A red blood cell five-methyl tetrahydrofolate level prediction system and application

Abstract

The application provides a red blood cell five-methyl tetrahydrofolic acid level prediction system, utilizes complete red blood cell folic acid clinical data to establish a half-life increment parameter library and a genotype correction coefficient parameter library, calls matched prediction half-life increments and genotype correction coefficients through individual drug use parameters of a to-be-tested object, calculates the current red blood cell folic acid level, and determines whether a standard is reached according to a prediction result and gives an abnormal index and adjustment suggestions, so as to provide data support for folic acid supplement guidance and disease risk prevention and control.

Description

Technical Field

[0001] This invention relates to the medical field, and more specifically, to a predictive system for predicting current erythrocyte pentamethyltetrahydrofolate levels based on folic acid usage regimens. Background Technology

[0002] Red blood cell folate levels are a core indicator reflecting the body's folate storage status. The level of folate is directly related to the risk of birth defects in fetuses during the perinatal period and the probability of diseases such as megaloblastic anemia. Accurate prediction of red blood cell folate levels is of great significance for personal health management and clinical intervention guidance.

[0003] Current predictive models still have certain problems. For example, the core metabolic kinetic parameters in existing models are fixedly assigned based on the population average of their respective research samples, rather than being adjusted according to individual differences, thus lacking the ability to perform dynamic calibration driven by real metabolic data. Furthermore, the model validation data in current technologies are derived from fixed supplementation cycle nodes (such as 4 weeks, 8 weeks, or short-term intensive sampling), which cannot handle supplementation and review data of the test subjects at arbitrary time intervals, nor can it utilize these scattered real-world data for parameter back-calculation and dynamic correction. In addition, the detection methodologies in current technology models have fundamental limitations; the selected microbiological experimental procedures are cumbersome and time-consuming. The long timeframe, high technical requirements, and large batch-to-batch variability make it difficult to achieve large-scale, standardized application in real-world clinical scenarios. Furthermore, other methods used in existing technologies cannot accurately characterize the long-term folic acid reserve status of red blood cells. In addition, the MTHFR gene is a human gene encoding methylenetetrahydrofolate reductase, which converts folic acid into its active form (5-methyltetrahydrofolate) and participates in key processes such as DNA synthesis and amino acid metabolism. It is divided into CC (wild-type), CT (heterozygous mutant), and TT (homozygous mutant), and different genotypes are also key factors affecting folic acid metabolism efficiency. Existing technologies do not consider this factor when building prediction models, thus reducing the reliability of prediction results.

[0004] Therefore, there is an urgent need in the clinical and health management fields for an accurate, convenient, personalized red blood cell folate level prediction system that takes into account MTHFR gene factors, so as to provide data support for folate supplementation guidance and disease risk prevention and control. Summary of the Invention

[0005] To address the problems existing in the prior art, this invention provides a red blood cell pentamethyltetrahydrofolate level prediction system. It utilizes complete clinical data on red blood cell folate to establish a half-life increment parameter library and a genotype correction coefficient parameter library. By calling the matched predicted half-life increment and genotype correction coefficient based on the individual medication parameters of the subject, it calculates the current red blood cell folate level and determines whether it meets the target level based on the prediction results, providing an abnormal index and adjustment suggestions. This provides data support for folate supplementation guidance and disease risk prevention and control.

[0006] Terminology Definition In this invention, red blood cell folate level is a core indicator reflecting the body's folate storage status, measured in nanomoles per liter (nmol / L), and is directly related to the risk of birth defects, megaloblastic anemia, and other diseases in the fetus during the perinatal period.

[0007] In this invention, the half-life increment (Δt) refers to the increase in erythrocyte folate levels over a basal half-life (60 days) after folate supplementation, expressed in nmol / L. This value is derived from clinical sample training, with different standard values ​​corresponding to different folate formulations and dosages. Mature erythrocytes do not exchange folate with plasma; folate is only loaded during erythrocyte formation and remains constant until erythrocyte death (lifespan 120 days). Therefore, changes in RBC folate are entirely determined by erythrocyte population turnover, not metabolic elimination.

[0008] In this invention, the initial folic acid level (C0) is the erythrocyte folic acid level before the user begins folic acid supplementation, measured in nmol / L, and can be obtained through detection, inference from detection values, or by averaging of clinical big data.

[0009] In this invention, the plateau phase refers to the stage where the erythrocyte folate level increases to its maximum value and no longer increases with the duration of supplementation. The maximum value is Cmax = C0 + 5 × Δt.

[0010] In this invention, the anomaly index is used to quantify the degree to which the predicted value of erythrocyte folate deviates from the target range. It is calculated when the predicted value fails to meet the target, and the larger the value, the greater the degree of deviation.

[0011] In this invention, the folic acid supplementation regimen includes folic acid dosage forms (regular folic acid, active folic acid), dosages (0.4mg, 0.8mg), and supplementation duration, including single dosage form dosages and combinations of two dosage forms. The chemical name of regular folic acid is pteroylglutamic acid, which is an unactivated folic acid precursor that requires two enzymatic conversions in the body to exert its effects. Active folic acid is pentamethyltetrahydrofolate, which is the final active form of folic acid that has already undergone metabolic conversion in the human body. It is itself a functional structure that can directly participate in metabolism and can be absorbed and utilized without additional conversion.

[0012] The above-mentioned objective of the present invention is achieved by providing the following technical solution: On one hand, the present invention provides a erythrocyte pentamethyltetrahydrofolate level prediction system, the prediction system including a parameter library calling module, the parameter library including a half-life increment parameter library, the parameter library is calculated by collecting and screening a complete erythrocyte folate clinical database to obtain the half-life increment under different clinical protocols and aggregating them to form the parameter library, the parameter library calling module can call the matching half-life increment according to the parameters of the object to be predicted.

[0013] In some approaches, the prediction system employs machine learning algorithms to establish a correlation model between parameters.

[0014] Furthermore, the machine learning algorithms are linear regression and random forest.

[0015] In real clinical scenarios, the timing of follow-up examinations for patients' erythrocyte pentamylic acid (PEA) levels is highly random. Existing predictive models often rely on data from fixed supplementation cycles, such as 4 weeks, 8 weeks, or short-term intensive sampling, with fixed values ​​for core parameters. This severely limits the model's generalization ability and real-world applicability. In contrast, this invention collects and filters a large number of complete erythrocyte PEA samples, expanding the coverage to different dosing regimens. This results in a library of calculated half-life increment parameters that encompasses various clinical situations and can be used to call up half-life increments for different prediction subjects, thereby achieving personalization of the erythrocyte PEA level prediction system.

[0016] In some embodiments, the parameter library calling module further includes a genotype correction coefficient parameter library, which can call matching genotype correction coefficients based on the different genotypes and half-life increments of the MTHFR 677 locus of the object to be predicted.

[0017] The MTHFR genotype represents different variants of the methylenetetrahydrofolate reductase gene, whose core function is to regulate folate metabolism. Clinically, the C677T locus is most commonly detected. There are three main genotypes: CC (wild-type), CT (heterozygous mutant), and TT (homozygous mutant). These three genotypes correspond to different folate metabolic capabilities, with the TT genotype exhibiting significant folate metabolic disorders. Therefore, adjusting folate supplementation regimens based on MTHFR genotype is crucial for accurately reducing health risks and avoiding indiscriminate supplementation. This invention incorporates MTHFR genotype as a factor in model construction, enabling the model to predict differentiated metabolic trajectories based on individual polymorphisms in key folate metabolism enzyme genes. The standard values ​​for half-life increments and genotype correction coefficients corresponding to folate supplementation regimens are shown in Table 1.

[0018] Table 1 Standard values ​​of half-life increment and genotype correction coefficients corresponding to folic acid supplementation regimens In some embodiments, the system comprises a data input module, a parameter library calling module, a mathematical calculation module, a standard determination module, and a result output module.

[0019] Five system modules form a cohesive whole, realizing the entire process from individual information input, data calculation and analysis, to result output. The data input module collects basic parameter information of the test subjects, providing individual data through parameter library retrieval and mathematical calculations. The parameter retrieval module retrieves matching core parameters based on the individual information obtained from the data input module. The mathematical calculation model combines the information collected by the data input module with the matching half-life increment and genotype correction coefficient retrieved by the parameter retrieval module to calculate the erythrocyte folate level at a specific time. The target assessment module determines whether the erythrocyte folate level meets the target based on the test subject's gestational age and erythrocyte folate level standards. The result output module outputs the final erythrocyte folate level and target achievement status. With this system, only basic information about the test subject needs to be input to calculate and determine the erythrocyte folate level and provide the corresponding conclusions.

[0020] In some embodiments, the data input module is used to receive the initial red blood cell folate level, folate formulation, folate dosage, duration of folate supplementation, number of days of folate supplementation interruption, and current gestational age information of the subject to be predicted.

[0021] Red blood cell folate levels and current gestational age are key parameters in developing a folate supplementation plan. Folate formulation, dosage, and duration of supplementation are key information for clinical medication regimens. The number of days of folate supplementation interruption is an important influencing factor in predicting red blood cell folate levels. Using all of the above information as data input can more accurately match the half-life increment, making the prediction results more accurate.

[0022] In some embodiments, the parameter calling module may call a matching half-life increment parameter from the half-life increment parameter library based on the parameters of the object to be predicted received by the data input module, or call a matching genotype correction parameter based on the MTHFR 677 locus genotype of the object to be predicted and other information collected by the data input module.

[0023] In existing technical models, the half-life increment parameter is derived from empirical values ​​in literature or small-sample experimental data and is fixedly assigned. This fixed assignment is not suitable for predicting different samples in actual clinical situations. Therefore, this invention uses the erythrocyte folate half-life increment as the core calibration parameter, achieving a paradigm shift from "empirical assignment" to "data-driven" approaches. This ensures that the parameters of the object to be predicted match the half-life increment parameter, further improving the system's personalization and accuracy. Furthermore, existing technologies do not incorporate the MTHFR 677 locus genotype into the prediction of erythrocyte folate levels. This invention, through experimental screening and verification, demonstrates the importance of the half-life increment parameter and genotype correction parameter in the prediction model system, which can improve the accuracy of the prediction model system.

[0024] In some methods, the mathematical calculation module acquires the initial red blood cell folate level, folate formulation, folate dosage, folate supplementation duration, number of days of folate supplementation interruption, and current gestational age information of the subject to be predicted collected by the data input module, as well as the matching half-life increment parameter retrieved by the parameter calling module, and substitutes them into the logistic regression equation to calculate and predict the red blood cell folate level at different times.

[0025] In some embodiments, the mathematical computation module includes the following logistic regression equation: When the object to be predicted does not have a folic acid interruption: C(k) = C0+S×Δt×(1-0.5^(N×k)) / (1-0.5); When the object to be predicted has a folic acid interruption: C(k) = C(k-1)-C(d), where C(d) = S×Δt×(1-0.5^(N×d)) / (1-0.5) When the subject of prediction changes their folic acid supplementation regimen and / or receives genotype information: According to C(k) = C(k-1) + Cs(k), on any day before changing the folic acid supplementation regimen and / or supplementing genotype information, C(k-1) = C0 + S_old × Δt_old × (1 - 0.5^(N × (k-1))) / (1 - 0.5). On the day of changing the folic acid supplementation regimen and / or supplementing genotype information, and on any subsequent day, Cs(k) = [C0 + S_new × Δt_new × (1 - 0.5^(N × k)) / (1 - 0.5)] - [C0 + S_new × Δt_new × (1 - 0.5^(N × (k-1))) / (1 - 0.5)]. Therefore, on the day of changing the folic acid supplementation regimen and / or supplementing genotype information, C(k) = C0+S_old×Δt_old×(1-0.5^(N×(k-1))) / (1-0.5)+[C0+S_new×Δt_new×(1-0.5^(N×k)) / (1-0.5)]-[C0+S_new×Δt_new×(1-0.5^(N×(k-1))) / (1-0.5)]; When the folate level in the red blood cells of the test subject reaches a plateau: When C(k) ≥ C0 + 5 × Δt, C(k) is fixed as Cmax = C0 + 5 × Δt. Where C(k) is the predicted level of erythrocyte folate after k days of supplementation; C0 is the initial level of erythrocyte folate; Δt is the standard value of half-life increment; N is the coefficient, with a value of 1 / 60 = 0.0167; S is the genotype correction coefficient, which is 1 when the genotype is not determined; k is the number of days of folate supplementation; d is the number of days of folate supplementation interruption; C(d) is the decrease in erythrocyte folate level after d consecutive days of no supplementation; C(k-1) is the level of erythrocyte folate after k-1 days of supplementation; Cs(k) is the daily increase in folate level after k days of supplementation; when changing the folate supplementation plan, the variable Δt needs to be reset, and the Δt of the new plan and the old plan are defined as Δt_new and Δt_old, respectively; when supplementing genotype information, the variable S needs to be reset, and the S before supplementing genotype information and after supplementing genotype information are defined as S_new and S_old, respectively; Cmax is the maximum value of erythrocyte folate level.

[0026] The mathematical calculation module substitutes the individual's actual parameters and the corresponding core calibration parameters into the regression equation for calculation. It also takes into account whether there are interruptions during medication, changes in the supplementation regimen, and whether the red blood cell folate level has reached a plateau. The calculation method is designed to be targeted so that the system can be applied to a variety of different real-world scenarios, maintain the accuracy of the system's predictions, and avoid prediction biases caused by complex situations resulting from calculation methods derived from a single case.

[0027] In some methods, the compliance determination module can determine whether C(k) meets the standard according to the determination criteria. If the module determines that the standard is not met, the determination module can perform an anomaly index calculation.

[0028] The judgment criteria are shown in Table 2. The lower limit refers to the red blood cell folate level required for the prevention of hyperhomocysteinemia and fetal neural tube defects in the "Multidisciplinary Expert Consensus on Rational Folic Acid Supplementation in Chinese Clinical Practice (2020)" and the total red blood cell folate in the perinatal population in the "Guidelines for Folic Acid Supplementation in the Periconception Period to Prevent Neural Tube Defects (2017)". The upper limit refers to the concentration of red blood cell folate in 97.5% of the population in the data of 120,000 perinatal women and the red blood cell folate level for the prevention of gestational diabetes. When the predicted red blood cell folate level C(k) is between the upper and lower limits of the corresponding gestational week (inclusive), it is judged as meeting the target. When C(k) is lower than the lower limit or higher than the upper limit, it is judged as not meeting the target, and the abnormality index is calculated. The outlier index (Index) is calculated as follows: when C(k) does not meet the standard, the outlier value outline = |C(k) - lower limit| (C(k) < lower limit) or outline = C(k) - upper limit (C(k) > upper limit); Index = 2 outline / Step Step is the step size, with a value of 50. The system adds a target achievement determination module, which can assess the target achievement of the test subjects and provide abnormality alerts. It not only provides predicted level values, but also provides a complete mechanism including abnormality determination, providing guidance for personal health management and data support for adjusting clinical folic acid supplementation programs, making it highly practical.

[0029] In some methods, the step size in the anomaly index calculation can be adjusted to 40 or 60, or Index = 2 can be used. outline / Step The calculation method is used to provide error messages.

[0030] Table 2. Criteria for Compliance Assessment Module In some methods, the result output module can output the predicted level of erythrocyte folate at a specific time, the status of compliance, the abnormality index, and adjustment suggestions based on the judgment conclusion obtained by the compliance judgment module. The adjustment suggestions include whether the current folate supplementation regimen needs to be adjusted.

[0031] Compared with existing technologies, this system upgrades from a simple numerical calculation tool to a complete decision support system that includes "assessment-judgment-early warning" functions. It directly links the prediction results with clinical decisions, providing a complete closed loop from "predicted value" to "action recommendations". In clinical practice, it can conduct personalized assessments for different patients and provide targeted adjustment suggestions.

[0032] On the other hand, the present invention provides a method for constructing a half-life incremental parameter library for a parameter library calling module of a pregnancy erythrocyte pentamethyltetrahydrofolate level prediction system, comprising the following steps: (1) Data screening: Screening clinical sample data containing complete folic acid supplementation information and two or more red blood cell folic acid test results; (2) Sample splitting: Clinical sample data with the same folic acid supplementation information are grouped together, and all clinical sample data are grouped. (3) Calculation of half-life increment for a single sample: For each sample in each group, obtain the erythrocyte folate level Ca of the first test, the folate supplementation time t, and the erythrocyte folate level Cb of the supplemented sample. Calculate the half-life increment Δt of the sample using the half-life calculation formula: Δt = (C(b)-C(a)) / [(1-0.5^(N×t)) / (1-0.5)], N: coefficient, with a value of 1 / 60 = 0.0167; (4) Within-group mean calibration: Statistical analysis is performed on the half-life increment Δt of each group sample. After removing outliers, the mean value within the group is calculated and used as the standard value of the half-life increment corresponding to the folic acid dosage form and dose. The result is stored in the half-life increment parameter library.

[0033] In another aspect, the present invention provides the use of the above-mentioned system for constructing a model for predicting the level of pentamethyltetrahydrofolate in erythrocytes during pregnancy.

[0034] The procedure for predicting erythrocyte folate levels using the erythrocyte pentamethyltetrahydrofolate level prediction system described in this invention is as follows: Data Input: Through the data input module, input the user's initial red blood cell folate level C0, folate supplementation plan (dosage form, dosage), supplementation duration k, number of days of supplementation interruption d, and current gestational week; Half-life increment call: The half-life increment call module calls the pre-stored half-life increment standard value Δt and genotype correction coefficient based on the input folic acid formulation and dosage and the MTHFR677 locus genotype (if any); Level prediction: Through the mathematical calculation module, based on the input parameters and Δt, the predicted value C(k) of erythrocyte folate at the current time point is calculated using the folate level growth formula and decline formula (if there is an interruption); if a plateau is reached, the predicted value is output according to Cmax. Target attainment determination: The target attainment determination module determines the attainment status of C(k) based on the target range corresponding to the current gestational week; if the target is not met, the abnormality index is calculated. Results output: The results output module outputs the predicted trajectory of red blood cell folate levels, the status of compliance, the abnormal index (if any), and personalized prompts.

[0035] Clinical sample data from June 2025 to May 2026 were obtained from the erythrocyte folate clinical database. The samples must include complete folate supplementation information (dosage form, dosage, and duration of supplementation), initial erythrocyte folate level, and follow-up erythrocyte folate level. Exclusion criteria include past vascular disease, any disease affecting intestinal absorption, cancer, diabetes, and switching of supplementation dosage form and dosage between two tests. The initial erythrocyte folate level, folate dosage form, dosage, duration of supplementation, and MTHFR 677 locus genotype (if any) of the validation samples were input into the analysis model described in this invention to obtain the predicted erythrocyte folate level. The predicted value was compared with the actual follow-up results of the validation samples to calculate the prediction accuracy. The allowable deviation range was set at ±15%. If the prediction accuracy was ≥80%, the model was considered to have passed validation and met the application requirements.

[0036] The beneficial effects of this invention are as follows: 1. The designed erythrocyte pentamethyltetrahydrofolate level prediction system has high accuracy in predicting erythrocyte folate levels.

[0037] 2. The designed erythrocyte pentamethyltetrahydrofolate level prediction system can achieve individualized prediction and adapt to the metabolic differences of different individuals.

[0038] 3. The designed erythrocyte pentamethyltetrahydrofolate level prediction system can achieve standard determination and abnormality alerts, and is highly practical.

[0039] 4. The designed erythrocyte pentamethyltetrahydrofolate level prediction system can handle a variety of complex supplementation scenarios.

[0040] 5. The erythrocyte folate level prediction system designed uses erythrocyte folate levels obtained by mass spectrometry, which is the same as the actual clinical detection method, ensuring the consistency between the model input and output indicators. Attached Figure Description

[0041] Figure 1 A schematic diagram illustrating the timeline division for training and validation; Figure 2 The effect of different ages on the increase in half-life; Figure 3 The effect of different body weights (BMI) on the increase in half-life; Figure 4 A comparison of the actual half-life increments in historical data for different folic acid formulations, dosages, and MTHFR 677 locus genotypes; Figure 5 This is an example of a module for predicting and determining erythrocyte folate levels. Detailed Implementation

[0042] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be noted that the embodiments described below are intended to facilitate the understanding of the present invention and do not limit it in any way.

[0043] Example 1: Erythrocyte Pentamethyltetrahydrofolate Level Prediction System The erythrocyte pentamethyltetrahydrofolate level prediction system provided by this invention consists of five core modules: a data input module, a parameter calling module, a mathematical calculation module, a target determination module, and a result output module. The parameter calling module contains a parameter library, including a half-life increment parameter library and a genotype correction parameter library. The parameter calling module needs to be trained on the model using clinical data. After the system is built, the model system is validated. The specific steps are as follows: Model training (1) Data screening: Sample data prior to June 2025 were screened from the red blood cell folic acid clinical database (training set information is as follows). Figure 1 As shown in the figure, the screening criteria were: complete folic acid supplementation information (dosage form, dosage, duration of supplementation), first red blood cell folic acid test results, follow-up test results after supplementation, age, weight, and MTHFR genotype. A total of 497 valid samples were screened. (2) Sample splitting: The samples were divided into 5 groups according to the folic acid dosage form and dosage: 0.4mg regular folic acid group, 0.4mg active folic acid group, 0.8mg regular folic acid group, 0.8mg active folic acid group, and 0.4mg regular + 0.4mg active folic acid mixed group, with sample sizes of 177, 86, 78, 64 and 92 cases in each group, respectively; (3) Calculation of half-life increment of single sample: Taking the 0.4mg ordinary folic acid group as an example, one sample was selected. The initial test of the red blood cell folic acid level C(a) = 350nmol / L, and 0.4mg ordinary folic acid was supplemented daily for t = 45 days. The MTHFR 677 locus was CC type, and the red blood cell folic acid level C(b) was 450nmol / L. Substituting into the formula: C(k) = C0 + Δt × (1 - 0.5^(N × k)) / (1 - 0.5); where N = 0.0167, the half-life increment of the sample was calculated to be Δt = (450 - 350) × (1 - 0.5^(0.75)) / 0.5 ≈ 80nmol / L; (4) Within-group mean calibration: The half-life increment Δt of each group sample was statistically analyzed. After removing outliers (values ​​deviating from the within-group mean ± 3 standard deviations), the average value of each group was calculated as the standard value of the half-life increment corresponding to that dosage form and dose. Further calculations were performed on the known genotype sample subgroups in each folic acid supplementation regimen group, which were divided into two subgroups: MTHFR gene CC / CT type and MTHFR gene TT type, as shown below. Figure 4As shown in Table 3, the gene correction parameters S1 and S2 were calculated by comparing the half-life increment under different genotype conditions with the average half-life increment of all samples. Table 3 Standard values ​​of half-life increment and genotype correction coefficients corresponding to folic acid supplementation regimens (5) Store the calculated standard values ​​of half-life increments and genotype correction coefficients in the model parameter library for use during prediction.

[0044] 2. Construction of a system for predicting erythrocyte pentamethyltetrahydrofolate levels (1) Data input module: used to input the user's initial red blood cell folic acid level, folic acid supplementation plan (including dosage form and dosage), supplementation duration, number of days of supplementation interruption and current gestational age information.

[0045] (2) Parameter calling module: The parameter calling module contains the standard value of half-life increment and genotype correction coefficient stored during model training. Based on the folic acid dosage form, dosage and MTHFR 677 site genotype (if any) input by the user, it calls the corresponding standard value of half-life increment and genotype correction coefficient pre-stored in the model training of step 1, which is trained based on clinical data before June 2025.

[0046] (3) Mathematical Calculation Module: Based on the input parameters and the standard value of the half-life increment, the predicted value of erythrocyte folate at different time points is calculated using the following formula: When the object to be predicted does not have a folic acid interruption: C(k) = C0+S×Δt×(1-0.5^(N×k)) / (1-0.5); When the object to be predicted has a folic acid interruption: C(k) = C(k-1)-C(d), where C(d) = S×Δt×(1-0.5^(N×d)) / (1-0.5) When the subject of prediction changes their folic acid supplementation regimen and / or receives genotype information: According to C(k) = C(k-1) + Cs(k), on any day before changing the folic acid supplementation regimen and / or supplementing genotype information, C(k-1) = C0 + S_old × Δt_old × (1 - 0.5^(N × (k-1))) / (1 - 0.5). On the day of changing the folic acid supplementation regimen and / or supplementing genotype information, and on any subsequent day, Cs(k) = [C0 + S_new × Δt_new × (1 - 0.5^(N × k)) / (1 - 0.5)] - [C0 + S_new × Δt_new × (1 - 0.5^(N × (k-1))) / (1 - 0.5)]. Therefore, on the day of changing the folic acid supplementation regimen and / or supplementing genotype information, C(k) = C0+S_old×Δt_old×(1-0.5^(N×(k-1))) / (1-0.5)+[C0+S_new×Δt_new×(1-0.5^(N×k)) / (1-0.5)]-[C0+S_new×Δt_new×(1-0.5^(N×(k-1))) / (1-0.5)]; When the folate level in the red blood cells of the test subject reaches a plateau: When C(k) ≥ C0 + 5 × Δt, C(k) is fixed as Cmax = C0 + 5 × Δt. Where C(k) is the predicted level of erythrocyte folate after k days of supplementation; C0 is the initial level of erythrocyte folate; Δt is the standard value of half-life increment; N is the coefficient, with a value of 1 / 60 = 0.0167; S is the genotype correction coefficient, which is 1 when the genotype is not determined; k is the number of days of folate supplementation; d is the number of days of folate supplementation interruption; C(d) is the decrease in erythrocyte folate level after d consecutive days of no supplementation; C(k-1) is the level of erythrocyte folate after k-1 days of supplementation; Cs(k) is the daily increase in folate level after k days of supplementation; when changing the folate supplementation plan, the variable Δt needs to be reset, and the Δt of the new plan and the old plan are defined as Δt_new and Δt_old, respectively; when supplementing genotype information, the variable S needs to be reset, and the S before supplementing genotype information and after supplementing genotype information are defined as S_new and S_old, respectively; Cmax is the maximum value of erythrocyte folate level.

[0047] (4) Target determination module: Based on the target range of red blood cell folic acid corresponding to the current gestational week, determine the target status of the predicted value. The criteria for determining the target status are shown in Table 4.

[0048] Table 4. Criteria for Compliance Assessment Module (5) Result output module: outputs the predicted trajectory of red blood cell folate level, the status of compliance and abnormal prompts.

[0049] For the valid samples selected from the training set, the above model was used to output the results, which were then compared with the actual review results. The results showed that the prediction accuracy was 94.2%.

[0050] 3. Model System Validation: Clinical sample data from June 2025 to May 2026 were obtained from the erythrocyte folic acid clinical database (validation set information is as follows). Figure 1 As shown, the sample must contain complete folic acid supplementation information (dosage form, dosage, and duration of supplementation), initial erythrocyte folate level, and retested erythrocyte folate level. Exclusion criteria include past vascular disease, any disease affecting intestinal absorption, cancer, diabetes, and exclusion of cases where the supplementation dosage form and dosage were switched between the two tests. The initial erythrocyte folate level, folic acid dosage form, dosage, duration of supplementation, and MTHFR 677 locus genotype (if any) of the validation sample are input into the analysis model described in this invention to obtain the predicted erythrocyte folate level. The predicted value is compared with the actual retest results of the validation sample to calculate the prediction accuracy. The allowable deviation range is set at ±15%. If the prediction accuracy is ≥80%, the model is considered to have passed validation and meets the application requirements.

[0051] According to statistics, the model system has a prediction accuracy of 93.1%, which meets the needs of clinical application, and the model has passed validation.

[0052] Example 2: Screening of Stratified Modeling Variables for a Predictive System of Pentamethyltetrahydrofolate Levels in Erythrocytes From the red blood cell folic acid clinical database, sample data prior to June 2025 were selected (training set information as follows). Figure 1 As shown in the figure, the screening criteria were: complete folic acid supplementation information (dosage form, dosage, and duration of supplementation), the first red blood cell folic acid test results, and the follow-up test results after supplementation. A total of 497 valid samples were selected.

[0053] Using four known factors influencing erythrocyte folate levels—half-life increment, MTHFR genotype, age, and weight—as the basis for segmentation, valid samples were segmented. For half-life increment, samples were segmented into 0.4 mg regular folate, 0.4 mg active folate, 0.8 mg regular folate, 0.8 mg active folate, and a mixture of 0.4 mg regular and 0.4 mg active folate. For MTHFR genotype, samples were segmented into MTHFR CC / CT and MTHFR TT genotypes. Samples were also segmented according to age and weight groups. Half-life increment, MTHFR genotype, age, and weight were selected as stratified modeling variables. Using the method described in Example 1, parameter libraries for half-life increment, MTHFR genotype, age, and weight were established, as shown in the figure. Age ( Figure 2 ) and weight ( Figure 3The effect on half-life increment is small (no significant correlation, R). 2 The values ​​were 0.0088 and 4E-06, respectively, while the increase in half-life was associated with folic acid supplementation regimens and MTHFR genotypes. Figure 4 Data are displayed as mean ± standard error. The half-life increment and genotype correction parameters for different folic acid supplementation regimens and genotype conditions are shown in Table 5.

[0054] Table 5 Standard values ​​of half-life increment and genotype correction coefficients corresponding to folic acid supplementation regimens Formulas for calculating erythrocyte folate levels were established for different time periods. When the half-life increment was used alone as the stratified modeling variable, S was 1 and directly applied to the formula in Example 1 for calculation. When genotype was used alone as the stratified modeling variable, Δt was fixed at 80 nmol / L (obtained from the average Δt of the total sample), and the half-life increment was corrected to S × Δt and applied to the formula in Example 1 for calculation. Weight and age had little effect on Δt, and the correction parameter was equivalent to 1 and applied to the formula in Example 1 for calculation. The screening results are shown in Table 6.

[0055] Table 6. Accuracy Results of Variable Selection in Hierarchical Modeling Experimental results show that the highest prediction accuracy (88.1%) was achieved when half-life increment was introduced as a stratified modeling variable. When MTHFR genotype was introduced, the accuracy reached 86.3%. Furthermore, the accuracy of using half-life increment or MTHFR genotype alone as stratified modeling variables was significantly higher than using age or weight alone. The AUC (Average Values) index showed that the AUC of using age or weight alone as a modeling variable was below 0.6, indicating a low model performance and weak discriminative ability. In contrast, the AUCs of the models introducing half-life increment and MTHFR genotype as stratified modeling variables were 0.89 and 0.87, respectively, indicating strong discriminative ability. Therefore, half-life increment and MTHFR genotype are key indicators for predicting erythrocyte folate levels in this model system.

[0056] Example 3: Screening of Multi-Stratified Modeling Variable Combinations for Erythrocyte Pentamethyltetrahydrofolate Level Prediction System From the red blood cell folic acid clinical database, sample data prior to June 2025 were selected (training set information as follows). Figure 1As shown in the figure, the half-life increment, MTHFR genotype, age, and weight were combined as stratified modeling variables. Using the parameter library established in Example 2, calculation formulas for calculating erythrocyte folate levels were established for different variable combinations at different time periods. The calculation formula for different groups was the correction parameter × Δt, which was then substituted into the calculation formula in Example 1. The screening results are shown in Table 7.

[0057] Table 7. Accuracy Results of Variable Combination Screening in Two-Tier Modeling Experimental results show that, in the combination of two stratified modeling variables, the highest prediction accuracy and AUC were achieved when half-life increment and MTHFR genotype were introduced as stratified modeling variables, reaching an accuracy of 94.2% and an AUC of 0.93. The prediction accuracy and AUC of other combinations of two stratified modeling variables were not superior to the above-mentioned combined model system. The prediction accuracy was basically consistent with the system model built by using half-life increment alone as a stratified modeling variable. This indicates that when half-life increment and MTHFR genotype are used together as stratified modeling variables in the prediction model system, the model system is the most accurate in predicting erythrocyte folate levels.

[0058] Example 4: Validation of the erythrocyte pentamethyltetrahydrofolate level prediction system 1. Using the method described in Example 1, erythrocyte folate levels were predicted and validated in a perinatal woman (Case 1) with an initial erythrocyte folate level of 10 weeks of gestation (C0 = 800 nmol / L), who received daily supplementation of 0.4 mg of active folate for 30 days (k = 30 days), had an unknown MTHFR 677 locus genotype, and did not interrupt supplementation. (1) Data input: Input C0 = 800 nmol / L, folic acid supplementation plan (0.4 mg active folic acid), k = 30 days, d = 0 days, gestational week = 10 weeks; (2) Half-life increment call: Call the standard value of half-life increment Δt = 310 nmol / L corresponding to 0.4 mg active folic acid; (3) Horizontal prediction: Substituting into the growth formula C(k) = C0 + Δt × (1 - 0.5^(N × k)) / (1 - 0.5), where N = 0.0167 and k = 30, we can calculate: At 14 weeks of gestation, C(30) = 800+310×(1-0.5^(0.0167×30)) / (1-0.5) ≈ 800+310×(1-0.707) / 0.5 ≈ 981.66 nmol / L, with a predicted value of 981.66 nmol / L. It can also predict the red blood cell folic acid level on any day during the supplementation period. For example, at 12 weeks of gestation, k = 15, C(15) = 800+310×(1-0.5^(0.0167×15)) / (1-0.5) ≈ 800+310×(1-0.84) / 0.5 ≈ 899.2 nmol / L; (4) Determination of compliance: The target range for 14 weeks of gestation is a lower limit of 906 nmol / L and an upper limit of 1500 nmol / L. The predicted value of 981.66 nmol / L is within the range and is therefore determined to be in compliance. (5) Output results: The predicted level of folate in the woman's red blood cells is 981.66 nmol / L. At this time, she is 14 weeks pregnant and falls within the range of w = 0~14. The predicted target status is met, and it is suggested to continue the current supplementation plan.

[0059] (6) Comparison of actual test data: The actual erythrocyte folate level of the woman after supplementing with 0.4mg of active folic acid for 30 days was 1036nmol / L, with a deviation of -5.25%. The predicted value met the prediction accuracy standard (within ±15%), the actual standard status was met, and the predicted standard status was accurate.

[0060] 2. Using the method described in Example 1, red blood cell folate levels were predicted and validated in a perinatal woman (Case 2) who was 4 weeks pregnant, had an initial red blood cell folate level C0 = 300 nmol / L, received daily supplementation of 0.8 mg of regular folate for 60 days, had the MTHFR 677 locus TT genotype, and did not interrupt supplementation. (1) Data input: Input C0 = 300 nmol / L, folic acid supplementation regimen (0.8 mg regular folic acid), k = 60 days, d = 0 days, gestational age = 4 weeks; (2) Half-life increment call: Call the standard value of half-life increment corresponding to 0.8 mg of ordinary folic acid and the genotype correction coefficient Δt = 310 nmol / L; (3) Horizontal prediction: Substituting into the growth formula C(k) = C0 + Δt × (1 - 0.5^(N × k)) / (1 - 0.5), where N = 0.0167 and k = 60, we can calculate: C(60) = 300+0.8×360×(1-0.5^(0.0167×60)) / (1-0.5) ≈ 300+288×(1-0.5) / 0.5 ≈ 588nmol / L, the predicted value is 588nmol / L, and C(45) can also be predicted as 536nmol / L; (4) Determination of compliance: At this time, the gestational age is 12 weeks (which is within the range of w = 0~14). The corresponding target range is a lower limit of 906 nmol / L and an upper limit of 1500 nmol / L. The predicted value of 588 nmol / L is not within the range and is judged as not meeting the target. (5) Calculation of the abnormal erythrocyte folate level index: When C(k) does not meet the standard, the outlier is defined as outline = |C(k) - lower limit| (C(k) < lower limit) or outline = C(k) - upper limit (C(k) > upper limit); Index = 2 outline / Step Where Step is the step size, with a value of 50. At this point, Index = 2. （906-588） / 50 ≈ The value was 82, indicating that the woman's red blood cell folate level abnormality index was 82.

[0061] (6) Prediction result output: The output shows that the current red blood cell folate prediction level of the woman is 588 nmol / L. At this time, she is 12 weeks pregnant and falls within the range of w = 0~14. The prediction status is not up to standard. The red blood cell folate level abnormality index is 82. It is suggested that the current supplementation plan be adjusted and a retest should be performed one month later.

[0062] (7) Comparison of actual test data: The actual red blood cell folic acid level of the woman after supplementing with 0.8mg of ordinary folic acid for 60 days was 604nmol / L, which was 2.65% different from the prediction. The predicted value met the accuracy standard (within ±15%). The actual target status was not achieved, which was consistent with the predicted target status. Based on the test results, the clinic suggested adjusting the current supplementation plan and retesting one month later, which was consistent with the prediction suggestion.

[0063] While the present invention has been disclosed above, it is not limited thereto. Any person skilled in the art can make various modifications and alterations without departing from the spirit and scope of the invention; therefore, the scope of protection of the present invention should be determined by the scope defined in the claims.

Claims

1. A system for predicting erythrocyte pentamethyltetrahydrofolate levels, characterized in that, The prediction system includes a data input module, a parameter library retrieval module, a mathematical calculation module, a target attainment judgment module, and a result output module. The parameter library retrieval module contains a half-life increment parameter library and a genotype correction coefficient parameter library. The half-life increment parameter library is calculated by collecting and screening a complete erythrocyte pentamethyltetrahydrofolate clinical database to obtain standard half-life increment values ​​for different folic acid supplementation regimens and then aggregating them to form the half-life increment parameter library. The parameter library retrieval module can, based on the folic acid supplementation regimen (including folic acid dosage form and dosage) received by the data input module for the subject to be predicted, retrieve the matching half-life increment standard value from the half-life increment parameter library, or based on the MTHFR of the subject to be predicted. The 677-locus genotype and half-life increment standard value are used to retrieve the matching genotype correction parameters; the mathematical calculation module obtains the initial erythrocyte pentamethyltetrahydrofolate level, folic acid formulation, folic acid dosage, folic acid supplementation duration, number of days of folic acid supplementation interruption, and current gestational age information of the subject to be predicted collected by the data input module, as well as the matching half-life increment standard value and genotype correction parameters retrieved by the parameter library retrieval module, and substitutes them into the logistic regression equation to calculate and predict the erythrocyte pentamethyltetrahydrofolate level at different time points; the mathematical calculation module includes the following logistic regression equation: When the object to be predicted does not have a folic acid interruption: C(k) = C0+S×Δt×(1-0.5^(N×k)) / (1-0.5); When the object to be predicted has a folic acid interruption: C(k) = C(k-1)-C(d), where C(d) = S×Δt×(1-0.5^(N×d)) / (1-0.5) When the subject of prediction changes their folic acid supplementation regimen and / or receives genotype information: According to C(k) = C(k-1) + Cs(k), on any day before changing the folic acid supplementation regimen and / or supplementing genotype information, C(k-1) = C0 + S_old × Δt_old × (1 - 0.5^(N × (k-1))) / (1 - 0.5). On the day of changing the folic acid supplementation regimen and / or supplementing genotype information, and on any subsequent day, Cs(k) = [C0 + S_new × Δt_new × (1 - 0.5^(N × k)) / (1 - 0.5)] - [C0 + S_new × Δt_new × (1 - 0.5^(N × (k-1))) / (1 - 0.5)]. Therefore, on the day of changing the folic acid supplementation regimen and / or supplementing genotype information, C(k) = C0+S_old×Δt_old×(1-0.5^(N×(k-1))) / (1-0.5)+[C0+S_new×Δt_new×(1-0.5^(N×k)) / (1-0.5)]-[C0+S_new×Δt_new×(1-0.5^(N×(k-1))) / (1-0.5)]; When the level of pentamethyltetrahydrofolate in the erythrocytes of the test subject reaches a plateau: When C(k) ≥ C0 + 5 × Δt, C(k) is fixed as Cmax = C0 + 5 × Δt. Where C(k) is the predicted level of erythrocyte pentamethyltetrahydrofolate after k days of vaccination; C0 is the initial erythrocyte pentamethyltetrahydrofolate level; Δt is the standard value of the half-life increment; and N is a coefficient, with a value of 1 / 60. 0.0167; S is the genotype correction coefficient, which is 1 when the genotype is not determined; k is the number of days of folic acid supplementation; d is the number of days of folic acid supplementation interruption; C(d) is the decrease in erythrocyte pentamethyltetrahydrofolate level after d consecutive days of no supplementation; C(k-1) is the erythrocyte pentamethyltetrahydrofolate level after k-1 days of supplementation; Cs(k) is the daily increase in erythrocyte pentamethyltetrahydrofolate level after k days of supplementation; when changing the folic acid supplementation plan, the variable Δt needs to be reset, and the Δt of the new plan and the old plan are defined as Δt_new and Δt_old, respectively; when supplementing genotype information, the variable S needs to be reset, and the S before supplementing genotype information and after supplementing genotype information are defined as S_old and S_new, respectively; Cmax is the maximum value of erythrocyte pentamethyltetrahydrofolate level.

2. The system as described in claim 1, characterized in that, The compliance determination module can determine whether C(k) meets the standard according to the determination criteria. If the compliance determination module determines that it does not meet the standard, the compliance determination module can perform anomaly index calculation.

3. The system as described in claim 2, characterized in that, The result output module can output the predicted level of erythrocyte pentamethyltetrahydrofolate, the status of compliance, the abnormality index, and adjustment suggestions at a specific time point based on the judgment conclusion obtained by the compliance judgment module.

4. Use of the system as described in any one of claims 1-3 for constructing a model to predict erythrocyte pentamethyltetrahydrofolate levels.

Images