An in-vitro blood glucose response prediction method and system based on dynamic gastric emptying kinetics

Abstract

The present application belongs to the field of in vitro biological detection and biomedical engineering technology, and particularly relates to an in vitro blood glucose response prediction method and system based on dynamic gastric emptying kinetics. The method introduces a gastric emptying rate constant ke to dynamically simulate the transfer process of food from the stomach cavity to the intestinal cavity by an exponential decay model Q(t)=Q0.exp(-ke.t); under the condition of standardization of gastrointestinal enzymolysis, the area under the glucose release curve (AUC) is calculated; the prediction model pGI-M coupled with the AUC ratio and the maximum glucose release rate (dG / dt_max) is used to calculate the blood glucose generation index prediction value, and the prediction value can be corrected by the resistant starch conversion rate (RI).

Description

Technical Field

[0001] This invention belongs to the field of in vitro biological detection and biomedical engineering technology, specifically relating to an in vitro blood glucose response prediction method and system based on dynamic gastric emptying dynamics. Background Technology

[0002] The glycemic index (GI) is an important physiological parameter that measures the blood glucose response after food intake, and it has significant value in clinical nutrition, metabolic disease risk assessment, and the development of functional foods. However, current in vivo GI measurement methods (ISO 26642:2010) require recruiting healthy subjects and conducting multiple blood tests, which not only has a long experimental cycle (generally 4–6 weeks), but also suffers from problems such as large individual variability, high cost, and cumbersome ethical review, making it difficult to meet the needs of large-scale food screening.

[0003] To overcome the limitations of in vivo assays, researchers have developed several alternative in vitro methods: (1) INFUGEST static enzymatic digestion protocol (Minekus et al., 2014, Food Funct.): This method uses standardized enzyme concentrations to enzymatically digest food at fixed time points and estimates the GI based on the amount of glucose released in the digestive products. Because the INFUGEST protocol keeps the food and enzymes statically mixed throughout the digestion process, it fails to simulate the dynamic emptying process of the stomach under real physiological conditions, resulting in a generally lower predicted value for high-GI foods by about 15–20 GI units. The correlation coefficient (R²) with the GI in vivo is usually below 0.80, making it difficult to use for accurate prediction.

[0004] (2) TNO gastrointestinal model (TIM, Minekus et al., 1995): The TIM-1 system simulates the physical dynamics of the stomach and small intestine through a mechanical peristalsis device and controls pH in real time, which can simulate the digestive process relatively realistically. However, the TIM system is complex (priced at over RMB 1.5 million) and requires a high level of professional expertise, making it difficult to popularize and apply. Furthermore, the standardized configuration of TIM does not have a built-in mathematical model that directly correlates gastric emptying dynamic parameters with in vitro GI prediction values. The prediction results rely on independent statistical regression, which limits reproducibility.

[0005] (3) Monogastric animal bionic digestive system (Chinese patent, Beijing Institute of Animal Husbandry and Veterinary Medicine): This system is mainly designed for nutritional evaluation of monogastric animals. Its enzymatic parameters are significantly different from human physiological parameters, and it lacks a method to directly model gastric emptying dynamic parameters and blood glucose response index, so it cannot be used for human blood glucose response prediction.

[0006] In summary, existing in vitro digestive assessment methods suffer from the following core technical deficiencies: first, they neglect the decisive influence of gastric emptying on starch digestion rate and glycemic peak time; second, they lack a mathematical prediction model that couples the gastric emptying rate constant (ke) with in vitro glucose release kinetics; and third, they fail to incorporate the actual absorption and conversion rate of resistant starch into the GI prediction correction framework. Therefore, a novel in vitro glycemic response prediction method that balances physiological realism, predictive accuracy, and operational accessibility is urgently needed. Summary of the Invention

[0007] This invention aims to provide an in vitro glycemic response prediction method and system based on dynamic gastric emptying kinetics. By introducing a gastric emptying rate constant (ke) to establish a dynamic chyme transfer model in the stomach, constructing a pGI-M prediction model coupled with AUC and maximum glucose release rate, providing a system calibration procedure for the alpha parameter, and introducing a resistant starch conversion rate (RI) correction term, it achieves high-precision, low-cost, and rapid in vitro prediction of the glycemic index (GI) of food.

[0008] In a first aspect, this invention provides an in vitro blood glucose response prediction method based on dynamic gastric emptying kinetics, characterized by comprising the following steps: a) Grind the food sample to be tested into a uniform powder and weigh the specified amount of dry matter; b) Based on the total dietary fiber content and physical state of the food, set the gastric emptying rate constant ke (0.005–0.100 min). - ¹), the amount of chyme transferred at each time step is calculated using the exponential decay model Q(t)=Q0·exp(-ke·t); c) Gastric enzymatic hydrolysis was carried out at pH 2.0±0.1 and temperature 37±0.5°C, and the chyme was gradually transferred to the intestinal reactor according to the kinetic transfer amount determined in step b). d) Intestinal enzymatic hydrolysis was performed at pH 6.8–7.2 and temperature 37±0.5°C, wherein the intestinal lumen contained pancreatic alpha-amylase, trypsin, pancreatic lipase and bile salts; e) Samples were taken at 0, 15, 30, 45, 60, 90, and 120 minutes, and the glucose concentration Gi was determined using the glucose oxidase method or an electrochemical biosensor. f) Calculate the area under the glucose release curve (AUC_sample) using the trapezoidal method; g) Calculate the predicted glycemic index (pGI) of food based on the pGI-M glycemic index prediction model.

[0009] Furthermore, the value of ke is determined based on the total dietary fiber content (TDF): when TDF > 5.0 g / 100g, ke is taken as 0.005–0.015 min. - ¹; When TDF is 2.0–5.0 g / 100g, ke is taken as 0.015–0.040 min. - ¹; When TDF < 2.0 g / 100g, ke should be taken as 0.040–0.100 min. - ¹.

[0010] Further, in step c), the concentration of pepsin added in the gastric enzymatic hydrolysis is 2000 U / mL; in step d), the concentration of pancreatic alpha-amylase is 200 U / mL, the concentration of trypsin is 100 U / mL, the concentration of pancreatic lipase is 40 U / mL, the concentration of bile salts is 3 mmol / L, and the pH is maintained by adding 0.1 mol / L NaOH dropwise using an automatic titrator.

[0011] Furthermore, the linear detection range for the glucose concentration determination in step e) is 0–500 mg / dL, with an intra-assay coefficient of variation (CV) < 3.5%. The formula for calculating AUC_sample in step f) is: AUC = summation[(Gi+G(i+1)) / 2×(t(i+1)-ti)], i=0,1,…,6.

[0012] Further, the calculation formula for pGI-M in step g) is: pGI=(AUC_sample / AUC_reference)×(1+alpha×dG / dt_max)×100; Where AUC_reference is the AUC of the reference substance (glucose solution containing 50 g of available carbohydrates) under the same experimental conditions; dG / dt_max is the maximum value of the quotient of the glucose concentration increment and the time interval between adjacent time points within 0–60 minutes of digestion; alpha is the rate correction coefficient, determined by the calibration procedure described in claim 8.

[0013] Furthermore, the method also includes step h): after 120 minutes of digestion, the digestate is dialyzed at 37°C for 2 hours through a dialysis membrane with a molecular weight cutoff of 1000 Da. The glucose concentrations C_dialysate in the dialysate and C_total in the total digestate are measured, and the resistant starch conversion rate RI = C_dialysate / C_total is calculated. The pGI value is then corrected: pGI_corrected = pGI × RI.

[0014] Furthermore, the total digestion time in the gastric cavity stage was 90 minutes, and the total digestion time in the intestinal cavity stage was 120 minutes; the entire experiment was conducted under constant temperature conditions of 37.0±0.5°C, and each food was independently repeated no less than 3 times.

[0015] In a second aspect, a method for calibrating the alpha parameter of the glycemic index prediction model pGI-M described in the first aspect is provided, characterized by comprising the following steps: I) Select at least 5 reference foods with known in vivo GI values ​​from literature, the in vivo GI values ​​of which range from 35 to 80 and cover different food categories (at least two of the following: cereals, tubers, and legumes). II) Perform the in vitro digestion experiment described in steps a) to f) of claim 1 (n ≥ 5 replicates) on each reference food, and determine AUC_sample and dG / dt_max; III) Calculate the baseline ratio R0 = AUC_sample / AUC_reference × 100 for each reference food; IV) Perform least squares linear regression with the in vivo GI of each reference food as the dependent variable and R0 × dG / dt_max as the independent variable, and the resulting regression coefficient is the alpha value; V) Use no less than 3 foods not included in the calibration as a validation set, calculate RMSE, and require RMSE ≤ 3.0 GI units.

[0016] Furthermore, the reference foods include white bread (GI=75±3), whole wheat bread (GI=71±3), boiled potatoes (GI=78±3), white rice (GI=73±3), and instant oatmeal (GI=79±3); in the least squares regression, the model is required to have R²≥0.95 and p<0.01.

[0017] Furthermore, when the validation set RMSE > 3.0 GI units, a reference food should be added for recalibration until RMSE ≤ 3.0; the alpha value is laboratory specific, and the calibration procedure should be repeated when changing enzyme batches or experimental instruments.

[0018] Thirdly, an in vitro digestion assessment system for implementing the method described in the first aspect is provided, characterized in that the system comprises: a gastric reactor unit equipped with a pH sensor, a temperature-controlled heating mantle, and a quantitative feed pump, used to dynamically transfer chyme to the intestinal lumen according to an exponential decay law under a set ke value control; an intestinal reactor unit equipped with an automatic titration module, a pH sensor, and a magnetic stirring device, used to perform multi-enzyme synergistic enzymatic hydrolysis under dynamic pH control conditions; a staged enzyme addition control module, which automatically adds various enzyme solutions and bile salts according to the time sequence and dosage set in the program; an online glucose detection module, which automatically samples and measures glucose concentration at each preset time point; and a data processing and pGI-M calculation module, which receives glucose concentration time-series data, automatically calculates AUC and dG / dt_max, and outputs pGI predicted values ​​based on pre-stored alpha parameters and ke input values.

[0019] Furthermore, the automatic titration module is equipped with a 0.1 mol / L NaOH storage bottle and a micro-peristaltic pump, with a pH control accuracy of ±0.1; the flow rate accuracy of the quantitative feed pump is ±0.5%; and the overall temperature control accuracy of the system is ±0.5°C.

[0020] Furthermore, the online glucose detection module employs an electrochemical biosensor or spectrophotometer, with a detection linear range of 0–500 mg / dL and a batch CV <3.5%. The detection module has an automatic sampling function, with a sampling volume of 1–3 mL, and automatically replenishes an equal volume of buffer solution after sampling to maintain a constant system volume.

[0021] Furthermore, the data processing and pGI-M calculation module includes: a trapezoidal integral submodule, a rate calculation submodule (for determining dG / dt_max), an alpha parameter storage submodule, and a report output submodule. The report output submodule generates a result report containing pGI predicted values, confidence intervals, and detection quality control information.

[0022] Fourthly, the method described in the first aspect or the system described in the third aspect is provided for the application of rapid in vitro screening and development of low glycemic index functional foods. The method or system is used to predict the pGI of candidate food ingredient combinations, screening out formulation combinations with a pGI_corrected value below 55, which are used to guide the formulation design and process optimization of low-GI functional foods. The screening and detection cycle does not exceed 4 hours, and the detection cost does not exceed 5% of similar in vivo GI determination methods.

[0023] The beneficial effects of this invention include: 1. For the first time, the gastric emptying rate constant ke is introduced as an adjustable core kinetic parameter into the in vitro digestion assessment system. The dynamic quantitative transfer of food from the gastric lumen to the intestinal lumen is achieved through the exponential decay model Q(t)=Q0·exp(-ke·t), which overcomes the inherent defects of static digestion in the INFOGEST protocol.

[0024] 2. A pGI-M prediction model coupling total glucose release (AUC) and maximum release rate (dG / dt_max) was established. By introducing a rate correction factor (1+alpha·dG / dt_max), the mechanism of blood glucose peak generation in vivo was more accurately reflected, and the R² was improved from 0.71 to 0.99 compared with the INFOGEST method.

[0025] 3. A standardized calibration procedure for the alpha parameter is provided—the pGI (uncorrected) and dG / dt_max are determined using five classic reference foods, and the alpha value is determined by a least-squares linear regression system, giving the parameter a clear biophysical meaning and a reproducible calibration method.

[0026] 4. The resistant starch conversion rate (RI) determined by ultrafiltration was introduced as a multiplicative correction factor for pGI (pGI_corrected=pGI×RI), which for the first time quantified the correction contribution of resistant starch to in vitro glycemic response prediction and significantly improved the prediction accuracy for foods containing high dietary fiber.

[0027] 5. Integrate the above methods into a programmable in vitro digestion assessment system to achieve full-process automation and standardization. Attached Figure Description

[0028] Figure 1 This is a flowchart of an in vitro blood glucose response prediction system based on dynamic gastric emptying kinetics, which shows the complete technical process from sample preparation, Ke parameter setting, dynamic enzymatic digestion in the gastric / intestinal lumen, glucose sampling detection to pGI-M model calculation.

[0029] Figure 2 The curves show the change in the mass of chyme in the stomach over time at different gastric emptying rate constants ke. The three curves correspond to ke = 0.01 min. - ¹(High-fiber foods), ke=0.03 min - ¹(Medium fiber food) and ke=0.05 min - ¹(Low-fiber foods).

[0030] Figure 3 The graph shows the glucose release concentration curves during the in vitro dynamic digestion of three representative foods. The shaded area represents the area of ​​the AUC integral.

[0031] Figure 4 This is a scatter plot showing the correlation between the pGI-M model predicted values ​​and the corresponding literature values ​​of GI in food, including the linear regression fit line and R² and RMSE statistics.

[0032] Figure 5 This is a bar chart comparing the accuracy (R²) of the pGI-M method of this invention with the INFUGEST static method for predicting the GI of six foods. Detailed Implementation

[0033] The following detailed embodiments further illustrate the concept and technical effects of the present invention to fully understand its purpose, features, and effects. Unless otherwise specified, all methods described are conventional methods. Unless otherwise specified, all materials are available from publicly available commercial sources. The illustrative embodiments and descriptions of the present invention are used to explain the invention and do not constitute an undue limitation thereof. It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other.

[0034] The model for the construction and application of this invention includes: (a) Dynamic gastric emptying model The gastric emptying process was described using a first-order exponential decay model (which is highly consistent with in vivo nucleoside-labeled gastric emptying studies): Q_stomach(t) = Q0 × exp(-ke × t) Where: Q_stomach(t) is the mass of chyme in the stomach at time t (g); Q0 is the initial mass of chyme (g), usually set to 5.0 g dry matter; ke is the gastric emptying rate constant (min). - ¹), with a value range of 0.005–0.100 min. - ¹, determined by the total dietary fiber content, physical state, and starch type of the food (see Table 1).

[0035] For each time step delta_t (usually 1 min), the mass of chyme transferred from the stomach to the intestine is delta_Q(t): delta_Q(t) = Q0 × exp(-ke×t) × (1 - exp(-ke×delta_t)) Table 1. Correspondence between food type and gastric emptying rate constant ke

[0036] (II) In vitro dynamic enzymatic digestion process of gastrointestinal tract

[0037] Gastric phase: 0 g of dry matter sample was dissolved in 50 mL of physiological saline (pH adjusted to 2.0±0.1, 37±0.5°C), and pepsin (2000 U / mL, Sigma-Aldrich P7012) was added. The chyme was quantitatively transferred to the intestinal reactor minutely according to the ke value, while simultaneously replenishing with an equal volume of physiological saline to maintain a constant gastric volume. The total duration of the gastric phase was 90 minutes.

[0038] Intestinal stage: The chyme transferred to the intestinal lumen was immediately placed in a buffer environment (phosphate buffer, 37°C) at pH 6.8 ± 0.1, and a mixture of pancreatic enzymes (pancreatic alpha-amylase 200 U / mL, trypsin 100 U / mL, pancreatic lipase 40 U / mL) and a mixture of bile salts (sodium taurocholate 3 mmol / L) were added. The intestinal lumen pH was maintained within the range of 6.8–7.2 using an automated titrator (0.1 mol / L NaOH).

[0039] Temperature control: The entire process is carried out in a constant temperature water bath at 37.0±0.5°C.

[0040] (III) Dynamic sampling of glucose concentration and AUC calculation

[0041] During digestion, 2 mL samples were taken from the intestinal reactor at 0, 15, 30, 45, 60, 90, and 120 minutes. After centrifugation at 8000 rpm for 3 min, the supernatant was collected, and the glucose concentration Gi (mg / dL) was determined using the glucose oxidase-peroxidase method (GOD-PAP method) or an electrochemical biosensor.

[0042] AUC is calculated using the trapezoidal rule: AUC = sum_i [(Gi + G(i+1)) / 2 × (t(i+1) - ti)] Where i = 0, 1, 2, …, 6, corresponding to 7 time points (0–120 min).

[0043] (iv) Glycemic Index Prediction Model (pGI-M)

[0044] 1. Basic pGI calculation: pGI = (AUC_sample / AUC_reference) × (1 + alpha × dG / dt_max) × 100 Where: AUC_sample is the AUC (mg / dL·min) of glucose release from the test food; AUC_reference is the AUC of the reference food (glucose solution, 50 g of available carbohydrates) under the same experimental conditions; dG / dt_max is the maximum glucose release rate (mg / dL·min) observed in the test food during digestion from 0 to 60 minutes. - ¹), defined as the maximum value of the quotient of the glucose concentration difference between adjacent time points and the time interval; alpha is the rate correction coefficient, determined by the standardized calibration procedure (see Example 2), with dimensions of min·dL / mg.

[0045] 2. Resistant starch correction (optional): pGI_corrected = pGI × RI RI = C_dialysate / C_total Where C_dialysate is the glucose concentration (mg / dL) in the dialysate after 120 minutes of digestion, C_total is the total glucose concentration (mg / dL) in the digestion solution, and the molecular weight cutoff of the dialysis membrane is 1000 Da.

[0046] The flowchart of the in vitro blood glucose response prediction system based on dynamic gastric emptying kinetics of the present invention is as follows: Figure 1 As shown.

[0047] Example 1: The effect of gastric emptying rate constant ke on glycemic response prediction

[0048] [Objective] To verify the decisive influence of the ke parameter on in vitro glucose release kinetics and pGI prediction values, and to establish the mapping relationship between ke and the physicochemical properties of food.

[0049] [Samples and Design] Three representative foods were selected: white bread (low fiber, ke=0.05 min) - ¹) Whole grain oatmeal (high fiber, ke=0.01 min) - ¹) Steamed sweet potatoes (medium fiber, ke=0.03 min) - ¹). 5.0 g of dry matter from each food (n=3 replicates) was subjected to in vitro dynamic digestion according to standard operating procedures. Glucose concentration was measured at 7 time points (0, 15, 30, 45, 60, 90, and 120 min) using the GOD-PAP method (wavelength 505 nm, linear range 0–500 mg / dL, CV < 3.5%). The reference food was 50 g of glucose solution (AUC_reference = 8,920 mg / dL·min).

[0050] [Experimental Results] The glucose concentration and AUC calculation results of each food at different time points are shown in Table 2.

[0051] Table 2 Summary of experimental data for Example 1 (alpha = 0.524 min·dL / mg)

[0052]

Results and Analysis

[0053] Example 2: System calibration procedure for rate correction coefficient alpha

[0054] [Objective] To establish a standardized calibration procedure for the alpha parameter, thereby giving the pGI-M model repeatability and transferability.

[0055] [Calibrated Samples] Five reference foods with reliable literature records of GI in vivo were selected: white bread (GI=75), whole wheat bread (GI=71), boiled potatoes (GI=78), white rice (GI=73), and instant oatmeal (GI=79).

[0056] [Calibration Method] In vitro digestion experiments were performed on each food (n=5 replicates), and AUC_sample and dG / dt_max were measured. The uncorrected baseline ratio R0 = AUC_sample / AUC_reference × 100 was calculated. With the GI of each food as the dependent variable and R0 × dG / dt_max as the independent variable, least squares linear regression was used to solve for alpha.

[0057] Table 3 Alpha parameter calibration dataset (5 reference foods, n=5 repetitions / food)

[0058]

alpha calibration results

[0059] Table 4. Prediction accuracy of the alpha calibration validation set

[0060]

Results and Analysis

[0061] Example 3: Validation of Corrected Resistant Starch Conversion Rate (RI)

[0062] [Objective] To verify the correction effect of the RI correction term on the GI prediction of high-resistant starch foods.

[0063]

Samples

[0064] Table 5. Comparison of pGI predicted values ​​and in vivo GI before and after RI correction.

[0065]

Results and Analysis

[0066] Example 4: Comparison and verification of the method of the present invention with the INFOOGS static method

[0067] [Objective] To quantify the prediction accuracy advantage of the method of this invention over the INFUGEST static method by comparing different methods.

[0068] [Design] Six foods were selected, and the pGI-M method of this invention and the INFOGEST 2014 standard protocol were used simultaneously to predict the in vitro GI values, which were then compared with the in vivo GI values ​​in the literature (n=3 replicates / food / method).

[0069] Table 6 Comparison of prediction accuracy between pGI-M method and INFogEST static method

[0070]

Results and Analysis

Claims

1. An in vitro blood glucose response prediction method based on dynamic gastric emptying kinetics, characterized in that, Includes the following steps: a) Prepare the food sample to be tested into a uniform powder, and weigh out the specified amount of dry matter as Q0; b) Set the gastric emptying rate constant ke (0.005–0.100 min) based on the total dietary fiber content and physical state of the food. - ¹), the amount of chyme transferred at each time step is calculated using the exponential decay model Q(t)=Q0·exp(-ke·t); c) Pepsin digestion was performed in the gastric cavity at pH 2.0±0.1 and 37±0.5°C, while the chyme was gradually transferred to the intestinal reactor according to the kinetics of step b). d) Perform multi-enzyme synergistic enzymatic hydrolysis in the intestinal lumen under conditions of pH 6.8–7.2 and 37±0.5°C; e) Samples were taken at 0, 15, 30, 45, 60, 90, and 120 minutes to determine the glucose concentration Gi, and the area under the glucose release curve AUC_sample was calculated using the trapezoidal method. f) Calculate the predicted glycemic index (pGI) of food based on the pGI-M model.

2. The method according to claim 1, characterized in that, The value of ke is determined based on the total dietary fiber content (TDF): when TDF > 5.0 g / 100g dry weight, ke is 0.005–0.015 min. - ¹; When TDF is 2.0–5.0 g / 100g dry weight, ke is 0.015–0.040 min. - ¹; When TDF < 2.0 g / 100g dry weight, ke is 0.040–0.100 min. - ¹.

3. The method according to claim 1, characterized in that, In gastric enzymatic hydrolysis, the concentration of pepsin was 1500–2500 U / mL; in intestinal enzymatic hydrolysis, the concentrations of pancreatic α-amylase were 100–300 U / mL, trypsin was 80–120 U / mL, pancreatic lipase was 30–50 U / mL, and bile salts were 2–5 mmol / L; the intestinal pH was maintained by automatically adding 0.05–0.15 mol / L NaOH; the total duration of the gastric stage was 60–120 minutes, and the total duration of the intestinal stage was 100–150 minutes; the temperature was controlled at 37.0 ± 0.5°C throughout the process, and each food was independently repeated at least 3 times.

4. The method according to claim 1, characterized in that, In step e), glucose concentration was determined using the glucose oxidase-peroxidase method or an electrochemical biosensor with a linear detection range of 0–500 mg / dL and an intra-assay coefficient of variation (CV) < 3.5%. The AUC_sample was calculated using the trapezoidal method: AUC = Σ[(Gi + G(i+1)) / 2 × (t(i+1) - ti)], i = 0, 1, …, 6.

5. A method for calibrating the rate correction coefficient alpha in the pGI-M model as described in claim 1, characterized in that, Includes the following steps: I) Select at least 5 reference foods with known GI values ​​in vivo, covering the GI range of 35–80 and at least two food categories; II) Perform the in vitro digestion experiment described in steps a) to e) of claim 1 on each reference food, determine AUC_sample and dG / dt_max, and calculate the basal ratio R0 = AUC_sample / AUC_reference × 100; III) Perform least squares linear regression with in vivo GI as the dependent variable and R0×dG / dt_max as the independent variable. The resulting regression coefficient is the alpha value. IV) Calculate RMSE using at least 3 validation foods that were not included in the calibration, with RMSE ≤ 3.0 GI units; if RMSE > 3.0, add reference foods and recalibrate; the alpha value needs to be recalibrated when changing enzyme batches or instruments.

6. The method according to claim 1, characterized in that, The calculation formula for the pGI-M model is as follows: pGI = (AUC_sample / AUC_reference) × (1 + alpha × dG / dt_max) × 100 where AUC_reference is the AUC of the reference food measured under the same conditions; dG / dt_max is the maximum value of the quotient of the glucose concentration increment and the time interval between adjacent time points within 0–60 minutes of digestion; and alpha is the rate correction coefficient, determined by the calibration method described in claim 5.

7. The method according to any one of claims 1 to 6, characterized in that, It also includes a step to correct the resistant starch conversion rate: After digestion, the digestate is dialyzed at 37°C for 2 hours through a dialysis membrane with a molecular weight cutoff of 1000 Da. The glucose concentrations of the dialysate (C_dialysate) and total digestate (C_total) are measured, and RI is calculated as C_dialysate / C_total. The corrected predicted value pGI_corrected = pGI × RI is then obtained.

8. An in vitro digestive assessment system for implementing the method of claim 1, characterized in that, The system includes: The gastric reactor unit includes a pH sensor, a temperature-controlled heating jacket, and a metering pump, used to dynamically transfer chyme according to the Ke index decay law; Intestinal reactor unit: includes an automatic titration module, pH sensor and magnetic stirring device, used for multi-enzyme hydrolysis under dynamic pH control; Staged enzyme addition control module; Online glucose detection module: automatically samples and measures glucose concentration; Data processing and pGI-M calculation module: Automatically calculates AUC, dG / dt_max and pGI predicted values.

9. The system according to claim 8, characterized in that, The automatic titration module is equipped with a 0.1 mol / L NaOH storage bottle and a micro-peristaltic pump, with pH control accuracy of ±0.1; quantitative feed pump flow accuracy of ±0.5%; and overall temperature control accuracy of ±0.5°C. The online glucose detection module has a linear range of 0–500 mg / dL, batch CV <3.5%, and a sampling volume of 1–3 mL with automatic buffer replenishment. The data processing module includes a trapezoidal integral submodule, a dG / dt_max rate calculation submodule, an alpha parameter storage submodule, and a report output submodule.

10. The application of the method according to claims 1-6 or the system according to any one of claims 8-9 in the rapid in vitro screening and development of functional foods with low glycemic index, characterized in that, The method or system described above is used to predict the pGI or pGI_corrected value of candidate food ingredient combinations, and to screen formulation combinations with a pGI_corrected value of less than 55, which can be used to guide the formulation design and process optimization of low-GI functional foods. The screening and detection cycle does not exceed 4 hours, and the detection cost does not exceed 5% of similar in vivo GI determination methods.

Images