Human physiological age prediction method based on telomere length

Abstract

According to the human physiological age prediction method based on the telomere length, the aging degree of an individual is quantified by the telomere loss degree on the basis of the telomere length index of white blood cells in crowd blood, and meanwhile, a machine learning model for predicting the physiological age is constructed on the basis of the telomere length and the actual calendar age; the method is used for evaluating the aging condition of the human body relative to the real age. In the prediction process, firstly, the abnormal distribution value of telomere data at each age is identified and filtered, and in the machine learning model construction process, physiological age models are constructed for men and women at different age groups so as to accurately evaluate the health condition and the aging degree of individuals. Therefore, the human physiological age prediction method based on the telomere length has the advantages of being good in prediction effect and high in accuracy.

Description

Technical Field

[0001] The present invention relates to the technical field of machine learning, and particularly to a method for predicting human physiological age based on telomere length with good prediction effect. Background Art

[0002] Telomeres are a small segment of DNA-protein complex existing at the ends of linear chromosomes in eukaryotes. Together with telomere-binding proteins, they form a special "cap" structure, whose function is to maintain the integrity of chromosomes and control the cell division cycle. When humans are just born, the telomere length is about 10,000 - 15,000 base pairs. During cell division, due to the end replication problem, one of the two DNA strands cannot be fully replicated, resulting in DNA shortening. Telomeres will lose several base pairs, shortening by about 30 - 200 base pairs per year. When the telomere shortens to a critical length, the cell will no longer be able to divide, triggering DNA damage, leading to cell aging and death, and causing human aging and various aging diseases. This is also known as the Hayflick limit, that is, fetal cells have limited replication potential and can only replicate 40 - 60 times. However, in addition to cell division, there are some other factors that affect the rate of telomere shortening, including genetic factors, lifestyle, environmental stress, etc. For example, the inhibition of telomerase activity causes telomeres to not be supplemented and extended, thus accelerating the rate of telomere shortening; mutations in telomere maintenance genes (such as TERT, TERC, DKC, etc.) will lead to accelerated telomere shortening; oxidative stress will damage the function of telomerase, resulting in abnormal repair of telomeres and thus accelerating shortening; excessive stress, obesity, bad lifestyle (such as smoking, alcoholism, lack of exercise) and environmental toxins, etc. will all accelerate telomere shortening; chronic inflammation and infections will increase oxidative stress, and thus accelerate telomere shortening. Therefore, quantifying the degree of telomere loss and the rate of shortening can evaluate whether the telomere function is normal and the factors that may affect the telomere rate.

[0003] Refer to the article published by Rossiello F et al. in *Nature Cell Biology*: Telomere dysfunction in ageing and age-related diseases. Telomere shortening activates cell cycle inhibitors (such as p16Ink4a and p21), leading to cell senescence. Moreover, too rapid telomere shortening increases the risk of various age-related diseases. For example, too rapid telomere shortening causes cells to enter a senescent state, resulting in decreased cell function and weakened tissue regeneration ability, thus accelerating cell senescence. At the same time, it activates the inflammatory pathway, releases inflammatory factors, exacerbates tissue damage, and promotes the occurrence of age-related diseases such as idiopathic pulmonary fibrosis (IPF), atherosclerosis, liver cirrhosis, type 2 diabetes, Alzheimer's disease, Parkinson's disease, and osteoporosis. It also leads to decreased stem cell function and the inability to effectively repair damaged tissues, thereby causing stem cell exhaustion. Therefore, by predicting the physiological age of the human body based on the actual age and telomere length of an individual, monitoring the aging speed of the individual, and customizing an individual's health management intervention plan, aging can be effectively delayed and the physical health level can be improved.

[0004] Refer to the article published by López-Otín C et al. in *Cell*: Hallmarks of aging: An expanding universe. Telomere attrition is one of the twelve hallmarks of aging, which can evaluate the physiological aging degree of an individual. The longer the telomere, the younger the individual; the shorter the telomere, the more senescent the individual. The telomere length of disease patients is significantly higher than that of healthy individuals. Unhealthy lifestyle habits (such as smoking, drinking, staying up late, etc.) are also associated with shorter telomere lengths. Therefore, quantifying the relationship between telomere attrition and the degree of aging provides reliable evidence for human health management intervention.

[0005] Currently, for the method of predicting physiological age based on telomere length, abnormal data caused by factors such as sample quality, stability, and accuracy of the technology for detecting telomere length are not filtered, which affects the learning of the model. Moreover, it does not distinguish between genders and does not explore the differences in different age groups. The established model does not have good generalization and adaptation abilities among different age groups and between men and women. Because there are significant differences in body metabolism ability, self-repair ability, and immune ability between different age groups and genders, a single overall model is difficult to accurately reflect the specific situation. Therefore, it is necessary to propose an improvement to overcome the defects of the existing technology. Summary of the Invention

[0006] The purpose of the present invention is to solve the problems in the existing technology and provide a method for predicting human physiological age based on telomere length.

[0007] The technical solution of the present invention is: A method for predicting human physiological age based on telomere length, comprising the following steps: S1. Telomere data collection and filtering: Collect telomere detection data and calendar age; group the telomere detection data, and filter out abnormal data in the grouped data: group the telomere detection data by age, calculate the interquartile range (IQR, the difference between the upper quartile and the lower quartile) of each age group, and then use the sum of the upper quartile value and 2.5 times the IQR as the upper threshold, and the difference between the lower quartile and 2.5 times the IQR as the lower threshold, and filter out the data higher than the upper threshold and lower than the lower threshold in each age group data as outliers; S2. Among the data after filtering out the outliers, divide them into a male data set and a female data set according to gender, and split the male data set and the female data set into male data subsets and female data subsets of different age groups according to age groups respectively; S3. Divide each male data subset and female data subset into a training set and a test set according to a ratio respectively; S4. Use a machine learning model to build a model in the training set respectively, and verify it in the test set after the model training is completed; S5. Perform human physiological age prediction correction: perform age correction according to the regression to the mean effect, use the calendar age as the independent variable and the original age difference as the dependent variable for linear regression; original age difference = slope * calendar age + intercept + residual; where the residual is the age difference after correction, that is, remove the part related to the calendar age from the original age difference, and the formula is: corrected age difference = original age difference – (slope * original age difference + intercept); where the slope and intercept are calculated, and the specific calculation method is: perform linear fitting on the original age difference and the calendar age in the training sets of the male data set and the female data set respectively to obtain the slope and intercept in the male data set and the female data set; then, substitute the original difference between the predicted physiological age and the calendar age in the test set into the formula to obtain the corrected age difference in the test set.

[0008] As a preferred technical solution, after step S5, the following steps are further included: S6. Calculate the aging index: The aging index calculation formula is: aging index = corrected age difference / calendar age.

[0009] As a preferred technical solution, after step S5, the following steps are further included: S7. Calculate the MAE value (mean absolute error) of the models built based on the data subsets of different age groups for men and women, select the model with the smallest MAE value in the models built based on the data subsets of different age groups for men and women respectively, perform steps S3 to S5, and perform physiological age prediction for this age group.

[0010] As a preferred technical solution, the telomere detection data in step S1 is the absolute telomere length.

[0011] As a preferred technical solution, the ratio of the data volume of the training set to the test set in step S3 is 2:1, 3:1, 4:1 or 5:1.

[0012] As a preferred technical solution, the following steps are further included between step S3 and step S4: S8. Perform normalization processing on the data of the training set; the normalization method can be any one of Z-score standardization, maximum-minimum normalization, median normalization, and maximum absolute value normalization.

[0013] As a preferred technical solution, the machine learning model in step S4 is one of Linear Regression model LinearRegression, SVR, LightGBM, CatBoost, Ridge, RandomForest, XGBoost, Huber, Lasso, ElasticNet, Neural Network.

[0014] As a further preferred technical solution, the machine learning model in step S4 adopts the Linear Regression model LinearRegression.

[0015] A method for predicting human physiological age based on telomere length according to the present invention is based on the telomere length index of white blood cells in human blood, quantifies the degree of individual aging by the degree of telomere loss, and simultaneously constructs a machine learning model for predicting physiological age based on telomere length and actual calendar age to evaluate the aging status of the human body relative to its true age. During the prediction process, first, the abnormal distribution values of telomere data at each age are identified and filtered. During the construction of the machine learning model, the present invention constructs physiological age models for men and women in different age groups respectively to accurately evaluate the health status and aging degree of individuals. Therefore, the method for predicting human physiological age based on telomere length according to the present invention has the advantages of good prediction effect and high accuracy. Description of the Drawings

[0016] Figure 1 It is a flow chart of the specific implementation manner of a method for predicting human physiological age based on telomere length according to the present invention; Figure 2 It is the prediction result of the male physiological age model before calibration in the test set of a method for predicting human physiological age based on telomere length in this embodiment; Figure 3 It is the difference between the physiological age and the true age of the male before calibration in a method for predicting human physiological age based on telomere length in this embodiment; Figure 4 It is the prediction result of the male physiological age after calibration through step S5 in a method for predicting human physiological age based on telomere length in this embodiment; Figure 5 The difference between the physiological age after correction for males and the true age in a method for predicting human physiological age based on telomere length in this embodiment; Figure 6 The prediction result before correction of the female physiological age model in a method for predicting human physiological age based on telomere length in this embodiment in the test set; Figure 7 The difference between the physiological age before correction for females and the true age in a method for predicting human physiological age based on telomere length in this embodiment; Figure 8 The prediction result of the female physiological age after correction through step S5 in a method for predicting human physiological age based on telomere length in this embodiment; Figure 9 The difference between the physiological age after correction for females and the true age in a method for predicting human physiological age based on telomere length in this embodiment. Detailed implementation manners

[0017] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Apparently, the described embodiments are some but not all of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without making creative efforts fall within the scope of protection of the present invention.

[0018] The terms used in the embodiments of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The singular forms "a", "the" and "said" used in the embodiments of the present invention and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. "Multiple" generally includes at least two, but does not exclude the case of including at least one.

[0019] It should be understood that the term " / and" used herein is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the character " / " herein generally represents an "or" relationship between the associated objects before and after.

[0020] Depending on the context, as used herein, the words "if" and "when" may be interpreted as "when" or "while" or "in response to determining" or "in response to detecting". Similarly, depending on the context, the phrase "if it is determined" or "if (a stated condition or event) is detected" may be interpreted as "when it is determined" or "in response to determining" or "when (the stated condition or event) is detected" or "in response to detecting (the stated condition or event)".

[0021] It should also be noted that the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, such that a commodity or system comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such commodity or system. Without further limitation, an element defined by the statement "comprising one..." does not exclude the presence of additional identical elements in the commodity or system comprising said element.

[0022] As Figure 1 Shown is a specific implementation manner of a method for predicting human physiological age based on telomere length according to the present invention. A method for predicting human physiological age based on telomere length in this embodiment includes the following steps: S1. Telomere data collection and filtering: Collect telomere detection data and calendar age; group the telomere detection data, and filter out abnormal data in the grouped data: group the telomere detection data by age, calculate the interquartile range (IQR, the difference between the upper quartile and the lower quartile) of each age group, and then use the sum of the upper quartile value and 2.5 times the IQR as the upper threshold, and the difference between the lower quartile and 2.5 times the IQR as the lower threshold. Filter out the data in each age group that is higher than the upper threshold and lower than the lower threshold as outliers; S2. Among the data after filtering out the outliers, divide them into a male data set and a female data set according to gender, and split the male data set and the female data set into male data subsets and female data subsets of different age groups according to age groups respectively; S3. Divide each male data subset and female data subset into a training set and a test set according to a ratio respectively; S4. Use a machine learning model to build a model in the training set respectively, and verify it in the test set after the model training is completed; S5. Perform correction for predicting human physiological age: Perform age correction according to the regression to the mean effect, use the calendar age as the independent variable and the original age difference as the dependent variable for linear regression; Original age difference = slope * calendar age + intercept + residual; Among them, the residual is the age difference after correction, that is, the part related to the calendar age is removed from the original age difference. The formula is: Corrected age difference = Original age difference – (Slope * Original age difference + Intercept); Among them, the slope and intercept are calculated. The specific calculation method is as follows: Perform linear fitting on the original age difference and calendar age in the training sets of the male dataset and the female dataset respectively to obtain the slope and intercept in the male dataset and the female dataset; Then, substitute the original difference between the predicted physiological age and the calendar age in the test set into the formula to obtain the corrected age difference in the test set.

[0023] In this embodiment, after step S5, the following steps are further included: S6. Calculate the senescence index: The formula for calculating the senescence index is: Senescence index = Corrected age difference / Calendar age.

[0024] In this embodiment, after step S5, the following steps are further included: S7. Calculate the MAE value (Mean Absolute Error, representing the average of the absolute errors between the predicted value and the observed value) of the models built based on the data subsets of different age groups for men and women. Select the model with the smallest MAE value among the models built for the data subsets of different age groups for men and women, and perform steps S3 to S5 to predict the physiological age of this age group. In this embodiment, select the age group with the smallest MAE value among the models for different age groups of men and women. It is found that the MAE is the smallest when building models based on telomere data in the age range of 29 - 70 years old for both men and women, that is, the performance of the built models is the best. Then, build models respectively according to steps 3 to 6 based on the male and female data in the age range of 29 - 70 years old to predict the physiological age of men and women in this age group. Step S7 of this embodiment is after step S6. In the actual application process, steps S6 and S7 can be selectively set according to actual needs, and neither affects the embodiment of the advantages of the present invention.

[0025] In this embodiment, the telomere detection data in step S1 is the absolute telomere length.

[0026] Specifically, in this embodiment, the telomere data is selected from the telomere database, and 3000 samples are collected, including 1500 male individual samples with an age distribution range of 2 - 80 years old and 1500 female individual samples with an age distribution range of 5 - 98 years old.

[0027] In step S2 of this embodiment, in step S3, "split the male dataset and the female dataset into male data subsets and female data subsets of different age groups according to age groups" is specifically to split them into datasets of different age groups by combining them in sequence between the ages of 1 - 30 years old and 70 - 100 years old, such as 1 - 70 years old, 1 - 71 years old, etc.

[0028] In this embodiment, the ratio of the data volume of the training set to the test set in step S3 is 3:1. In practical applications, the ratio of the training set to the test set can be 4:1, 3:1, 2:1 or 5:1 according to needs, which will not affect the implementation of the solution of the present invention.

[0029] In the actual application process, the following steps are further included between step S3 and step S4: S8. Perform normalization processing on the data of the training set; the normalization method can be any one of Z-score standardization, maximum-minimum normalization, median normalization, and maximum absolute value normalization. Similar effects can be achieved after the data normalization processing. Since the telomere physiological age model has only one index of telomere absolute length, and there are very few outliers, and the outliers have been filtered, step S8 is not set in this embodiment to avoid increasing complexity and affecting prediction accuracy.

[0030] In practical applications, the machine learning model in step S4 is one of LinearRegression, SVR, LightGBM, CatBoost, Ridge, RandomForest, XGBoost, Huber, Lasso, ElasticNet, NeuralNetwork, and similar effects can be achieved. Based on the specific practice of the modeling process of telomere absolute length data, the linear regression model LinearRegression is adopted in this embodiment, which has the highest accuracy and the best stability between men and women.

[0031] The specific implementation method of a human physiological age prediction method based on telomere length in this embodiment, taking the linear regression model LinearRegression as an example, is as follows: Import of python packages: import pandas as pd import numpy as np from sklearn.model_selection import train_test_split from sklearn.linear_model import LinearRegression from sklearn.metrics import mean_squared_error, mean_absolute_error import matplotlib.pyplot as plt from sklearn.decomposition import PCA from sklearn.preprocessing import StandardScaler Set the font to enable normal display of Chinese plt.rcParams['font.sans-serif'] = ['SimHei'] Correctly display negative signs plt.rcParams['axes.unicode_minus'] = False Read in the modeling data - telomere detection data of healthy people. The data columns to be used are TL (absolute telomere length) and Age (actual calendar age) df = pd.read_excel('Telomere.xlsx') Group by age grouped = df.groupby('Age') Define a function to detect and remove outliers for each age group def remove_outliers_grouped(group): q1 = group['TL'].quantile(0.25) q3 = group['TL'].quantile(0.75) iqr = q3 - q1 lower_bound = q1 - 2.5 * iqr upper_bound = q3 + 2.5 * iqr return group[(group['TL']>= lower_bound)&(group['TL']<= upper_bound)] Define a function to calculate the MAE value def mean_absolute_error(y_true, y_pred): return np.mean(np.abs(y_true - y_pred)) Define a function for linear regression modeling def linear_regression_model(X,y): Split the training set and test set X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=42) Train the model and predict the physiological age model = LinearRegression() model.fit(X_train, y_train) y_pred = model.predict(X_test) return y_test, y_pred,X_train,y_train Define a function to correct the age difference between the predicted physiological age and the actual calendar age def adjust_age_gap(y_pred, y_test,X_train,y_train): age_gap = y_pred - y_test model = LinearRegression() model.fit(X_train, y_train) y_train_pred = model.predict(X_train) age_gap_train = y_train_pred - y_train model = LinearRegression() model.fit(pd.DataFrame(y_train), age_gap_train.values) intercept = model.intercept_ coef = model.coef_[0] adjusted_age_gap = age_gap - (intercept + coef * y_test) adjusted_pred = y_test + adjusted_age_gap return adjusted_pred, adjusted_age_gap Apply the function to remove outliers df_cleaned = grouped.apply(remove_outliers_grouped).reset_index(drop=True) Generate all possible age combinations using itertools.product age_range_1 = range(1, 30) age_range_2 = range(70, 100) age_combinations = list(itertools.product(age_range_1, age_range_2)) Initialize an empty DataFrame to store the results dataframes = [] results_df = pd.DataFrame(columns=['Gender', 'Model', 'MSE', 'MAE']) For data of different age groups, model separately for male and female, predict the physiological age, and calculate MAE for combination in age_combinations: df_healthy_1 = df_cleaned[df_cleaned['Age']<= combination[1]] df_healthy_1 = df_healthy_1[df_healthy_1['Age']>= combination[0]] Perform operations for each Gender for gender, group in df_healthy_1.groupby('Gender'): Build a model y_test, y_pred,X_train,y_train = linear_regression_model(group[['TL']], group['Age']) Calibrate the age difference between the predicted physiological age and the actual calendar age adjusted_pred, adjusted_age_gap = adjust_age_gap(y_pred, y_test,X_train,y_train) Calculate the MAE value to evaluate the quality of the model mae = mean_absolute_error(y_test, adjusted_pred) results_df = pd.concat([results_df, pd.DataFrame({ 'AgeRange': [combination], 'Gender': [gender], 'MAE': [mae] })], ignore_index=True) Evaluate the correlation and significance between the predicted physiological age and the true age after correction correlation_coefficient, p_value = pearsonr(y_test,adjusted_pred) Output the MAE results for each age group of men and women, and visualize them as two line charts for men and women. It is found that the MAE values of the models for the male and female datasets are the smallest in the age range of 29 - 70 results_df.to_excel('MAE.xlsx', index=False) Based on the dataset of the age range of 29 - 70, build a model to predict the physiological age, calculate the telomere senescence index, and save the model Here, take the female data as an example On the basis of filtering out the outliers, select the female data of Chinese healthy people in the age range of 29 - 70 data = df_cleaned[df_cleaned['Age']<= 70] data = data[data['Age']>= 29] data = data[data['Gender']=='female']] Build a model y_test, y_pred,X_train,y_train = linear_regression_model(data[['TL']], data['Age']) Age difference between the corrected predicted physiological age and the true calendar age adjusted_pred, adjusted_age_gap = adjust_age_gap(y_pred, y_test,X_train,y_train) Calculate the telomere senescence index telomere_index = adjusted_age_gap / y_test Calculate the MAE value to evaluate the quality of the model mae = mean_absolute_error(y_test, adjusted_pred) Evaluate the correlation and significance between the predicted physiological age and the true age after correction correlation_coefficient, p_value = pearsonr(y_test,adjusted_pred) In this embodiment, the prediction performance of the male telomere physiological age model in the test set is shown in Figure 2 , and the Pearson correlation coefficient between the predicted physiological age and the calendar age reaches 0.44; the prediction performance of the female telomere physiological age model in the test set is shown in Figure 6 , and the correlation coefficient between the predicted female physiological age and the calendar age reaches 0.48, both showing relatively good modeling effects. However, both models show the "regression to the mean effect" in statistics, that is, for young individuals, the predicted biological age is often higher than the calendar age; for older individuals, the predicted biological age is often lower than the calendar age. The regression to the mean effects of men and women can be clearly seen in Figure 3 and Figure 7 respectively, and the mean absolute error MAE values of the models are 8.55 and 8.04 respectively

[0032] Use the method in step S5 to correct the difference between the physiological age and the calendar age. The age differences after correction for men and women are shown in Figure 5 and Figure 9 respectively. The MAE after correction for men is 3.99, and the MAE after correction for women is 3.44, achieving very good results. And the prediction performances of the male and female telomere physiological age models after correcting the age difference in the test set are shown in Figure 4 and Figure 8 respectively, and the correlation coefficients between the predicted physiological age and the calendar age reach 0.92 and 0.93 respectively, showing a great improvement in performance compared with before correction

[0033] Additionally, the formula in step 6 is used to calculate the telomere senescence index. The senescence index measures the degree of change in physiological age relative to calendar age and can evaluate the senescence rate. If the senescence index is greater than 0, it indicates that the individual's senescence speed is higher than the normal speed, and the senescence speed is faster; if the senescence index is less than 0, it represents that the individual's senescence speed is slower than the normal speed, and the senescence speed is slower.

[0034] A method for predicting human physiological age based on telomere length in this embodiment identifies and filters abnormal distribution values at each age, avoiding the influence of a very small number of abnormal data caused by force majeure factors such as sample quality, the stability and accuracy of the technology for detecting telomere length, and human factors on modeling; separate models are built for men and women, which is more accurate and will not ignore the obvious individual differences between men and women; the model performances after modeling with telomere data of different age groups are compared, and the age group with the best performance (the smallest mean absolute error MAE value) is selected. It is found that the models based on telomere data of the age group from 29 to 70 years old have the best performance for both men and women. Therefore, predicting the physiological age of humans in this age group based on the telomere database in this age group has higher accuracy and stability between men and women; the senescence index indicator is added, which can evaluate the senescence rate of individuals.

[0035] A method for predicting human physiological age based on telomere length of the present invention quantifies the senescence degree of an individual based on the telomere length index of white blood cells in human blood, and at the same time constructs a machine learning model for predicting physiological age based on telomere length and actual calendar age to evaluate the senescence status of the human body relative to its own true age. During the prediction process, first, the abnormal distribution values of telomere data at each age are identified and filtered. During the construction of the machine learning model, the present invention constructs physiological age models for men and women at different age groups respectively to accurately evaluate the health status and senescence degree of individuals. Therefore, a method for predicting human physiological age based on telomere length of the present invention has the advantages of good prediction effect and high accuracy.

[0036] In this embodiment, separate models are built for men and women. A small amount of abnormal data is filtered based on the data distribution at different ages. At the same time, the accuracy of predicting physiological age with data of different age groups and the stability between men and women are compared. Finally, the age group with uniform data distribution, significant trend, accurate model prediction ability and little difference between men and women is determined. On this basis, a machine learning model for predicting physiological age based on telomere length is constructed to accurately evaluate the senescence degree and senescence rate of individuals, providing a scientific basis for realizing precise health management.

[0037] The above are only the preferred embodiments of the present invention and are not intended to limit the scope of implementation of the present invention. That is, all equivalent changes and modifications made to the content within the scope of the patent application of the present invention shall fall within the technical scope of the present invention.

Claims

1. A method for predicting human physiological age based on telomere length, characterized in that: The following steps are involved: S1. Telomere data collection and filtering: collect telomere detection data and calendar age; group the telomere detection data, and filter abnormal data in the grouped data: group the telomere detection data by age, calculate the interquartile range (IQR, the difference between the upper quartile and the lower quartile) of each age group, and then use the sum of the upper quartile value and 2.5 times the IQR as the upper threshold, and the difference between the lower quartile and 2.5 times the IQR as the lower threshold, and filter out the data in each age group that is higher than the upper threshold and lower than the lower threshold as abnormal values; S2. The data from which outliers are filtered out are divided into male and female data sets according to gender, and the male and female data sets are split into male data subsets and female data subsets of different age groups according to age groups; S3, dividing each male data subset and female data subset into a training set and a test set according to the proportion; S4. Use machine learning models to build models in the training set, and verify them in the test set after the model training is completed; S5. Perform prediction and correction of human physiological age: perform age correction based on the regression center effect, use calendar age as the independent variable and original age difference as the dependent variable for linear regression; Original age difference = slope * calendar age + intercept + residual; Among them, the residual is the age difference after correction, that is, the part related to calendar age is removed from the original age difference. The formula is: Corrected age difference = original age difference – (slope * original age difference + intercept); The slope and intercept are calculated by linear fitting the original age difference and calendar age in the training sets of the male and female data sets, respectively, to obtain the slope and intercept in the male and female data sets; Then, in the test set, the original difference between the predicted physiological age and the calendar age is substituted into the formula to obtain the corrected age difference in the test set.

2. The method for predicting human physiological age based on telomere length according to claim 1, characterized in that: After step S5, the following steps are also included: S6. Calculate the aging index: The calculation formula for the aging index is: Aging index = corrected age difference / calendar age.

3. The method for predicting human physiological age based on telomere length according to claim 1, characterized in that: After step S5, the following steps are also included: S7. Calculate the MAE value (mean absolute error) of the model built based on the data subsets of different age groups of men and women, select the model with the smallest MAE value from the models built based on the data subsets of different age groups of men and women, and perform steps S3 to S5 to predict the physiological age of the age group.

4. The method for predicting human physiological age based on telomere length according to claim 1, characterized in that: The telomere detection data in step S1 is the absolute length of the telomere.

5. The method for predicting human physiological age based on telomere length according to claim 1, characterized in that: In step S3, the ratio of the data volume of the training set to that of the test set is 2:1, 3:1, 4:1 or 5:

1.

6. The method for predicting human physiological age based on telomere length according to claim 1, characterized in that: The following steps are also included between step S3 and step S4: S8. Perform homogenization on the data of the training set; the homogenization method can be any one of Z-score normalization, maximum and minimum value normalization, median normalization, and maximum absolute value normalization.

7. The method for predicting human physiological age based on telomere length according to claim 1, characterized in that: In step S4, the machine learning model is one of the linear regression models LinearRegression, SVR, LightGBM, CatBoost, Ridge, RandomForest, XGBoost, Huber, Lasso, ElasticNet, and Neural Network.

8. The method for predicting human physiological age based on telomere length according to claim 7, characterized in that: In step S4, the machine learning model adopts the linear regression model LinearRegression.

Images