An analytical system for the impact of oral weakness on sarcopenia

Abstract

This invention relates to the field of data analysis technology and discloses an analysis system for the impact of oral frailty on sarcopenia, including a data acquisition module, an impact analysis module, a data visualization module, and an interactive feedback module. The impact analysis module includes a correlation analysis module, a causal relationship analysis module, a longitudinal change analysis module, and a mechanism of action analysis module. The causal relationship analysis module is used to analyze the causal relationship between genetic instrumental variables and the risk of oral frailty and sarcopenia. The longitudinal change analysis module is used to analyze the changing trends of oral frailty and sarcopenia in the elderly population at different time points. This invention allows for in-depth research on the impact of oral frailty on sarcopenia, not only enhancing the depth and breadth of research on the relationship between oral frailty and sarcopenia but also providing a scientific, precise, and personalized solution for the health management of the elderly population.

Description

Technical Field

[0001] This invention relates to the field of data analysis technology, and more specifically, to an analysis system for the impact of oral weakness on sarcopenia. Background Technology

[0002] With the increasing aging of the global population, the health problems of the elderly are receiving growing attention. Oral health and muscle health, as important components of overall health in the elderly, are often interrelated and have a profound impact on their quality of life, independence, and lifespan. Oral frailty mainly refers to the decline in oral function in the elderly, usually manifested as tooth loss, difficulty chewing, and swallowing difficulties. These problems not only affect the elderly's ability to eat but may also lead to insufficient nutrient intake, thereby affecting their overall health. Sarcopenia, on the other hand, refers to the gradual decline in muscle mass and strength with age. This decline is usually closely related to the elderly's motor ability, metabolic health, and quality of life. Because the occurrence and development of sarcopenia involve multiple physiological and pathological factors, its research is not limited to the muscles themselves but also needs to take into account nutritional status, inflammation levels, nervous system function, and other health-related factors.

[0003] In recent years, an increasing number of studies have focused on the relationship between oral frailty and sarcopenia. Some scholars have proposed that oral frailty may affect muscle mass and function by influencing food intake and altering nutritional status. Simultaneously, the decline in oral function may also affect the social interaction and mental health of older adults, such as increasing the risk of depression and reducing cognitive function; these factors may also indirectly influence the development of sarcopenia. Furthermore, some studies suggest that chronic inflammation may play a crucial role in this process. Due to poor oral health, older adults may develop chronic inflammation such as periodontitis, which is closely related to muscle atrophy and may accelerate the progression of sarcopenia. However, although existing research reveals a possible interaction and influence between oral frailty and sarcopenia, the specific mechanisms have not been fully elucidated, and many key questions still lack systematic exploration and validation.

[0004] Traditional research on the relationship between oral frailty and sarcopenia primarily relies on single statistical analysis methods, such as regression analysis and correlation analysis. While these methods can reveal connections between variables to some extent, they often overlook the complex causal relationships and potential mediating effects between the two. For example, traditional studies typically use cross-sectional data for analysis, lacking observation of long-term trends, which limits the causal inference ability of the research conclusions. Furthermore, because health problems in the elderly usually involve multiple interrelated biological, psychological, and social factors, single statistical analysis methods cannot fully reveal the complex interaction mechanisms involved, thus limiting the depth and breadth of the research results.

[0005] No effective solutions have yet been proposed to address the problems in the relevant technologies. Summary of the Invention

[0006] In response to the problems in related technologies, this invention proposes an analytical system for the impact of oral weakness on sarcopenia, in order to overcome the aforementioned technical problems existing in the existing related technologies.

[0007] Therefore, the specific technical solution adopted by the present invention is as follows:

[0008] An analysis system for the impact of oral weakness on sarcopenia includes a data acquisition module, an impact analysis module, a data visualization module, and an interactive feedback module connected in sequence.

[0009] The data acquisition module is used to collect relevant data on oral weakness and sarcopenia;

[0010] The data visualization module is used to generate data visualization charts;

[0011] The interactive feedback module is used to interact with users and provide personalized feedback;

[0012] The impact analysis module includes a correlation analysis module, a causal relationship analysis module, a longitudinal change analysis module, and a mechanism of action analysis module, which are connected in sequence.

[0013] The correlation analysis module is used to analyze the relationship between oral frailty indicators and sarcopenia indicators using statistical methods.

[0014] The causal relationship analysis module is used to analyze the causal relationship between genetic instrumental variables and the risk of oral frailty and sarcopenia using a two-sample Mendelian randomization method combined with genome-wide association study data;

[0015] The longitudinal change analysis module is used to analyze the changing trends of oral frailty and sarcopenia in the elderly population at different time points using a deep learning model combined with time-varying confounding factor adversarial learning technology.

[0016] The mechanism of action analysis module is used to analyze the specific mechanisms by which oral weakness affects sarcopenia using structural equation modeling.

[0017] Furthermore, the data acquisition module includes an oral frailty index acquisition module, a sarcopenia index acquisition module, a health index acquisition module, and a data preprocessing module connected in sequence.

[0018] Among them, the oral frailty index collection module is used to collect oral frailty indicators of the elderly population, and the oral frailty indicators include the number of teeth, chewing function and swallowing function.

[0019] The sarcopenia index collection module is used to collect sarcopenia indicators in the elderly population, including grip strength, gait speed, and muscle mass.

[0020] The health indicator collection module is used to collect other relevant health data of the elderly population, including inflammatory markers, cognitive function, depressive state, nutritional status and activity level.

[0021] The data preprocessing module is used to perform missing value imputation, outlier detection, and data standardization on the collected data to ensure data quality and consistency.

[0022] Furthermore, the correlation analysis module includes the following when statistically analyzing the relationship between oral frailty indicators and sarcopenia indicators:

[0023] Calculate the basic statistics for each indicator variable, and draw frequency distribution plots, histograms, and box plots based on the calculation results to check the distribution of the data;

[0024] Assess the individual distribution of oral frailty and sarcopenia indicators, and select statistical methods based on the assessment results. Parametric statistical methods are selected when the assessment results conform to a normal distribution, and nonparametric statistical methods are selected when the assessment results do not conform to a normal distribution.

[0025] We used selected statistical methods to analyze the linear or nonlinear relationship between oral weakness and sarcopenia indicators, and combined this with regression analysis to analyze the linear effect of oral weakness on sarcopenia.

[0026] Furthermore, the causal relationship analysis module includes a genetic instrument variable selection module, a Mendelian randomization analysis module, an analysis result testing module, and a causal relationship output module, which are connected in sequence.

[0027] Among them, the genetic instrument variable selection module is used to obtain the association data between oral frailty and sarcopenia from two independent genome-wide association study datasets, and extract single nucleotide polymorphism sites associated with oral frailty and sarcopenia as genetic instrument variables respectively.

[0028] The Mendelian randomization analysis module was used to assess the effects of genetic instrumental variables on oral asthenia and sarcopenia, and to infer the causal relationship between oral asthenia and sarcopenia using various Mendelian randomization methods.

[0029] The analysis results verification module is used to evaluate whether the assumptions of Mendelian randomization analysis hold, in order to ensure the reliability of causal inference;

[0030] The causal relationship output module is used to interpret the analysis results and output the causal relationship between oral weakness and sarcopenia.

[0031] Furthermore, the genetic instrumental variable selection module is also used to evaluate the strength of genetic instrumental variables in order to avoid bias and inefficiency caused by weak instrumental variables, enhance the statistical power and robustness of causal inference, and improve the credibility of research results.

[0032] The formula for assessing the strength of genetic instrumental variables is as follows:

[0033]

[0034] In the formula, F Indicates the strength of genetic instrumental variables. R 2 This represents the degree of variability in the explanation of the exposure variable by the genetic instrumental variable. n Indicates the number of samples. k This indicates the number of genetic instrumental variables.

[0035] Furthermore, the Mendelian randomization analysis module includes the following for assessing the effects of genetic instrumental variables on oral asthenia and sarcopenia, and for inferring the causal relationship between oral asthenia and sarcopenia using various Mendelian randomization methods:

[0036] Effect estimates, standard errors, and statistical significance of genetic instrumental variables associated with oral frailty were obtained from genome-wide association study data, and effect allele orientations were aligned to ensure that the effect orientations of genetic instrumental variables were consistent across different datasets.

[0037] Effect estimates, standard errors, and statistical significance of genetic instrumental variables associated with sarcopenia were obtained from genome-wide association study data of sarcopenia, and the direction of effect alleles was checked to ensure that the direction of effect alleles in oral frailty and sarcopenia data were consistent.

[0038] The weighted average effect value of genetic instrumental variables was calculated using the inverse variance weighting method, and the preliminary causal effect of oral frailty on sarcopenia was analyzed based on the weighted average effect value. The regression adjustment method was used to detect level pleiotropic bias in order to check whether genetic instrumental variables affect other pathways, and the weighted median method was used to improve robustness when some genetic instrumental variables were ineffective.

[0039] The outlier adjustment method was used to identify and remove abnormal genetic instrumental variables that had a significant impact on the results. Sensitivity analysis was performed on the removed genetic instrumental variables, and the causal effects were summarized to obtain the causal relationship between oral weakness and sarcopenia.

[0040] Furthermore, the analysis results verification module includes a correlation hypothesis evaluation module, an independence hypothesis evaluation module, and an exclusivity hypothesis evaluation module connected in sequence.

[0041] The correlation hypothesis assessment module is used to assess the effectiveness of genetic instrument variables based on their effect estimates, strength, and significance tests.

[0042] The independence hypothesis assessment module is used to test whether genetic instrumental variables are related to confounding factors and whether they meet the independence hypothesis.

[0043] The exclusion hypothesis assessment module is used to assess whether genetic instrumental variables affect outcome variables only through exposure variables.

[0044] Furthermore, the longitudinal change analysis module, utilizing deep learning models combined with time-varying confounding factor adversarial learning techniques, analyzes the changing trends of oral frailty and sarcopenia in the elderly population at different time points, including:

[0045] We obtained longitudinal data on oral frailty and sarcopenia at multiple time points in the database of the elderly population, and preprocessed the data to select key features that affect changes in oral frailty and sarcopenia. The longitudinal data included exposure variables, outcome variables and time-varying confounding factors.

[0046] The time-varying confounding factor is input into the generator to generate an exposure variable that removes the influence of the confounding factor. The discriminator receives the generated exposure variable and the original confounding factor data, evaluates the impact of the time-varying confounding factor on the exposure variable, and trains the generator and discriminator alternately against each other.

[0047] Using the exposure variables, outcome variables, and confounding factors at each time point as input data, and the predicted values ​​of the exposure variables and outcome variables at future time points as output, a long short-term memory network time series model is constructed; an adversarial regularization term is added to the loss function of the long short-term memory network, and the long short-term memory network time series model is trained using adversarial regularization.

[0048] The trained long short-term memory network time series model was used to output the predicted values ​​of exposure variables and outcome variables at future time points, and the changing trends of oral frailty and sarcopenia in the elderly population at different time points were analyzed based on the prediction results.

[0049] Furthermore, the expression for the loss function of Long Short-Term Memory (LSTM) networks is as follows:

[0050]

[0051] In the formula, L Represents the loss function. E Indicates the expectation symbol, y t Indicates the first t The actual target value at a given time point. Indicates the first t Predictions from point-in-time models , These are used to adjust the hyperparameters of the adversarial loss and the regularization term with respect to the total loss, respectively. D Represents the discriminator model. This represents the output prediction of a long short-term memory network temporal model. f This represents a temporal model of the Long Short-Term Memory network. Indicates the first t Exposure variables at specific points in time Indicates the first t Time-varying confounding factor at a given time point. This represents the L2 regularization term.

[0052] Furthermore, the mechanism of action analysis module, when using structural equation modeling to analyze the specific mechanisms by which oral frailty affects sarcopenia, includes:

[0053] Identify latent variables and determine corresponding observed variables based on them. The latent variables include oral frailty, sarcopenia, inflammatory factors, cognitive function, depressive state, and nutritional status. Define the causal relationship and pathway between the latent variables and the observed variables.

[0054] Factor analysis was used to verify the matching degree between latent variables and observed variables to ensure the reliability and validity of the measurement results. Causal paths between latent variables were defined, and path analysis was performed to estimate the coefficients of each path and obtain the structural equation model.

[0055] The goodness of fit of the structural equation model was tested, the path coefficients were interpreted, and the mechanism of the influence of oral weakness on sarcopenia was analyzed based on the path coefficients.

[0056] The beneficial effects of this invention are as follows:

[0057] 1) This invention provides a systematic analytical framework by comprehensively utilizing technologies such as data acquisition, impact analysis, longitudinal change analysis, causal inference, and structural equation modeling. This framework enables in-depth research on the impact of oral frailty on sarcopenia, not only enhancing the depth and breadth of research on the relationship between oral frailty and sarcopenia, but also providing scientific, precise, and personalized solutions for the health management of the elderly population.

[0058] 2) This invention employs statistical methods and Mendelian randomization to reveal the potential causal relationship between oral atony and sarcopenia through in-depth analysis. In particular, the selection of genetic instrumental variables and the support of multi-sample data improve the accuracy and scientific rigor of causal inference, avoiding the interference of confounding factors.

[0059] 3) This invention utilizes a Long Short-Term Memory (LSTM) network model from deep learning, combined with adversarial learning techniques for time-varying confounding factors, to predict the changing trends of oral frailty and sarcopenia at different time points. This technique effectively reduces the influence of time-varying confounding factors, providing a more accurate basis for future prediction and intervention.

[0060] 4) This invention utilizes structural equation modeling (SEM) to deeply analyze how oral weakness affects sarcopenia through multiple pathways. Factor analysis and pathway analysis reveal the direct impact of oral weakness on sarcopenia and its potential mediating mechanisms, leading to a better understanding of the influence of oral weakness on sarcopenia. Attached Figure Description

[0061] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments 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.

[0062] Figure 1 This is a structural block diagram of an analysis system for the effect of oral weakness on sarcopenia according to an embodiment of the present invention.

[0063] In the picture:

[0064] 1. Data acquisition module; 2. Impact analysis module; 3. Data visualization module; 4. Interactive feedback module. Detailed Implementation

[0065] To further illustrate the various embodiments, the present invention provides accompanying drawings, which are part of the disclosure of the present invention. These drawings are mainly used to illustrate the embodiments and can be used in conjunction with the relevant descriptions in the specification to explain the operating principles of the embodiments. With reference to these drawings, those skilled in the art should be able to understand other possible implementation methods and the advantages of the present invention. The components in the drawings are not drawn to scale, and similar component symbols are generally used to represent similar components.

[0066] According to an embodiment of the present invention, an analytical system for the effect of oral weakness on sarcopenia is provided.

[0067] The present invention will now be further described in conjunction with the accompanying drawings and specific embodiments, such as... Figure 1 As shown, the analysis system for the impact of oral weakness on sarcopenia according to an embodiment of the present invention includes a data acquisition module 1, an impact analysis module 2, a data visualization module 3, and an interactive feedback module 4 connected in sequence.

[0068] Data acquisition module 1 is used to collect relevant data on oral weakness and sarcopenia;

[0069] Specifically, the data acquisition module 1 includes an oral frailty index acquisition module, a sarcopenia index acquisition module, a health index acquisition module, and a data preprocessing module connected in sequence.

[0070] Among them, the oral frailty index collection module is used to collect oral frailty indicators of the elderly population (e.g., through questionnaires, oral examinations, etc.), and the oral frailty indicators include the number of teeth, chewing function, and swallowing function.

[0071] The sarcopenia index collection module is used to collect sarcopenia indicators (such as grip strength, gait test, muscle mass measurement, etc.) of the elderly population, and the sarcopenia indicators include grip strength, gait speed and muscle mass.

[0072] The health indicator collection module is used to collect other relevant health data of the elderly population, including inflammatory markers, cognitive function, depressive state, nutritional status and activity level.

[0073] The data preprocessing module is used to perform missing value imputation, outlier detection, and data standardization on the collected data to ensure data quality and consistency.

[0074] Handling missing data: Missing values ​​can be filled using methods such as mean imputation or interpolation, or the method of deletion or imputation can be chosen based on the degree of missingness.

[0075] Outlier detection: Use box plots, standard deviation methods, etc. to detect and handle outliers.

[0076] Data standardization: For variables with different dimensions, standardization or normalization can be performed to ensure the comparability between data.

[0077] Among them, the impact analysis module 2 includes a correlation analysis module, a causal relationship analysis module, a longitudinal change analysis module, and a mechanism of action analysis module connected in sequence;

[0078] The correlation analysis module is used to statistically analyze the relationship between oral frailty indicators and sarcopenia indicators. This module includes the following steps when performing statistical analysis on the relationship between oral frailty indicators and sarcopenia indicators:

[0079] 1) Calculate the basic statistics for each indicator variable, and draw frequency distribution plots, histograms, and box plots based on the calculation results to check the distribution of the data;

[0080] 11) Calculate descriptive statistics

[0081] Calculate the mean, median, standard deviation (SD), skewness, and kurtosis.

[0082] Skewness: Measures whether data is symmetrical. If the skewness is far from 0 (e.g., >1 or <-1), it indicates that the data is highly skewed.

[0083] Kurtosis: Measures the steepness of a data distribution. A high kurtosis suggests that the data may have a heavy-tailed distribution (long-tail effect).

[0084] 12) Draw a data distribution map

[0085] Histogram: A visual representation of the distribution of data, whether it follows a normal distribution or has a significant skewness.

[0086] QQ plot (Quantile-Quantile Plot): If the data points on the QQ plot are basically a straight line, it means that the data is close to a normal distribution; if the data points deviate from the straight line, it means that the data may be skewed or have heavy tails.

[0087] Boxplots can help identify outliers, and too many outliers can affect the analysis.

[0088] 2) Assess the individual distribution of oral frailty and sarcopenia indicators, determine whether further transformation (e.g., logarithmic transformation) is needed to meet the assumptions of subsequent analysis, and select statistical methods based on the assessment results;

[0089] Methods for testing the normality of data: The Shapiro-Wilk test (suitable for small samples of less than 50; if p < 0.05, the data deviates from a normal distribution) and the Kolmogorov-Smirnov test (suitable for large samples, but sensitive to outliers) can be used to test whether the data conforms to a normal distribution. Parametric statistical methods (such as Pearson correlation) are chosen when the evaluation results conform to a normal distribution; nonparametric statistical methods (such as Spearman rank correlation) are chosen when the evaluation results do not conform to a normal distribution.

[0090] Log transformation:

[0091] Applicable situations: Data distribution is right-skewed (positively skewed), meaning most data points are small and a few data points are large.

[0092] Method: Take the logarithm of the data.

[0093] Applicable scenarios: number of missing teeth (if some people are completely edentulous); muscle mass (if the distribution is wide).

[0094] 3) Use selected statistical methods to analyze the linear or nonlinear relationship between oral weakness and sarcopenia indicators, and combine regression analysis to analyze the linear effect of oral weakness on sarcopenia.

[0095] 31) Correlation analysis: used to analyze the linear or non-linear relationship between oral weakness and sarcopenia indicators;

[0096] Pearson correlation coefficient: used to assess the linear correlation between oral weakness and sarcopenia indicators, provided that the data conforms to a normal distribution.

[0097] Spearman rank correlation coefficient: used to assess the monotonic relationship between oral weakness and sarcopenia, and is suitable for non-normally distributed data.

[0098] Output: Correlation coefficient and its significance level (p-value). If the p-value is less than the significance level (e.g., 0.05), then a significant statistical relationship is considered to exist between the two.

[0099] 32) Regression analysis: used to further analyze the relationship between oral weakness and sarcopenia indicators, and to evaluate the predictive role of oral weakness in sarcopenia.

[0100] Linear regression analysis: used to assess the linear effect of oral weakness (as an independent variable) on sarcopenia indicators (as a dependent variable). If the data are continuous variables, simple linear regression or multiple linear regression can be performed. The predictive value of oral weakness for sarcopenia is assessed through the regression coefficients.

[0101] Multiple regression analysis: If there are multiple potential confounding factors (such as age, gender, chronic diseases, etc.), multiple regression analysis can be performed to control these confounding factors in the model and further explore the independent relationship between oral weakness and sarcopenia.

[0102] Output: regression coefficient (β value), R² value, significance level (p-value), etc. If the p-value is less than the significance level (0.05), oral weakness is considered to have a significant impact on sarcopenia.

[0103] 33) Model testing and diagnosis: used to evaluate the effectiveness of regression models and perform hypothesis testing.

[0104] Residual analysis: Checks whether the residuals of the regression model conform to a normal distribution and whether they satisfy assumptions such as independence and homoscedasticity.

[0105] VIF test: Checks for multicollinearity among independent variables. If the VIF value is too high (generally greater than 10), it may be necessary to remove some variables.

[0106] The causal relationship analysis module is used to analyze the causal relationship between genetic instrumental variables and the risk of oral frailty and sarcopenia using a two-sample Mendelian randomization method combined with genome-wide association study data;

[0107] Specifically, the causal relationship analysis module includes a genetic instrument variable selection module, a Mendelian randomization analysis module, an analysis result testing module, and a causal relationship output module, which are connected in sequence.

[0108] Among them, the genetic instrument variable selection module is used to obtain the association data between oral frailty and sarcopenia from two independent genome-wide association study datasets, and extract single nucleotide polymorphism sites associated with oral frailty and sarcopenia as genetic instrument variables respectively.

[0109] Specifically, the single nucleotide polymorphisms (SNPs) associated with oral weakness were screened as follows:

[0110] Selecting single nucleotide polymorphisms (SNPs) associated with oral atrophy: SNPs associated with oral atrophy were screened from the GWAS dataset. These SNPs typically showed significant statistical association (p-value less than a predefined significance threshold, e.g., ...). These single nucleotide polymorphisms should have a strong effect on genes or pathways related to oral weakness.

[0111] Effect size and direction: Collect the effect size (e.g., β value) and effect direction (synergistic or detrimental effect of alleles) of these single nucleotide polymorphisms.

[0112] The following single nucleotide polymorphisms were screened for in relation to sarcopenia:

[0113] Selection of single nucleotide polymorphisms (SNPs) associated with sarcopenia: Significant SNPs were screened from the GWAS dataset for sarcopenia. These SNPs are expected to play an important role in the pathogenesis of sarcopenia and are statistically significant.

[0114] Effect size and direction: Similarly, effect size and direction of sarcopenia-related single nucleotide polymorphisms were collected, and significance criteria were ensured when they were selected.

[0115] The genetic instrumental variables are determined as follows:

[0116] Independence test: To avoid correlations (such as chain effects) between instrumental variables, a linkage disequilibrium (LD) test is performed on the single nucleotide polymorphisms (SNPs) of oral asthenia and sarcopenia to select independent SNPs. The r² value can be used to assess the degree of LD, and the threshold is usually set at r² < 0.1.

[0117] Effect consistency: Ensure that the selected single nucleotide polymorphisms (SNPs) have consistent effect directions across different datasets. SNPs with opposite effect directions in different datasets should be selected with caution.

[0118] The genetic instrument variable selection module is also used to evaluate the strength of genetic instrument variables in order to avoid bias and inefficiency caused by weak instrument variables, enhance the statistical power and robustness of causal inference, and improve the credibility of research results.

[0119] The formula for assessing the strength of genetic instrumental variables is as follows:

[0120]

[0121] In the formula, F Indicates the strength of genetic instrumental variables. R 2 This represents the degree of variability in the explanation of the exposure variable by the genetic instrumental variable. n Indicates the number of samples. k This indicates the number of genetic instrumental variables.

[0122] The Mendelian randomization analysis module was used to assess the effects of genetic instrumental variables on oral asthenia and sarcopenia, and to infer the causal relationship between oral asthenia and sarcopenia using various Mendelian randomization methods.

[0123] Specifically, the Mendelian randomization analysis module includes the following for assessing the effects of genetic instrumental variables on oral asthenia and sarcopenia, and for inferring the causal relationship between oral asthenia and sarcopenia using various Mendelian randomization methods:

[0124] 1) Obtain the effect estimates, standard errors, and statistical significance (p-values) of genetic instrumental variables related to oral frailty from genome-wide association study data, and perform effect allele alignment to ensure that the effect directions of genetic instrumental variables are consistent in different datasets;

[0125] Specifically, alignment of effector allele orientations includes:

[0126] 11) Understanding effect alleles:

[0127] Effect alleles: These are the variant alleles associated with the trait (in this case, oral asthenia). Each SNP typically has two alleles (reference allele and variant allele), and the effect size is calculated based on the variant allele.

[0128] 12) Examine the effect alleles in different datasets:

[0129] The issue of inconsistent effect allele orientations: In different GWAS datasets (e.g., from different populations or studies), the effect alleles of the same SNP may have different orientations (i.e., positive or negative effects). Therefore, it is necessary to ensure that the effect allele orientations of the selected SNPs are consistent across all datasets in the analysis.

[0130] 13) Allele alignment method:

[0131] Finding allele orientation: For each SNP, examine the effect alleles in different datasets. Each GWAS dataset provides the reference and effect alleles for the SNP.

[0132] Alignment effect alleles: If the directions of effect alleles are inconsistent, they need to be aligned using the following methods:

[0133] Checking effect size: For different datasets, if the allele directions of the effect are inconsistent, the effect directions can usually be reversed to align them. For example, if an SNP shows a positive effect in one dataset but a negative effect in another, the effect size of the latter can be reversed to make the effect directions consistent.

[0134] Alignment effect size: Once the orientations are aligned, effect consistency is ensured by calculating SNP effect estimates (β values). If the orientations of the effect alleles are assumed to be consistent, effect sizes can be used directly for comparison.

[0135] 14) Perform genotype alignment:

[0136] When aligning effector alleles, it is essential to ensure that the alignment methods used for the genotype data are consistent. Genotype alignment can typically be performed using the following steps:

[0137] Verify the reference genome for alleles: Ensure that the reference genome version (e.g., GRCh37 or GRCh38) for all SNPs is consistent.

[0138] Same allele alignment: For different research datasets, ensure that the selected reference allele and effect allele have consistent meanings across all datasets.

[0139] 2) Obtain the effect estimates, standard errors, and statistical significance of genetic instrumental variables related to sarcopenia from genome-wide association study data of sarcopenia, and check the consistency of effect allele directions to ensure that the effect allele directions are consistent in the oral frailty and sarcopenia data.

[0140] 3) The weighted average effect value of the genetic instrumental variables was calculated using the inverse variance weighting method, and the preliminary causal effect of oral weakness on sarcopenia was analyzed based on the weighted average effect value; the regression adjustment method was used to detect level pleiotropic bias in order to check whether the genetic instrumental variables affect other pathways, and the weighted median method was used to improve robustness when some genetic instrumental variables were ineffective.

[0141] 31) Inverse variance weighted method:

[0142] The weighted average effect of instrumental variables was calculated by weighting the effect values ​​of each instrumental variable according to their standard errors, using the inverse variance weighting method to calculate the overall effect of oral weakness on sarcopenia.

[0143] Significance test: By calculating the p-value, if the p-value is less than 0.05, it is considered that oral weakness may have a significant causal effect on sarcopenia.

[0144] Results Analysis: This method typically provides the most accurate effect estimate, but it relies on all instrumental variables satisfying the Mendelian randomization assumption. If the p-value of this method is less than 0.05, it suggests a possible causal effect of oral weakness on sarcopenia, but further verification is needed.

[0145] 32) Regression adjustment method (used to detect level pleiotropic bias):

[0146] Regression intercept test: Based on the inverse variance weighted method, the regression adjustment method is further used to test whether there is level pleiotropy in the instrumental variables; the instrumental variables are adjusted using the regression model, and the intercept of the regression model is tested to see if it is significantly different from zero. If the intercept is not zero and the p-value is less than 0.05, it indicates the existence of level pleiotropy bias, which may affect the accuracy of causal inference.

[0147] Results Analysis: If there is no level pleiotropy (intercept ≈ 0, p > 0.05), the results of the inverse variance weighted method are relatively reliable, and sensitivity analysis can be continued. If level pleiotropy exists (intercept significantly ≠ 0, p < 0.05), a more robust method is needed to further estimate the causal effect in order to reduce the influence of bias.

[0148] Further adjustments: If pleiotropic bias is found, the analysis can be repeated by removing or adjusting specific instrumental variables.

[0149] 33) Weighted median method (applicable when some instrumental variables are invalid):

[0150] Median regression: If some instrumental variable effects are invalid, the weighted median method can be used, which uses the median of the instrumental variable effects as an estimate of the causal effect, thereby enhancing the robustness of the model to invalid instrumental variables.

[0151] Robustness verification: The robustness of causal inference is verified by comparing the results of the weighted median method with those of other methods.

[0152] Results Analysis: If the results of the weighted median method are similar to those of the inverse variance weighted method, it indicates that the causal inference is robust and the impact of level pleiotropy is relatively small. If the results of the weighted median method are significantly different from those of the inverse variance weighted method, it indicates that some instrumental variables are biased, and it may be necessary to further remove outlier instrumental variables and reanalyze.

[0153] 4) Based on the outlier adjustment method, identify and remove abnormal genetic instrumental variables that have a significant impact on the results, and perform sensitivity analysis on the removed genetic instrumental variables to summarize the causal effects and obtain the causal relationship between oral weakness and sarcopenia.

[0154] 41) Outlier Detection and Adjustment:

[0155] Outlier identification: Identify and remove outlier genetic instrumental variables that have a significant impact on the results, thereby reducing the interference of outliers on causal inference.

[0156] Reanalysis: After removing outliers, perform causal inference analysis again to ensure the robustness of the results.

[0157] Results Analysis: If the results remain unchanged after removing outliers, the causal inference is robust, and the causal effect of oral weakness on sarcopenia is reliable. If the results change significantly after removing outliers, the causal inference is affected by individual genetic instrumental variables, and the results need to be interpreted with caution.

[0158] 42) Conduct sensitivity analysis:

[0159] Sensitivity analysis of removing instrumental variables one by one: By removing each genetic instrumental variable one by one, we can test whether any single genetic instrumental variable has a significant impact on the estimation of causal effects, thus ensuring the stability of causal inference.

[0160] Comparison of multiple methods: Causal inference was performed using different Mendelian randomization methods (such as inverse variance weighting, regression adjustment, median method, etc.), and the consistency of the results of different methods was compared to further verify the robustness of causal relationships.

[0161] 43) Interpretation of results and inference of causal relationships:

[0162] Summarize the causal effects: Summarize the results of different methods (inverse variance weighted method, regression adjustment method, weighted median method, etc.) to obtain the estimated causal effect of oral weakness on sarcopenia, and report the 95% confidence interval.

[0163] To verify the robustness of the causal effect: if all methods yield consistent results without pleiotropic bias and the p-value is less than 0.05, then oral weakness can be considered to have a significant causal effect on sarcopenia.

[0164] The analysis results verification module is used to evaluate whether the assumptions of Mendelian randomization analysis hold, in order to ensure the reliability of causal inference;

[0165] Specifically, the analysis results verification module includes a correlation hypothesis evaluation module, an independence hypothesis evaluation module, and an exclusivity hypothesis evaluation module connected in sequence;

[0166] The correlation hypothesis assessment module is used to assess the effectiveness of genetic instrument variables based on their effect estimates, strength, and significance tests.

[0167] When the strength is greater than or equal to 10, the instrumental variable is generally considered to be strong and can effectively predict the exposure variable. A strength less than 10 may indicate that the instrumental variable is weak and there may be weak instrumental bias.

[0168] The independence hypothesis assessment module is used to test whether genetic instrumental variables are related to confounding factors and whether they meet the independence hypothesis.

[0169] Evaluation steps:

[0170] Examine the correlation between instrumental variables and confounding factors: Use known confounding factors (such as age, gender, socioeconomic status, lifestyle habits, etc.) to perform correlation tests. Correlation analysis or regression analysis can be used to assess whether a significant relationship exists between instrumental variables and confounding factors.

[0171] Assessing the random allocation properties of instrumental variables: Since Mendelian randomization is based on the random allocation properties of genetic variation, theoretically, genetic instrumental variables should be independent of confounding factors. This hypothesis can be assessed using level pleiotropic testing.

[0172] Using IVW and MR-Egger methods: The MR-Egger regression method is used to examine whether the instrumental variable affects the outcome variable through other paths, thereby detecting potential confounding factors. If the intercept of the MR-Egger is significantly non-zero (p<0.05), it indicates that the instrumental variable may have level pleiotropy (i.e., the instrumental variable may affect the outcome variable through other paths, violating the independence assumption).

[0173] Interpretation of results: If there is no significant relationship between the instrumental variable and the confounding factors (p>0.05), the independence assumption holds. If regression analysis reveals a correlation between the instrumental variable and the confounding factors, or if the MR-Egger method detects a significant intercept, it may be necessary to reconsider the choice of instrumental variable or use a robust method for analysis.

[0174] The exclusion hypothesis assessment module is used to assess whether genetic instrumental variables affect outcome variables only through exposure variables.

[0175] Evaluation steps:

[0176] Test whether instrumental variables exhibit pleiotropy:

[0177] MR-Egger regression: assess whether the regression intercept is significantly non-zero. If the intercept is significant (p<0.05), it indicates that the instrumental variable has pleiotropic effects and may influence the outcome variable through other mechanisms.

[0178] Plexiactivity test: If there are multiple instrumental variables, the Q statistic can be used to test whether the effects of each instrumental variable are consistent. If p < 0.05, it indicates that there is heterogeneity among the instrumental variables, which may affect causal inference.

[0179] Using the MR-PRESSO method: The MR-PRESSO (Mendelian Randomization Pleiotropy RESidual Sum and Outlier) method can identify and eliminate abnormal SNPs with pleiotropy, reducing the impact of pleiotropy bias.

[0180] Remove pleiotropic instrumental variables: Pleiotropic instrumental variables identified by the above methods should be removed or adjusted to ensure that the effects of instrumental variables only affect the outcome variable through the exposure variable.

[0181] Interpretation of results: If the intercept of the MR-Egger regression is not significant (p>0.05) and the instrumental variable does not show pleiotropic effects (p-value of Q statistic is greater than 0.05), then the exclusivity hypothesis holds, and the instrumental variable affects the outcome only through the exposure variable. If a significant intercept exists (p<0.05), then the instrumental variable may affect the outcome variable through other pathways, violating the exclusivity hypothesis, and further adjustments or the use of other methods are needed.

[0182] The causal relationship output module is used to interpret the analysis results and output the causal relationship between oral weakness and sarcopenia.

[0183] The longitudinal change analysis module is used to analyze the changing trends of oral frailty and sarcopenia in the elderly population at different time points using a deep learning model combined with time-varying confounding factor adversarial learning technology.

[0184] Specifically, the longitudinal change analysis module, when using a deep learning model combined with time-varying confounding factor adversarial learning technology to analyze the changing trends of oral frailty and sarcopenia in the elderly population at different time points, includes:

[0185] 1) Obtain longitudinal data on oral frailty and sarcopenia of the elderly population at multiple time points in the database, and perform preprocessing to select key features that affect changes in oral frailty and sarcopenia. The longitudinal data includes exposure variables, outcome variables and time-varying confounding factors.

[0186] 2) Input the time-varying confounding factor into the generator to generate exposure variables that have been freed from the influence of the confounding factor. The discriminator receives the generated exposure variables and the original confounding factor data, evaluates the impact of the time-varying confounding factor on the exposure variables, and trains the generator and discriminator alternately against each other.

[0187] 3) Using the exposure variables, outcome variables, and confounding factors at each time point as input data, and the predicted values ​​of the exposure variables and outcome variables at future time points as output, construct a long short-term memory network time series model; add an adversarial regularization term to the loss function of the long short-term memory network, and use adversarial regularization to train the long short-term memory network time series model.

[0188] The expression for the loss function of a Long Short-Term Memory (LSTM) network is:

[0189]

[0190] In the formula, L Represents the loss function. E Indicates the expectation symbol, y t Indicates the first t The actual target value at a given time point. Indicates the first t Predictions from point-in-time models , These are used to adjust the hyperparameters of the adversarial loss and the regularization term with respect to the total loss, respectively. D Represents the discriminator model. This represents the output prediction of a long short-term memory network temporal model. f This represents a temporal model of the Long Short-Term Memory network. Indicates the first t Exposure variables at specific points in time Indicates the first t Time-varying confounding factor at a given time point. This represents the L2 regularization term.

[0191] 4) Utilize the trained long short-term memory network time series model to output the predicted values ​​of exposure variables and outcome variables at future time points, and analyze the changing trends of oral frailty and sarcopenia in the elderly population at different time points based on the prediction results.

[0192] Specifically, by using a trained Long Short-Term Memory (LSTM) network time series model to output predicted values ​​of exposure and outcome variables for future time points, and analyzing the changing trends of oral frailty and sarcopenia in the elderly population at different time points based on the prediction results, the following steps can be followed:

[0193] 41) Use the trained LSTM model for prediction:

[0194] After the LSTM model is trained, we can use it to make predictions for future time points. Specifically, the LSTM model has learned the temporal dependencies in the data, and it can predict future exposure variables (such as the progression of oral frailty) and outcome variables (such as changes in sarcopenia) based on the current time point and historical information.

[0195] Predictive exposure variables and outcome variables:

[0196] Exposure variables (e.g., oral frailty): By inputting the exposure variables and related features (e.g., lifestyle, diet, chronic diseases, etc.) at the current time point into the LSTM model, the model will output the predicted value of oral frailty in the future.

[0197] Outcome variables (such as sarcopenia): Similarly, the LSTM model will output a predicted value for sarcopenia based on the outcome variables and related features at the current time point.

[0198] 42) Analyze the changing trends of oral weakness and sarcopenia:

[0199] Based on the predicted values ​​of exposure and outcome variables at future time points output by the LSTM model, trend analysis can be performed. Specifically, this can be done from the following aspects:

[0200] Visualizing and predicting trends:

[0201] Time series plot: A time series plot is used to show the predicted trends of oral atrophy and sarcopenia. The horizontal axis represents time, and the vertical axis represents the predicted value. By comparing the predicted results at different time points, the changing trends of oral atrophy and sarcopenia over a future period can be clearly seen.

[0202] If the predictive value for oral weakness continues to rise while the predictive value for sarcopenia declines, this may mean that the negative impact of oral weakness on sarcopenia is intensifying.

[0203] If the trends of change of the two show synchronous or opposite trends, the interaction or causal relationship between the two can be further explored.

[0204] Rate of change analysis:

[0205] Rate of change: Calculating the rate of change of oral weakness and sarcopenia within each time period (e.g., using the difference or percentage change between two time points) can help quantify the speed and magnitude of change. This allows us to determine the trend and rate of change between different time points.

[0206] Comparing the changing trends of different groups:

[0207] Based on different group characteristics (such as age, gender, chronic diseases, etc.), the changing trends of different groups can be analyzed in groups. For example:

[0208] Compare the differences in oral frailty and sarcopenia between healthy individuals and those with chronic diseases.

[0209] This study analyzes the progression of oral frailty and sarcopenia in different age groups and explores the impact of age on the trends of both conditions.

[0210] For example, younger older adults may experience slower frailty and muscle loss than older older adults.

[0211] Causal inference based on prediction results:

[0212] If the model has already established causal relationships, further analysis could explore the causal relationship between oral frailty and sarcopenia. By observing the prediction results at different time points and combining them with time-varying confounding factors removed through adversarial learning, we could explore which exposure factors (such as oral health, diet, exercise, etc.) have a significant causal effect on changes in sarcopenia.

[0213] For example, the long-term effects of oral weakness on sarcopenia can be quantified by estimating the causal effects (such as ATE, Average Treatment Effect) output by the model.

[0214] The mechanism of action analysis module is used to analyze the specific mechanisms by which oral weakness affects sarcopenia using structural equation modeling.

[0215] Specifically, the mechanism of action analysis module, when using structural equation modeling to analyze the specific mechanisms by which oral frailty affects sarcopenia, includes:

[0216] 1) Identify latent variables and determine the corresponding observed variables based on the latent variables. The latent variables include oral frailty, sarcopenia, inflammatory factors, cognitive function, depressive state, and nutritional status; define the causal relationship and pathway between the latent variables and the observed variables.

[0217] 11) Identify latent variables:

[0218] Oral weakness: Oral health status (measured by oral examination score, number of teeth, chewing function, etc.).

[0219] Sarcopenia: Muscle mass and function (measured by muscle mass, grip strength, gait speed, etc.).

[0220] Inflammatory factors: such as C-reactive protein (CRP), serum albumin and other indicators.

[0221] Cognitive function: Assessed using a cognitive function scale (such as MMSE).

[0222] Depressive state: assessed using a depression scale (such as the PHQ-9).

[0223] Nutritional status: assessed through dietary intake, weight changes, and malnutrition screening tools.

[0224] 12) Determine the observed variables:

[0225] For each latent variable, determine the corresponding observed variable (i.e., the actual measured data). For example:

[0226] Oral health indicators include the number of missing teeth and chewing difficulty scores.

[0227] Indicators of sarcopenia include grip strength and muscle mass (measured via DXA or BIA).

[0228] Inflammatory factors: such as serum CRP, white blood cell count, etc.

[0229] Cognitive function: such as MMSE score.

[0230] Depressive state: such as PHQ-9 score.

[0231] Nutritional status: such as weight, BMI, dietary survey data, etc.

[0232] Confounding factors such as age, gender, education level, bone density, and history of chronic diseases should be included as control variables in the model.

[0233] 13) Determine the causal relationship and path:

[0234] Define the causal relationships between latent variables and determine the paths of mediating and confounding variables.

[0235] Causal relationship hypothesis:

[0236] Oral weakness → inflammatory factors: Oral health problems may trigger or exacerbate inflammatory responses.

[0237] Oral weakness → Cognitive function: Oral weakness may affect the nutritional intake of older adults, thereby affecting their cognitive function.

[0238] Oral health problems may increase the risk of depression.

[0239] Oral health problems can affect the ability to eat, leading to malnutrition.

[0240] Mediating pathways: Inflammatory factors, cognitive function, depression, and nutritional status may further influence the occurrence of sarcopenia.

[0241] Control paths: Variables such as age, gender, education level, bone density, and chronic disease history can affect all paths, so they need to be included in the model control.

[0242] 2) Use factor analysis to verify the matching degree between latent variables and observed variables to ensure the reliability and validity of measurement results, set causal paths between latent variables, perform path analysis, estimate the coefficients of each path, and obtain structural equation models;

[0243] Specifically, before establishing a structural equation model, it is necessary to evaluate the measurement model to ensure that the observed variables of each latent variable accurately reflect its potential meaning.

[0244] 21) Check the data fit:

[0245] Data distribution: Check whether the data conforms to a normal distribution.

[0246] Missing value handling: Fill in missing values ​​in the data, or use appropriate imputation methods (such as multiple imputation).

[0247] Multicollinearity: Ensure that there is no serious multicollinearity problem among the independent variables.

[0248] 22) Evaluation of the measurement model:

[0249] Confirmatory Factor Analysis (CFA) is used to examine the relationship between each latent variable and the observed variable. Specific checks include:

[0250] Factor loadings: Whether each observed variable can significantly reflect the latent variables.

[0251] Reliability and validity: Whether the measurement tool is reliable (e.g., Cronbach's α coefficient) and effective (e.g., construct validity).

[0252] 23) Constructing a structural equation model: After the measurement model is validated, construct a structural model, that is, define the causal relationships (paths) between latent variables. Use software such as AMOS, Mplus, LISREL, or the lavaan package in R to perform path analysis.

[0253] Path analysis:

[0254] Direct effects: Analysis of the direct impact of oral weakness on sarcopenia (i.e., path coefficient).

[0255] Indirect effects: The indirect impact of oral weakness on sarcopenia was analyzed using mediating variables (such as nutrition, activity, and psychological state).

[0256] Total effect: The total effect is the sum of the direct and indirect effects, reflecting the overall impact of oral weakness on sarcopenia.

[0257] 3) Detect the goodness of fit of the structural equation model, interpret the path coefficients, and analyze the mechanism of oral weakness on sarcopenia based on the path coefficients.

[0258] 31) Model fit evaluation:

[0259] Common metrics for checking model fit include:

[0260] Chi-square test: Used to test the goodness of fit between a model and data; a larger p-value indicates a better fit.

[0261] RMSEA (Root Mean Square Error of Approximation): A value less than 0.08 indicates a good fit.

[0262] CFI (Comparative Fit Index): A value greater than 0.90 indicates that the model fits well.

[0263] TLI (Tucker-Lewis Index): A value greater than 0.90 indicates a good model fit.

[0264] If the model fits poorly, adjustments can be made by modifying the indices, or by considering modifying the paths and relationships.

[0265] 32) Interpretation and Analysis of Results:

[0266] Interpreting the path coefficient:

[0267] Path coefficient (β): Used to explain the causal relationship between latent variables. For example, the direct path coefficient of oral weakness to sarcopenia is 0.3, indicating that oral weakness has a positive effect on sarcopenia, and the degree of influence is 30%.

[0268] Indirect effects: For example, oral weakness affects the indirect pathway coefficient of sarcopenia through nutritional status, indicating that oral health may indirectly affect the occurrence of sarcopenia by influencing the nutritional status of the elderly.

[0269] Data visualization module 3 is used to generate data visualization charts;

[0270] Specifically, various graphics libraries, such as Matplotlib, Seaborn, and Plotly, can be used to generate data visualization charts. These charts can help clearly show the relationships between data related to oral frailty, sarcopenia, genetic instrumental variables, and other health-related factors.

[0271] Interactive feedback module 4 is used to interact with users and provide personalized feedback;

[0272] Specifically, the core function of the interactive feedback module is to enhance the user experience by improving the accuracy of analysis and the user's understanding through dynamic interaction and personalized feedback.

[0273] The goals of the interactive feedback module are as follows:

[0274] Real-time user interaction: Allows users to input specific needs, such as adjusting statistical thresholds, selecting specific SNPs, and filtering data for specific groups of people.

[0275] Personalized feedback: Based on user-input data, specific analysis results are provided, such as SNP statistics, significance analysis, causal inference results, etc.

[0276] Dynamic parameter adjustment: Users can adjust parameters such as effect size threshold, significance level, instrument variable screening criteria, etc., and the system will dynamically update the results.

[0277] Visual feedback: Charts and tables help users understand the analysis results, such as the effect size of genetic instrumental variables, p-value distribution, F-statistic, etc.

[0278] In summary, by utilizing the technical solutions described above in this invention and by comprehensively employing techniques such as data acquisition, impact analysis, longitudinal variation analysis, causal inference, and structural equation modeling, a systematic analytical framework is provided. This framework enables in-depth research into the impact of oral frailty on sarcopenia, not only enhancing the depth and breadth of research on the relationship between oral frailty and sarcopenia but also providing a scientific, precise, and personalized solution for the health management of the elderly population.

[0279] Furthermore, this invention employs statistical methods and Mendelian randomization to reveal a potential causal relationship between oral atony and sarcopenia through in-depth analysis. In particular, the selection of genetic instrumental variables and the support of multi-sample data improve the accuracy and scientific rigor of causal inference, avoiding the interference of confounding factors.

[0280] Furthermore, this invention utilizes a Long Short-Term Memory (LSTM) network model from deep learning, combined with adversarial learning techniques for time-varying confounding factors, to predict the changing trends of oral frailty and sarcopenia at different time points. This technique effectively reduces the influence of time-varying confounding factors, providing a more accurate basis for future prediction and intervention.

[0281] Furthermore, this invention utilizes structural equation modeling (SEM) to deeply analyze how oral weakness affects sarcopenia through multiple pathways. Factor analysis and pathway analysis reveal the direct impact of oral weakness on sarcopenia and its potential mediating mechanisms, leading to a better understanding of the influence of oral weakness on sarcopenia.

[0282] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An analytical system for the impact of oral weakness on sarcopenia, characterized in that, It includes a data acquisition module, an impact analysis module, a data visualization module, and an interactive feedback module connected in sequence; The data acquisition module is used to collect relevant data on oral weakness and sarcopenia; The data visualization module is used to generate data visualization charts; The interactive feedback module is used to interact with users and provide personalized feedback; The impact analysis module includes a correlation analysis module, a causal relationship analysis module, a longitudinal change analysis module, and a mechanism of action analysis module connected in sequence. The correlation analysis module is used to analyze the relationship between oral frailty indicators and sarcopenia indicators using statistical methods. The causal relationship analysis module is used to analyze the causal relationship between genetic instrumental variables and the risk of oral frailty and sarcopenia by using a two-sample Mendelian randomization method combined with genome-wide association study data. The longitudinal change analysis module is used to analyze the changing trends of oral frailty and sarcopenia in the elderly population at different time points using a deep learning model combined with time-varying confounding factor adversarial learning technology; specifically including: We obtained longitudinal data on oral frailty and sarcopenia at multiple time points in the database of the elderly population, and preprocessed the data to select key features that affect changes in oral frailty and sarcopenia. The longitudinal data included exposure variables, outcome variables and time-varying confounding factors. The time-varying confounding factor is input into the generator to generate an exposure variable that removes the influence of the confounding factor. The discriminator receives the generated exposure variable and the original confounding factor data, evaluates the impact of the time-varying confounding factor on the exposure variable, and trains the generator and discriminator alternately against each other. Using the exposure variables, outcome variables, and confounding factors at each time point as input data, and the predicted values ​​of the exposure variables and outcome variables at future time points as output, a long short-term memory network time series model is constructed; an adversarial regularization term is added to the loss function of the long short-term memory network, and the long short-term memory network time series model is trained using adversarial regularization. The trained long short-term memory network time series model was used to output the predicted values ​​of exposure variables and outcome variables at future time points, and the changing trends of oral frailty and sarcopenia in the elderly population at different time points were analyzed based on the prediction results. The mechanism of action analysis module is used to analyze the specific mechanism by which oral frailty affects sarcopenia using structural equation modeling; specifically, it includes: Identify latent variables and determine corresponding observed variables based on them. The latent variables include oral frailty, sarcopenia, inflammatory factors, cognitive function, depressive state, and nutritional status. Define the causal relationship and pathway between the latent variables and the observed variables. Factor analysis was used to verify the matching degree between latent variables and observed variables to ensure the reliability and validity of the measurement results. Causal paths between latent variables were defined, and path analysis was performed to estimate the coefficients of each path and obtain the structural equation model. The goodness of fit of the structural equation model was tested, the path coefficients were interpreted, and the mechanism of the influence of oral weakness on sarcopenia was analyzed based on the path coefficients.

2. The analytical system for the effect of oral weakness on sarcopenia according to claim 1, characterized in that, The data acquisition module includes an oral frailty index acquisition module, a sarcopenia index acquisition module, a health index acquisition module, and a data preprocessing module connected in sequence. The oral frailty index collection module is used to collect oral frailty indicators of the elderly population, and the oral frailty indicators include the number of teeth, chewing function and swallowing function. The sarcopenia index collection module is used to collect sarcopenia indicators of the elderly population, and the sarcopenia indicators include grip strength, gait speed and muscle mass. The health indicator collection module is used to collect other relevant health data of the elderly population, including inflammatory indicators, cognitive function, depressive state, nutritional status and activity level. The data preprocessing module is used to perform missing value imputation, outlier detection, and data standardization on the collected data to ensure data quality and consistency.

3. The analytical system for the effect of oral weakness on sarcopenia according to claim 1, characterized in that, The correlation analysis module, when analyzing the relationship between oral frailty indicators and sarcopenia indicators using statistical methods, includes: Calculate the basic statistics for each indicator variable, and draw frequency distribution plots, histograms, and box plots based on the calculation results to check the distribution of the data; Assess the individual distribution of oral frailty and sarcopenia indicators, and select statistical methods based on the assessment results. Parametric statistical methods are selected when the assessment results conform to a normal distribution, and nonparametric statistical methods are selected when the assessment results do not conform to a normal distribution. We used selected statistical methods to analyze the linear or nonlinear relationship between oral weakness and sarcopenia indicators, and combined this with regression analysis to analyze the linear effect of oral weakness on sarcopenia.

4. The analytical system for the effect of oral weakness on sarcopenia according to claim 1, characterized in that, The causal relationship analysis module includes a genetic instrument variable selection module, a Mendelian randomization analysis module, an analysis result verification module, and a causal relationship output module connected in sequence. The genetic instrument variable selection module is used to obtain association data between oral frailty and sarcopenia from two independent genome-wide association study datasets, and extract single nucleotide polymorphism sites associated with oral frailty and sarcopenia as genetic instrument variables, respectively. The Mendelian randomization analysis module is used to evaluate the effects of genetic instrumental variables on oral asthenia and sarcopenia, and to infer the causal relationship between oral asthenia and sarcopenia using multiple Mendelian randomization methods. The analysis result verification module is used to evaluate whether the assumptions of Mendelian randomization analysis are valid, so as to ensure the reliability of causal inference. The causal relationship output module is used to interpret the analysis results and output the causal relationship between oral weakness and sarcopenia.

5. The analytical system for the effect of oral weakness on sarcopenia according to claim 4, characterized in that, The genetic instrument variable selection module is also used to evaluate the strength of genetic instrument variables in order to avoid bias and inefficiency caused by weak instrument variables, enhance the statistical power and robustness of causal inference, and improve the credibility of research results. The formula for assessing the strength of genetic instrumental variables is as follows: In the formula, F Indicates the strength of genetic instrumental variables. R 2 This represents the degree of variability in the explanation of the exposure variable by the genetic instrumental variable. n Indicates the number of samples. k This indicates the number of genetic instrumental variables.

6. The analytical system for the effect of oral weakness on sarcopenia according to claim 4, characterized in that, The Mendelian randomization analysis module, when assessing the effects of genetic instrumental variables on oral asthenia and sarcopenia, and using various Mendelian randomization methods to infer the causal relationship between oral asthenia and sarcopenia, includes: Effect estimates, standard errors, and statistical significance of genetic instrumental variables associated with oral frailty were obtained from genome-wide association study data, and effect allele orientations were aligned to ensure that the effect orientations of genetic instrumental variables were consistent across different datasets. Effect estimates, standard errors, and statistical significance of genetic instrumental variables associated with sarcopenia were obtained from genome-wide association study data of sarcopenia, and the direction of effect alleles was checked to ensure that the direction of effect alleles in oral frailty and sarcopenia data were consistent. The weighted average effect value of genetic instrumental variables was calculated using the inverse variance weighting method, and the preliminary causal effect of oral frailty on sarcopenia was analyzed based on the weighted average effect value. The regression adjustment method was used to detect level pleiotropic bias in order to check whether genetic instrumental variables affect other pathways, and the weighted median method was used to improve robustness when some genetic instrumental variables were ineffective. The outlier adjustment method was used to identify and remove abnormal genetic instrumental variables that had a significant impact on the results. Sensitivity analysis was performed on the removed genetic instrumental variables, and the causal effects were summarized to obtain the causal relationship between oral weakness and sarcopenia.

7. The analytical system for the effect of oral weakness on sarcopenia according to claim 4, characterized in that, The analysis result verification module includes a correlation hypothesis evaluation module, an independence hypothesis evaluation module, and an exclusion hypothesis evaluation module connected in sequence. The correlation hypothesis assessment module is used to assess the effectiveness of genetic instrument variables based on their effect estimates, strength, and significance tests. The independence hypothesis assessment module is used to test whether the genetic instrumental variable is related to confounding factors and whether it satisfies the independence hypothesis. The exclusion hypothesis assessment module is used to assess whether genetic instrumental variables affect outcome variables only through exposure variables.

8. The analytical system for the effect of oral weakness on sarcopenia according to claim 1, characterized in that, The expression for the loss function of a Long Short-Term Memory (LSTM) network is: In the formula, L Represents the loss function. E Indicates the expectation symbol, y t Indicates the first t The actual target value at a given time point. Indicates the first t Predictions from point-in-time models λ a , These are used to adjust the hyperparameters of the adversarial loss and the regularization term with respect to the total loss, respectively. D Represents the discriminator model. f ( X t , Z t ) represents the output prediction of the Long Short-Term Memory network temporal model. f This represents a temporal model of the Long Short-Term Memory network. X t Indicates the first t Exposure variables at specific points in time Z t Indicates the first t Time-varying confounding factor at a given time point. This represents the L2 regularization term.

Images