An information-based quantitative evaluation method for progression trend of diabetic microvascular complications

Abstract

The application discloses a kind of informationization quantification evaluation methods of the progress trend of diabetic microvascular complications, belong to biomedical signal processing field, including, step 1, identify the point of slope in blood glucose sequence exceeds predetermined slope threshold as incentive anchor point, search microvessel response point in preset time window, the response value of this point exceeds predetermined deviation threshold relative to the offset of baseline before excitation, and it is as response anchor point, output the time difference between incentive anchor point and response anchor point as original time offset;Step 2, based on double exponential decay recovery model from original time offset, response measurement and baseline value before excitation;The present application can realize by accurate identification of subclinical stage microvessel occult injury, effectively avoid subclinical progress period patient misjudgment as low-risk population, provide reliable data support for early intervention of microvascular complications, improve the timeliness and effectiveness of early intervention.

Description

Technical Field

[0001] This invention relates to the field of biomedical signal processing, and more specifically, to an information-based quantitative assessment method for the progression trend of diabetic microvascular complications. Background Technology

[0002] Currently, information processing methods used in clinical practice for assessing the risk of diabetic microvascular complications mainly rely on static or threshold-based indicators such as urinary microalbumin, glycated hemoglobin, and fundus grading to construct risk stratification models. The data processing characteristics of these methods are: statically comparing the test results of a single or sparse sample with a preset threshold, thereby classifying patients into different risk levels. However, this information processing model has limited ability to identify data features corresponding to diffuse microvascular function changes in the subclinical stage.

[0003] In the subclinical stage, although patients have no obvious clinical symptoms, microvascular damage has already occurred, especially early pathological changes such as a decline in the dynamic regulation of blood perfusion. These changes are manifested in the data as subtle fluctuations and response delays in time-series signals. Existing risk stratification technologies based on static indicators cannot effectively capture the above-mentioned time-series signal characteristics in their data processing framework. As a result, a large number of patients in the subclinical progression stage who have not yet shown abnormal static indicators are incorrectly classified as low-risk individuals at the data processing level, missing the best intervention opportunity.

[0004] A significant contributing factor to microvascular damage is blood glucose fluctuation, including dynamic changes such as postprandial blood glucose spikes, abnormal diurnal blood glucose rhythms, and stress-induced blood glucose shifts. However, in routine clinical data collection scenarios, limited by sampling conditions, healthcare professionals can only obtain sparse, discontinuous point-in-time data, making it impossible to construct a complete, continuous time series of blood glucose fluctuations. To ensure the standardization and comparability of risk stratification results, existing assessment paradigms are forced to use averaging parameters, such as glycated hemoglobin, as the analysis object. The information processing of such parameters essentially involves smoothing and compressing the time axis, deliberately ignoring short-term blood glucose fluctuation details and retaining only the average blood glucose level over a longer period.

[0005] Early pathological signals in microvessels manifest precisely as phase differences in the dynamic response to blood glucose fluctuations, such as the delayed recovery of retinal blood flow after postprandial hyperglycemia and the lag in glomerular filtration rate following blood pressure fluctuations. These dynamic response relationships are the earliest measurable data characteristics of microvascular reserve depletion in the subclinical stage and are also key information for achieving individualized risk stratification. However, in the standardization of detection indicators in existing risk stratification technologies, the averaging operation completely loses the temporal correlation information between blood glucose fluctuations and microvascular responses. This makes it impossible for existing risk stratification technologies to extract early pathological signals from subclinical stage data, making it difficult to achieve accurate quantitative assessment and risk classification of the progression trend of diabetic microvascular complications. Summary of the Invention

[0006] To address the problems existing in the prior art, the present invention aims to provide an information-based quantitative assessment method for the progression trend of diabetic microvascular complications. This method can accurately identify occult damage to microvessels in the subclinical stage, effectively avoid misclassifying patients in the subclinical progression stage as low-risk individuals, provide reliable data support for early intervention of microvascular complications, and improve the timeliness and effectiveness of early intervention.

[0007] To solve the above problems, the present invention adopts the following technical solution:

[0008] An information-based quantitative assessment method for the progression trend of diabetic microvascular complications, the method comprising:

[0009] Step 1: Identify points in the blood glucose sequence whose slope exceeds a predetermined slope threshold as excitation anchor points. Search for microvascular response points within a preset time window. If the offset of the response value of a point relative to the baseline before excitation exceeds a predetermined deviation threshold, use it as a response anchor point. Output the time difference between the excitation anchor point and the response anchor point as the original time offset.

[0010] Step 2: Based on the double exponential decay recovery model, the first decay time constant and the second decay time constant are inversely derived from the original time offset, response measurement value and baseline value before excitation, where the first decay time constant is less than the second decay time constant, and the ratio of the two is output as the response delay exponent.

[0011] Step 3: Dynamically update the adaptive reference delay interval based on the patient's own historical data, perform nonlinear mapping on the response delay exponent, with the boundary mapping being a constant and the intermediate nonlinear compression transformation being used to output the exhaustion coefficient.

[0012] Step 4: Calculate the weight coefficient for each stimulus anchor point. This coefficient is determined based on the ratio of the magnitude of the blood glucose peak exceeding the fasting baseline to the pathological threshold. Multiply the weight of each event by a decay factor that decreases exponentially with the time of the event. Accumulate the weighted exhaustion sum and the expected baseline load sum, and output the subclinical progression index of the ratio of the two.

[0013] Step 5: Perform a signed-rank test on the longitudinally collected subclinical progression index sequence to determine its monotonic trend. Based on the sum of the ranks of the positive and negative differences, determine whether it is an increasing trend or a non-increasing trend, and output the sign of the trend slope.

[0014] Step 6: Combine the current subclinical progression index with the trend slope symbol to generate a risk color block code. Define three regions according to the threshold and symbol, and output the color block code.

[0015] Further, step 1 includes:

[0016] Step 11: Calculate the equivalent fluctuation energy from the non-continuous blood glucose sequence according to the ratio of the time interval between two adjacent points to the blood glucose difference. Perform pattern recognition on three consecutive fluctuation segments. If the absolute value of the slope of the middle segment is greater than the threshold multiple of the sum of the absolute values ​​of the slopes of the two segments before and after, mark the start and end points of the middle segment as the rising start point and peak point of the candidate excitation event, and output the list of candidate excitation events.

[0017] Step 12: Based on the peak time of the candidate excitation event, obtain the microvascular response value within the subsequent time window, arrange the deviation direction of each response value relative to the baseline into a symbol sequence, calculate the ratio of the maximum length of consecutive identical signs to the total length, if the ratio exceeds the threshold and the last sign is consistent with the expected abnormal direction, then confirm that the point is a valid response anchor point, and output the valid response anchor point and its time point.

[0018] Step 13: Use the time difference between the response anchor point and the peak point as the original time offset and the maximum delay, and the time difference between the first sign-reversed measurement point after the response anchor point and the response anchor point as the minimum delay. Take the geometric midpoint between the minimum and maximum delays as the original time offset output after noise suppression.

[0019] Further, step 2 includes:

[0020] Step 21: Using the maximum and minimum delays in the original time offset, locate the response anchor point measurement value corresponding to the maximum delay and the first sign reversal point measurement value corresponding to the minimum delay. Subtract the baseline value before excitation from each of the two measurements and take the ratio to output the normalized response pair.

[0021] Step 22: Using the normalized response pairs and their corresponding time points, construct two constraint equations for the double exponential decay recovery model. Each equation constrains the double exponential term to be equal to the normalized response value at that time point, and the first decay time constant and the second decay time constant in the two equations share the same value. Output the simultaneous constraint relationship between the first decay time constant and the second decay time constant.

[0022] Furthermore, step 2 also includes:

[0023] Step 23: Using the simultaneous constraint relationship, the interval shrinkage approximation method is adopted: the physiological upper and lower limits of the first decay time constant are preset and divided into sub-intervals. In each sub-interval, the simultaneous constraint relationship is transformed into a single variable equation about the second decay time constant. The absolute value of the equation residual in each sub-interval is compared to narrow the candidate interval until the residual is lower than the preset threshold. The first decay time constant and the second decay time constant at this time are output.

[0024] Step 24: Using the first decay time constant and the second decay time constant, calculate the ratio of the second decay time constant to the first decay time constant, then perform logarithmic compression on the ratio with the natural constant as the base, and output the compressed ratio as the response delay index.

[0025] Further, step 3 includes:

[0026] Step 31: Based on the response delay exponential sequence, extract the values ​​in the sequence that are within the predetermined physiological cutoff range, sort them by value size, remove the extreme values ​​at both ends by a fixed proportion, calculate the sliding median and sliding absolute deviation of the remaining values, and output the patient's own reference center value and reference dispersion.

[0027] Step 32: Relying on the reference center value and the reference dispersion, the lower and upper limits of the adaptive reference delay interval are determined by the multiple of the reference center value plus or minus the reference dispersion. The value of this multiple is adjusted according to the offset direction of the event response delay index and the reference center value: when the offset direction is positive, the first preset step size is increased, and when it is negative, the second preset step size is decreased. The adjusted upper and lower limits of the adaptive reference delay interval are then output.

[0028] Furthermore, step 3 also includes:

[0029] Step 33: Relying on the upper and lower limits of the adaptive reference delay interval, compare the response delay exponent of this event with the interval: if it is less than or equal to the lower limit, it is mapped to the first constant; if it is greater than or equal to the upper limit, it is mapped to the second constant; if it is within the interval, a composite operation of piecewise linear stretching followed by logarithmic shrinking based on the interval width is used to output the intermediate mapping value.

[0030] Step 34: Depending on the intermediate mapping value and whether it comes from within the interval, if it comes from nonlinear compression within the interval, calculate the difference between the response delay exponent and the reference center value, divide it by the reference dispersion as a local fine-tuning coefficient, multiply the intermediate mapping value by this coefficient, and output the final exhaustion coefficient; if it comes from boundary constant mapping, directly output the constant as the exhaustion coefficient.

[0031] Further, step 4 includes:

[0032] Step 41: Based on the blood glucose peak and fasting baseline value of each excitation anchor point, calculate the absolute range of the peak exceeding the baseline, divide the range by the pathological threshold to obtain the relative ratio, then perform logarithmic stretching of the relative ratio with the natural constant as the base and add one, and output the stretched original weight coefficient.

[0033] Step 42: Based on the difference between the time point of each excitation anchor point and the current evaluation time point, divide the difference by the preset half-life constant to obtain the dimensionless duration ratio, and then calculate the decay factor using an exponential function with the natural constant as the base, and output the time decay factor of each event.

[0034] Step 43: The weighted exhaustion contribution value is obtained by multiplying the original weight coefficient after stretching and the time decay factor. The weighted exhaustion amount is obtained by multiplying the exhaustion coefficient of each event and the weighted exhaustion contribution value. The expected baseline load is obtained by multiplying the weighted exhaustion contribution value and the average exhaustion coefficient of the healthy population. The weighted exhaustion amount sequence and the expected baseline load sequence of each event are output.

[0035] Furthermore, step 4 also includes:

[0036] Step 44: Based on the weighted exhaustion sequence and the expected baseline load sequence, sum them up in the order of event occurrence to obtain the cumulative weighted exhaustion sum and the cumulative expected baseline load sum. If the total number of events is lower than the minimum sample threshold, multiply the cumulative sum by the compensation factor. The compensation factor is equal to the minimum sample threshold divided by the actual total number of events. Output the compensated cumulative sum.

[0037] Step 45: Based on the ratio of the cumulative weighted exhaustion after compensation to the cumulative expected baseline load after compensation, the ratio is then logarithmically compressed to the base of the natural constant and then incremented by one to output the final subclinical progress index.

[0038] Further, step 5 includes:

[0039] Step 51: Based on the subclinical progression index sequence acquired longitudinally, calculate the difference between two adjacent indices in chronological order, assign the absolute values ​​of each difference to natural number ranks in ascending order and retain the original sign of each difference, and output the signed rank sequence.

[0040] Step 52: Based on the signed rank sequence, sum all positive ranks to obtain the positive rank sum, sum all negative ranks to obtain the negative rank sum, calculate the difference between the positive rank sum and the negative rank sum. If the difference is greater than the preset monotonic trend threshold and the positive rank sum is greater than the negative rank sum, it is determined to be an increasing trend; otherwise, it is determined to be a non-increasing trend, and the trend slope sign is output.

[0041] Further, step 6 includes:

[0042] Step 61: Based on the current subclinical progression index and the trend slope sign, convert the trend slope sign into a sign weight factor, where a positive sign corresponds to a weight of 1 and a negative sign corresponds to a weight of 0. Calculate the risk boundary offset, which is equal to the preset first index threshold multiplied by the sign weight factor and then multiplied by the preset nonlinear coupling coefficient. Use this offset to correct the preset first index threshold and second index threshold respectively to obtain the dynamic lower threshold and dynamic upper threshold. Compare the current subclinical progression index with the dynamic lower threshold and dynamic upper threshold in sequence. If it is less than the dynamic lower threshold, it is determined to be in the first region; if it is between the two, it is determined to be in the second region; if it is greater than or equal to the dynamic upper threshold, it is determined to be in the third region. Output the region determination result.

[0043] Step 62: Based on the region determination result, extract the color block code that is uniquely associated with the region from the preset color code correspondence rules, where the color code correspondence rules are: the first region corresponds to the first color code, the second region corresponds to the second color code, and the third region corresponds to the third color code. Output the color block code.

[0044] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0045] (1) This approach effectively avoids misjudging patients in the subclinical stage of microvascular lesions as low-risk individuals by accurately identifying the occult damage of microvessels in the subclinical stage, providing reliable data support for the early intervention of microvascular complications and improving the timeliness and effectiveness of early intervention.

[0046] (2) This scheme relies on the patient's own historical data to construct an adaptive reference delay interval, and combines nonlinear mapping to realize the individualized calculation of the exhaustion coefficient. It fully takes into account the differences in the basic microvascular function of different patients, avoids the assessment bias caused by a unified reference standard, and greatly improves the accuracy and pertinence of the assessment results.

[0047] (3) This scheme quantifies the damage intensity of blood glucose stimulation by weighting coefficients, reflects the time effect of stimulation events by combining time decay factors, and judges the progression trend by sign-rank test, so as to achieve a comprehensive quantification of the subclinical progression of complications and provide a scientific basis for the formulation of individualized clinical intervention schemes.

[0048] (4) This scheme combines the current progress index with the trend slope symbol, divides the risk area through dynamic threshold correction and generates a visual color block code, which intuitively presents the patient's risk level and development trend, making it easier for medical staff to quickly identify and interpret the risk, and improving the convenience and efficiency of clinical risk assessment and case management. Attached Figure Description

[0049] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0050] Figure 1 This is a flowchart illustrating the overall method of the present invention;

[0051] Figure 2 This is a flowchart of the calculation process for the double-exponential decay recovery and response delay exponent of the present invention;

[0052] Figure 3 This is a flowchart of the adaptive reference delay interval construction and exhaustion coefficient mapping of the present invention. Detailed Implementation

[0053] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0054] Please see Figures 1 to 3 An information-based quantitative assessment method for the progression trend of diabetic microvascular complications, the method comprising:

[0055] Step 1: Identify points in the blood glucose sequence whose slope exceeds a predetermined slope threshold as excitation anchor points. Search for microvascular response points within a preset time window. The response value at a point whose offset relative to the baseline before excitation exceeds a predetermined deviation threshold is used as the response anchor point. Output the time difference between the excitation anchor point and the response anchor point as the original time offset. The specific operation is as follows:

[0056] In the subclinical stage of diabetic microvascular complications, abnormal blood glucose fluctuations are the core triggering factor for occult microvascular damage. The degree of delay in the microvascular response to this blood glucose stimulus is a key early measurable characteristic reflecting the functional status of microvessels and judging whether their reserve capacity has declined. Therefore, accurately obtaining this time delay data is a prerequisite for subsequent assessment. In the process, the slope of the collected blood glucose sequence is first analyzed segment by segment and point by point. The slope calculation must strictly follow a fixed logic: for any two adjacent data points in the blood glucose sequence, the blood glucose value of the next data point is subtracted from the blood glucose value of the previous data point to obtain the blood glucose difference between the two adjacent points; then the acquisition time of the next data point is subtracted from the acquisition time of the previous data point to obtain the time interval between the two adjacent points; finally, the calculated blood glucose difference is divided by the time interval to obtain the slope of the line segment formed by the two adjacent points. This slope directly reflects the rate and intensity of blood glucose fluctuations during this period.

[0057] After calculating the slopes of all adjacent points, each slope is compared with a preset slope threshold. When the value of a certain slope is greater than the preset slope threshold, it indicates that there has been a fluctuation in blood glucose within that period that is sufficient to generate detectable stimulation to the microvessels. At this time, the next data point corresponding to that segment is marked as the stimulation anchor point, serving as a clear time marker for the blood glucose stimulation event. After determining the stimulation anchor point, a preset time window is opened, starting from the acquisition time corresponding to the stimulation anchor point. The setting of this time window needs to be combined with the physiological laws of clinical microvascular response to ensure that the entire response process of microvessels to blood glucose stimulation can be completely captured. Within this preset time window, the response data points of all microvascular-related detection indicators are retrieved one by one. For each response data point, its offset relative to the baseline before stimulation is calculated. The calculation logic for the offset is as follows:

[0058] The response offset is calculated by subtracting the baseline values ​​of microvascular indicators before the activation anchor point from the measured values ​​of the current response data point. This offset reflects the degree of change in microvessels after glucose stimulation. The response offset of each response data point is compared with a preset deviation threshold. When the response offset of a response data point is greater than the preset deviation threshold, it indicates that the response point can truly reflect the clear response of microvessels to the corresponding glucose stimulation. At this time, the response data point is marked as a response anchor point, serving as a clear time marker for the microvascular response event. Finally, the time difference between the activation anchor point and the response anchor point is calculated. Specifically, the time difference is calculated by subtracting the acquisition time corresponding to the activation anchor point from the acquisition time corresponding to the response anchor point. This time difference is the original time offset, and its magnitude directly reflects the initial response delay of microvessels to glucose stimulation.

[0059] Step 1 further includes the following steps:

[0060] Step 11: From the discontinuous blood glucose sequence, calculate the equivalent fluctuation energy based on the ratio of the time interval between two adjacent points to the blood glucose difference. Perform pattern recognition on three consecutive fluctuation segments. If the absolute value of the slope of the middle segment is greater than a threshold multiple of the sum of the absolute values ​​of the slopes of the preceding and following segments, mark the start and end points of the middle segment as the rising start point and peak point of the candidate excitation event, and output the candidate excitation event list. The specific operations are as follows:

[0061] In clinical settings, blood glucose sampling is limited by testing conditions, making continuous collection impossible. The resulting blood glucose sequences are often sparse, discontinuous, and discrete data points. Directly judging effective excitation based on a single-point slope can easily misinterpret random fluctuations as effective excitations, leading to biases in subsequent response anchor point retrieval. Therefore, multi-segment fluctuation analysis and pattern recognition are needed for accurate screening. During implementation, for every two adjacent sampling points in a discontinuous blood glucose sequence, the equivalent fluctuation energy of that fluctuation segment is calculated. The calculation logic for the equivalent fluctuation energy is consistent with the slope calculation, specifically:

[0062] The difference between two adjacent sampling points is obtained by subtracting the blood glucose value of the previous sampling point from the blood glucose value of the next sampling point. The time interval between two adjacent sampling points is obtained by subtracting the collection time of the previous sampling point from the collection time of the next sampling point. The calculated blood glucose difference is divided by the time interval, and the result is the equivalent fluctuation energy of the fluctuation segment. This energy value can accurately quantify the intensity and rate of blood glucose fluctuation within this period. The larger the energy value, the more intense the blood glucose fluctuation, and the more likely it is to constitute an effective stimulus.

[0063] After calculating the equivalent wave energy of all adjacent wave segments, three consecutive adjacent wave segments are used as a group of analysis units. Pattern recognition is then performed on each group of analysis units in turn. The aim is to screen out the middle segment where the wave intensity is significantly higher than the preceding and following waves, thereby locating the effective excitation event. The specific recognition logic is as follows:

[0064] Calculate the absolute value of the slope for each of the three consecutive fluctuation segments, using the same method as in step 1: subtract the previous blood glucose value from the subsequent value, then divide by the time interval. Next, sum the absolute values ​​of the slopes of the preceding and following fluctuation segments, and multiply this sum by a preset threshold to obtain a slope comparison threshold. Compare the absolute value of the slope of the intermediate fluctuation segment with this slope comparison threshold. If the absolute value of the slope of the intermediate fluctuation segment is greater than the slope comparison threshold, it indicates that the blood glucose fluctuation intensity of the intermediate fluctuation segment is significantly higher than that of the fluctuations before and after it, conforming to the fluctuation characteristics of an effective stimulus event, and capable of producing a clear stimulus to microvessels. The stimulation effect is as follows: At this time, the data point before the intermediate fluctuation segment, that is, the connection point between the intermediate fluctuation segment and the previous fluctuation segment, is marked as the rising start point of the candidate stimulation event, representing the starting time when blood glucose begins to show a significant upward fluctuation; the data point after the intermediate fluctuation segment, that is, the connection point between the intermediate fluctuation segment and the next fluctuation segment, is marked as the peak point of the candidate stimulation event, representing the time when blood glucose fluctuation reaches its peak; all rising start points and peak points that meet the above conditions are matched one by one to form a candidate stimulation event list and output it. Each candidate stimulation event contains a clear rising start time and peak time.

[0065] Step 12: Based on the peak time of the candidate excitation event, obtain the microvascular response value within the subsequent time window. Arrange the deviation direction of each response value relative to the baseline into a sign sequence. Calculate the ratio of the longest consecutive sign with the same sign to the total length. If this ratio exceeds the threshold and the last sign is consistent with the expected abnormal direction, then the point is confirmed as a valid response anchor point. Output the valid response anchor point and its time point. The specific operation is as follows:

[0066] Since microvascular response signals are often accompanied by detection noise, some response data points may be spurious fluctuations caused by noise interference rather than a true response to glucose stimulation. Directly using all response points with offsets exceeding the threshold as response anchors would lead to significant errors in time offset calculations. Therefore, symbolic sequence analysis is needed to confirm effective response anchors. During implementation, the specific acquisition time corresponding to the peak point of each candidate stimulation event output in step 11 is extracted. Using this peak time as a benchmark, a preset time window is opened afterward. The duration of this time window needs to be set in conjunction with the physiological characteristics of clinical microvascular responses to ensure complete coverage of the microvascular response cycle to glucose stimulation, without omitting true responses or including excessive irrelevant subsequent data. Within this preset time window, the response measurements of all microvascular-related detection indicators are acquired one by one. These measurements reflect the dynamic changes in the functional state of microvessels after the appearance of the glucose stimulation peak. The deviation direction of each response measurement is judged, and the specific judgment logic is as follows:

[0067] The baseline values ​​of microvascular indicators before the stimulus are subtracted from the current response measurement. A positive result indicates that the response measurement is higher than the baseline value, and the deviation direction is positive; a negative result indicates that the response measurement is lower than the baseline value, and the deviation direction is negative. The deviation directions of all response measurements are arranged sequentially according to their acquisition time to form a continuous symbol sequence. This sequence can intuitively reflect the continuity and change pattern of the microvascular response. Then, the symbol sequence is statistically analyzed to calculate the longest segment length that continuously maintains the same symbol, that is, the number of symbols contained in the longest segment in the symbol sequence that continuously contains positive or negative symbols. At the same time, the total length of the entire symbol sequence is calculated, that is, the total number of symbols contained in the symbol sequence. The longest segment length with consecutive same signs is divided by the total length of the symbol sequence to obtain the proportion of consecutive same signs. This proportion can quantify the persistence of the microvascular response. The higher the proportion, the more stable the response, and the more likely it is a true response to blood glucose stimulus rather than random noise.

[0068] The calculated percentage of consecutive identical signs is compared with a preset percentage threshold. At the same time, it is checked whether the sign at the end of the sign sequence is consistent with the preset expected abnormal direction. The expected abnormal direction is preset based on the pathological characteristics of microvascular complications. That is, after microvessels are stimulated by abnormal blood glucose, their response value should deviate from the baseline in a specific direction, such as increasing or decreasing. When the percentage of consecutive identical signs is greater than the preset percentage threshold, and the sign at the end of the sign sequence is consistent with the preset expected abnormal direction, it indicates that the response point can truly reflect the abnormal response of microvessels to blood glucose stimulation and there is no obvious noise interference. At this time, the response measurement point corresponding to the end sign is confirmed as an effective response anchor point, and the specific measurement value of the effective response anchor point and its corresponding acquisition time are output synchronously.

[0069] Step 13: Use the time difference between the response anchor point and the peak point as the original time offset and the maximum delay, and the time difference between the first sign-reversed measurement point after the response anchor point and the response anchor point as the minimum delay. Take the geometric midpoint between the minimum and maximum delays as the original time offset output after noise suppression. The specific operation is as follows:

[0070] Since the microvascular response signal is inevitably affected by factors such as the accuracy of the detection instrument and fluctuations in the patient's physiological state during the acquisition process, it contains a certain amount of random noise. If the time difference between the effective response anchor point and the peak point determined in step 12 is directly used as the original time offset, the data may be biased due to noise interference, thus affecting the accuracy of subsequent evaluation results. Therefore, noise suppression is required through dual delay calculation and geometric midpoint optimization. First, the initial original time offset and the maximum delay are calculated. The specific calculation logic is as follows:

[0071] The initial raw time offset is obtained by subtracting the acquisition time corresponding to the peak point of the candidate excitation event determined in step 11 from the acquisition time corresponding to the effective response anchor point confirmed in step 12. This time difference also serves as the maximum delay, which physically represents the longest delay time for the detectable response generated by microvessels to glucose excitation to reach the threshold, reflecting the slowest speed of the microvascular response. Subsequently, after the acquisition time corresponding to the effective response anchor point, the microvascular response symbol sequence formed in step 12 is retrieved, and the first measurement point in the symbol sequence showing a sign reversal is located. This is the first response measurement point whose deviation direction is opposite to that of the effective response anchor point. The appearance of this point indicates that the microvascular response has begun to show a reverse change, effectively eliminating the interference of later noise on the response delay judgment. The time difference between this sign reversal point and the effective response anchor point is calculated using the following logic:

[0072] The time difference obtained by subtracting the acquisition time corresponding to the effective response anchor point from the acquisition time corresponding to the sign reversal point is used as the minimum delay. Its physical meaning is the shortest delay time after the microvascular response reaches its peak and begins to show an inverse change, reflecting the fastest speed of the microvascular response. Finally, the calculated minimum and maximum delays are geometrically averaged to obtain the original time offset after noise suppression. The formula for calculating the geometric average is: ;

[0073] The derivation logic of this formula is as follows: the minimum delay and the maximum delay correspond to the fastest and slowest speeds of microvascular response, respectively. The geometric mean of the two can balance the errors of the two, effectively offsetting the bias caused by random detection noise, and reflecting the true level of microvascular response delay better than the arithmetic mean; in the formula, This represents the original time offset after noise suppression; Represents minimum delay; This represents the maximum delay. The geometric midpoint calculated using this formula is used as the original time offset output after noise suppression, ensuring that the first and second decay time constants derived from this data can accurately reflect the true response delay characteristics of microvessels.

[0074] In a preferred embodiment of the present invention, step 2 is further included: a first decay time constant and a second decay time constant are inversely calculated from the original time offset, response measurement value, and pre-excitation baseline value based on the double exponential decay recovery model, wherein the first decay time constant is less than the second decay time constant; the ratio of the two is output as the response delay exponent. The specific operation is as follows:

[0075] The recovery process of the response signal generated by microvessels after glucose stimulation is not a single-rate decay, but rather exhibits two stages: rapid decay and slow decay. These two stages correspond to the functional states of different layers of microvessels. Therefore, the double exponential decay recovery model can accurately fit this dynamic process. Based on the original time offset output in step 13, the maximum and minimum delays contained therein are extracted. Combined with the response measurements corresponding to these two delays and the baseline values ​​of microvascular indicators before stimulation, the data is preprocessed and normalized to construct the input parameters required for the model. Based on the mathematical characteristics of the double exponential decay recovery model, the preprocessed parameters are substituted into the model to construct a simultaneous constraint relationship. This constraint relationship is solved using the interval contraction approximation method to obtain two decay time constants, ensuring that the first decay time constant is less than the second decay time constant. The first decay time constant corresponds to the rapid decay stage of the microvascular response, and the second decay time constant corresponds to the slow decay stage. Finally, the two decay time constants are processed through specific mathematical operations to obtain a response delay index that comprehensively reflects the degree of microvascular response delay. This index can effectively quantify the recovery ability of microvessels to glucose stimulation and indirectly reflect the damage state of microvascular reserve function.

[0076] Step 2 further includes the following steps:

[0077] Step 21: Using the maximum and minimum delays in the original time offset, locate the response anchor point measurement corresponding to the maximum delay and the first sign reversal point measurement corresponding to the minimum delay. Subtract the pre-excitation baseline value from each measurement and take the ratio to output the normalized response pair. The specific operation is as follows:

[0078] First, extract the maximum and minimum delays from the raw time offset output in step 13. These two delay values ​​have been processed by noise suppression and can accurately reflect the key time nodes of the microvascular response. Based on the time point corresponding to the maximum delay, locate the response measurement value corresponding to the effective response anchor point confirmed in step 12. This measurement value is the true response state of the microvessel at the maximum delay time. Based on the time point corresponding to the minimum delay, locate the response measurement value corresponding to the first sign reversal point determined in step 13. This measurement value is the state at the moment when the microvascular response begins to show a reverse change. Baseline correction is then performed on these two response measurement values. The specific calculation logic is as follows:

[0079] The response deviation value corresponding to the maximum delay is obtained by subtracting the baseline value of the microvascular index before excitation from the measured value of the response anchor point corresponding to the maximum delay; the response deviation value corresponding to the minimum delay is obtained by subtracting the baseline value of the microvascular index before excitation from the measured value of the sign reversal point corresponding to the minimum delay. To eliminate the influence of the difference between the measurement scale and the baseline, the two response deviation values ​​are normalized. Specifically, the response deviation value corresponding to the minimum delay is divided by the response deviation value corresponding to the maximum delay, and the ratio obtained is the normalized response pair. This normalized response pair is output together with the corresponding maximum and minimum delay time points to form the standardized data pair required for model construction.

[0080] Step 22: Using the normalized response pairs and their corresponding time points, construct two constraint equations for the double exponential decay recovery model. Each equation constrains the double exponential term to be equal to the normalized response value at that time point, and the first decay time constant and the second decay time constant in the two equations share the same value. Output the simultaneous constraint relationship between the first decay time constant and the second decay time constant. The specific operation is as follows:

[0081] The recovery process of microvascular response to glucose stimulation exhibits two independent phases: rapid decay and slow decay. A single exponential model cannot accurately fit this complex process; therefore, a double exponential decay recovery model is used. This model can capture the characteristics of each decay phase separately, thus accurately deducing the corresponding decay time constant. The formula for the double exponential decay recovery model is: ;

[0082] The derivation logic of this formula is as follows: Combining the physiological mechanism of microvascular response, the response recovery process consists of the superposition of a rapid decay component and a slow decay component. The rapid decay component corresponds to the rapid recovery of the superficial function of microvessels, and the slow decay component corresponds to the slow recovery of the deep function of microvessels. Both components follow the exponential decay law. Therefore, the two exponential decay terms are superimposed to form a double exponential decay recovery model. Since the response value normalization and baseline correction have been completed in step 21, there is no need to add an extra constant term to the model. Only the two exponential decay terms need to be retained. In the formula, y(t) represents the normalized response value; t represents the time calculated from the excitation peak point. This represents the first decay time constant, corresponding to the rapid decay phase; represents the second decay time constant, corresponding to the slow decay stage; A and B represent the amplitude coefficients of the two decay components, respectively.

[0083] During implementation, the normalized response pairs and their corresponding time points output in step 21 are substituted into the model to obtain two constraint equations: the time corresponding to the maximum delay and the normalized response value are substituted into the model to obtain the first constraint equation; the time corresponding to the minimum delay and the normalized response value are substituted into the model to obtain the second constraint equation. In the two constraint equations, the first decay time constant and the second decay time constant share the same set of values, and the amplitude coefficients A and B are also shared by the two sets of equations. This forms a simultaneous constraint relationship about the first decay time constant, the second decay time constant, and the amplitude coefficients. This simultaneous constraint relationship is then output.

[0084] Step 23: Using the simultaneous constraint relationship, the interval shrinkage approximation method is adopted: the physiological upper and lower limits of the first decay time constant are preset and divided into equal sub-intervals. In each sub-interval, the simultaneous constraint relationship is transformed into a single-variable equation about the second decay time constant. The absolute value of the equation residuals in each sub-interval is compared to narrow down the candidate interval until the residuals are lower than the preset threshold. The first decay time constant and the second decay time constant at this time are output. The specific operation is as follows:

[0085] Since the simultaneous constraint relationship involves multiple variables, it is difficult to obtain a unique solution by direct solution, and the decay time constant has a clear physiological range. Therefore, the interval contraction approximation method is adopted to achieve accurate solution by gradually narrowing down the candidate interval. In the implementation process, combined with the clinical microvascular physiological characteristics, the physiological upper and lower limits of the first decay time constant are preset. These upper and lower limits need to cover the time range of the rapid decay stage of normal human microvessels to ensure that the solution results are consistent with the physiological reality and avoid abnormal values ​​that exceed the reasonable range. The preset physiological upper and lower limits of the first decay time constant are divided into several sub-intervals. The number of sub-intervals can be set according to the solution accuracy requirements. The more sub-intervals, the higher the solution accuracy.

[0086] For each subinterval, any value within that subinterval is selected as a candidate value for the first decay time constant. This candidate value is substituted into the simultaneous constraint relationship formed in step 22. At this point, the simultaneous constraint relationship is transformed into a single-variable equation only concerning the second decay time constant. By solving this single-variable equation, the corresponding candidate value for the second decay time constant is obtained. Then, the obtained candidate values ​​for the first and second decay time constants are substituted into the double exponential decay recovery model. The difference between the normalized response value output by the model and the actual normalized response value output in step 21 is calculated. The absolute value of this difference is the equation residual. The smaller the residual, the closer the candidate value is to the true value.

[0087] Compare the absolute values ​​of the residuals within all sub-intervals, retain the sub-interval with the smallest absolute residual value as the new candidate interval, and discard the remaining sub-intervals with larger residuals. This completes one interval shrinkage. Repeat the above process, continuously subdividing the candidate interval into smaller intervals, solving for the residuals, and narrowing the candidate interval until the calculated absolute value of the residual is lower than the preset residual threshold. At this point, the corresponding first decay time constant and second decay time constant are the desired results. At the same time, it is necessary to ensure that the output first decay time constant is less than the second decay time constant, which conforms to the physiological logic of fast decay and slow decay in the double exponential decay recovery model. If the first decay time constant is greater than or equal to the second decay time constant, the sub-interval division or physiological upper and lower limits need to be readjusted, and the approximation solution is performed again until two decay time constants that meet the requirements are obtained and output.

[0088] Step 24: Using the first decay time constant and the second decay time constant, calculate the ratio of the second decay time constant to the first decay time constant, then perform logarithmic compression on this ratio with the natural constant as the base, and output the compressed ratio as the response delay exponent. The specific operation is as follows:

[0089] The first decay time constant corresponds to the rapid decay phase of the microvascular response, and the second decay time constant corresponds to the slow decay phase. The ratio of the two can reflect the degree of difference between the two decay phases, and thus indirectly reflect the state of microvascular reserve function. The larger the ratio, the higher the proportion of the slow decay phase, the weaker the microvascular recovery ability, and the more obvious the response delay. The specific calculation logic for the ratio is as follows: the second decay time constant output in step 23 is divided by the first decay time constant to obtain the ratio of the two decay time constants. This ratio can initially reflect the relative degree of microvascular response delay. However, since this ratio may have problems such as an excessively large numerical range and uneven distribution, it is not conducive to the subsequent calculation of nonlinear mapping and exhaustion coefficient. Therefore, it is necessary to perform logarithmic compression on this ratio.

[0090] Logarithmic compression employs logarithmic operations with the natural constant as the base. Specifically, the ratio of two decay time constants is taken as the natural logarithm to obtain the compressed ratio. This compression process can scale excessively large ratios to a reasonable range while preserving the relative differences between the two decay time constants, thus avoiding deviations in subsequent evaluation results due to extreme values. The ratio after natural logarithmic compression is output as the response delay index, which can accurately quantify the degree of response delay of microvessels to glucose stimulation.

[0091] In a preferred embodiment of the present invention, step 3 is further included: dynamically updating the adaptive reference delay interval based on the patient's own historical data, performing a nonlinear mapping on the response delay exponent, with the boundary mapping being a constant and the intermediate part using a nonlinear compression transformation, and outputting the exhaustion coefficient. The specific operation is as follows:

[0092] Because of individual differences in the basic microvascular function of different patients, using a uniform and fixed reference standard can lead to biases in the assessment results. Therefore, constructing an adaptive reference interval based on the patient's own historical data can achieve individualized assessment and improve the accuracy of the assessment. By extracting effective data from the patient's own historical response delay index sequence and calculating the reference center value and reference dispersion, the adaptive reference delay interval is initially constructed. According to the offset direction of the response delay index of the current event and the reference center value, the upper and lower limits of the reference interval are dynamically adjusted so that the reference interval can adapt to the dynamic changes of the patient's own microvascular function. Then, the current response delay index is compared with the adjusted adaptive reference interval. Different mapping methods are used according to the position of the interval. Constant mapping is used for the boundary region and nonlinear compression transformation is used for the middle region to obtain the intermediate mapping value. Finally, according to the source of the intermediate mapping value, the final depletion coefficient is obtained through local fine-tuning or direct output. This coefficient can accurately quantify the degree of depletion of microvascular reserve function.

[0093] Step 3 further includes the following steps:

[0094] Step 31: Based on the response delay exponential sequence, extract the values ​​within the predetermined physiological cutoff range from the sequence, sort them by value size, remove the extreme values ​​at both ends by a fixed proportion, calculate the sliding median and sliding absolute deviation of the remaining values, and output the patient's own reference center value and reference dispersion. The specific operation is as follows:

[0095] The response delay index sequence is formed by arranging multiple historical response delay indices output in step 24 in chronological order. It contains the microvascular response delay characteristics of the patient at different times, but it may contain abnormal values ​​that exceed the physiological range. These values ​​need to be screened before they can be used as a reference. During the implementation, the entire response delay index sequence is traversed, and the values ​​in the sequence that are within the preset physiological cutoff range are extracted. This physiological cutoff range is set in combination with the physiological characteristics of microvessels, which can exclude abnormal response delay indices caused by detection errors or extreme physiological states, and ensure that the screened data are consistent with the patient's true microvascular function status. The filtered values ​​are sorted in ascending order. After sorting, extreme values ​​at both ends of the sequence are removed at a fixed ratio. The setting of the fixed ratio needs to balance the amount of data and the effect of outlier removal to avoid removing too much valid data or missing outliers. After removing extreme values, the moving median and moving absolute deviation are calculated for the remaining valid values. The logic for calculating the moving median is as follows: a preset moving window length is set, and the remaining valid value sequence is traversed point by point in the moving window as a unit. The median of the values ​​in each window is calculated. The medians of all windows constitute the moving median sequence. The mean of this sequence is taken as the patient's own reference center value. The reference center value reflects the baseline level of the patient's own microvascular response delay.

[0096] The calculation logic of the sliding absolute deviation is as follows: First, calculate the difference between each remaining valid value and the sliding median, and take the absolute value of the difference to obtain the absolute deviation sequence; then, calculate the sliding median using the same sliding window on the absolute deviation sequence, and take the mean of the sliding median sequence as the patient's own reference dispersion, which reflects the fluctuation range of the patient's own response delay index.

[0097] Step 32: Relying on the reference center value and reference dispersion, the lower and upper limits of the adaptive reference delay interval are determined by a multiple of the reference center value plus or minus the reference dispersion. The value of this multiple is adjusted according to the offset direction of the event response delay index relative to the reference center value: increasing the first preset step size when the offset direction is positive, and decreasing the second preset step size when the offset direction is negative. The adjusted upper and lower limits of the adaptive reference delay interval are then output. The specific operations are as follows:

[0098] The adaptive reference delay interval is constructed based on the patient's own data, rather than using a uniform fixed interval. This fully considers individual patient differences and the dynamic evolution of microvascular function, improving the accuracy of subsequent mapping results. First, an initial adaptive reference delay interval is constructed. The specific calculation logic is as follows: the lower limit of the initial adaptive reference delay interval is obtained by subtracting the product of the reference dispersion and a preset multiple from the reference center value. The upper limit of the initial adaptive reference delay interval is obtained by adding the product of the reference dispersion and the preset multiple to the reference center value. The preset multiple is set in combination with the clinical data statistics to ensure that the initial interval can cover the response delay index range under normal patient conditions.

[0099] Subsequently, the offset direction of the response delay index from the reference center value is determined. The specific logic is as follows: the response delay index of the current event is subtracted from the reference center value. If the result is positive, it indicates that the response delay index is higher than the reference center value, and the offset direction is positive; if the result is negative, it indicates that the response delay index is lower than the reference center value, and the offset direction is negative. The value of the preset multiple is adjusted according to the offset direction. If the offset direction is positive, the preset multiple is increased by a first preset step size. The first preset step size is set to a small, fixed value to ensure the smoothness of the interval adjustment and avoid interval distortion due to excessive single adjustment. If the offset direction is negative, the preset multiple is decreased by a second preset step size. The second preset step size can be the same as the first preset step size or set according to clinical needs, also ensuring the smoothness of the adjustment. After adjusting the preset multiple, the upper and lower limits of the adaptive reference delay interval are recalculated. The calculation logic is the same as the initial interval: the lower limit is the reference center value minus the product of the adjusted multiple and the reference dispersion, and the upper limit is the reference center value plus the product of the adjusted multiple and the reference dispersion.

[0100] Step 33: Relying on the upper and lower limits of the adaptive reference delay interval, compare the response delay exponent of this event with the interval: if it is less than or equal to the lower limit, it is mapped to the first constant; if it is greater than or equal to the upper limit, it is mapped to the second constant; if it is within the interval, a composite operation of piecewise linear stretching followed by logarithmic shrinking based on the interval width is used to output the intermediate mapped value. The specific operation is as follows:

[0101] The response delay index in different intervals corresponds to different states of microvascular reserve function. The boundary region corresponds to the extreme state of depletion, and constant mapping ensures the uniformity of the evaluation standard. The intermediate region corresponds to the gradual state of depletion, and nonlinear compression transformation can accurately capture subtle changes while suppressing the interference of extreme values. The response delay index of this event is compared one by one with the upper and lower limits of the adjusted adaptive reference delay interval to determine its interval position. If the response delay index is less than or equal to the lower limit of the adaptive reference delay interval, it indicates that the microvascular reserve function is depleted at a low level, and it is mapped to the first constant, which is a preset fixed value corresponding to the standardized value of the low depletion state. If the response delay index is greater than or equal to the upper limit of the adaptive reference delay interval, it indicates that the microvascular reserve function is depleted at a high level, and it is mapped to the second constant, which is also a preset fixed value and is greater than the first constant, corresponding to the standardized value of the high depletion state.

[0102] If the response delay index falls between the upper and lower limits of the adaptive reference delay interval, it indicates that the depletion of microvascular reserve function is in a gradual state. A composite operation of piecewise linear stretching followed by logarithmic contraction based on the interval width is used to obtain an intermediate mapping value. The formula for the composite operation is: ;

[0103] The derivation logic of this formula is as follows: Linear stretching maps the response delay index within the interval to a unified interval of 0 to 1, eliminating the influence of interval width differences. Then, logarithmic contraction with a base of the natural constant suppresses extreme fluctuations in the stretched value while preserving relative differences between values, making the mapping result more closely match the gradual change in microvascular depletion. In the formula, y represents the intermediate mapped value; x represents the response delay index of the current event; L represents the lower limit of the adaptive reference delay interval; and U represents the upper limit of the adaptive reference delay interval. The linear stretching operation logic is to subtract the lower limit of the interval from the current response delay index, and then divide by the difference between the upper and lower limits of the interval to obtain the stretched value. Subsequently, one is added to this stretched value, and the natural logarithm is taken to obtain the intermediate mapped value. This intermediate mapped value is output, and it is marked as originating from nonlinear compression within the interval.

[0104] Step 34: Depending on the intermediate mapping value and whether it comes from within the interval, if it comes from nonlinear compression within the interval, calculate the difference between the response delay exponent and the reference center value, divide it by the reference dispersion as a local fine-tuning coefficient, multiply the intermediate mapping value by this coefficient, and output the final exhaustion coefficient; if it comes from boundary constant mapping, directly output the constant as the exhaustion coefficient. The specific operation is as follows:

[0105] The source of intermediate mapping values ​​differs, leading to different fine-tuning requirements. Nonlinear mappings within an interval require local fine-tuning based on the reference center value and reference dispersion to adapt to the patient's own fluctuation characteristics. Constant mappings in the boundary region already meet the standardization requirements and require no additional fine-tuning. During implementation, the source identifier of the intermediate mapping value is first determined to confirm whether it originates from nonlinear compression within the interval or from constant mapping at the boundary. If the intermediate mapping value originates from nonlinear compression within the interval, local fine-tuning coefficients need to be calculated. The calculation logic for local fine-tuning coefficients is as follows:

[0106] The local fine-tuning coefficient is obtained by subtracting the reference center value output in step 31 from the response delay index of this event, and dividing the difference by the reference dispersion output in step 31. This coefficient reflects the degree of deviation of the response delay index from the patient's own baseline level, enabling individualized fine-tuning of the mapping value. The final exhaustion coefficient is obtained by multiplying the intermediate mapping value by this local fine-tuning coefficient. This exhaustion coefficient retains the standardized characteristics after nonlinear mapping and incorporates the patient's own fluctuation pattern, accurately quantifying the degree of exhaustion of microvascular reserve function in this event. If the intermediate mapping value comes from the boundary constant mapping, it indicates that the response delay index is in an extreme exhaustion state, and no additional fine-tuning is required. This constant is directly used as the final exhaustion coefficient output. In any case, the output exhaustion coefficient quantifies the degree of exhaustion of microvascular reserve function, and the larger the value, the more severe the exhaustion of microvascular reserve function.

[0107] In a preferred embodiment of the present invention, step 4 is further included: calculating the weight coefficient of each excitation anchor point. This coefficient is determined based on the ratio of the magnitude of the blood glucose peak exceeding the fasting baseline to the pathological threshold. The weight of each event is multiplied by a decay factor that decreases exponentially with the time of the event. The total weighted exhaustion and the total expected baseline load are accumulated, and the subclinical progression index of the ratio of the two is output. The specific operation is as follows:

[0108] Different triggering anchors have varying effects on microvascular damage; the greater the blood glucose peak exceeds the fasting baseline, the stronger the damage. Simultaneously, the further back in time the triggering event occurred, the weaker its impact on current microvascular function. Therefore, weighting coefficients and time decay factors are needed to reflect this difference. The weighting coefficient for each triggering anchor is calculated, derived from the ratio of the blood glucose peak exceeding the fasting baseline to the pathological threshold, reflecting the intensity of damage from the triggering event. A time decay factor is calculated for each triggering anchor to reflect the attenuation of the event's impact over time. Based on the weighting coefficients and decay factors, the weighted exhaustion and expected baseline load for each triggering event are calculated. These two values ​​are then summed chronologically. If the sample size is insufficient, compensation is performed. Finally, the ratio of the accumulated values ​​and logarithmic compression yield the final subclinical progression index, which comprehensively reflects the cumulative progression of microvascular complications in patients.

[0109] Step 4 further includes the following steps:

[0110] Step 41: Based on the peak blood glucose value and fasting baseline value of each excitation anchor point, calculate the absolute magnitude of the peak value exceeding the baseline. Divide this magnitude by the pathological threshold to obtain the relative ratio. Then, perform logarithmic stretching of the relative ratio with the natural constant as the base and add one to it. Output the stretched original weight coefficients. The specific operation is as follows:

[0111] The peak blood glucose value at the excitation anchor point is the blood glucose value corresponding to the peak point of the candidate excitation event determined in step 11. The fasting baseline value is the blood glucose benchmark value of the patient in a fasting state. The difference between the two directly reflects the abnormal blood glucose level of the excitation event. The larger the abnormal level, the higher the risk of damage to microvessels, and the larger the corresponding weighting coefficient should be. The absolute magnitude of the blood glucose peak exceeding the fasting baseline is calculated. The specific calculation logic is as follows: the blood glucose peak value of each excitation anchor point is subtracted from the fasting baseline value, and the difference obtained is the absolute magnitude. This value intuitively reflects the degree of abnormality of the blood glucose peak relative to the fasting baseline.

[0112] The absolute amplitude is standardized by dividing it by a preset pathological threshold to obtain a relative ratio. The pathological threshold is a clinically set threshold for blood glucose abnormality damage. The relative ratio eliminates the influence of individual fasting baseline differences, making the damage intensity of different stimulus events comparable. Since the relative ratio may have a narrow range and unclear differences, which is not conducive to the accuracy of subsequent weighted calculations, it needs to be logarithmically stretched with the natural constant as the base, and then one is added to ensure that the stretched weight coefficient is positive and can amplify the intensity differences of different stimulus events. The specific calculation is as follows: take the natural logarithm of the relative ratio, add one to the logarithmic result, and the resulting value is the original weight coefficient after stretching.

[0113] Step 42: Based on the difference between the time point of each excitation anchor point and the current evaluation time point, divide the difference by the preset half-life constant to obtain the dimensionless duration ratio, and then calculate the decay factor using an exponential function with the natural constant as the base, outputting the time decay factor for each event. The specific operation is as follows:

[0114] The longer the excitation event occurred, the weaker its damaging effect on current microvascular function. The time decay factor should decrease as the time difference increases, which conforms to the natural decay law. Therefore, an exponential function with the natural constant as the base is used to construct the decay factor. The time difference between the time point of occurrence of each excitation anchor point and the current evaluation time point is calculated. The specific calculation logic is as follows: the difference between the current evaluation time point and the time point of occurrence of each excitation anchor point is the time difference. This value reflects the time since the excitation event. The longer the time, the smaller the decay factor.

[0115] The time difference is converted into a dimensionless duration ratio. The specific calculation is as follows: the time difference is divided by a preset half-life constant, which is the time required for the clinically defined stimulus effect to decay by half. The dimensionless duration ratio eliminates the influence of differences in time units, providing a standardized basis for calculating the decay factor. The formula for calculating the time decay factor is: ;

[0116] The derivation logic of this formula is as follows: Combining the natural decay law of the excitation effect, the decay factor should decrease exponentially with the increase of the time difference. An exponential function with the natural constant as the base can accurately fit this trend. The half-life constant determines the decay rate. Therefore, by combining the dimensionless duration ratio with the exponential function, a formula for the time decay factor is constructed; in the formula, Represents the time decay factor; The value represents the difference between the time of the excitation anchor point occurrence and the current evaluation time; T represents the preset half-life constant; substituting the dimensionless duration ratio into this formula, that is, taking the natural exponent for the negative dimensionless duration ratio, yields the time decay factor for each excitation event.

[0117] Step 43: The product of the stretched original weight coefficients and the time decay factor is used as the weighted exhaustion contribution value. The product of the exhaustion coefficient of each event and the weighted exhaustion contribution value is used as the weighted exhaustion amount. The product of the weighted exhaustion contribution value and the average exhaustion coefficient of the healthy population is used as the expected baseline load. The weighted exhaustion amount sequence and the expected baseline load sequence for each event are output. The specific operations are as follows:

[0118] The original weighting coefficients reflect the damage intensity of the stimulus event, while the time decay factor reflects the time effect of the stimulus event. Combining the two yields the effective damage intensity of each stimulus event. By combining the exhaustion coefficient with the baseline of healthy individuals, the actual exhaustion and expected load can be quantified. The weighted exhaustion contribution value of each stimulus event is calculated. The specific calculation logic is as follows: the stretched original weighting coefficients output in step 41 are multiplied by the time decay factor output in step 42. The product obtained is the weighted exhaustion contribution value. This value comprehensively reflects the effective damage intensity of each stimulus event, taking into account both the impact of abnormal blood glucose levels and the time decay effect.

[0119] Subsequently, the weighted exhaustion amount for each stimulus event is calculated. Specifically, the exhaustion coefficient corresponding to each stimulus event output in step 3 is multiplied by the weighted exhaustion contribution value of that event, and the resulting product is the weighted exhaustion amount. This value reflects the actual degree of microvascular reserve function exhaustion caused by each stimulus event. The larger the exhaustion coefficient and the larger the weighted exhaustion contribution value, the higher the weighted exhaustion amount. At the same time, the expected baseline load for each stimulus event is calculated. Specifically, the weighted exhaustion contribution value of that event is multiplied by the preset average exhaustion coefficient of the healthy population, and the resulting product is the expected baseline load. The average exhaustion coefficient of the healthy population is the baseline value of microvascular exhaustion in the healthy population as statistically observed in clinical practice. The expected baseline load reflects the theoretical exhaustion degree of the healthy population under the same effective damage intensity. The weighted exhaustion amounts of all stimulus events are arranged in chronological order to form a weighted exhaustion amount sequence. The expected baseline load amounts of all stimulus events are also arranged in chronological order to form an expected baseline load sequence.

[0120] Step 44: Based on the weighted exhaustion sequence and the expected baseline load sequence, sum the cumulative weighted exhaustion sum and the cumulative expected baseline load sum according to the event occurrence time sequence. If the total number of events is lower than the minimum sample threshold, multiply the cumulative sum by a compensation factor, which is equal to the minimum sample threshold divided by the actual total number of events. Output the compensated cumulative sum. The specific operation is as follows:

[0121] Cumulative summation reflects the cumulative damage effect of all excitation events on microvessels, while compensation processing corrects statistical bias caused by insufficient sample size, making the cumulative result more consistent with the patient's true microvascular functional status. The cumulative summation is performed on all values ​​in the weighted exhaustion sequence according to the chronological order of the excitation events. The specific calculation logic is as follows: starting from the weighted exhaustion corresponding to the earliest excitation event, the weighted exhaustion of each subsequent excitation event is added in sequence, and the sum obtained is the cumulative weighted exhaustion sum. This sum reflects the cumulative degree of microvascular exhaustion caused by all excitation events.

[0122] Following the same chronological order, all values ​​in the expected baseline load sequence are cumulatively summed. The specific calculation logic is as follows: starting with the expected baseline load corresponding to the earliest stimulus event, the expected baseline load of each subsequent stimulus event is added sequentially. The sum obtained is the cumulative expected baseline load, which reflects the theoretical cumulative exhaustion level of healthy individuals under the same stimulus sequence. The total number of stimulus events is counted and compared with a preset minimum sample threshold. If the total number of events is greater than or equal to the minimum sample threshold, the sample size is sufficient, and no compensation processing is needed; the cumulative weighted exhaustion sum and the cumulative expected baseline load sum are directly output. If the total number of events is less than the minimum sample threshold, the sample size is insufficient, and compensation processing is required. The compensation factor is calculated by dividing the preset minimum sample threshold by the actual total number of stimulus events. The ratio obtained is the compensation factor. A compensation factor greater than one amplifies the cumulative sum and corrects the bias caused by insufficient sample size. The cumulative weighted exhaustion sum is multiplied by the compensation factor to obtain the compensated cumulative weighted exhaustion sum; the cumulative expected baseline load sum is multiplied by the compensation factor to obtain the compensated cumulative expected baseline load sum.

[0123] Step 45: Based on the ratio of the compensated cumulative weighted exhaustion to the compensated cumulative expected baseline load, this ratio is logarithmically compressed to the base of the natural constant and then incremented by one to output the final subclinical progression index. The specific operation is as follows:

[0124] The compensated cumulative weighted exhaustion sum reflects the patient's actual exhaustion level, while the compensated cumulative expected baseline load sum reflects the theoretical exhaustion level of healthy individuals. The ratio of the two directly reflects the exhaustion difference between the patient and the healthy population, and then logarithmic compression is used to standardize the values. The specific calculation logic for the ratio of the two compensated cumulative sums is as follows: the compensated cumulative weighted exhaustion sum is divided by the compensated cumulative expected baseline load sum, and the resulting ratio is the exhaustion cumulative ratio. When the ratio is greater than one, it indicates that the patient's exhaustion level is higher than that of healthy individuals, and the larger the ratio, the more obvious the progression. When the ratio is less than one, it indicates that the patient's exhaustion level is lower than that of healthy individuals.

[0125] Because the cumulative exhaustion ratio may have an excessively large numerical range and uneven distribution, which is not conducive to subsequent trend judgment and risk coding, it needs to be logarithmically compressed to the base of the natural constant, and then one is added to ensure that the index is positive and within a reasonable range. The formula for calculating the subclinical progression index is: ;

[0126] The derivation logic of this formula is as follows: the cumulative exhaustion ratio can reflect the difference in exhaustion between patients and healthy individuals, but the value fluctuates significantly. Natural logarithmic compression can scale the ratio to a reasonable range, eliminating the interference of extreme values. Adding a '-' ensures that the compressed values ​​are all positive, facilitating subsequent threshold judgment and risk coding. In the formula, S represents the subclinical progression index; B represents the cumulative weighted exhaustion after compensation; and C represents the cumulative expected baseline load after compensation. Substituting the cumulative exhaustion ratio into this formula, i.e., taking the natural logarithm of the cumulative exhaustion ratio, and adding one to the logarithm result, yields the final subclinical progression index.

[0127] In a preferred embodiment of the present invention, step 5 is further included: performing a signed-rank test on the longitudinally acquired subclinical progression index sequence to determine its monotonic trend; determining whether it is an increasing trend or a non-increasing trend based on the sum of the ranks of the positive and negative differences; and outputting the sign of the trend slope. The specific operation is as follows:

[0128] The longitudinally acquired subclinical progression index sequence is formed by arranging multiple subclinical progression indices output in step 45 in chronological order of acquisition time. This sequence reflects the degree of subclinical progression of microvascular complications at different time points in the patient. By judging the trend, the development trend of complications can be clearly identified. An increasing trend means that the complications continue to progress, while a non-increasing trend means that the progression has slowed down or stabilized. The sequence is first preprocessed, and the difference between adjacent indices is calculated and assigned a signed rank. The trend characteristics of the original sequence are transformed into quantifiable rank information. By summing and comparing the signed ranks, the monotonic trend of the sequence is judged, and finally the trend slope sign is output. This sign intuitively reflects the trend direction.

[0129] Step 5 further includes the following steps:

[0130] Step 51: Based on the longitudinally acquired subclinical progression index sequence, calculate the difference between two adjacent indices in chronological order. Assign natural number ranks to the absolute values ​​of each difference in ascending order and retain the original sign of each difference. Output the signed rank sequence. The specific operations are as follows:

[0131] The subclinical progression index sequence is arranged in chronological order. The difference between two adjacent indices directly reflects the change in the degree of progression between two time points. A positive difference indicates an increase in the degree of progression, a negative difference indicates a decrease in the degree of progression, and a zero difference indicates a stable degree of progression. Assigning a rank can eliminate the interference of the absolute value of the difference and focus on the statistical characteristics of the direction of change. The subclinical progression index sequence collected longitudinally is extracted, and the difference between two adjacent subclinical progression indices is calculated in chronological order. The specific calculation logic is as follows: the subclinical progression index at the later time point is subtracted from the subclinical progression index at the previous time point. The difference obtained is the change between two adjacent indices. The sign of this change directly reflects the direction of change in the degree of progression.

[0132] After calculating all adjacent differences, the absolute value of each difference is taken to eliminate the influence of the direction of change, retaining only the magnitude of the change. Then, all the absolute values ​​of the differences are arranged in ascending order from smallest to largest. Each absolute value is assigned a natural number rank according to the order of arrangement, with the smallest absolute value corresponding to rank 1, the second smallest to rank 2, and so on. If there are differences with equal absolute values, they are assigned the same average rank to ensure the rationality and objectivity of the rank assignment. After the assignment is completed, the original sign of each difference is restored to the corresponding rank. That is, if the original difference is positive, the corresponding rank retains a positive sign; if the original difference is negative, the corresponding rank retains a negative sign; if the original difference is zero, the corresponding rank is recorded as zero, finally forming a signed rank sequence.

[0133] Step 52: Relying on the signed rank sequence, sum all positive ranks to obtain the positive rank sum, and sum all negative ranks to obtain the negative rank sum. Calculate the difference between the positive and negative rank sums. If this difference is greater than a preset monotonicity threshold and the positive rank sum is greater than the negative rank sum, it is determined to be an increasing trend; otherwise, it is determined to be a non-increasing trend. Output the sign of the trend slope. The specific operation is as follows:

[0134] In a signed rank sequence, positive ranks correspond to changes in the degree of progress that increase, while negative ranks correspond to changes in the degree of progress that decrease. The magnitude of the rank sum reflects the cumulative effect of different directions of change. By comparing the positive and negative rank sums, the overall trend of the sequence can be clearly identified. The specific logic for classifying and summing the signed rank sequence is as follows: traverse the entire signed rank sequence and sum the values ​​of all positive ranks. The sum obtained is the positive rank sum. The magnitude of the positive rank sum reflects the cumulative effect of all changes in the degree of progress that increase; the larger the positive rank sum, the more significant the increase. Simultaneously, sum the values ​​of all negative ranks. The sum obtained is the negative rank sum. The magnitude of the negative rank sum reflects the cumulative effect of all changes in the degree of progress that decrease; the larger the absolute value of the negative rank sum, the more significant the decrease.

[0135] The difference between the positive and negative rank sums is calculated as follows: the positive rank sum is subtracted from the negative rank sum. The difference reflects the relative strength of the two trends. A positive difference indicates that the upward trend is dominant, while a negative difference indicates that the downward trend is dominant. This difference is compared with a preset monotonic trend threshold, and the relationship between the positive and negative rank sums is determined. If the calculated difference is greater than the preset monotonic trend threshold, and the positive rank sum is greater than the negative rank sum, it indicates that the subclinical progression index sequence shows a significant monotonically increasing trend, meaning that the subclinical progression of microvascular complications in patients is continuously increasing. In this case, a positive trend slope sign is output. If neither of the above two conditions is met, i.e., the difference is less than or equal to the monotonic trend threshold, or the positive rank sum is less than or equal to the negative rank sum, it indicates that the subclinical progression index sequence does not show a significant monotonically increasing trend and may be a stable or downward trend. In this case, a non-positive trend slope sign is output.

[0136] In a preferred embodiment of the present invention, step 6 is further included: combining the current subclinical progression index with the trend slope symbol to generate a risk color block code; defining three regions according to the threshold and the symbol; and outputting the color block code. The specific operation is as follows:

[0137] The current subclinical progression index reflects the patient's current level of microvascular depletion, while the trend slope sign reflects the development trend of complications. Combining these two indicators allows for a more comprehensive assessment of the patient's risk status, avoiding misjudgments based on a single indicator. The risk assessment threshold is dynamically adjusted based on the trend slope sign, ensuring it aligns with the development trend of complications and improving the accuracy of risk area segmentation. The current subclinical progression index is compared with the adjusted dynamic threshold to delineate corresponding risk areas. Finally, according to a pre-defined color code correspondence rule, the risk area is converted into a unique color block code. This code visually distinguishes different risk levels, facilitating rapid identification of the patient's risk status by healthcare professionals and providing a convenient reference for clinical intervention decisions.

[0138] Step 6 further includes the following steps:

[0139] Step 61: Based on the current subclinical progression index and the trend slope sign, convert the trend slope sign into a sign weight factor, where a positive sign corresponds to a weight of 1 and a negative sign corresponds to a weight of 0. Calculate the risk boundary offset, which is equal to a preset first index threshold multiplied by the sign weight factor and then multiplied by a preset nonlinear coupling coefficient. Use this offset to correct the preset first and second index thresholds respectively, obtaining the dynamic lower threshold and dynamic upper threshold. Compare the current subclinical progression index with the dynamic lower threshold and dynamic upper threshold sequentially. If it is less than the dynamic lower threshold, it is determined to be in the first region; if it is between the two, it is determined to be in the second region; if it is greater than or equal to the dynamic upper threshold, it is determined to be in the third region. Output the region determination result. The specific operation is as follows:

[0140] The trend slope sign reflects the development trend of complications. A positive sign indicates that complications are progressing continuously, requiring adjustment of the risk threshold to improve the sensitivity of risk identification. A negative sign indicates that progress is slowing down or stabilizing, and a basic threshold can be used for regional division. First, the trend slope sign output in step 52 is weighted and converted into a sign weight factor that can be used for threshold correction. The specific conversion rule is as follows: if the trend slope sign is positive, it indicates that complications are increasing, and it is converted to a sign weight factor of 1; if the trend slope sign is negative, it indicates that complications are not increasing, and it is converted to a sign weight factor of 0. This conversion quantifies the trend direction into a weight parameter that can participate in the calculation, realizing the dynamic influence of the trend on the threshold. The risk boundary offset is calculated using the following formula: ;

[0141] The derivation logic of this formula is as follows: To ensure that the risk threshold adapts to the trend of complications, when the trend is increasing, the threshold boundary needs to be adjusted through an offset. The magnitude of the offset should be positively correlated with the base threshold. Simultaneously, the offset amplitude is adjusted through a non-linear coupling coefficient to avoid excessive or insufficient adjustment. Therefore, an offset formula is constructed by combining the first exponential threshold, the sign weight factor, and the non-linear coupling coefficient. In the formula... Represents the risk boundary offset; The first exponential threshold is represented by w; the sign weight factor is represented by w; and the nonlinear coupling coefficient is represented by k. Substituting each parameter into the formula, the first exponential threshold is multiplied by the sign weight factor, and then multiplied by the nonlinear coupling coefficient to obtain the risk boundary offset.

[0142] Next, the offset is used to correct the preset first exponential threshold and the second exponential threshold to obtain the dynamic lower threshold and the dynamic upper threshold. The specific correction logic is as follows: the preset first exponential threshold is added to the risk boundary offset to obtain the dynamic lower threshold; the preset second exponential threshold is added to the risk boundary offset to obtain the dynamic upper threshold, where the second exponential threshold is greater than the first exponential threshold. The corrected dynamic threshold can be dynamically adjusted according to the complication trend, making the risk judgment under the increasing trend more stringent.

[0143] After correction, the current subclinical progression index output in step 45 is compared with the dynamic lower threshold and the dynamic upper threshold in sequence. The current subclinical progression index is subtracted from the dynamic lower threshold. If the difference is less than 0, it means that the current subclinical progression index is less than the dynamic lower threshold, and it is determined to be in the first region. If the current subclinical progression index minus the dynamic lower threshold is greater than or equal to 0, and the dynamic upper threshold minus the current subclinical progression index is greater than 0, it means that the current subclinical progression index is between the dynamic lower threshold and the dynamic upper threshold, and it is determined to be in the second region. If the dynamic upper threshold minus the current subclinical progression index is less than or equal to 0, it means that the current subclinical progression index is greater than or equal to the dynamic upper threshold, and it is determined to be in the third region.

[0144] Step 62: Based on the region determination result, extract the color block code uniquely associated with the region from the preset color code correspondence rules, where the color code correspondence rules are: the first region corresponds to the first color code, the second region corresponds to the second color code, and the third region corresponds to the third color code. Output the color block code. The specific operation is as follows:

[0145] The preset color code correspondence rules are set in advance according to clinical risk assessment needs. Each risk area corresponds to a unique color block code, and the color differences in the codes can clearly distinguish different risk levels, ensuring rapid transmission of risk information. First, the preset color code correspondence rules are invoked. These rules clearly define the one-to-one correspondence between the three risk areas and the color block codes: the first area corresponds to the first color code, the second area to the second color code, and the third area to the third color code. The three color codes have clear color differentiation and can be set to different gradients according to clinical practice to visually reflect the differences in risk levels from low to high or from high to low. According to the steps... The region determination result output in step 61 is searched in the preset color code correspondence rules to extract the color block code that is uniquely associated with the region determination result. If the region determination result is the first region, the first color code is extracted; if it is the second region, the second color code is extracted; if it is the third region, the third color code is extracted. After extraction, the color block code is output. This code serves as the final risk visualization result, which can intuitively and clearly reflect the patient's current risk level and development trend of microvascular complications. It provides a concise and clear reference for clinical medical staff to quickly formulate individualized intervention plans, and also facilitates subsequent case management and risk tracking.

[0146] The above description is merely a preferred embodiment of the present invention; however, the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and its improved concepts, should be covered within the scope of protection of the present invention.

Claims

1. An information-based quantitative assessment method for the progression trend of diabetic microvascular complications, characterized in that, include: Step 1: Identify points in the blood glucose sequence whose slope exceeds a predetermined slope threshold as excitation anchor points. Search for microvascular response points within a preset time window. If the offset of the response value of a point relative to the baseline before excitation exceeds a predetermined deviation threshold, use it as a response anchor point. Output the time difference between the excitation anchor point and the response anchor point as the original time offset. Step 2: Based on the double exponential decay recovery model, the first decay time constant and the second decay time constant are inversely derived from the original time offset, response measurement value and baseline value before excitation, where the first decay time constant is less than the second decay time constant, and the ratio of the two is output as the response delay exponent. Step 3: Dynamically update the adaptive reference delay interval based on the patient's own historical data, perform nonlinear mapping on the response delay exponent, with the boundary mapping being a constant and the intermediate nonlinear compression transformation being used to output the exhaustion coefficient. Step 4: Calculate the weight coefficient for each stimulus anchor point. This coefficient is determined based on the ratio of the magnitude of the blood glucose peak exceeding the fasting baseline to the pathological threshold. Multiply the weight of each event by a decay factor that decreases exponentially with the time of the event. Accumulate the weighted exhaustion sum and the expected baseline load sum, and output the subclinical progression index of the ratio of the two. Step 5: Perform a signed-rank test on the longitudinally collected subclinical progression index sequence to determine its monotonic trend. Based on the sum of the ranks of the positive and negative differences, determine whether it is an increasing trend or a non-increasing trend, and output the sign of the trend slope. Step 6: Combine the current subclinical progression index with the trend slope symbol to generate a risk color block code. Define three regions according to the threshold and symbol, and output the color block code.

2. The information-based quantitative assessment method for the progression trend of diabetic microvascular complications according to claim 1, characterized in that, Step 1 includes: Step 11: Calculate the equivalent fluctuation energy from the non-continuous blood glucose sequence according to the ratio of the time interval between two adjacent points to the blood glucose difference. Perform pattern recognition on three consecutive fluctuation segments. If the absolute value of the slope of the middle segment is greater than the threshold multiple of the sum of the absolute values ​​of the slopes of the two segments before and after, mark the start and end points of the middle segment as the rising start point and peak point of the candidate excitation event, and output the list of candidate excitation events. Step 12: Based on the peak time of the candidate excitation event, obtain the microvascular response value within the subsequent time window, arrange the deviation direction of each response value relative to the baseline into a symbol sequence, calculate the ratio of the maximum length of consecutive identical signs to the total length, if the ratio exceeds the threshold and the last sign is consistent with the expected abnormal direction, then confirm that the point is a valid response anchor point, and output the valid response anchor point and its time point. Step 13: Use the time difference between the response anchor point and the peak point as the original time offset and the maximum delay, and the time difference between the first sign-reversed measurement point after the response anchor point and the response anchor point as the minimum delay. Take the geometric midpoint between the minimum and maximum delays as the original time offset output after noise suppression.

3. The information-based quantitative assessment method for the progression trend of diabetic microvascular complications according to claim 2, characterized in that, Step 2 includes: Step 21: Using the maximum and minimum delays in the original time offset, locate the response anchor point measurement value corresponding to the maximum delay and the first sign reversal point measurement value corresponding to the minimum delay. Subtract the baseline value before excitation from each of the two measurements and take the ratio to output the normalized response pair. Step 22: Using the normalized response pairs and their corresponding time points, construct two constraint equations for the double exponential decay recovery model. Each equation constrains the double exponential term to be equal to the normalized response value at that time point, and the first decay time constant and the second decay time constant in the two equations share the same value. Output the simultaneous constraint relationship between the first decay time constant and the second decay time constant.

4. The information-based quantitative assessment method for the progression trend of diabetic microvascular complications according to claim 3, characterized in that, Step 2 also includes: Step 23: Using the simultaneous constraint relationship, the interval shrinkage approximation method is adopted: the physiological upper and lower limits of the first decay time constant are preset and divided into sub-intervals. In each sub-interval, the simultaneous constraint relationship is transformed into a single variable equation about the second decay time constant. The absolute value of the equation residual in each sub-interval is compared to narrow the candidate interval until the residual is lower than the preset threshold. The first decay time constant and the second decay time constant at this time are output. Step 24: Using the first decay time constant and the second decay time constant, calculate the ratio of the second decay time constant to the first decay time constant, then perform logarithmic compression on the ratio with the natural constant as the base, and output the compressed ratio as the response delay index.

5. The information-based quantitative assessment method for the progression trend of diabetic microvascular complications according to claim 4, characterized in that, Step 3 includes: Step 31: Based on the response delay exponential sequence, extract the values ​​in the sequence that are within the predetermined physiological cutoff range, sort them by value size, remove the extreme values ​​at both ends by a fixed proportion, calculate the sliding median and sliding absolute deviation of the remaining values, and output the patient's own reference center value and reference dispersion. Step 32: Relying on the reference center value and the reference dispersion, the lower and upper limits of the adaptive reference delay interval are determined by the multiple of the reference center value plus or minus the reference dispersion. The value of this multiple is adjusted according to the offset direction of the event response delay index and the reference center value: when the offset direction is positive, the first preset step size is increased, and when it is negative, the second preset step size is decreased. The adjusted upper and lower limits of the adaptive reference delay interval are then output.

6. The information-based quantitative assessment method for the progression trend of diabetic microvascular complications according to claim 5, characterized in that, Step 3 also includes: Step 33: Relying on the upper and lower limits of the adaptive reference delay interval, compare the response delay exponent of this event with the interval: if it is less than or equal to the lower limit, it is mapped to the first constant; if it is greater than or equal to the upper limit, it is mapped to the second constant; if it is within the interval, a composite operation of piecewise linear stretching followed by logarithmic shrinking based on the interval width is used to output the intermediate mapping value. Step 34: Depending on the intermediate mapping value and whether it comes from within the interval, if it comes from nonlinear compression within the interval, calculate the difference between the response delay exponent and the reference center value, divide it by the reference dispersion as a local fine-tuning coefficient, multiply the intermediate mapping value by this coefficient, and output the final exhaustion coefficient; if it comes from boundary constant mapping, directly output the constant as the exhaustion coefficient.

7. The information-based quantitative assessment method for the progression trend of diabetic microvascular complications according to claim 6, characterized in that, Step 4 includes: Step 41: Based on the blood glucose peak and fasting baseline value of each excitation anchor point, calculate the absolute range of the peak exceeding the baseline, divide the range by the pathological threshold to obtain the relative ratio, then perform logarithmic stretching of the relative ratio with the natural constant as the base and add one, and output the stretched original weight coefficient. Step 42: Based on the difference between the time point of each excitation anchor point and the current evaluation time point, divide the difference by the preset half-life constant to obtain the dimensionless duration ratio, and then calculate the decay factor using an exponential function with the natural constant as the base, and output the time decay factor of each event. Step 43: The weighted exhaustion contribution value is obtained by multiplying the original weight coefficient after stretching and the time decay factor. The weighted exhaustion amount is obtained by multiplying the exhaustion coefficient of each event and the weighted exhaustion contribution value. The expected baseline load is obtained by multiplying the weighted exhaustion contribution value and the average exhaustion coefficient of the healthy population. The weighted exhaustion amount sequence and the expected baseline load sequence of each event are output.

8. The information-based quantitative assessment method for the progression trend of diabetic microvascular complications according to claim 7, characterized in that, Step 4 also includes: Step 44: Based on the weighted exhaustion sequence and the expected baseline load sequence, sum them up in the order of event occurrence to obtain the cumulative weighted exhaustion sum and the cumulative expected baseline load sum. If the total number of events is lower than the minimum sample threshold, multiply the cumulative sum by the compensation factor. The compensation factor is equal to the minimum sample threshold divided by the actual total number of events. Output the compensated cumulative sum. Step 45: Based on the ratio of the cumulative weighted exhaustion after compensation to the cumulative expected baseline load after compensation, the ratio is then logarithmically compressed to the base of the natural constant and then incremented by one to output the final subclinical progress index.

9. The information-based quantitative assessment method for the progression trend of diabetic microvascular complications according to claim 8, characterized in that, Step 5 includes: Step 51: Based on the subclinical progression index sequence acquired longitudinally, calculate the difference between two adjacent indices in chronological order, assign the absolute values ​​of each difference to natural number ranks in ascending order and retain the original sign of each difference, and output the signed rank sequence. Step 52: Based on the signed rank sequence, sum all positive ranks to obtain the positive rank sum, sum all negative ranks to obtain the negative rank sum, calculate the difference between the positive rank sum and the negative rank sum. If the difference is greater than the preset monotonic trend threshold and the positive rank sum is greater than the negative rank sum, it is determined to be an increasing trend; otherwise, it is determined to be a non-increasing trend, and the trend slope sign is output.

10. The information-based quantitative assessment method for the progression trend of diabetic microvascular complications according to claim 9, characterized in that, Step 6 includes: Step 61: Based on the current subclinical progression index and the trend slope sign, convert the trend slope sign into a sign weight factor, where a positive sign corresponds to a weight of 1 and a negative sign corresponds to a weight of 0. Calculate the risk boundary offset, which is equal to the preset first index threshold multiplied by the sign weight factor and then multiplied by the preset nonlinear coupling coefficient. Use this offset to correct the preset first index threshold and second index threshold respectively to obtain the dynamic lower threshold and dynamic upper threshold. Compare the current subclinical progression index with the dynamic lower threshold and dynamic upper threshold in sequence. If it is less than the dynamic lower threshold, it is determined to be in the first region; if it is between the two, it is determined to be in the second region; if it is greater than or equal to the dynamic upper threshold, it is determined to be in the third region. Output the region determination result. Step 62: Based on the region determination result, extract the color block code that is uniquely associated with the region from the preset color code correspondence rules, where the color code correspondence rules are: the first region corresponds to the first color code, the second region corresponds to the second color code, and the third region corresponds to the third color code. Output the color block code.

Images