An algorithm construction method and system of a lightweight functional biological age indicator

Abstract

The application provides a lightweight functional biological age index algorithm construction method and system, and relates to the technical field of biological age, which comprises the following steps: obtaining multi-dimensional functional index data of an individual; homogenizing each functional index obtained, so that the higher the value of all indexes is, the more serious the biological aging degree is; filling the missing values in each index after homogenization by using a gender-based median interpolation logic; standardizing each index after filling according to gender; performing weighted linear combination on each standardized index to construct a functional age score; and calibrating the functional age score to an age scale to obtain a lightweight biological age index in years. The application constructs a lightweight biological age algorithm through non-invasive functional indexes, has a performance comparable to that of a DNA methylation clock and a very low cost, and can realize popular application of large-scale community population aging assessment.

Description

Technical Field

[0001] This invention relates to the field of biological age technology, specifically to an algorithmic construction method and system for a lightweight functional biological age index. Background Technology

[0002] Among existing technologies, biological age models used to assess an individual's biological aging primarily fall into two categories: one is based on the epigenetic clock using DNA methylation molecular markers, and the other is based on biological age assessment models constructed using clinical biochemical blood indicators (e.g., albumin, creatinine, etc.). These models have been confirmed in academic research to be associated with mortality risk and various age-related diseases.

[0003] However, existing biological age assessment models still have significant shortcomings in practical applications. On the one hand, these models all rely on high-throughput gene sequencing or specialized biochemical laboratory testing, requiring expensive and invasive equipment (such as blood sampling), resulting in extremely high costs per test and making it difficult to widely implement in scenarios such as grassroots community screening, resource-scarce areas, or home health monitoring. On the other hand, existing algorithms often employ complex multiple imputation logic when dealing with missing values ​​commonly found in real-world data, which not only increases the computational burden but also affects the robustness of the model under incomplete data conditions. Furthermore, existing models lack systematic standardization in handling physiological differences and indicator dimensional differences between genders during their construction, limiting the comparability of assessment results across populations. In addition, the weight allocation of existing models is mostly based on empirical settings, lacking a direct risk-driven association with aging endpoint events (such as mortality risk), and the output results are mostly abstract scores, making it difficult to intuitively reflect the degree of aging deviation in an individual.

[0004] Therefore, there is an urgent need for a low-cost, non-invasive, computationally efficient, and lightweight bio-age assessment solution with clear risk prediction capabilities to address the aforementioned technical problems of existing technologies in large-scale population applications. Summary of the Invention

[0005] To address the problems in related technologies, this invention provides an algorithmic construction method for a lightweight functional biological age index.

[0006] To achieve the above objectives, the technical solution adopted by the present invention includes:

[0007] A lightweight algorithm for constructing a functional biological age indicator includes the following steps:

[0008] Acquire multidimensional functional index data of individuals, including grip strength and walking speed reflecting muscle and motor function, cognitive score reflecting cognitive state, self-rated health status reflecting the individual's overall perception, and body mass index and waist circumference reflecting metabolic and morphological characteristics.

[0009] The obtained functional indicators are homogenized so that the higher the value of all indicators, the more severe the degree of biological aging.

[0010] For missing values ​​in each indicator after homogenization, a gender-based median imputation logic is used to fill them.

[0011] The indicators after filling were standardized according to gender to eliminate the influence of dimensional differences caused by physiological differences between genders.

[0012] The standardized indicators are weighted and linearly combined to construct a functional age score, where the weight of each indicator is determined based on all-cause mortality risk.

[0013] The functional age score was calibrated to an age scale to obtain a lightweight biological age index in years.

[0014] Optionally, the alignment process includes:

[0015] For indicators such as grip strength, walking speed, cognitive score, and self-rated health status, the higher the original value, the better the health status. Through reverse encoding logic, the indicators are transformed so that the higher the value, the worse the health function or the more severe the degree of biological aging.

[0016] Optionally, the gender-based median imputation logic is as follows: for each indicator with missing values, the median value of that indicator in the same gender group in the development queue is used to fill the missing values.

[0017] Optionally, the standardization process by gender is performed by gender. Score standardization, for each individual and indicators The standardized score is calculated using the following formula:

[0018]

[0019] In the formula, Represents an individual The Standardized indicators Fraction, Represents an individual The Each indicator observation value, and These represent the genders in the development queue. The participants The mean and standard deviation of each indicator.

[0020] Optionally, the weighted linear combination of the functional age scores is constructed according to the following formula:

[0021]

[0022] In the formula, Represents an individual Functional age rating, Indicates the first The weights corresponding to each indicator This indicates the total number of indicators included.

[0023] Optionally, the weight of each indicator The Cox proportional hazards model was obtained by fitting a multivariate Cox proportional hazards model to the development cohort. This model simultaneously incorporates all indicators, with all-cause mortality as the outcome event. The weights of each indicator in the Cox proportional hazards model are as follows: The value is taken as the log-hazard ratio of the indicator in the Cox model.

[0024] Optionally, calibrating the functional age score to an age scale specifically includes:

[0025] In the development cohort, a linear regression model was fitted with chronological age as the dependent variable and functional age score as the independent variable to obtain the intercept and slope. These intercept and slope were then used to map an individual's functional age score to a lightweight biological age in years, specifically calibrated according to the following formula:

[0026]

[0027] In the formula, Represents an individual The age of lightweight organisms Show individual i's functional age score, This represents the intercept of the linear regression model. This represents the slope of the linear regression model.

[0028] Optionally, the algorithm construction method for the lightweight functional biological age index further includes the step of calculating a lightweight biological aging acceleration index:

[0029] Calculate the individual according to the following formula. Lightweight biological aging acceleration value:

[0030]

[0031] In the formula, Represents an individual The lightweight biological aging acceleration value, Represents an individual Actual age;

[0032] When the acceleration value is positive, it indicates that the individual's biological aging rate is faster than that of their peers.

[0033] Optionally, the algorithmic construction method for the lightweight functional biological age indicator further includes the step of constructing a community assessment framework:

[0034] The lightweight biological age acceleration value is compared with a preset risk stratification threshold to generate a graded assessment result and corresponding intervention recommendations, forming a closed loop from algorithm output to community health management application; wherein, the risk stratification threshold has three levels: ≤-5 years, -5 to +5 years, and ≥+5 years.

[0035] According to a second aspect of the present invention, a lightweight functional biometrics algorithm construction system is also provided, for executing the lightweight functional biometrics algorithm construction method described in any of the technical solutions of the first aspect of the present invention, the system comprising:

[0036] The data acquisition module is used to acquire multidimensional functional indicator data of an individual;

[0037] The homogenization module is used to homogenize the acquired functional indicators so that the higher the value of all indicators, the more severe the degree of biological aging.

[0038] The missing value imputation module is used to fill in the missing values ​​in each indicator after homogenization using gender-based median imputation logic.

[0039] The standardization module is used to standardize the filled indicators by gender to eliminate the influence of dimensional differences caused by physiological differences between genders.

[0040] The functional age score construction module is used to construct a functional age score by weighted linear combination of standardized indicators, wherein the weight of each indicator is determined based on all-cause mortality risk.

[0041] The bio-age calibration module is used to calibrate the functional age score to an age scale, resulting in a lightweight bio-age index in years.

[0042] Beneficial effects:

[0043] 1. First, this invention boasts extremely high accessibility and extremely low implementation costs. Specifically, this invention requires no blood samples, biochemical testing, or high-throughput sequencing equipment; only simple tools such as a grip strength meter, measuring tape, and stopwatch, along with a standardized questionnaire, are needed to complete all data collection. Compared to DNA methylation clocks (which have high costs per test) and clinical biochemical indicator models (which rely on specialized laboratories and blood collection procedures), the implementation cost of this invention is reduced to a level affordable for ordinary community screening. This fundamentally breaks through the application bottleneck of bio-age assessment in resource-scarce areas, primary healthcare institutions, and home settings, making it feasible and affordable for large-scale global promotion.

[0044] Second, this invention exhibits strong robustness and generalization ability across populations and ethnicities. Specifically, the technical solution of this invention has been developed and validated in five national cohorts with independent population representation, including CHARLS (n=7308) in China, ELSA (n=7023) in the UK, KLoSA (n=8930) in South Korea, MHAS (n=16178) in Mexico, and NHANES (n=10280) in the US, with a total sample size exceeding 49,000. Without changing the algorithm parameters (indicator definition, standardization coefficients, weights, calibration intercept, and slope), each cohort stably reproduced the significant association between the lightweight biological aging acceleration indicator and all-cause mortality risk. This demonstrates that this invention has robust generalization ability and portability across countries with different ethnicities, cultural backgrounds, healthcare systems, and economic development levels, overcoming the technical limitations of previous models that required recalibration when constructed in a single population and transferred across populations.

[0045] Third, this invention demonstrates strong predictive power for all-cause mortality risk. Specifically, based on a Cox weighting mechanism driven by all-cause mortality risk, the resulting lightweight biological aging acceleration index exhibits highly sensitive predictive power for mortality. Experimental data show that for every 5-year increase in the lightweight biological aging acceleration value, the all-cause mortality risk increases by 30% to 89% in various national cohorts (HR range: 1.30–1.89, P<0.001 for all cohorts). This predictive power is comparable to existing models such as PhenoAge, which rely on expensive blood indicators, but the data acquisition cost of this invention is significantly lower. Furthermore, restricted cubic spline analysis shows that this association exhibits a monotonically increasing dose-response relationship (or approximately linear relationship) in all cohorts, and the highest quartile in the quartile analysis shows a mortality risk increase of up to 6 times compared to the lowest quartile (MHAS cohort: HR=6.01, 95% CI: 4.59–7.87), confirming the sensitivity and stability of this invention as an early warning tool for mortality risk.

[0046] Fourth, this invention demonstrates a significant correlation with the risk of multiple coexisting diseases and chronic diseases. Specifically, in addition to mortality risk, the lightweight biological aging acceleration index of this invention is also highly correlated with the risk of developing multiple chronic diseases. Experimental results show that for every 5-year increase in the lightweight biological aging acceleration value, the odds ratio (OR) for multiple coexisting diseases (≥2 chronic diseases) ranges from 1.13 to 7.84 in different cohorts; it also shows a significant association for specific chronic diseases, such as stroke (ELSA cohort OR=2.22, 95%CI: 2.05-2.39) and mental illness. Furthermore, in cohorts where frailty can be assessed, the lightweight biological aging acceleration value is strongly associated with frailty (assessed by the continuous frailty index) (KLoSA cohort: OR=7.29, 95%CI: 6.56-8.11). These results indicate that this invention is not only effective for the mortality endpoint but also comprehensively reflects the decline in the function of multiple systems in an individual, providing a powerful quantitative tool for community-based multi-disease screening and stratified health management.

[0047] Fifth, this invention demonstrates performance comparable to or even superior to the DNA methylation "gold standard." Specifically, in a direct comparative analysis with 12 epigenetic clocks based on DNA methylation, the per-standard-deviation mortality hazard ratio (HR=1.27, 95% CI: 1.21-1.34) of the lightweight biological aging acceleration index produced by this invention is comparable to PhenoAge acceleration (HR=1.25) and GrimAgeMort acceleration (HR=1.62), and significantly superior to methylation-based acceleration indices such as WeidnerAge, LinAge, and SkinBloodAge. More importantly, in models that simultaneously adjust for both the lightweight biological aging acceleration value and various epigenetic indices, the index of this invention maintains significant and independent predictive power (adjusted HR=1.20-1.27). In other words, this invention captures unique functional aging signals missing in molecular marker assessments, proving its irreplaceable incremental predictive value.

[0048] Sixth, the non-invasive nature of this invention leads to high compliance and dynamic monitoring advantages. Specifically, this invention requires no blood sampling, making it particularly suitable for the elderly, those with needle phobia, and those with coagulation disorders, effectively increasing acceptance and participation rates among the target population. Furthermore, since all assessment procedures can be performed repeatedly at low frequency in home or community settings without the assistance of professional medical personnel, it supports the establishment of long-term longitudinal health function trajectory files for individuals. This provides a feasible technical solution for dynamically monitoring changes in the rate of aging, timely warning of sudden functional decline, and evaluating the effectiveness of intervention measures.

[0049] Seventh, the lightweight algorithm design of this invention ensures both computational efficiency and applicability to real-world data. Specifically, addressing the common problem of missing values ​​in real-world screening, this invention employs a gender-based median imputation strategy to replace traditional complex multiple imputation or full-information maximum likelihood estimation methods. This significantly reduces the algorithm's dependence on complete data and computational complexity while preserving gender physiological distribution characteristics, enabling the model to run stably on mobile terminals or basic devices with limited computing power. Furthermore, gender-stratified Z-score standardization helps eliminate the interference of natural physiological differences between genders and different units on model fusion, further enhancing the algorithm's robustness in practical application scenarios with gender imbalance or skewed indicator distributions.

[0050] Eighth, this invention forms a complete closed loop from algorithm to community application. Specifically, this invention not only provides a method for constructing a bio-age algorithm, but also further constructs a standardized assessment framework and toolkit for community scenarios (including a community implementation flowchart, a resident-friendly result feedback form, and a quick interpretation card for community doctors). Based on a lightweight bio-age acceleration value, it sets clear three-level risk stratification thresholds (≤-5 years, -5 to +5 years, ≥+5 years) and corresponding action recommendations. In this way, the complex algorithm output can be directly transformed into a tiered follow-up path with operational guidance, enabling both non-professional community workers and individual residents to accurately understand the assessment results and take corresponding health management measures. This effectively solves the "last mile" problem in translating cutting-edge bio-age assessment technology from academic research to grassroots public health practice, possessing extremely high social public health application value.

[0051] 2. Other beneficial effects or advantages of the present invention will be described in detail in conjunction with specific structures in specific embodiments. Attached Figure Description

[0052] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In addition, it should be understood that the proportional relationship of each component in the drawings of this specification does not represent the proportional relationship in the actual material selection and design, but is only a schematic diagram of the structure or position, wherein:

[0053] Figure 1 This is a flowchart of the steps of an algorithm construction method for a lightweight functional bio-age index provided by an exemplary embodiment of the present invention;

[0054] Figure 2This is a schematic diagram of a simplified community implementation process for biological age provided by an exemplary embodiment of the present invention;

[0055] Figure 3 This is provided as an exemplary embodiment of the present invention. A comparison of the mortality prediction capabilities of this method with those of several commonly used DNA methylation-based molecular clocks. Detailed Implementation

[0056] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0057] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0058] Furthermore, the terms "comprising" and "having," and any variations thereof, used in the description of this invention are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include other steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or devices. It should also be noted that in embodiments of this invention, the words "exemplary" or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design described as "exemplary" or "for example" in embodiments of this invention should not be construed as preferred or advantageous over other embodiments or designs. Specifically, the use of the words "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0059] To facilitate understanding by those skilled in the art, the technical concept of this invention will be explained below.

[0060] The technical concept of this invention is as follows: First, by screening non-invasive indicators from three dimensions—physical function (grip strength, gait speed), cognitive psychology, and anthropometry—the problem of data accessibility is solved from the source. Second, a gender-stratified standardization and median imputation strategy is introduced at the algorithm's underlying level to eliminate biases caused by physiological heterogeneity and the lack of real-world data at the lowest computational cost. On this basis, driven by the all-cause mortality endpoint, the risk contribution of each indicator is extracted using the Cox model as a fusion weight to ensure that the score has a clear biological endpoint orientation. Furthermore, through linear calibration, the abstract statistical score is transformed into an intuitive "age scale" and "acceleration offset".

[0061] Overall, this invention aims to deconstruct the complex aging assessment into a standardized tool that integrates "non-invasive data collection, lightweight computation, and intuitive interpretation," thereby accurately addressing the inherent shortcomings of existing technologies that are difficult to promote to grassroots communities and large populations due to their high cost, invasiveness, and computational complexity.

[0062] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings.

[0063] like Figure 1 As shown, this embodiment provides a lightweight algorithm construction method for functional biological age indicators, including the following steps:

[0064] Acquire multidimensional functional index data for individuals, including grip strength and walking speed reflecting muscle and motor function, cognitive scores reflecting cognitive state, self-rated health status reflecting an individual's overall perception, and body mass index and waist circumference reflecting metabolic and morphological characteristics.

[0065] The obtained functional indicators are homogenized so that the higher the value of all indicators, the more severe the degree of biological aging.

[0066] For missing values ​​in each indicator after homogenization, a gender-based median imputation logic is used to fill them.

[0067] The indicators after filling were standardized according to gender to eliminate the influence of dimensional differences caused by physiological differences between genders.

[0068] The standardized indicators are weighted and linearly combined to construct a functional age score, where the weight of each indicator is determined based on all-cause mortality risk.

[0069] The functional age score was calibrated to an age scale to obtain a lightweight biological age index in years.

[0070] Through the above technical solution, firstly, in existing related technologies, biological age assessment models generally rely on DNA methylation sequencing or blood biochemical indicators (such as albumin, creatinine), the detection process requires expensive equipment and professional laboratory support, and is invasive (blood collection), resulting in severely limited application scenarios. In contrast, the required input data in this invention consists of only six indicators: physical function (grip strength, gait speed), cognitive psychology (cognitive score, self-reported health), and anthropometry (BMI, waist circumference). All of these indicators can be obtained through simple physical tests (such as grip strength meters, stopwatches, and measuring tapes) or short questionnaires. This completely eliminates the dependence on biological laboratories and invasive blood collection, significantly reducing assessment costs and technical barriers from the data source, making the method of this invention feasible for large-scale deployment in grassroots communities, home self-testing, and resource-scarce areas.

[0071] Second, in existing related technologies, the original indicators have different biological orientations (for example, higher grip strength indicates better function, while higher BMI often indicates increased risk). Directly fusing these indicators with opposite orientations can lead to confusion in the physical meaning of the weighted model, making it difficult to interpret the actual meaning of the comprehensive score. In this invention, all indicators are uniformly converted into a one-dimensional variable where "higher values ​​represent more severe biological aging." This eliminates the heterogeneity of the original indicators in terms of biological orientation, ensuring that the score obtained from the subsequent weighted linear combination has a clear and consistent physical orientation (i.e., a monotonically increasing score represents a more severe degree of aging), thereby guaranteeing the logical consistency and interpretability of the model during the mathematical fusion process.

[0072] Third, real-world screening data commonly contains missing values. Existing technologies often employ complex statistical methods such as multiple imputation, which are not only computationally intensive and inefficient but also require strict assumptions about data distribution, making them unsuitable for lightweight computing environments. In this invention, however, the missing values ​​are replaced with a single value using the median value of the same-sex group. This approach preserves the physiological distribution characteristics of different sexes through gender stratification, avoiding bias caused by coarse imputation. Furthermore, median imputation has extremely low computational complexity and does not require iterative convergence, effectively improving the processing efficiency and robustness of the method on incomplete datasets, enabling it to run stably on mobile devices or low-powered infrastructure devices.

[0073] Fourth, in this invention, each indicator is standardized independently by gender. This gender-stratified standardization effectively eliminates the interference of gender physiological heterogeneity on the original values, making indicators within the same gender mathematically comparable. Simultaneously, standardization effectively eliminates weight bias caused by different units of measurement (e.g., kilograms, seconds, minutes), ensuring that the subsequent weighted combination results accurately reflect the relative contribution of each indicator to aging, rather than the influence of unit size. This effectively enhances the fairness and universality of the method in cross-gender and cross-population applications.

[0074] Fifth, in existing related technologies, the construction of many biological age or health indices relies heavily on expert-based scoring or principal component analysis—purely data-driven methods—lacking a direct risk correlation with the ultimate endpoint of aging (death). This results in scores that, while describing the indicator's state, lack sufficient sensitivity in predicting future health deterioration and survival risk. In this invention, however, the weights are determined based on all-cause mortality risk and are integrated using a weighted linear combination. By using the risk contribution of each indicator to mortality as the fusion weight, the constructed functional age score is essentially a "risk-driven" comprehensive exposure variable. Compared to experience-based weighting or statistical variance weighting, this score is optimized from the algorithm's inception as a highly sensitive quantitative indicator to the endpoint event. This ensures that the final lightweight biological age index not only reflects the current functional state but also the individual's future systemic mortality risk information, guaranteeing the core predictive power of the method in predicting all-cause mortality.

[0075] Sixth, in existing related technologies, the output of existing models (especially survival models or factor analysis models) is often an abstract risk score or factor score, lacking an intuitive physical dimension. Non-professionals find it difficult to understand the actual meaning of this score and cannot directly compare it with an individual's actual age, making it less practical for public health education and clinical communication. In this invention, the abstract functional age score is mapped to the age scale of "years" through linear regression. This gives the originally dimensionless risk score an intuitive physical dimension of "age," allowing the assessment results to be presented in a commonly understood form of "how many years old." Through this transformation, individuals can intuitively quantify and compare their biological age with their birth age, greatly improving the readability and acceptability of the assessment results in community screening, resident feedback, and doctor-patient communication.

[0076] In one embodiment of the present invention, the homogenization process may specifically include: for indicators in which the original value of grip strength, walking speed, cognitive score and self-rated health status is higher and the health status is better, the indicators are converted through reverse encoding logic so that the higher the value, the worse the health function or the more serious the degree of biological aging.

[0077] This implementation clarifies which indicators need to be reversed to ensure the operability of the alignment operation in practice. Specifically, by precisely reversing the indicator that "higher values ​​represent better health," all input variables are truly within a unified semantic space before entering the weighted combination. This ensures the logical correctness of subsequent fusion modeling from the algorithm implementation level, avoiding incorrect weight symbols or distorted scoring interpretations caused by directional confusion.

[0078] In one embodiment of the present invention, the gender-based median imputation logic can be as follows: for each index with missing values, the median value of the index of the same gender group in the development queue is used to fill the missing values.

[0079] In this implementation, using the development queue as the specific source of the imputation reference group helps ensure that the method of the present invention has a clear reference standard in actual deployment, eliminating the need for online dynamic calculation. Simultaneously, the limitation to a "same-sex group" helps preserve the physiological distribution differences between genders, while the single-value imputation method of the "median value" contributes to the algorithm's extremely high operating efficiency, without making any assumptions about the data distribution pattern. Overall, this implementation method simultaneously achieves optimizations in gender adaptability, computational simplicity, and implementation standardization in the missing data processing stage.

[0080] In one embodiment of the present invention, the standardization process by gender is performed by gender. Score standardization, for each individual and indicators The standardized score is calculated using the following formula:

[0081]

[0082] In the formula, Represents an individual The Standardized indicators Fraction, Represents an individual The Each indicator observation value, and These represent the genders in the development queue. The participants The mean and standard deviation of each indicator.

[0083] In this implementation, a standardized mathematical execution logic is defined, wherein the mean ( ) and standard deviation ( All of these parameters originate from the development cohort, which helps ensure the consistency and reproducibility of standardized parameters throughout the analysis. In this way, after standardization, each indicator is transformed into a dimensionless pure quantity with a mean of 0 and a standard deviation of 1. This can completely eliminate the interference of the original indicator dimension differences on the weighted combination. At the same time, it can also retain the relative difference information between individuals within the same sex, so as to facilitate subsequent equal-weighted or weighted fusion.

[0084] In one embodiment of the present invention, the weighted linear combination of the functional age score of the present invention is constructed according to the following formula:

[0085]

[0086] In the formula, Represents an individual Functional age rating, Indicates the first The weights corresponding to each indicator This indicates the total number of indicators included.

[0087] In this implementation, the functional age score is the sum of the products of each standardized indicator and its corresponding weight. This construction method has several advantages. First, the mathematical form of the linear combination is simple, the computation is extremely low, and it can be processed in parallel, making it suitable for efficient execution in a lightweight computing environment. Second, the contribution of each indicator is reflected independently through the weight coefficient, making the influence of each indicator in the final score quantifiable and traceable, which helps to ensure the interpretability and transparency of the evaluation results.

[0088] In one embodiment of the present invention, the weights of each indicator of the present invention are... The Cox proportional hazards model was obtained by fitting a multivariate Cox proportional hazards model to the development cohort. This model incorporates all indicators simultaneously, with all-cause mortality as the outcome event. The weights of each indicator are as follows: The value is taken as the log-hazard ratio of this indicator in the Cox model.

[0089] In this implementation, a multivariate Cox proportional hazards model that simultaneously includes all indicators is fitted to the development cohort, with the log hazard ratio of each indicator used as the weight. The "simultaneous inclusion of all indicators" setting in the multivariate Cox model ensures that the weight of each indicator represents its independent risk contribution after controlling for the influence of other indicators, rather than the confounding effects found in univariate analysis. Using the "log hazard ratio" as the weight ensures that the fusion coefficient of each indicator directly corresponds to its effect size on the mortality endpoint; a larger weight indicates a stronger impact of that indicator on survival prognosis. Based on this, the contribution of each component of the functional age score has a clear clinical epidemiological interpretation (i.e., the logarithmic increment of relative mortality risk corresponding to each unit change in standardized level), thus enabling the final score weight allocation to have an endpoint-driven causal orientation, rather than empirical or statistically descriptive weights.

[0090] In one embodiment of the present invention, calibrating functional age scores to an age scale specifically includes: in a development cohort, fitting a linear regression model with actual age as the dependent variable and functional age score as the independent variable to obtain the intercept and slope; using the intercept and slope to map an individual's functional age score to a lightweight biological age in years, specifically calibrated according to the following formula:

[0091]

[0092] In the formula, Represents an individual The age of lightweight organisms Show individual i's functional age score, This represents the intercept of the linear regression model. This represents the slope of the linear regression model.

[0093] In this embodiment, the intercept ( ) and slope ( All data points are derived from linear regression of actual age against functional age scores in the development cohort, which helps ensure the stability of calibration parameters and cross-sample portability. Linear mapping is mathematically a first-order linear transformation, involving only one addition and one multiplication operation, resulting in extremely low computational cost. This enables the conversion of abstract risk scores to intuitive physical dimensions, placing the subsequent output of lightweight biological age on the same metric scale as actual age. This provides a direct quantitative basis for the intuitive comparison of "different ages for the same person" at the individual level, effectively improving the cognitive efficiency of non-experts in understanding the assessment results.

[0094] In one embodiment of the present invention, the algorithm construction method for the lightweight functional biological age index of the present invention further includes the step of calculating a lightweight biological age acceleration index:

[0095] Calculate the individual according to the following formula. Lightweight biological aging acceleration value:

[0096]

[0097] In the formula, Represents an individual The lightweight biological aging acceleration value, Represents an individual Actual age;

[0098] When the acceleration value is positive, it indicates that the individual's biological aging rate is faster than that of their peers.

[0099] In this implementation, a lightweight separation between biological age and chronological age can be achieved through a simple subtraction operation. The calculation involves only one subtraction operation, and the computational cost is negligible. The rule for determining accelerated aging, with a positive acceleration value, is simple and intuitive, requiring no additional conversions or reference tables. Thus, the "aging offset," independent of chronological age, is precisely quantified, enabling accurate identification of individuals with different intrinsic aging rates within the same chronological age group. Furthermore, the positive or negative sign provides an intuitive risk warning direction, thus offering a highly sensitive and interpretable reference for subsequent early warning and intervention.

[0100] In one embodiment of the present invention, the algorithm construction method for the lightweight functional biological age index of the present invention further includes the step of constructing a community assessment framework:

[0101] The lightweight biological aging acceleration value is compared with the preset risk stratification threshold to generate a graded assessment result and corresponding intervention recommendations, forming a closed loop from algorithm output to community health management application; the risk stratification threshold has three levels: ≤-5 years, -5 to +5 years, and ≥+5 years.

[0102] In this implementation, three stratified thresholds discretize continuous acceleration values ​​into levels with different risk implications, and each level is linked to specific action recommendations. This allows non-professional community workers and individual residents to directly take appropriate health management measures based on the stratification results. This enables the technical solution to be better applied from algorithmic modeling to community public health scenarios, making this lightweight bio-age indicator a practical community screening tool with standardized operating procedures and tiered decision-making logic. It effectively solves the "last mile" problem in translating cutting-edge bio-age assessment technology into grassroots practice.

[0103] For example, the method of the present invention will be further described below with reference to an exemplary embodiment.

[0104] In this exemplary embodiment, the algorithm construction method for a lightweight functional biological age index of the present invention may include the following steps:

[0105] Step 1: Indicator selection and definition.

[0106] In this exemplary embodiment, six key indicators were selected and integrated: grip strength and walking speed, reflecting muscle and motor function; comprehensive cognitive score, reflecting cognitive state; self-rated health status, reflecting an individual's overall perception; and body mass index (BMI) and waist circumference, reflecting metabolic and morphological characteristics. These indicators can all be obtained through simple physical tests or questionnaires and together constitute the underlying input variables of the method of this invention.

[0107] Step 2: Data preprocessing and homogenization.

[0108] To ensure that the model scores have clear biological meaning, all input components are homogenized. For indicators such as grip strength, walking speed, cognitive score, and self-reported health, a reverse encoding logic is used to uniformly represent poorer health function or more severe biological aging by making higher values.

[0109] Step 3: Fill in missing values.

[0110] To address the common issue of missing data in real-world screening, a gender-specific median imputation logic is introduced, using the median value of the same-sex group in the development cohort for imputation. This significantly reduces the computational complexity of the algorithm and its reliance on high-quality, complete data while preserving the physiological distribution characteristics of gender.

[0111] Step 4: Gender-based standardization.

[0112] Due to significant gender differences in physiological functions, all homogenized components were processed according to gender. z-score standardization makes indicators with different dimensions comparable. For individuals... and indicators The standardized score is calculated as follows:

[0113]

[0114] In the formula, Represents an individual The Standardized indicators Fraction, Represents an individual The Each indicator observation value, and These represent the genders in the development queue. The participants The mean and standard deviation of each indicator.

[0115] Step 5: Functional Age Score The construction of ).

[0116] Functional age rating ( The method for constructing the study involves a weighted linear combination of standardized component scores. The weights of each component are derived from a multivariate Cox proportional hazards model (used to predict all-cause mortality) fitted to a subset of a large cohort development program, which incorporates all components. Participants of The calculation formula is:

[0117]

[0118] In the formula, Represents an individual Functional age rating, Indicates the first The weights corresponding to each indicator This indicates the total number of indicators included.

[0119] Step Six: Lightweight Biological Age ( )calibration.

[0120] In order to represent the overall score in an interpretable age scale, in the developmental subset, Linear calibration was performed against actual age to arrive at a lightweight biological age in years. Specifically, with actual age as the dependent variable, A linear regression model was fitted to the independent variable:

[0121]

[0122] In the formula, This represents the actual age of individual i. This represents the random error term (i.e., the residual term) for individual i. This represents the intercept of the linear regression model. This represents the slope of the linear regression model. The functional age score of individual i is displayed;

[0123] Then, the estimated intercept is applied ( ) and slope ( ) Individual Value mapping to :

[0124]

[0125] In the formula, Represents an individual The age of lightweight organisms;

[0126] Lightweight biological aging acceleration ( ) is defined as:

[0127]

[0128] A positive value indicates that the rate of biological aging is faster compared to people of the same age.

[0129] In this exemplary embodiment, applying the method of the present invention, a simplified, community-implementable assessment framework for biological age may include:

[0130] In order to Acceleration Indicator ( By combining theoretical research with practical community assessment in public health settings, a simplified, biologically age-based, community-implementable assessment toolkit can be constructed. This toolkit can be used to support... Acceleration Indicator ( The feasibility and reproducibility assessment of this study does not constitute a clinical diagnosis or treatment recommendation.

[0131] For detailed instructions on how to deploy the toolkit, please refer to [link / reference]. Figure 2 The community implementation flowchart shown is included; the toolkit may also include a resident-friendly version of the results feedback form as shown in Table 1 below:

[0132] Table 1: Resident-Friendly Version Feedback Form

[0133]

[0134] Understandably, when implementing this in a real community, the following prompts can be given to participants:

[0135] What does your biological age mean?

[0136] Biological age reflects your physical functioning relative to a population of the same chronological age. It is estimated based on simple physical / functional tests and a short questionnaire, without the need for blood tests.

[0137] Your results

[0138] Chronological age: ____ years old;

[0139] Biological age (BA): ____ years old;

[0140] Biological aging acceleration (BA acceleration): ____ years;

[0141] In this exemplary implementation, it is understood that biological age and its acceleration indicators are not diagnostic indicators. Their purpose is to help understand an individual's relative aging level across functional and health dimensions, supporting increased health awareness and preventative attention. The aforementioned stratification thresholds are for communication and illustrative purposes only; actual interpretation should be based on the characteristics of the research subjects, population background, and specific application scenarios, and should not be considered a fixed standard.

[0142] Correspondingly, the toolkit may also include quick-read cards for community doctors, as shown in Table 2 below:

[0143] Table 2. Quick Interpretation Cards for Community Doctors

[0144]

[0145] The community doctor's quick interpretation card displays stratified interpretations and suggested follow-up pathways. However, it is understandable that accelerated biological aging comprehensively reflects the aging status of multiple systems and is mainly used for population-level screening and stratification signals, and should not be used as the sole basis for individual-level clinical decision-making.

[0146] The technical solution of the present invention will be further described below with reference to a more specific exemplary embodiment.

[0147] I. Research Design and Population.

[0148] In this exemplary embodiment, an observational cohort study was conducted to develop and validate a non-blood-based biological age assessment method. (This is used for risk stratification at the population level.)

[0149] This biological age model is built upon data from the China Health and Retirement Longitudinal Study (CHARLS), a nationally representative cohort study that included adults aged 40 and older residing in Chinese communities. For internal validation, the CHARLS cohort was randomly divided into a development subset and an internal validation subset at a 1:1 ratio. CHARLS employed a multi-stage stratified probability sampling design and collected standardized information on sociodemographic characteristics, physical function, cognition, health behaviors, and chronic diseases.

[0150] CHARLS participants were followed up every two years until 2020; when participants could not be contacted, their living status was verified through family members, neighbors or local community organizations, and deaths were recorded through off-site (proxy) interviews and cross-checked where administrative records were available.

[0151] To assess the applicability of the findings across different healthcare systems and sociodemographic contexts, external validation was conducted in independent cohorts in the UK, US, South Korea, and Mexico. All participating studies were approved by their respective institutional review boards, and written informed consent was obtained from all participants.

[0152] II. Selection of candidate indicators and components.

[0153] The candidate metrics are pre-defined based on the following three criteria:

[0154] 1. Previous literature has confirmed its association with aging, mortality, or functional decline;

[0155] 2. Routine assessments can be conducted in communities or standard health assessments without the need for invasive procedures or dedicated infrastructure;

[0156] 3. It can be applied and is comparable across multiple national queues.

[0157] Candidate indicators covered physical function and muscle strength, including the time (in seconds) required to complete a 2.5-meter walk and the maximum grip strength (in kilograms) of the dominant hand. Functional status was assessed by limitation counts of activities of daily living (ADL; six items, including dressing, bathing, and eating) and instrumental activities of daily living (IADL; five items, including housework, meal preparation, and shopping).

[0158] Cognitive and mental health status was assessed using an overall cognitive score (range 0-21), while depressive symptoms were assessed using the 10-item Center for Epidemiological Studies Depression Scale (CES-D-10; range 0-30). Health burden and behavior were assessed through self-reported health status (5-point scale), the number of chronic diseases diagnosed by a physician, and current smoking status.

[0159] To ensure the interpretability of the overall score, all indicators were directionally standardized so that higher values ​​consistently reflect poorer health or more severe biological aging; for indicators that indicate better health, such as grip strength, cognition, and self-rated health, they were reverse-coded before standardization.

[0160] three, The construction of the rating system.

[0161] First, directional harmonization is performed on routinely available functional and behavioral indicators, and then they are standardized to be gender-specific within a subset of CHARLS development. Scoring was performed and integrated using mortality-based weights to generate a comprehensive functional age score. This score was then linearly calibrated to actual age to obtain a lightweight biological age (BIA). ), accelerated biological aging ( ) is defined as The difference between the actual age and the actual age.

[0162] The internal validation subset and all external validation queues use fixed metric definitions, standardized procedures, weights, and calibration parameters derived from the development queues, without any changes.

[0163] IV. Determination of Outcome Indicators

[0164] The primary outcome measure was all-cause mortality. Secondary outcomes included multiple diseases (defined as ≥2 physician-diagnosed chronic diseases) and individual-specific chronic diseases, which were uniformly defined across all cohorts. In some cohorts, other age-related outcomes, including frailty (assessed using the Continuous Frailty Index [FI]) and poor self-reported health status, were also examined as secondary or exploratory endpoints.

[0165] V. External validation and portability.

[0166] External validation was conducted in independent population cohorts from the UK Longitudinal Study of Ageing (ELSA), the US National Health and Nutrition Examination Survey (NHANES), the Korean Longitudinal Study of Ageing (KLoSA), and the Mexican Health and Ageing Study (MHAS).

[0167] In each queue, and All calculations used the same component definitions and fixed parameters derived from the development cohort, eliminating the need for re-estimation or recalibration. A uniform analytical strategy was used to assess associations with mortality and secondary outcomes in each cohort.

[0168] VI. Comparison with the epigenetic clock.

[0169] In order to Compared with DNAm-based aging indicators, this exemplary embodiment only compared and analyzed the NHANES1999-2002 data, which is the only dataset containing epigenetic data. Twelve epigenetic aging indicators were evaluated (see [reference]). Figure 3): HorvathAge Acceleration (i.e., HorvathAgeAcceleration, a first-generation clock primarily used to predict chronological age), HannumAge Acceleration (i.e., HannumAgeAcceleration, a first-generation clock primarily used to predict chronological age), PhenoAge Acceleration (i.e., PhenoAgeAcceleration, a second-generation clock primarily used to predict all-cause mortality and age-related disease risk), GrimAgeMort Acceleration (i.e., GrimAge Acceleration, a second-generation clock primarily used to predict all-cause mortality), GrimAge2Mort Acceleration (i.e., GrimAge2 Acceleration, an updated version of GrimAge), DunedinPoAm (i.e., Dunedin Paceof Aging Methylation, used to measure the rate of biological aging in individuals), ZhangAge Acceleration (i.e., ZhangAgeAcceleration, developed based on blood and saliva samples, used to predict chronological age), WeidnerAge Acceleration (i.e., WeidnerAge Acceleration, an epigenetic clock), LinAge Acceleration (i.e., LinAge Acceleration (an age-predicting clock), Vidal-BraloAge Acceleration (i.e., Vidal-BraloAge Acceleration, an age-predicting clock), SkinBloodAge Acceleration (i.e., SkinBloodAge Acceleration, a skin and blood clock), and HorvathTelo Acceleration (i.e., Horvath Telomere Length, a telomere length predictor developed based on DNA methylation data).

[0170] To ensure comparability between indicators of different scales, for each aging indicator (including...) Each epigenetic marker was standardized within the analyzed sample and assessed in increments of one standard deviation (SD). The association with all-cause mortality was estimated using a Cox proportional hazards model, employing the same covariate adjustment strategy as the primary analysis.

[0171] In order to evaluate Whether mortality risk can be captured independently of methylation-based indicators was also determined in this exemplary embodiment by fitting a model, wherein... Adjustments were made for each epigenetic marker individually.

[0172] VII. Statistical Analysis.

[0173] Baseline characteristics were summarized as mean (standard deviation), median (interquartile range), or proportion, depending on the circumstances. Analysis was conducted separately within each cohort.

[0174] 8. Primary and Secondary Analysis.

[0175] Assessment using the Cox proportional hazards regression model The association with all-cause mortality was determined, and age was used as the time scale where applicable. Participants were followed from the start of baseline assessment to the date of death or the cutoff date at the last follow-up. The proportional hazard assumption was assessed using Schoenfeld residuals, and the results showed that the assumption was acceptable.

[0176] For secondary outcomes, including multiple diseases and individual chronic diseases, logistic regression was used to estimate the odds ratio (OR) and 95% confidence interval (CI). For these non-fatal outcomes, multicollinearity was assessed using variance inflation factor (VIF), which was found to be within acceptable limits. Multiple diseases were defined as the presence of two or more physician-diagnosed chronic diseases. Logistic regression analysis was used to assess self-reported poor health (fair or poor vs. good / very good / excellent). Frailty was assessed using a continuous frailty index (FI) (if available); FI was standardized within the cohort and modeled in increments of one standard deviation (SD), and its association with all-cause mortality was assessed using a Cox proportional hazards model (including models incorporating both FI and...). (model).

[0177] 9. Model specification and covariate adjustment.

[0178] All major models were adjusted for a set of common covariates, including age, gender, education level, marital status, rural residence, body mass index, waist circumference, alcohol consumption, and physical activity. It is primarily modeled as a continuous variable and scaled in increments of one standard deviation (SD) or five years as needed. In supplementary analyses, It is also divided into quantiles for easier interpretation.

[0179] 10. Dose-response and discrimination analysis.

[0180] In this exemplary embodiment, a restricted cubic spline model is employed, with a pre-defined... Using percentiles as nodes, the test was performed. The potential non-linear association between mortality and all-cause mortality was investigated. Results are based on the median.

[0181] To evaluate the model's discriminative power in predicting mortality, a time-dependent receiver operating characteristic (ROC) curve was constructed, and the inclusion and exclusion parameters were calculated. The area under the time dependence curve (AUC) of the model is calculated. Cumulative / dynamic AUC is used to assess time dependence discrimination, with right censoring considered using inverse probability weighting. AUCs are estimated at pre-defined time points (2, 4, 6, 8, and 10 years), and 95% confidence intervals are calculated using 1000 bootstrap resampling. The difference in AUCs (ΔAUC) and its 95% confidence intervals at multiple follow-up time points are estimated.

[0182] XI. External Validation and Benchmark Analysis.

[0183] In the external validation cohort, the same modeling strategy and covariate adjustment specification were applied, eliminating the need to re-estimate model parameters. For the NHANES baseline analysis from 1999–2002, mortality rates and each aging indicator (including...) were assessed. The association between the DNA methylation-based aging markers and 12 other aging indicators increased by one standard deviation for each additional standard deviation. Other models evaluated the association between these markers. Individual adjustments for each epigenetic marker.

[0184] 12. Missing Data and Sensitivity Analysis.

[0185] In the development cohort, missing values ​​for component variables were handled using sex-specific median imputation. Participants with missing outcome data or inappropriate follow-up times were excluded from the survival analysis. Sensitivity analyses included recalculating the composite score after removing individual components one by one, comparing mortality-weighted composite scores with equally weighted composite scores, and subgroup analyses stratified by sex, age group, and baseline health status.

[0186] Thirteen, Results.

[0187] 1. Study population.

[0188] The baseline characteristics of the five cohort study populations are summarized in Table 3 below. The final analysis sample included 49,719 participants from five national cohorts: China (CHARLS, n=7308), the United Kingdom (ELSA, n=7023), South Korea (KLoSA, n=8930), Mexico (MHAS, n=16178), and the United States (NHANES, n=10280). During the 500,126 person-years of follow-up, 8,433 deaths were recorded.

[0189]

[0190] In Table 3, CHARLS represents the China Health and Retirement Longitudinal Study; ELSA represents the UK Longitudinal Study of Aging; KLoSA represents the South Korean Longitudinal Study of Aging; MHAS represents the Mexico Health and Aging Study; and NHANES represents the US National Health and Nutrition Examination Survey.

[0191] 2. The components of each part Contributions and Distribution of phenotypes.

[0192] In the developmental cohort, mortality-based weighted analysis indicated that grip strength (β=0.304), difficulties with activities of daily living (IADL) (β=0.261), and self-reported health status (β=0.220) were the most significant risk factors. The main contributing factors to the comprehensive indicators.

[0193] In all queues, All of these increase with age. The distribution range within the queue is relatively wide, and the values ​​of each queue extend to both sides of the zero value. Association with all-cause mortality

[0194] Of all five countries, All were associated with all-cause mortality. In the multivariate adjusted model, The risk of death increases with each additional 5 years of age: HR for CHARLS is 1.58 (95% CI, 1.44-1.73); HR for ELSA is 1.30 (95% CI, 1.26-1.35); HR for KLoSA is 1.31 (95% CI, 1.26-1.36); HR for MHAS is 1.89 (95% CI, 1.72-2.09); and HR for NHANES is 1.43 (95% CI, 1.35-1.52). Here, HR refers to the Hazard Ratio.

[0195] When considering each increase of 1 standard deviation and each increase of 1 year... Consistent correlations were observed during modeling. Quartile analysis showed that, with... As the level increased, the risk of death increased in a gradient (the trend p-values ​​for all cohorts were <0.001). In the MHAS cohort, the highest quartile (Q4) participants had a higher risk of death than the lowest quartile (Q1) participants (HR 6.01 [95% CI, 4.59–7.87]).

[0196] Subgroup analyses stratified by age, sex, and baseline health status showed a statistically significant interaction between age and mortality risk in multiple cohorts (all interaction P values ​​< 0.05), with a stronger association observed in participants under 65 years of age.

[0197] 3. The dose-response relationship with mortality.

[0198] Restricted cubic spline analysis showed that in each cohort The higher the level, the greater the risk of death. In some cohorts, The higher the level, the more pronounced the increase in risk; while in other cohorts, The association with mortality rate is approximately linear.

[0199] 4. Association with a variety of diseases and non-fatal health outcomes.

[0200] In addition to the mortality rate, the higher It was also associated with an increased likelihood of multiple diseases and multiple nonfatal health outcomes. When modeled on a 5-year increment, the odds ratio for multiple diseases ranged from 1.13 in the KLoSA cohort to 7.84 in the MHAS cohort. The risk was also increased for some specific diseases, including stroke (e.g., OR 2.22 [95% CI, 2.05–2.39] in the ELSA study) and mental illness. In the KLoSA cohort, a higher risk was observed. It was associated with a weakened state (OR = 7.29 [95% CI, 6.56–8.11]). Here, OR stands for Odds Ratio, which is the core effect indicator in the Logistic Regression model and is used to measure... The strength of the association between the disease prevalence or probability of occurrence.

[0201] When considering each increase of 1 standard deviation and each increase of 1 year... Consistent associations were observed during modeling.

[0202] 5. Comparison with the epigenetic clock.

[0203] Please see Figure 3 In the NHANES queue, Its mortality prediction capability is comparable to or better than several molecular clocks based on DNA methylation. The hazard ratio (HR) per standard deviation was 1.27 (95% CI, 1.21–1.34), similar to PhenoAge (1.25) and GrimAgeMort (1.62), but higher than WeidnerAge accel (HR=1.05, 95% CI: 1.00–1.11), LinAge accel (HR=1.08, 95% CI: 1.03–1.14), SkinBloodAge accel (HR=1.08, 95% CI: 1.02–1.14), and HorvathTelo (HR=0.77, 95% CI: 0.72–0.83). It is noteworthy that after mutual adjustment… And in each epigenetic clock model, It remains a significant and independent predictor (adjusted HR of 1.20–1.27), indicating that it captures unique functional aging signals that are missing in molecular assessments.

[0204] 6. Incremental discriminative power in mortality prediction.

[0205] Time-dependent discriminant analysis shows that adding time-dependent discriminant analysis to age- and gender-based models... It can improve the mortality prediction ability of each cohort. The degree of improvement in time-dependent AUC varies depending on the cohort and the prediction time range. The ELSA cohort shows the most significant improvement, while the other cohorts show smaller but consistently positive improvements. The NHANES cohort shows very little improvement in discrimination, reflecting that the discrimination based solely on age baseline is already strong.

[0206] 7. Analysis of independence from weakness and robustness.

[0207] In the analysis of the adjusted decay index (FI), In both the CHARLS and ELSA cohorts, frailty index was independently associated with all-cause mortality. Although frailty index itself is closely associated with mortality, its inclusion in the cohort... The association with mortality rate weakened only slightly.

[0208] According to a second aspect of the present invention, a lightweight functional biometrics algorithm construction system is also provided, for executing the lightweight functional biometrics algorithm construction method of any of the technical solutions in the first aspect of the present invention, the system comprising:

[0209] The data acquisition module is used to acquire multidimensional functional indicator data of an individual;

[0210] The homogenization module is used to homogenize the acquired functional indicators so that the higher the value of all indicators, the more severe the degree of biological aging.

[0211] The missing value imputation module is used to fill in the missing values ​​in each indicator after homogenization using gender-based median imputation logic.

[0212] The standardization module is used to standardize the filled indicators by gender to eliminate the influence of dimensional differences caused by physiological differences between genders.

[0213] The functional age score construction module is used to construct a functional age score by weighted linear combination of standardized indicators, wherein the weight of each indicator is determined based on all-cause mortality risk.

[0214] The bio-age calibration module is used to calibrate functional age scores to an age scale, resulting in a lightweight bio-age index in years.

[0215] In this implementation, firstly, the data acquisition module serves as the system's data entry point, specifically configured to acquire data on six non-invasive functional indicators (grip strength, walking speed, cognitive score, self-reported health, BMI, and waist circumference). This ensures the system is compatible with simple physical measurement tools and questionnaires at the data collection end, eliminating the need for any laboratory equipment or blood collection devices. This facilitates hardware compatibility and site flexibility for deployment in non-professional environments.

[0216] Second, the homogenization module enables the data transmitted between various functional modules within the system to have a consistent biological semantic orientation, so that all subsequent calculation modules (standardization, weighted combination, calibration) process "unidimensional aging signals", ensuring that the output results have a clear and unified physical meaning (i.e., the monotonically increasing value of the final output of the system always corresponds to the aggravation of aging), and eliminating system-level semantic ambiguity caused by heterogeneous index directions.

[0217] Third, the missing value imputation module employs a single-value replacement strategy based on the gender median to handle missing values. This means that when processing missing data, the system only needs to perform simple grouped median lookup and value replacement operations, resulting in extremely low computational complexity, predictable runtime, and no impact from missing data patterns. This effectively improves the system's robustness and response efficiency in real-world sparse data scenarios. Simultaneously, the gender-stratified imputation strategy simplifies calculations while retaining information about the physiological differences in male and female gender distribution, avoiding system bias caused by coarse-grained global median imputation.

[0218] Fourth, the standardization module plays a crucial role in the system by standardizing Z-scores for each indicator independently by gender. This allows the system to establish gender-stratified data processing channels, ensuring that male and female data are expressed within the same mathematical framework (mean 0 and standard deviation 1 within the same gender). This effectively eliminates system bias in evaluating individuals of different genders. Simultaneously, the standardization process converts all indicators into dimensionless pure quantities, addressing the weight distortion caused by dimensional differences. This ensures that the output of subsequent weighted combination modules reflects only the physiological contribution of each indicator, not the numerical range of the measurement scale, enhancing the fairness and mathematical rigor of the system in cross-gender applications.

[0219] Fifth, the functional age score construction module is responsible for linearly fusing standardized indicators according to weights determined based on mortality risk. This allows the functional age score generation process to carry all-cause mortality risk information, improving the system output from a "state description" to a "risk prediction" model. The linear weighted combination calculation is mathematically concise, and the contribution of each indicator can be independently quantified, ensuring the system can quickly complete core calculations even in computationally limited environments. Simultaneously, it maintains the transparency and traceability of the source of influence for each input indicator in the system output.

[0220] Sixth, the biometric age calibration module converts functional age scores into biometric age values ​​in "years" through linear mapping. This makes the system's final output easily understandable to the average person, giving the generated assessment report a physical dimension directly comparable to actual age. This significantly improves the system's communication efficiency and user-friendliness in scenarios such as community screening, resident feedback, and health education, ensuring the system is not only "accurate" but also "readable."

[0221] 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. A lightweight algorithmic construction method for functional biological age indicators, characterized in that, Includes the following steps: Acquire multidimensional functional index data of individuals, including grip strength and walking speed reflecting muscle and motor function, cognitive score reflecting cognitive state, self-rated health status reflecting the individual's overall perception, and body mass index and waist circumference reflecting metabolic and morphological characteristics. The obtained functional indicators are homogenized so that the higher the value of all indicators, the more severe the degree of biological aging. For missing values ​​in each indicator after homogenization, a gender-based median imputation logic is used to fill them. The indicators after filling were standardized according to gender to eliminate the influence of dimensional differences caused by physiological differences between genders. The standardized indicators are weighted and linearly combined to construct a functional age score, where the weight of each indicator is determined based on all-cause mortality risk. The functional age score was calibrated to an age scale to obtain a lightweight biological age index in years.

2. The algorithm construction method for the lightweight functional biological age index according to claim 1, characterized in that, The alignment process includes: For indicators such as grip strength, walking speed, cognitive score, and self-rated health status, the higher the original value, the better the health status. Through reverse encoding logic, these indicators are transformed so that the higher the value, the worse the health function or the more severe the degree of biological aging.

3. The algorithm construction method for the lightweight functional biological age index according to claim 1, characterized in that, The gender-based median imputation logic is as follows: for each indicator with missing values, the median value of that indicator in the same gender group in the development queue is used to fill the missing values.

4. The algorithm construction method for the lightweight functional biological age index according to claim 1, characterized in that, The standardization process by gender is performed by gender. Score standardization, for each individual and indicators The standardized score is calculated using the following formula: ； In the formula, Represents an individual The Standardized indicators Fraction, Represents an individual The Each indicator observation value, and These represent the genders in the development queue. The participants The mean and standard deviation of each indicator.

5. The algorithm construction method for the lightweight functional biological age index according to claim 4, characterized in that, The weighted linear combination of the functional age scores is constructed according to the following formula: ； In the formula, Represents an individual Functional age rating, Indicates the first The weights corresponding to each indicator This indicates the total number of indicators included.

6. The algorithm construction method for the lightweight functional biological age index according to claim 5, characterized in that, Weights of each indicator The Cox proportional hazards model was obtained by fitting a multivariate Cox proportional hazards model to the development cohort. This model simultaneously incorporates all indicators, with all-cause mortality as the outcome event. The weights of each indicator in the Cox proportional hazards model are as follows: The value is taken as the log-hazard ratio of the indicator in the Cox model.

7. The algorithm construction method for the lightweight functional biological age index according to claim 1, characterized in that, The process of calibrating functional age scores to an age scale specifically includes: In the development cohort, a linear regression model was fitted with chronological age as the dependent variable and functional age score as the independent variable to obtain the intercept and slope. These intercept and slope were then used to map an individual's functional age score to a lightweight biological age in years, specifically calibrated according to the following formula: ； In the formula, Represents an individual The age of lightweight organisms Show individual i's functional age score, This represents the intercept of the linear regression model. This represents the slope of the linear regression model.

8. The algorithm construction method for the lightweight functional biological age index according to claim 7, characterized in that, The algorithm construction method for the lightweight functional biological age index also includes the step of calculating a lightweight biological aging acceleration index: Calculate the individual using the following formula. Lightweight biological aging acceleration value: ； In the formula, Represents an individual The lightweight biological aging acceleration value, Represents an individual Actual age; When the acceleration value is positive, it indicates that the individual's biological aging rate is faster than that of their peers.

9. The algorithm construction method for the lightweight functional biological age index according to claim 8, characterized in that, The algorithmic construction method for the lightweight functional biological age indicator also includes the step of constructing a community assessment framework: The lightweight biological age acceleration value is compared with a preset risk stratification threshold to generate a graded assessment result and corresponding intervention recommendations, forming a closed loop from algorithm output to community health management application; wherein, the risk stratification threshold has three levels: ≤-5 years, -5 to +5 years, and ≥+5 years.

10. A lightweight algorithmic construction system for functional biological age indicators, characterized in that, A method for constructing an algorithm for executing any one of the lightweight functional biological age indicators according to claims 1-9, the system comprising: The data acquisition module is used to acquire multidimensional functional indicator data of an individual; The homogenization module is used to homogenize the acquired functional indicators so that the higher the value of all indicators, the more severe the degree of biological aging. The missing value imputation module is used to fill in the missing values ​​in each indicator after homogenization using gender-based median imputation logic. The standardization module is used to standardize the filled indicators by gender to eliminate the influence of dimensional differences caused by physiological differences between genders. The functional age score construction module is used to construct a functional age score by weighted linear combination of standardized indicators, wherein the weight of each indicator is determined based on all-cause mortality risk. The bio-age calibration module is used to calibrate the functional age score to an age scale, resulting in a lightweight bio-age index in years.

Images