content + height + weight + apolipoprotein A content + platelet-to-volume ratio + diastolic blood pressure + creatinine content + mean corpuscular volume + glucose content + aspartate aminotransferase + total cholesterol content + total protein content + monocyte percentage + low-density lipoprotein cholesterol content + red blood cell distribution width + body mass index + hematocrit + mean corpuscular hemoglobin + triglyceride content + neutrophil count + uric acid content + monocyte count + hemoglobin concentration + red blood cell count + platelet count + lymphocyte count.

[0201] By utilizing any combination of the above indicators, it is possible to construct an accurate and usable model. It is important to emphasize that although the above indicators are known in existing technologies, current technologies for models predicting human aging levels only demonstrate the use of a large number of indicators for model construction (Tian, ​​Ye Ella, et al. "Heterogeneous aging across multiple organ systems and prediction of chronic disease and mortality." Nature medicine 29.5 (2023): 1221-1231.), without disclosing how to select a specific set of indicators to construct the model.

[0202] In one optional embodiment, the aforementioned human aging level includes biological age; optionally, before using the aforementioned machine learning training, the aforementioned dataset is divided into a training set and a test set, the aforementioned training set is used to build a model, and the aforementioned test set is used to verify the model's performance; optionally, the aforementioned machine learning training method includes any one or more of the following: XGBoost regression model, Ridge regression model, Lasso regression model, or Ealstic Net regression model; optionally, the objective function of the aforementioned XGBoost regression model is reg:squarederror.

[0203] In the above model, the "biological age" output by the model represents the aging level of an individual. That is, after a usable model is constructed, the subject's indicators are input into the model, and the model can output the biological age. The biological age can reflect the actual degree of aging of the subject, which may be different from the value of calendar age. This quantifiable data provides a more accurate assessment of the aging level.

[0204] In the above construction method, before machine learning training, the dataset is scientifically divided into a training set and a test set. The former is used to build the model, while the latter serves as an independent validation set to ensure the model's generalization ability. The machine learning training methods described are flexible and diverse, including but not limited to XGBoost regression models, Ridge regression models, Lasso regression models, or Ealstic Net regression models. Optionally, in the XGBoost regression model, the squared error (reg:squarederror) is used as the objective function to minimize the mean squared error between the predicted and true values, ensuring the model's accuracy.

[0205] In an optional embodiment, the method for dividing the above dataset into a training set and a test set includes: randomly splitting the above sample population into the above training set or the above test set; optionally, before performing the above machine learning training, the above dataset is subjected to any one or more of the following processing: i) missing value processing, including: calculating the average value of each of the above indicators in the above dataset, and using the average value to fill the missing indicators in the above dataset; ii) standardization, including: converting all data of each of the above biomarkers in the above dataset into relative values ​​respectively; optionally, the above standardization includes: performing Z-score standardization processing on all data of each of the above biomarkers in the above dataset respectively.

[0206] In the above construction method, the training set and the test set are strictly separated. Data in the test set is not used in the model building process, and data in the training set is not used in the model validation process.

[0207] Processing the dataset before machine learning training can improve the accuracy of model building and reduce training difficulty. Processing methods include, but are not limited to:

[0208] i) Handle missing values. Taking a single metric as an example, if not all samples in the dataset have the corresponding metric, meaning that some samples in the dataset have missing data for that metric (empty), first sum the data for that metric in the dataset and calculate the average (average = total data ÷ number of data points, excluding the number of individuals with missing data). Then, use the average to fill in the missing data for that metric. Perform this missing value handling for each metric to ensure that the accuracy of the normal model is not affected by missing data during machine learning training.

[0209] ii) Standardization. This involves converting each indicator into a relative value to eliminate the influence of dimensions and ensure the model learns on a uniform scale. Relative values ​​reflect an individual's relative ranking within the entire population for that indicator. Optionally, Z-score standardization can be used to process the indicators, which involves subtracting the mean from each biomarker data and then dividing by the standard deviation, thus converting it into standardized data with zero mean and unit standard deviation.

[0210] Optionally, when selecting metrics, metrics with a missing rate ≤20% in the dataset are selected for model training. Optionally, the missing rate of specific samples is not screened, because such samples and situations may exist in the real world. Therefore, retaining such samples during training and testing can further improve the robustness of the model.

[0211] In an optional embodiment, step b) above further includes: dividing the dataset into three overlapping age groups according to the calendar age, and independently training a corresponding age segmentation sub-model for each age group. The predicted value of samples in the test set that fall into the overlapping age groups is obtained by weighted summation of two adjacent age segmentation sub-models according to the linear weight of the distance from the boundary. Optionally, the three overlapping age groups include 0-A years old, BC years old, and D-100 years old; A includes 47-50 (including 47, 48, 49, or 50), B includes 42-50 (including 42, 43, 44, 45, 46, 47, 48, 49, or 50), C includes 60-63 (including 60, 61, 62, or 63), and D includes 57-60 (including 57, 58, 59, or 60); wherein A≥B, C≥D. Optionally, the three overlapping age groups mentioned above can be any of the following groups: 0-48 years old, 42-63 years old and 57-100 years old; 0-50 years old, 50-60 years old and 60-100 years old; or 0-47 years old, 43-62 years old and 58-100 years old.

[0212] In the model building process, a segmented modeling approach can be optionally adopted, dividing the chronological age of the sample population into age segments (intervals). Each interval is trained independently with an age segment sub-model to better capture the biological changes at different life stages. Optionally, these intervals may overlap. If the chronological age of an individual in the training set falls into two adjacent intervals, that individual's data is used independently during the training of the models corresponding to both intervals.

[0213] When validating the model using the test set, if the calendar age of an individual in the test set falls in the overlapping region of two intervals, the predicted value is obtained through a linear weighted fusion strategy. That is, the predicted value is obtained by weighted summation of the prediction results of two adjacent models, ensuring the continuity and stability of the model in the transition age range.

[0214] The calculation of linear weights includes, but is not limited to, the following methods: For samples whose ages fall within overlapping intervals, the linear weights are first calculated based on their positional relationship with the boundaries of the two age intervals. Assuming the sample age is x, and the overlapping interval is [c, b], where c is the upper limit of the left interval and b is the lower limit of the right interval (c < b), then the weights w_left of the left model and w_right of the right model can be calculated using the following formulas: w_left = (x - c) / (b - c), w_right = 1 - w_left. This means that samples closer to the left boundary have a more dependent prediction on the output of the left model; conversely, samples closer to the right boundary have a more dependent prediction on the right model.

[0215] After obtaining the two weights mentioned above, the final predicted value y_hat(x) of the sample is obtained by weighted summation of the predicted values ​​M_left(x) and M_right(x) of the two piecewise models: y_hat(x) = w_left × M_left(x) + w_right × M_right(x). This ensures a smooth and continuous transition of the predicted value between different models, avoiding abrupt changes or inconsistencies in the prediction results caused by hard boundary divisions.

[0216] For samples in non-overlapping intervals, i.e., when the age falls entirely within the coverage of a single model, the predicted value is directly generated by the model for that interval, without the need for weight fusion.

[0217] The linear weight fusion strategy is based on the fact that although age is artificially divided into different intervals, an individual's biological age change is a continuous process. In overlapping regions, an individual may be influenced by biological change patterns in both age intervals simultaneously. Therefore, by calculating the relative position to the boundaries of the two intervals and linearly assigning the prediction weights of the two models, this continuous change trend can be more accurately reflected, improving the accuracy and reliability of predictions.

[0218] Those skilled in the art can flexibly divide age groups, and are not limited to the division methods described in this application. When the above three overlapping age groups are divided into: 0-48 years old, 42-63 years old, and 57-100 years old, compared with the division into: 0-47 years old, 43-62 years old, and 58-100 years old, or into: 0-50 years old, 50-60 years old, and 60-100 years old, the performance and interpretability of the division are better, and the prediction results have better smoothness at the segment boundaries.

[0219] In an optional embodiment, step b) above includes: dividing the dataset into a male training set and a female training set according to gender, and using the machine learning training described above to train male sub-models and female sub-models respectively.

[0220] Furthermore, step b) above employs a gender-differentiated modeling strategy, classifying the dataset by gender and establishing separate training sets for men and women. This is because men and women differ in physiological structure, metabolic pathways, and other aspects, which may affect the expression of biomarkers. By training male and female sub-models separately, the aging process specific to each gender can be more accurately reflected, enhancing the model's predictive ability and interpretability.

[0221] In an optional embodiment, before constructing the model using the training set, the samples are weighted based on the distribution of the calendar ages in the training set; optionally, the weight of the weighting is A ÷ (B × C), where A is the total number of samples in the training set, B is the total number of unique calendar ages in the training set, and C is the number of times the calendar age appears in the training set.

[0222] In the above construction method, a sample weighting strategy is introduced before model construction to balance the imbalance in the number of samples across age groups. The weight calculation formula for each calendar age is: Weight = Total number of samples A ÷ (Number of unique calendar age items (and total number) B × Number of samples for that calendar age (and frequency of occurrence) C). This ensures that samples from each age group are fully considered during training, avoiding the neglect of features from a few age groups. By weighting the samples in the training set, the impact of samples from different calendar ages on the model varies. This weighting helps improve the fairness and accuracy of the model's predictions for each age group.

[0223] In an optional embodiment, step b) above includes: dividing the sample population into a training set and a test set; establishing a mapping relationship between the above-mentioned indicators and the above-mentioned calendar age in the training set or the test set, respectively obtaining a training dataset and a test dataset; constructing a model on the training dataset using the above-mentioned machine learning method, including determining the optimal hyperparameters in the training dataset using five-fold cross-validation combined with random search, to obtain a preliminary model; validating the preliminary model using the above-mentioned test dataset, calculating and evaluating any one or more of the following parameters of the preliminary model: mean absolute error, coefficient of determination, or Pearson correlation coefficient, to obtain the above-mentioned prediction model.

[0224] Step b) includes: First, the sample population is divided into training and test sets to ensure data independence and the model's generalization ability. Then, based on the training dataset, the model's hyperparameters are optimized using a built-in five-fold cross-validation method combined with a random search strategy to obtain a preliminary model. This step effectively avoids model overfitting and ensures good model performance on unknown data. Finally, the preliminary model is rigorously evaluated using the test dataset by calculating the mean absolute error (MAE) and coefficient of determination (R²). 2 One or more of the following can be used to evaluate the predictive accuracy and stability of the model, thereby obtaining a high-quality predictive model for human aging levels: the Pearson correlation coefficient (r).

[0225] In a second typical embodiment of this application, a predictive model for human aging level is provided, wherein the model is obtained by using the method for constructing the predictive model for human aging level.

[0226] This application also provides a predictive model for human aging levels, which is a model obtained using the above-described construction method. Using this model, the cost of predicting biological age can be significantly reduced because only routine physical examinations are required to obtain the necessary data, and the prediction process is rapid, generating reports in a short time. More importantly, compared with existing technologies, the model of this invention exhibits better predictive performance, particularly demonstrating the effectiveness of the segmented modeling strategy in ablation experiments, with lower mean absolute error (MAE) and lower coefficient of determination (R²). 2 The model shows significant improvements in parameters such as the Pearson correlation coefficient (r) compared to existing technologies, demonstrating its remarkable advantages in predicting biological age.

[0227] In a third typical embodiment of this application, a device for predicting the level of human aging is provided. The device includes an input module and a prediction module. The input module is used to acquire an index of a sample to be tested and input the index into the prediction module. The index is the aforementioned index. The prediction module is used to calculate the level of human aging based on the index. The prediction module contains a prediction model, which is a model obtained using the method for constructing a prediction model for the level of human aging.

[0228] The core of the aforementioned device lies in the prediction model contained in its prediction module. This prediction model is obtained using the method described above for constructing a prediction model of human aging levels. In using the device, simply obtaining the subject's indicators using the input module and inputting these indicators into the prediction module is sufficient to determine the subject's aging level.

[0229] In a fourth typical embodiment of this application, a method for predicting the level of human aging is provided. The method includes: inputting the indicators of the sample to be tested into a prediction model and obtaining the level of human aging output by the prediction model; wherein the prediction model is a model obtained by using the method for constructing the prediction model of the level of human aging.

[0230] Among the methods described above for predicting the level of human aging, the models obtained based on these construction methods can predict or assess the level of human aging. It should be noted that this method for predicting the level of human aging is not a diagnostic or treatment method for diseases; it is merely an assessment of an individual's physiological condition and is unrelated to disease.

[0231] In a fifth typical embodiment of this application, an application of a biomarker in predicting the level of human aging is provided, and the biomarker used includes the aforementioned biomarker.

[0232] In a sixth typical embodiment of this application, a method for establishing a dataset is provided, the method comprising: obtaining the calendar age and / or indicators of a target sample from a dataset to obtain the dataset; wherein the indicators used include the aforementioned indicators.

[0233] In an optional embodiment, the method further includes: i) missing value processing, including: calculating the average value of each of the above indicators in the above dataset, and using the average value to fill in the missing indicators in the above dataset; and / or ii) standardization, including: performing Z-score standardization on all data of each of the above biomarkers in the above dataset to convert them into relative values; optionally, the standardization includes: performing Z-score standardization on all data of each of the above biomarkers in the above dataset.

[0234] This application provides a method for building a dataset, which involves obtaining the calendar age and specific indicators of target samples from a dataset. Using the above-described method for building a dataset, specific data can be obtained from a dataset containing a large amount of data, providing a foundation for subsequent data applications, including but not limited to model building, or using the model and the target sample data to predict the aging level of the target sample.

[0235] In a seventh typical embodiment of this application, a training method for a predictive model of human aging level is provided. The training method includes: obtaining the calendar age and indicators of a sample population, establishing a dataset of the sample population, and using the indicators to construct the model on the dataset using the machine learning method.

[0236] In an optional embodiment, the machine learning method includes determining the optimal hyperparameters using five-fold cross-validation combined with random search within the training dataset to obtain the model. Optionally, before the machine learning training, the dataset is subjected to any one or more of the following processing: i) missing value processing, including: calculating the average value of each of the above indicators in the dataset and using the average value to fill in the missing indicators in the dataset; ii) standardization, including: converting all data of each of the above biomarkers in the dataset into relative values; Optionally, the standardization includes: performing Z-score standardization on all data of each of the above biomarkers in the dataset.

[0237] Optionally, the above training method further includes: dividing the above dataset into three overlapping age segments according to the above calendar age, and training the corresponding age segment sub-model independently for each age segment;

[0238] Optionally, the three overlapping age groups mentioned above include 0-A years, BC years, and D-100 years; A includes 47-50 years (including 47, 48, 49, or 50), B includes 42-50 years (including 42, 43, 44, 45, 46, 47, 48, 49, or 50), C includes 60-63 years (including 60, 61, 62, or 63), and D includes 57-60 years (including 57, 58, 59, or 60); where A≥B, C≥D. Optionally, the three overlapping age groups mentioned above may include any one of the following groups: 0-48 years, 42-63 years and 57-100 years; 0-50 years, 50-60 years and 60-100 years; or 0-47 years, 43-62 years and 58-100 years.

[0239] Optionally, the above training method further includes: dividing the above dataset into male training set and female training set according to gender, and using the above machine learning training to train male sub-model and female sub-model respectively.

[0240] Optionally, the above training method further includes: before constructing the model using the above training set, weighting the above samples based on the distribution of the above calendar ages in the above training set; optionally, the weight of the above weighting is A ÷ (B × C), where A is the total number of samples in the above training set, B is the total number of non-repeating above calendar ages in the above training set, and C is the number of times the above calendar age appears in the above training set.

[0241] Using the above training method, a model that can be used to predict the level of human aging can be trained using a dataset of the sample population. Optionally, after using the above training method, those skilled in the art can use data different from the sample population to perform a posterior test on the model, thereby improving the reliability of the model and verifying the model's usability for the sample.

[0242] In the eighth typical embodiment of this application, a computer device is provided, including a memory, a processor, and a computer program stored in the memory. The processor executes the computer program to implement the steps of the method for constructing a prediction model of human aging level, the method for predicting human aging level, the method for establishing a dataset, or the method for training a model for predicting human aging level.

[0243] In a ninth typical embodiment of this application, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, it implements the steps of the method for constructing the prediction model of human aging level, the method for predicting human aging level, the method for establishing a dataset, or the method for training the model for predicting human aging level.

[0244] In a tenth typical embodiment of this application, a computer program product is provided, including a computer program, which, when executed by a processor, implements the steps of constructing a prediction model for human aging level, predicting human aging level, establishing a dataset, or training a model for predicting human aging level.

[0245] The beneficial effects of this application will be explained in more detail below with reference to specific embodiments.

[0246] Example 1

[0247] 1. Research Objectives

[0248] A model is built based on a small number of blood test indicators and general body measurements to predict biological age.

[0249] 2. Data Source

[0250] a) Study subjects: 10,000 physical examination data;

[0251] b) Inclusion criteria: The dataset includes samples that simultaneously include calendar age, blood cell count, biochemical indicators, metabolic indicators, and general body measurements; samples that simultaneously include blood tests and general body measurements.

[0252] 3. Indicators

[0253] Includes any combination of five or more of the following indicators:

[0254] The blood samples of the sample population include mean corpuscular volume, platelet count, hematocrit, red blood cell distribution width, monocyte percentage, mean corpuscular hemoglobin, neutrophil count, monocyte count, red blood cell count, platelet count or lymphocyte count, total cholesterol content, triglyceride content, low-density lipoprotein cholesterol content, total protein content, uric acid content, urea content, hemoglobin content, alkaline phosphatase content, albumin content, alanine aminotransferase content, glucose content, aspartate aminotransferase content, apolipoprotein A content, or creatinine content, as well as diastolic blood pressure, systolic blood pressure, height, weight or body mass index.

[0255] Note: The levels of cholesterol, triglycerides, direct low-density lipoprotein cholesterol, total protein, uric acid, urea, hemoglobin, alkaline phosphatase, albumin, alanine aminotransferase, glucose, aspartate aminotransferase, apolipoprotein A, or creatinine are all corresponding components in the subject's blood.

[0256] 4. Gender-specific modeling strategy (optional)

[0257] When constructing the overall biological age prediction model, different models are built for men and women.

[0258] 5. Segmented modeling strategy (optional)

[0259] When constructing an overall biological age prediction model, different models are built for different age groups.

[0260] a) Age segmentation: Divide the calendar age range into 3 overlapping intervals: 0-48 years old, 42-63 years old, and 57-100 years old.

[0261] b) Overlapping interval fusion: For samples falling into overlapping age zones, the predicted values ​​are weighted and summed by the two adjacent segmented models according to the linear weights of the distance from the boundary.

[0262] 6. Model Evaluation Metrics

[0263] MAE: Mean Absolute Error. MAE is the preferred core indicator for evaluating "aging clock" models. It represents the average difference between the biological age predicted by the model and the individual's actual physiological age (calendar age).

[0264] R 2 Coefficient of determination;

[0265] r: Pearson correlation coefficient.

[0266] While the difference between predicted biological age and calendar age reflects an individual's aging level, a smaller difference at the individual level does not necessarily indicate a more accurate model. However, for large-scale datasets, since the aging level of a population also follows a similar normal distribution, it is generally accepted in the art that a smaller MAE indicates a weaker predictive performance. Existing technologies often use MAE to evaluate the predictive performance of such models.

[0267] 7. Comparison

[0268] A single full-coverage model (without segmentation) serves as a contrast to segmented modeling.

[0269] 8. Interpretiveness of prediction results

[0270] The SHAP value is used to calculate the impact of different indicators on an individual's biological age.

[0271] Example 2

[0272] Based on the description in Example 1, a specific experiment was conducted, and the method is as follows:

[0273] 1. Data processing flow

[0274] a) Missing value handling: Fill missing values ​​with the mean of each indicator, including calculating the mean of each indicator in the dataset and using the mean to fill missing indicators in the data.

[0275] b) Standardization: Standardize all indicators to a mean of 0 and a variance of 1, including performing Z-score standardization on all data for each biomarker in the dataset.

[0276] c) Training and test set splitting: Subjects were randomly divided into training and test sets in an 8:2 ratio.

[0277] 2. Build a model using the training set.

[0278] a) Model: A model for predicting human aging levels was built using XGBoost regression (xgboost reg:squarederror) and training set data. The software and package versions included: Python 3.9, xgboost (2.1.4), and scikit_learn (1.5.1).

[0279] b) Train a sub-model for each gender in different calendar age ranges (0-48 years, 42-63 years, 57-100 years). The training data is samples within the range (if the calendar age of a sample falls in the overlapping range of the above ranges, such as 42-48 years or 57-63 years, then the sample is used as the training set in the training of the sub-models in the two adjacent calendar age ranges).

[0280] c) Setting sample weights: In order to address the uneven distribution of samples across different calendar ages, weights are set for the samples of subjects of different calendar ages, and a weighted calculation is performed before modeling.

[0281] Weight = Total number of samples / (Number of unique ages in the data × Number of samples of that age).

[0282] d) Hyperparameters are determined on the training set through five-fold cross-validation via random search.

[0283] Example of candidate range:

[0284] i."n_estimators": (100, 500);

[0285] ii. "max_depth": (3, 10);

[0286] iii."learning_rate": (0.01, 0.2);

[0287] iv."subsample": (0.6, 1.0);

[0288] v."colsample_bytree": (0.6, 1.0);

[0289] vi."min_child_weight": (1, 10);

[0290] vii."gamma": (0, 0.5);

[0291] viii."reg_alpha": (0, 10);

[0292] ix."reg_lambda": (0.1, 10).

[0293] 3. Test the model using the test set.

[0294] a) Biological age output:

[0295] Using the model obtained by training the training set, predict the biological age of the test set.

[0296] For samples in the test set whose calendar ages fall within the overlapping ranges mentioned above (e.g., 42-48 years old, 57-63 years old), their biological ages predicted by the model are obtained using a weighted calculation method, as follows:

[0297] i. Define adjacent calendar age segments (including three segments: 0-48 years old, 42-63 years old, and 57-100 years old) as [a,b] and [c,d], where b > c indicates an overlapping segment [c,b].

[0298] ii. Let x be the calendar age of the sample. For samples x∈[c, b], calculate the linear weight of the sample, including:

[0299] w_left = (x - c) / (b - c), w_right = 1 - w_left;

[0300] y_hat(x) = w_left × M_left(x) + w_right × M_right(x);

[0301] Where y_hat(x) represents the final predicted value of the sample, w_left represents the weight of the left model, w_right represents the weight of the right model, and M_left(x) and M_right(x) represent the predicted values ​​of the two segmented models, respectively.

[0302] iii. Samples located in non-overlapping intervals are predicted only by the corresponding segmented model.

[0303] 4. Explanatory

[0304] The SHAP value of each indicator on each sample is calculated to represent the indicator's impact on biological age.

[0305] 5. Ablation test

[0306] Repeat steps 2-4 in this embodiment, except that the calendar age range modeling (also known as segmented modeling) in step 2b) is no longer performed. Instead, a model is built for all age ranges and the performance is compared.

[0307] Example 3

[0308] 1. The model constructed using steps 2-4 of Example 2 (without gender-specific modeling, but with segmented modeling, and 30 indicators selected in the model) has the following overall performance: MAE = 2.50 for the test set, and R² = 0.50 for the test set. 2 =0.85, the correlation coefficient of the test set r=0.93, the scatter plot is as follows Figure 1 As shown, True Age is the calendar age, and Predicted Age is the biological age.

[0309] 2. Using the model constructed in step 5 of Example 2 (without gender-specific or segmented modeling, and using 30 indicators), its overall performance is as follows: MAE = 5.47 on the test set, and R² = 0.5%. 2 =0.31, and the correlation coefficient of the test set r=0.57.

[0310] 3. The model obtained by step 2-4 of Example 2 (which performs gender-specific modeling and segmented modeling, and selects 30 indicators in the model) is compared with the model disclosed in relevant literature in the prior art. The results are shown in Table 1.

[0311] Literature: Tian, ​​Ye Ella, et al. "Heterogeneous aging across multiple organsystems and prediction of chronic disease and mortality." Nature medicine29.5 (2023): 1221-1231.

[0312] Table 1

[0313]

[0314] Example 4

[0315] Predictive models were constructed using different indicators, with the same construction methods as in Examples 1 and 2. Gender-specific modeling was performed, and segmented and non-segmented modeling were attempted respectively.

[0316] The results of the test set are shown in Tables 2 and 3 below, where the MAE value is the overall result that combines the male and female models.

[0317] Table 2

[0318] .

[0319] Table 3

[0320]

[0321]

[0322] The “content” in Tables 2 and 3 above refers to the content in the blood, that is, the molar concentration of the corresponding substance in the blood, for example, in mmol / L.

[0323] As can be seen from the above description, the embodiments of the present invention achieve the following technical effects: The technical solution of this application only needs to rely on blood test indicators and basic body measurement data widely used in routine physical examinations to achieve highly accurate prediction of an individual's biological age. Based on widely available and economical health examination data, and supplemented by refined gender- and age-group modeling strategies, this application improves the accuracy and reliability of aging prediction without the need for additional expensive testing methods, making it more conducive to expanding its application and promoting aging prediction as an indicator for daily health monitoring. Compared with existing technologies, this solution has lower costs, faster speed, and better predictive performance. MAE, coefficient of determination R², and correlation coefficient r all show better predictive accuracy compared to existing technologies.

[0324] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. The application of biomarkers in the preparation of reagent kits for predicting human aging levels, characterized in that, The biometrics include: Mean corpuscular volume, platelet count, hematocrit, red blood cell distribution width, monocyte percentage, mean corpuscular hemoglobin, neutrophil count, monocyte count, red blood cell count, platelet count, lymphocyte count, total cholesterol, triglyceride, low-density lipoprotein cholesterol, total protein, uric acid, urea, hemoglobin, alkaline phosphatase, albumin, alanine aminotransferase, glucose, aspartate aminotransferase, apolipoprotein A, or creatinine, as well as diastolic blood pressure, systolic blood pressure, height, weight, or body mass index of the sample population.

2. The application according to claim 1, characterized in that, The bioindicators include any one, two, three, four, five, six, seven, eight, nine, or ten of the following bioindicators: The levels of urea, alkaline phosphatase, albumin, alanine aminotransferase, apolipoprotein A, platelet-to-volume ratio, as well as systolic blood pressure, height, weight, and diastolic blood pressure.

3. The application according to claim 1, characterized in that, The biomarkers contain at least total cholesterol; The bioindicators may also optionally include: The diastolic blood pressure, systolic blood pressure, height, weight, or body mass index of the sample population, and / or The blood of the sample population contained any one or more of the following biological indicators: mean corpuscular volume, platelet count, hematocrit, red blood cell distribution width, monocyte percentage, mean corpuscular hemoglobin, neutrophil count, monocyte count, red blood cell count, platelet count, lymphocyte count, triglyceride level, low-density lipoprotein cholesterol level, total protein level, uric acid level, urea level, hemoglobin level, alkaline phosphatase level, albumin level, alanine aminotransferase level, glucose level, aspartate aminotransferase level, apolipoprotein A level, or creatinine level; Optionally, the biomarkers include: total cholesterol, mean corpuscular volume, urea content, alkaline phosphatase content, albumin content, alanine aminotransferase content, apolipoprotein A content, platelet-to-volume ratio, glucose content, aspartate aminotransferase content, or creatinine content in the blood of the sample population, as well as systolic blood pressure, height, weight, and diastolic blood pressure, or a combination thereof.

4. The application according to any one of claims 1-3, characterized in that, The biometrics include mean corpuscular volume (MCV), platelet count, hematocrit, red blood cell distribution width (RBF), monocyte percentage, mean corpuscular hemoglobin (MCH), neutrophil count, monocyte count, red blood cell count, platelet count, lymphocyte count, total cholesterol, triglyceride, low-density lipoprotein cholesterol, total protein, uric acid, urea, hemoglobin, alkaline phosphatase, albumin, alanine aminotransferase (ALT), glucose, aspartate aminotransferase (AST), apolipoprotein A, or creatinine, as well as diastolic blood pressure, systolic blood pressure, height, weight, and body mass index (BMI) of the sample population.

5. A method for constructing a predictive model for human aging levels, characterized in that, The construction method includes: Step a) Obtain the calendar age and biometrics of the sample population and establish a dataset of the sample population; wherein, the biometrics include the biometrics described in any one of claims 1-4. Step b) Use the dataset above to construct the model using machine learning training methods.

6. The construction method according to claim 5, characterized in that, The level of human aging includes biological age; Optionally, before training with the machine learning method, the dataset is divided into a training set and a test set, the training set is used to build a model, and the test set is used to verify the model's performance. Optionally, the machine learning training method includes any one or more of the following: XGBoost regression model, Ridge regression model, Lasso regression model, or Ealstic Net regression model; Optionally, the objective function of the XGBoost regression model is reg:squarederror; Optionally, the method for dividing the dataset into a training set and a test set includes: randomly splitting the sample population into the training set or the test set; Optionally, before the machine learning training, the dataset may undergo any one or more of the following processing: i) Missing value handling, including: calculating the average value of each of the biological indicators in the dataset, and using the average value to fill in the missing indicators in the dataset; ii) Standardization, including: converting all data for each of the biometrics in the dataset into relative values; Optionally, the standardization includes performing Z-score standardization on all data for each of the biological indicators in the dataset.

7. The construction method according to claim 5 or 6, characterized in that, Step b) further includes: dividing the dataset into three overlapping age segments according to the calendar age, and independently training a corresponding age segmentation sub-model for each age segment. For samples in the test set that fall into the overlapping age bands, the corresponding predicted values ​​are obtained by weighted summation of two adjacent age segment sub-models according to the linear weights of their distance from the boundary. Optionally, the three overlapping age groups include 0-A years, BC years, and D-100 years, where A≥B and C≥D; A includes 47-50, B includes 42-50, C includes 60-63, and D includes 57-60; Optionally, the three overlapping age groups include any one of the following groups: 0-48 years old, 42-63 years old, and 57-100 years old; 0-50 years old, 50-60 years old, and 60-100 years old; or 0-47 years old, 43-62 years old, and 58-100 years old; Optionally, step b) includes: dividing the dataset into a male training set and a female training set according to gender, and using the machine learning training to train male sub-models and female sub-models respectively; Optionally, before constructing the model using the training set, the samples are weighted based on the distribution of calendar ages in the training set. Optionally, the weighted weight = A ÷ (B × C). A is the total number of samples in the training set. B is the total number of unique calendar ages in the training set; C represents the number of times the calendar age appears in the training set.

8. The construction method according to any one of claims 5-7, characterized in that, Step b) includes: The sample population is divided into a training set and a test set. A mapping relationship between the biometrics and calendar age of the training set or the test set is established to obtain the training dataset and the test dataset, respectively. The machine learning method is used to build a model on the training dataset, including determining the optimal hyperparameters within the training dataset using five-fold cross-validation combined with random search, to obtain a preliminary model. The preliminary model is validated using the test dataset. Any one or more of the following parameters of the preliminary model are calculated and evaluated: mean absolute error, coefficient of determination, or Pearson correlation coefficient, to obtain the prediction model.

9. A predictive model for human aging levels, characterized in that, The model is obtained using the method for constructing a predictive model of human aging level as described in any one of claims 5-8.

10. A device for predicting the level of human aging, characterized in that, The device includes an input module and a prediction module. The input module is used to acquire the indicators of the sample to be tested and input the biological indicators into the prediction module; wherein, the biological indicators are the biological indicators in any of the applications described in claims 1-4; The prediction module is used to calculate the human aging level based on the bioindicators; wherein the prediction module contains a prediction model, and the prediction model is a model obtained by constructing a prediction model for human aging level according to any one of claims 5-8.

11. A method for predicting the level of human aging, characterized in that, The method includes: inputting the indicators of the sample to be tested into a prediction model, and obtaining the human aging level output by the prediction model; The prediction model is a model obtained using the method for constructing a prediction model for human aging level according to any one of claims 5-8.

12. A method for establishing a dataset, characterized in that, The method includes: obtaining the calendar age and / or metrics of the target sample from the dataset to obtain the dataset; Wherein, the biometrics include the biometrics in any of the applications described in claims 1-4; Optionally, the method further includes: i) Missing value handling, including: calculating the average value of each of the biometrics in the dataset, and using the average value to fill in the missing biometrics in the dataset; and / or ii) Standardization, including: converting all data for each of the biometrics in the dataset into relative values; Optionally, the standardization includes performing Z-score standardization on all data for each of the biological indicators in the dataset.

13. A training method for a predictive model of human aging level, characterized in that, The training method includes: obtaining the calendar age and indicators of the sample population, establishing a dataset of the sample population, wherein the biometrics include the biometrics in any one of the applications of claims 1-4; The model is constructed using machine learning methods on the dataset; Optionally, the machine learning method includes determining the optimal hyperparameters using five-fold cross-validation combined with random search within the training dataset to obtain the model; Optionally, before the machine learning training, the dataset may undergo any one or more of the following processing: i) Missing value handling, including: calculating the average value of each of the biometrics in the dataset, and using the average value to fill in the missing biometrics in the dataset; ii) Standardization, including: converting all data for each of the biomarkers in the dataset into relative values; Optionally, the standardization includes: performing Z-score standardization on all data for each of the biological indicators in the dataset; Optionally, the training method further includes: dividing the dataset into three overlapping age segments according to the calendar age, and independently training a corresponding age segment sub-model for each age segment; Optionally, the three overlapping age groups include 0-A years, BC years, and D-100 years, where A≥B and C≥D; A includes 47-50, B includes 42-50, C includes 60-63, and D includes 57-60. Optionally, the three overlapping age groups include any one of the following groups: 0-48 years old, 42-63 years old and 57-100 years old; 0-50 years old, 50-60 years old and 60-100 years old; or 0-47 years old, 43-62 years old and 58-100 years old; Optionally, the training method further includes: dividing the dataset into a male training set and a female training set according to gender, and using the machine learning training to train male sub-models and female sub-models respectively; Optionally, the training method further includes: before constructing the model using the training set, weighting the samples based on the distribution of calendar ages in the training set; Optionally, the weighted weight = A ÷ (B × C). A is the total number of samples in the training set. B is the total number of unique calendar ages in the training set; C represents the number of times the calendar age appears in the training set.

14. A computer device comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the steps of the method for constructing a predictive model of human aging level according to any one of claims 5 to 8, the method for predicting human aging level according to claim 11, the method for establishing a dataset according to claim 12, or the method for training a model for predicting human aging level according to claim 13.

15. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the method for constructing a predictive model of human aging level according to any one of claims 5 to 8, the method for predicting human aging level according to claim 11, the method for establishing a dataset according to claim 12, or the method for training a model for predicting human aging level according to claim 13.

16. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the method for constructing a predictive model of human aging level according to any one of claims 5 to 8, the method for predicting human aging level according to claim 11, the method for establishing a dataset according to claim 12, or the method for training a model for predicting human aging level according to claim 13.

Images