Calibration method for non-invasive blood glucose monitor based on joint optimization

Abstract

The present invention relates to the field of data processing technology, and in particular to a calibration method for a non-invasive blood glucose monitor based on joint optimization, comprising: segmenting the blood glucose monitoring data sequence according to cycles, obtaining all bands of the blood glucose monitoring data sequence set in each cycle; obtaining the fluctuation difference of the blood glucose monitoring data sequence at each acquisition moment in each cycle; obtaining the difference degree of the blood glucose monitoring data sequence at each acquisition moment in each cycle; obtaining the abnormality degree at each acquisition moment in each cycle; obtaining the correction amplitude of the blood glucose concentration at each acquisition moment in each cycle; obtaining an optimized correction factor of the non-invasive blood glucose monitor based on the correction amplitude of the blood glucose concentration at all acquisition moments in all acquisition time periods in all cycles; and optimizing and correcting the non-invasive blood glucose monitor based on the optimized correction factor. The present invention makes the measurement data of the non-invasive blood glucose monitor more accurate.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a calibration method for a non-invasive blood glucose monitor based on joint optimization. Background Art

[0002] Non-invasive blood glucose monitors use optical, thermal, electrical or other physical principles to assess blood glucose levels in real time without puncturing the skin to collect blood samples. The measurement accuracy of non-invasive blood glucose monitors is affected by individual physiological differences, changes in the external environment and device stability. Since changes in blood glucose concentration may be very subtle, the measurement data often contains noise and interference, and effective data processing technology is required to extract accurate data. The working principle of non-invasive blood glucose monitors is to use multi-source data to measure blood glucose levels, and then optimize and correct the non-invasive blood glucose monitor by performing anomaly analysis on the multi-source data. The conventional data correction method is to simply denoise the measurement data of the non-invasive blood glucose monitor without considering correction based on the working principle of the non-invasive blood glucose monitor, which results in the optimized and corrected measurement values ​​being inaccurate. Summary of the Invention

[0003] In order to solve the above problems, the present invention provides a calibration method for a non-invasive blood glucose monitor based on joint optimization, the method comprising:

[0004] Obtaining a set of blood glucose monitoring data sequences within a certain collection time period;

[0005] By analyzing the noise distribution in the blood glucose monitoring data sequence, all cycles of the blood glucose monitoring data sequence set within each acquisition time period are obtained; the blood glucose monitoring data sequence is segmented according to the cycle to obtain all bands of the blood glucose monitoring data sequence set in each cycle; based on the difference between the corresponding band of the blood glucose monitoring data sequence at each acquisition moment and other bands, the fluctuation difference amount of the blood glucose monitoring data sequence at each acquisition moment in each cycle is obtained;

[0006] Based on the difference in the fluctuation difference of the blood glucose monitoring data sequence between each collection moment and other collection moments, the difference degree of the blood glucose monitoring data sequence at each collection moment in each cycle is obtained; based on the difference in the difference degree between each collection time period and adjacent collection time periods, the abnormality degree at each collection moment in each cycle is obtained; based on the difference in the abnormality degree between each collection moment and surrounding collection moments, the correction amplitude of the blood glucose concentration at each collection moment in each cycle is obtained;

[0007] According to the correction amplitude of the blood glucose concentration at all sampling moments in all cycles in all sampling time periods, an optimized correction factor of the non-invasive blood glucose monitor is obtained; and the non-invasive blood glucose monitor is optimized and corrected according to the optimized correction factor.

[0008] Preferably, the method of obtaining all cycles of the blood glucose monitoring data sequence set within each acquisition time period by analyzing the noise distribution in the blood glucose monitoring data sequence includes the following specific methods:

[0009] Get the Fluctuation curve of blood glucose monitoring data within a collection time period;

[0010] For the first Perform Fourier transform on the fluctuation curve of blood glucose monitoring data within the acquisition time period, and take the inverse of the frequency with the largest amplitude in the spectrum as the One cycle of the blood glucose monitoring data fluctuation curve within the collection time period is recorded as the target cycle;

[0011] According to the maximum amplitude in the spectrum and the frequency of the target cycle, the curve obtained is recorded as the fluctuation curve of the target cycle; The difference curve between the fluctuation curve of the blood glucose monitoring data in the acquisition time period and the fluctuation curve of the target period is recorded as the curve to be analyzed; the curve to be analyzed is decomposed using the EDM decomposition algorithm to obtain several IMF components and residual terms; the possibility of each IMF component belonging to high-frequency noise is obtained;

[0012] The IMF component with the highest probability of being high-frequency noise is recorded as the noise component, and the other IMF components and the residual term are added to obtain the curve to be analyzed after the noise is removed. The curve to be analyzed after the noise is removed is subjected to Fourier transform, and the inverse of the frequency corresponding to each peak in the spectrum is taken as the first One cycle of the blood glucose monitoring data fluctuation curve within a collection time period.

[0013] Preferably, the obtaining The blood glucose monitoring data fluctuation curve within a collection time period includes the following specific methods:

[0014] Construct a four-dimensional space based on the acquisition time, skin reflected light intensity data, skin electrical impedance data, and skin temperature data; The blood glucose monitoring data sequence set within the collection time period is input into the four-dimensional space to obtain the Several blood glucose monitoring data points within a collection time period; in four-dimensional space, the least square method is used to transform the first The curve fitting is performed on all the blood glucose monitoring data points within the collection time period to obtain the Fluctuation curve of blood glucose monitoring data within a collection time period.

[0015] Preferably, the method of obtaining the possibility that each IMF component belongs to high-frequency noise includes:

[0016] For any IMF component, a frequency spectrum of the IMF component is obtained by Fourier transform; a peak detection method is used to obtain the frequency corresponding to each peak in the frequency spectrum, and the frequency is recorded as the peak frequency;

[0017] The cumulative sum of the absolute values ​​of the differences between the mean of all peak frequencies of the IMF component and the mean of all peak frequencies of all other IMF components is taken as the possibility that the IMF component belongs to high-frequency noise.

[0018] Preferably, the method of obtaining the fluctuation difference of the blood glucose monitoring data sequence at each collection moment in each cycle according to the difference between the corresponding band of the blood glucose monitoring data sequence at each collection moment and other bands includes the following specific methods:

[0019] In the In the blood glucose monitoring data fluctuation curve within the collection time period, The band corresponding to the blood glucose monitoring data point at each collection moment is recorded as the target band; all bands except the target band are recorded as reference bands;

[0020] According to The similarity between the blood glucose monitoring data sequence at the time of collection and the blood glucose monitoring data sequence at other collection times is obtained. All similar data sequences at the time of collection;

[0021] The first The position of the blood glucose monitoring data point in the target band at each collection moment is recorded as the target position;

[0022] The first The first Dimensional data and The first The absolute value of the difference between the two dimensional data is recorded as the target band and the The first reference band The data difference value of the first dimension data; the target band is The DTW distance between the reference bands is recorded as the DTW distance between the target band and the The sequence difference value between the reference bands; the target band and the The sum of the data differences of all dimensional data between the reference bands is recorded as the difference between the target band and the The overall data difference value between the reference bands; the product of the overall data difference value and the sequence difference value is recorded as the target band and the The sum of the corrected difference values ​​between the target band and all reference bands is recorded as the first The fluctuation difference factor of the blood glucose monitoring data sequence at the first collection moment; the fluctuation difference factor is combined with the The ratio between the number of all similar data sequences at the collection time is used as the The fluctuation difference of the blood glucose monitoring data sequence at each collection moment.

[0023] Preferably, the obtaining All similar data sequences at the collection moment, including the specific methods:

[0024] Preset a similarity threshold parameter , if The blood glucose monitoring data sequence at the collection moment is The cosine similarity between the blood glucose monitoring data sequences at the target position in the reference band is greater than or equal to the similarity threshold parameter , will The blood glucose monitoring data sequence at the target position in the reference band is used as the first Similar data sequences at the same collection moment.

[0025] Preferably, the method of obtaining the difference degree of the blood glucose monitoring data sequence at each collection moment in each cycle according to the difference in the fluctuation difference amount of the blood glucose monitoring data sequence between each collection moment and other collection moments includes the following specific methods:

[0026] In the During the collection period, The sum of the absolute values ​​of the differences between the fluctuation difference of the blood glucose monitoring data sequence at the first collection moment and the fluctuation difference of the blood glucose monitoring data sequence at other collection moments is taken as the first The difference of blood glucose monitoring data series at each collection moment.

[0027] Preferably, the method of obtaining the abnormality degree at each collection moment in each cycle according to the difference between each collection time period and the adjacent collection time periods includes the following specific methods:

[0028] Preset a time period neighborhood parameter , will Before the collection period The collection period and Each collection time period is taken as The reference collection time period of each collection time period;

[0029] The first The first The first The difference between the blood glucose monitoring data sequence at the left adjacent collection time and the The first The first The absolute value of the difference between the difference of the blood glucose monitoring data sequence at the right adjacent collection time is recorded as The adjacent abnormal difference of the reference collection time period; The cumulative sum of the adjacent abnormal differences of all reference collection time periods in the collection time period is recorded as The abnormal correction factor of the first acquisition period; the abnormal correction factor is combined with the The product of the differences between the blood glucose monitoring data sequences at the time of collection is used as the The abnormality degree of the blood glucose monitoring data sequence at each collection moment.

[0030] Preferably, the method of obtaining the correction amplitude of the blood glucose concentration at each collection moment in each cycle according to the difference in the degree of abnormality between each collection moment and surrounding collection moments includes the following specific methods:

[0031] Preset a collection time neighborhood parameter , in During the collection period, before the collection time The collection time and The time sequence composed of the collection moments is recorded as Neighborhood reference time sequence of each acquisition moment;

[0032] The first The mean of the patient blood glucose concentration data at all collection moments in the neighborhood reference time sequence of the collection moment is recorded as the first mean; During the collection period The absolute value of the difference between the patient's blood glucose concentration data at the time of collection and the first mean is recorded as the correction factor; The normalized value of the product of the abnormality degree of the blood glucose monitoring data sequence at the time of collection and the correction factor is used as the first The correction range of blood glucose concentration at each collection moment.

[0033] Preferably, the method of obtaining the optimized correction factor of the non-invasive blood glucose monitor according to the correction amplitude of the blood glucose concentration at all acquisition moments in all acquisition time periods in all cycles includes the following specific methods:

[0034] The first During the collection period The patient's blood glucose concentration data at the time of collection is During the collection period The difference between the mean values ​​of the patient's blood glucose concentration data at all collection moments in the neighborhood reference time sequence of the collection moment is recorded as the first difference;

[0035] The first difference and the The product of the correction amplitude of the blood glucose concentration at the time of collection is used as the Correction value at each collection moment;

[0036] If the first difference is greater than or equal to 0, The difference between the patient's blood glucose concentration data at the time of collection and the corrected value is taken as the Corrected patient blood glucose concentration data at each collection moment;

[0037] If the first difference is less than 0, the The sum of the patient's blood glucose concentration data at the time of collection and the correction value is taken as the Corrected patient blood glucose concentration data at each collection moment;

[0038] The ratio between the mean of the corrected patient blood glucose concentration data at all collection moments in all cycles of all collection time periods and the mean of the patient blood glucose concentration data at all collection moments in all cycles of all collection time periods is used as the optimized correction factor of the non-invasive blood glucose monitor.

[0039] The beneficial effects of the technical solution of the present invention are as follows: the present invention obtains the difference degree of the blood glucose monitoring data sequence at each collection moment in each cycle according to the difference in the fluctuation difference amount of the blood glucose monitoring data sequence between each collection moment and other collection moments; The difference between the difference between each collection time period and the adjacent collection time periods is used to obtain the degree of abnormality at each collection moment in each cycle; the correction amplitude of the blood glucose concentration at each collection moment in each cycle is obtained based on the difference in the degree of abnormality between each collection moment and the surrounding collection moments; the optimized correction factor of the non-invasive blood glucose monitor is obtained based on the correction amplitude of the blood glucose concentration at all collection moments in all cycles of all collection time periods; the non-invasive blood glucose monitor is optimized and corrected according to the optimized correction factor, so as to achieve optimized correction of the non-invasive blood glucose monitor through the working principle of the non-invasive blood glucose monitor, so that the measurement data of the non-invasive blood glucose monitor is more accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0041] Figure 1 Flowchart of the steps of the calibration method of the non-invasive blood glucose monitor based on joint optimization of the present invention;

[0042] Figure 2 The figure is a flow chart showing the characteristic relationships of the calibration method of the non-invasive blood glucose monitor based on joint optimization according to the present invention. DETAILED DESCRIPTION

[0043] To further illustrate the technical means and effects employed by the present invention to achieve the intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, describes in detail the specific implementation, structure, features, and effects of the calibration method for a non-invasive blood glucose monitor based on joint optimization proposed by the present invention. In the following description, different references to "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.

[0044] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0045] The specific scheme of the calibration method of the non-invasive blood glucose monitor based on joint optimization provided by the present invention is described in detail below with reference to the accompanying drawings.

[0046] See also Figure 1 , which shows a flowchart of a calibration method for a non-invasive blood glucose monitor based on joint optimization provided by one embodiment of the present invention, the method comprising the following steps:

[0047] Step S001: Acquire a set of blood glucose monitoring data sequences within a certain collection time period.

[0048] It should be noted that non-invasive blood glucose monitors use different non-invasive technologies to measure blood glucose levels. Their working principles are: estimating blood glucose levels by emitting near-infrared light to the skin and analyzing changes in the intensity of the reflected light; estimating blood glucose levels by measuring changes in the electrical impedance of the skin and tissue; estimating blood glucose levels by measuring changes in skin temperature.

[0049] Specifically, it is necessary to first collect a set of blood glucose monitoring data sequences within a certain collection time period. The specific process is as follows:

[0050] An infrared light source is used to emit near-infrared light onto the skin. An optical receiver is installed near the surface of the patient's skin to receive the near-infrared light reflected from the skin. The intensity of the received reflected light is measured by a sensor. Two electrodes are placed on the patient's skin, one for applying a low-frequency alternating current and the other for measuring voltage. The electrical impedance data is calculated according to Ohm's law. The patient's skin temperature data is measured using an infrared thermometer. The patient's blood glucose concentration data is measured using a non-invasive blood glucose monitor.

[0051] The data were collected every 30 minutes for a total of 5 hours. For any collection period, a sampling moment was taken every 1 minute, and the three dimensional data types, namely, the patient's skin reflected light intensity data, skin electrical impedance data, and skin temperature data, were collected in sequence each time for a total of 30 minutes. The three dimensional data, namely, the patient's skin reflected light intensity data, skin electrical impedance data, and skin temperature data at each sampling moment were used as the blood glucose monitoring data sequence set within the collection period.

[0052] At this point, a set of blood glucose monitoring data sequences within several collection time periods is obtained through the above method.

[0053] Step S002: By analyzing the noise distribution in the blood glucose monitoring data sequence, all cycles of the blood glucose monitoring data sequence set within each acquisition time period are obtained; the blood glucose monitoring data sequence is segmented according to the cycle, and all bands of the blood glucose monitoring data sequence set in each cycle are obtained; based on the difference between the corresponding band of the blood glucose monitoring data sequence at each acquisition moment and other bands, the fluctuation difference amount of the blood glucose monitoring data sequence at each acquisition moment in each cycle is obtained.

[0054] 1. Obtain all cycles of the blood glucose monitoring data sequence set within each collection time period.

[0055] It should be noted that since the intervals between collection moments within the collection time period are certain, and the patient's heartbeat, respiratory rhythm and other physiological processes are all periodic changes, under normal circumstances, the data fluctuations of the blood glucose monitoring data within each collection time period show periodic characteristics; since the collected blood glucose monitoring data may contain high-frequency noise, resulting in the inability to accurately obtain other periods, it is necessary to remove the high-frequency noise to obtain more accurate blood glucose monitoring data.

[0056] Preferably, in some implementations of the embodiments of the present invention, the specific method for obtaining all cycles of the blood glucose monitoring data sequence set within each acquisition time period by analyzing the noise distribution in the blood glucose monitoring data sequence is:

[0057] Get the Fluctuation curve of blood glucose monitoring data within a collection time period;

[0058] For the first Perform Fourier transform on the fluctuation curve of blood glucose monitoring data within the acquisition time period, and take the inverse of the frequency with the largest amplitude in the spectrum as the One cycle of the blood glucose monitoring data fluctuation curve within the collection time period is recorded as the target cycle;

[0059] According to the maximum amplitude in the spectrum and the frequency of the target cycle, the obtained curve is recorded as the fluctuation curve of the target cycle; The difference curve between the fluctuation curve of the blood glucose monitoring data in the acquisition time period and the fluctuation curve of the target period is recorded as the curve to be analyzed; the curve to be analyzed is decomposed using the EDM decomposition algorithm to obtain several IMF components and residual terms; the possibility of each IMF component belonging to high-frequency noise is obtained;

[0060] The IMF component with the highest probability of being high-frequency noise is recorded as the noise component, and the other IMF components and the residual term are added to obtain the curve to be analyzed after the noise is removed. The curve to be analyzed after the noise is removed is subjected to Fourier transform, and the inverse of the frequency corresponding to each peak in the spectrum is taken as the first One cycle of the blood glucose monitoring data fluctuation curve within a collection time period.

[0061] It should be noted that high-frequency noise only appears in part of the original signal, unlike periodic signals that exist in the entire original signal, and the EDM decomposition algorithm can decompose the high-frequency noise. Therefore, the high-frequency noise signal can be determined by analyzing all frequency distributions of different IMF components in the spectrum diagram; the EDM decomposition algorithm is an existing technology and will not be described in detail in this embodiment.

[0062] Preferably, in some implementations of the embodiments of the present invention, obtaining the The specific method of the blood glucose monitoring data fluctuation curve within a collection time period is:

[0063] Construct a four-dimensional space based on the acquisition time, skin reflected light intensity data, skin electrical impedance data, and skin temperature data; The blood glucose monitoring data sequence set within the collection time period is input into the four-dimensional space to obtain the Several blood glucose monitoring data points within a collection time period; in four-dimensional space, the least square method is used to transform the first The curve fitting is performed on all the blood glucose monitoring data points within the collection time period to obtain the Fluctuation curve of blood glucose monitoring data within a collection time period.

[0064] Among them, the least square method is an existing technology and will not be described in detail in this embodiment.

[0065] Preferably, in some implementations of the embodiments of the present invention, the specific method for obtaining the possibility that each IMF component belongs to high-frequency noise is:

[0066] For any IMF component, a frequency spectrum of the IMF component is obtained by Fourier transform; a peak detection method is used to obtain the frequency corresponding to each peak in the frequency spectrum, and the frequency is recorded as the peak frequency;

[0067] The cumulative sum of the absolute values ​​of the differences between the mean of all peak frequencies of the IMF component and the mean of all peak frequencies of all other IMF components is used as the possibility that the IMF component belongs to high-frequency noise;

[0068] The specific formula is:

[0069]

[0070] Where, Indicates the possibility that the IMF component belongs to high-frequency noise; represents the total number of all IMF components; Indicates the The mean of all peak frequencies of the IMF components; represents the mean of all peak frequencies of the IMF components; Indicates taking the absolute value.

[0071] It should be noted that the greater the absolute value of the difference between the mean of all peak frequencies of the IMF component and the mean of all peak frequencies of all other IMF components, the greater the waveform difference between the IMF component and all other IMF components, and the greater the possibility that the IMF component belongs to high-frequency noise; Fourier transform is a prior art and will not be described in detail in this embodiment.

[0072] At this point, all cycles of the blood glucose monitoring data sequence set within each acquisition time period are obtained.

[0073] 2. Obtain the fluctuation difference of the blood glucose monitoring data sequence at each collection moment in each cycle.

[0074] It should be noted that the greater the difference in the fluctuation difference of the blood glucose monitoring data sequence between each collection moment and other collection moments, the greater the fluctuation difference of the blood glucose monitoring data sequence at each collection moment, and the greater the degree of abnormality of the blood glucose monitoring data sequence at each collection moment.

[0075] Preferably, for Any cycle of the blood glucose monitoring data sequence set within the collection time period, The blood glucose monitoring data fluctuation curve within the collection time period is divided according to the cycle to obtain the first All bands of the blood glucose monitoring data fluctuation curve within a collection time period.

[0076] Preferably, in some implementations of the embodiments of the present invention, based on the difference between the corresponding band of the blood glucose monitoring data sequence at each collection moment and other bands, a specific method for obtaining the fluctuation difference amount of the blood glucose monitoring data sequence at each collection moment in each cycle is:

[0077] In the In the blood glucose monitoring data fluctuation curve within the collection time period, The band corresponding to the blood glucose monitoring data point at each collection moment is recorded as the target band; all bands except the target band are recorded as reference bands;

[0078] According to The similarity between the blood glucose monitoring data sequence at the time of collection and the blood glucose monitoring data sequence at other collection times is obtained. All similar data sequences at the time of collection;

[0079] The first The position of the blood glucose monitoring data point in the target band at each collection moment is recorded as the target position;

[0080] The first The first Dimensional data and The first The absolute value of the difference between the two dimensional data is recorded as the target band and the The first reference band The data difference value of the first dimension data; the target band is The DTW distance between the reference bands is recorded as the DTW distance between the target band and the The sequence difference value between the reference bands; the target band and the The sum of the data differences of all dimensional data between the reference bands is recorded as the difference between the target band and the The overall data difference value between the reference bands; the product of the overall data difference value and the sequence difference value is recorded as the target band and the The sum of the corrected difference values ​​between the target band and all reference bands is recorded as the first The fluctuation difference factor of the blood glucose monitoring data sequence at the first collection moment; the fluctuation difference factor is combined with the The ratio between the number of all similar data sequences at the collection time is used as the The fluctuation difference of the blood glucose monitoring data sequence at each collection moment;

[0081] The specific formula is:

[0082]

[0083] Where, Indicates the The collection time period is the first The fluctuation difference of the blood glucose monitoring data sequence at each collection moment; Indicates the first The number of all reference bands of the blood glucose monitoring data fluctuation curve within a collection time period; Represents the data type quantity of all dimensional data in the blood glucose monitoring data sequence at each collection moment; Indicates the During the collection period The first Dimensional data; Indicates the During the collection period The first Dimensional data; Indicates the During the collection period The corresponding band of the blood glucose monitoring data point at the first collection moment is DTW distance between reference bands; Indicates the During the collection period The number of all similar data sequences at the collection moment; Indicates taking the absolute value.

[0084] It should be noted that Indicates the target band and the The first reference band The data difference value of the dimension data; Indicates the target band and the The overall data difference value between the reference bands; Indicates the target band and the Corrected difference value between reference bands; Indicates the The fluctuation difference factor of the blood glucose monitoring data sequence at each collection moment; The larger the During the collection period The blood glucose monitoring data sequence at the collection moment is The greater the difference between the blood glucose monitoring data sequences at the same position in other reference bands within the acquisition time period, the greater the difference between the blood glucose monitoring data sequences at the same position in the reference bands within the acquisition time period. The greater the possibility of abnormality of the blood glucose monitoring data sequence at the time of collection; the smaller the fluctuation difference factor, the greater the possibility of abnormality of the blood glucose monitoring data sequence at the time of collection; The more the blood glucose monitoring data sequences within a collection period conform to the periodic distribution, the The less likely the blood glucose monitoring data within a collection period is to be abnormal; The larger the The smaller the fluctuation difference of the blood glucose monitoring data point at each collection moment, the smaller the fluctuation difference of the data point. The smaller the anomaly of the blood glucose monitoring data sequence at each collection moment is likely to be.

[0085] Preferably, in some implementations of the embodiments of the present invention, obtaining the The specific method for all similar data sequences at a collection moment is:

[0086] Preset a similarity threshold parameter , wherein this embodiment is based on This example is described as an example, and this embodiment is not specifically limited. Depends on the specific implementation situation;

[0087] Jordi The blood glucose monitoring data sequence at the collection moment is The cosine similarity between the blood glucose monitoring data sequences at the target position in the reference band is greater than or equal to the similarity threshold parameter , will The blood glucose monitoring data sequence at the target position in the reference band is used as the first Similar data sequences at the same collection moment.

[0088] At this point, the fluctuation difference of the blood glucose monitoring data sequence at each collection moment in each cycle is obtained through the above method.

[0089] Step S003: Based on the difference in the fluctuation difference of the blood glucose monitoring data sequence between each collection moment and other collection moments, the difference of the blood glucose monitoring data sequence at each collection moment in each cycle is obtained; based on the difference in the difference between each collection time period and the adjacent collection time periods, the abnormality degree at each collection moment in each cycle is obtained; based on the difference in the abnormality degree between each collection moment and the surrounding collection moments, the correction amplitude of the blood glucose concentration at each collection moment in each cycle is obtained.

[0090] 1. The difference between the blood glucose monitoring data series at each collection time in each cycle.

[0091] It should be noted that under normal circumstances, the blood glucose monitoring data in each collection time period fluctuates uniformly and periodically. When an abnormality occurs, because the collection time interval is short, the abnormal collection time will not exist alone, and abnormalities will occur in the linked collection time. Therefore, the greater the abnormality, the greater the difference between the adjacent sampling times.

[0092] Preferably, in some implementations of the embodiments of the present invention, based on the difference in the fluctuation difference of the blood glucose monitoring data sequence between each collection moment and other collection moments, a specific method for obtaining the difference degree of the blood glucose monitoring data sequence at each collection moment in each cycle is:

[0093] In the During the collection period, The sum of the absolute values ​​of the differences between the fluctuation difference of the blood glucose monitoring data sequence at the first collection moment and the fluctuation difference of the blood glucose monitoring data sequence at other collection moments is taken as the first The difference of blood glucose monitoring data sequence at each collection moment;

[0094] The specific formula is:

[0095]

[0096] Where, Indicates the The collection time period is the first The difference of blood glucose monitoring data sequence at each collection moment; Indicates the The number of all collection moments in a collection time period; Indicates the The collection time period is the first The fluctuation difference of the blood glucose monitoring data sequence at each collection moment; Indicates the The collection time period is the first The fluctuation difference of the blood glucose monitoring data sequence at each collection moment; Indicates taking the absolute value.

[0097] At this point, the difference of the blood glucose monitoring data sequence at each collection moment in each cycle is obtained.

[0098] 2. Obtain the degree of abnormality at each collection moment in each cycle.

[0099] It should be noted that since abnormal collection moments do not exist alone, abnormalities will occur at collection moments at adjacent positions in adjacent collection time periods. Therefore, the difference degree is corrected according to the difference in the difference degree at adjacent collection moments in adjacent collection time periods to obtain the degree of abnormality at the collection moment.

[0100] Preferably, in some implementations of the embodiments of the present invention, the specific method for obtaining the abnormality degree at each collection moment in each cycle according to the difference between each collection time period and the adjacent collection time periods is:

[0101] Preset a time period neighborhood parameter , wherein this embodiment is based on This example is described as an example, and this embodiment is not specifically limited. Depends on the specific implementation situation;

[0102] The first Before the collection period The collection period and Each collection time period is taken as The reference collection time period of each collection time period;

[0103] The first The first The first The difference between the blood glucose monitoring data sequence at the left adjacent collection time and the The first The first The absolute value of the difference between the difference of the blood glucose monitoring data sequence at the right adjacent collection time is recorded as The adjacent abnormal difference of the reference collection time period; The cumulative sum of the adjacent abnormal differences of all reference collection time periods in the collection time period is recorded as The abnormal correction factor of the first acquisition period; the abnormal correction factor is combined with the The product of the differences between the blood glucose monitoring data sequences at the time of collection is used as the The abnormality of the blood glucose monitoring data sequence at each collection moment;

[0104] The specific formula is:

[0105]

[0106] Where, Indicates the The collection time period is the first The abnormality of the blood glucose monitoring data sequence at each collection moment; Indicates the The collection time period is the first The difference of blood glucose monitoring data sequence at each collection moment; Indicates the The number of all reference collection time periods for a collection time period; Indicates the The first The reference collection time period is the first The difference between the blood glucose monitoring data sequences at the left adjacent collection time of a collection time; Indicates the The first The reference collection time period is the first The difference between the blood glucose monitoring data sequences at the right adjacent collection time of a collection time; Indicates taking the absolute value.

[0107] It should be noted that Indicates the anomaly correction factor.

[0108] At this point, the degree of abnormality at each acquisition moment in each cycle is obtained.

[0109] 3. The correction range of blood glucose concentration at each collection moment in each cycle.

[0110] Preferably, in some implementations of the embodiments of the present invention, the specific method for obtaining the correction amplitude of the blood glucose concentration at each collection moment in each cycle according to the difference in abnormality between each collection moment and surrounding collection moments is:

[0111] Preset a collection time neighborhood parameter , wherein this embodiment is based on This example is described as an example, and this embodiment is not specifically limited. Depends on the specific implementation situation;

[0112] In the During the collection period, before the collection time The collection time and The time sequence composed of the collection moments is recorded as Neighborhood reference time sequence of each acquisition moment;

[0113] The first The mean of the patient blood glucose concentration data at all collection moments in the neighborhood reference time sequence of the collection moment is recorded as the first mean; During the collection period The absolute value of the difference between the patient's blood glucose concentration data at the time of collection and the first mean is recorded as the correction factor; The normalized value of the product of the abnormality degree of the blood glucose monitoring data sequence at the time of collection and the correction factor is used as the first The correction range of blood glucose concentration at each collection moment;

[0114]

[0115] Where, Indicates the The collection time period is the first The correction range of blood glucose concentration at each collection moment; Indicates the The collection time period is the first The abnormality of the blood glucose monitoring data sequence at each collection moment; Indicates the During the collection period The patient's blood glucose concentration data at each collection moment; Indicates the During the collection period The number of all acquisition moments in the neighborhood reference moment sequence of the acquisition moment; Indicates the During the collection period The first in the neighborhood reference time sequence of the collection time The patient's blood glucose concentration data at each collection moment; Indicates taking the absolute value; represents the linear normalization function.

[0116] It should be noted that Represents the correction factor.

[0117] So far, the correction range of the blood glucose concentration at each collection moment in each cycle is obtained through the above method.

[0118] Step S004: Obtain an optimized correction factor of the non-invasive blood glucose monitor according to the correction amplitude of the blood glucose concentration at all acquisition moments in all acquisition time periods in all cycles; and optimize and correct the non-invasive blood glucose monitor according to the optimized correction factor.

[0119] Preferably, in some implementations of the embodiments of the present invention, the specific method for obtaining the optimized correction factor of the non-invasive blood glucose monitor according to the correction amplitude of the blood glucose concentration at all acquisition moments in all acquisition time periods in all cycles is:

[0120] The first During the collection period The patient's blood glucose concentration data at the time of collection is During the collection period The difference between the mean values ​​of the patient's blood glucose concentration data at all collection moments in the neighborhood reference time sequence of the collection moment is recorded as the first difference;

[0121] The first difference and the The product of the correction amplitude of the blood glucose concentration at the time of collection is used as the Correction value at each collection moment;

[0122] If the first difference is greater than or equal to 0, The difference between the patient's blood glucose concentration data at the time of collection and the corrected value is taken as the Corrected patient blood glucose concentration data at each collection moment;

[0123] If the first difference is less than 0, the The sum of the patient's blood glucose concentration data at the time of collection and the correction value is taken as the Corrected patient blood glucose concentration data at each collection moment;

[0124] The ratio between the mean of the corrected patient blood glucose concentration data at all collection moments in all cycles of all collection time periods and the mean of the patient blood glucose concentration data at all collection moments in all cycles of all collection time periods is used as the optimized correction factor of the non-invasive blood glucose monitor.

[0125] Preferably, the specific method for optimizing and correcting the non-invasive blood glucose monitor according to the optimized correction factor is:

[0126] For each measurement value of the non-invasive blood glucose monitor, the sum of the optimized correction factor of the non-invasive blood glucose monitor and 1 is recorded as the correction value; the product of the measurement value and the correction value is taken as the optimized corrected measurement value of the non-invasive blood glucose monitor.

[0127] See also Figure 2 , which shows a characteristic relationship flow chart of a calibration method for a non-invasive blood glucose monitor based on joint optimization;

[0128] At this point, this embodiment is completed.

[0129] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A calibration method for a non-invasive blood glucose monitor based on joint optimization, characterized in that: The method comprises the following steps: Obtaining a set of blood glucose monitoring data sequences within a certain collection time period; By analyzing the noise distribution in the blood glucose monitoring data sequence, all cycles of the blood glucose monitoring data sequence set within each acquisition time period are obtained; the blood glucose monitoring data sequence is segmented according to the cycle to obtain all bands of the blood glucose monitoring data sequence set in each cycle; based on the difference between the corresponding band of the blood glucose monitoring data sequence at each acquisition moment and other bands, the fluctuation difference amount of the blood glucose monitoring data sequence at each acquisition moment in each cycle is obtained; Based on the difference in the fluctuation difference of the blood glucose monitoring data sequence between each collection moment and other collection moments, the difference degree of the blood glucose monitoring data sequence at each collection moment in each cycle is obtained; based on the difference in the difference degree between each collection time period and adjacent collection time periods, the abnormality degree at each collection moment in each cycle is obtained; based on the difference in the abnormality degree between each collection moment and surrounding collection moments, the correction amplitude of the blood glucose concentration at each collection moment in each cycle is obtained; Obtaining an optimized correction factor of the non-invasive blood glucose monitor according to the correction amplitude of the blood glucose concentration at all sampling moments in all sampling time periods; and optimizing and correcting the non-invasive blood glucose monitor according to the optimized correction factor; The specific method for obtaining the fluctuation difference of the blood glucose monitoring data sequence at each collection moment in each cycle is as follows: In the In the blood glucose monitoring data fluctuation curve within the collection time period, The band corresponding to the blood glucose monitoring data point at each collection moment is recorded as the target band; all bands except the target band are recorded as reference bands; According to The similarity between the blood glucose monitoring data sequence at the time of collection and the blood glucose monitoring data sequence at other collection times is obtained. All similar data sequences at the time of collection; The first The position of the blood glucose monitoring data point in the target band at each collection moment is recorded as the target position; The first The first Dimensional data and The first The absolute value of the difference between the two dimensional data is recorded as the target band and the The first reference band The data difference value of the first dimension data; the target band is The DTW distance between the reference bands is recorded as the DTW distance between the target band and the The sequence difference value between the reference bands; the target band and the The sum of the data differences of all dimensional data between the reference bands is recorded as the difference between the target band and the The overall data difference value between the reference bands; the product of the overall data difference value and the sequence difference value is recorded as the target band and the The sum of the corrected difference values ​​between the target band and all reference bands is recorded as the first The fluctuation difference factor of the blood glucose monitoring data sequence at the first collection moment; the fluctuation difference factor is combined with the The ratio between the number of all similar data sequences at the collection time is used as the The fluctuation difference of the blood glucose monitoring data sequence at each collection moment; The specific formula is: Where, Indicates the The collection time period is the first The fluctuation difference of the blood glucose monitoring data sequence at each collection moment; Indicates the first The number of all reference bands of the blood glucose monitoring data fluctuation curve within a collection time period; Represents the data type quantity of all dimensional data in the blood glucose monitoring data sequence at each collection moment; Indicates the During the collection period The first Dimensional data; Indicates the During the collection period The first Dimensional data; Indicates the During the collection period The corresponding band of the blood glucose monitoring data point at the first collection moment is DTW distance between reference bands; Indicates the During the collection period The number of all similar data sequences at the collection moment; Indicates taking the absolute value; Indicates the The fluctuation difference factor of the blood glucose monitoring data series at each collection moment.

2. The calibration method of the non-invasive blood glucose monitor based on joint optimization according to claim 1, characterized in that: The method of obtaining all cycles of the blood glucose monitoring data sequence set within each acquisition time period by analyzing the noise distribution in the blood glucose monitoring data sequence includes: Get the Fluctuation curve of blood glucose monitoring data within a collection time period; For the first Perform Fourier transform on the fluctuation curve of blood glucose monitoring data within the acquisition time period, and take the inverse of the frequency with the largest amplitude in the spectrum as the One cycle of the blood glucose monitoring data fluctuation curve within the collection time period is recorded as the target cycle; According to the maximum amplitude in the spectrum and the frequency of the target cycle, the curve obtained is recorded as the fluctuation curve of the target cycle; The difference curve between the fluctuation curve of the blood glucose monitoring data in the acquisition time period and the fluctuation curve of the target period is recorded as the curve to be analyzed; the curve to be analyzed is decomposed using the EDM decomposition algorithm to obtain several IMF components and residual terms; the possibility of each IMF component belonging to high-frequency noise is obtained; The IMF component with the highest probability of being high-frequency noise is recorded as the noise component, and the other IMF components and the residual term are added to obtain the curve to be analyzed after the noise is removed. The curve to be analyzed after the noise is removed is subjected to Fourier transform, and the inverse of the frequency corresponding to each peak in the spectrum is taken as the first One cycle of the blood glucose monitoring data fluctuation curve within a collection time period.

3. The calibration method of the non-invasive blood glucose monitor based on joint optimization according to claim 2, characterized in that: The acquisition The blood glucose monitoring data fluctuation curve within a collection time period includes the following specific methods: Construct a four-dimensional space based on the acquisition time, skin reflected light intensity data, skin electrical impedance data, and skin temperature data; The blood glucose monitoring data sequence set within the collection time period is input into the four-dimensional space to obtain the Several blood glucose monitoring data points at collection times within a collection time period; In four-dimensional space, the least square method is used to transform The curve fitting is performed on all the blood glucose monitoring data points within the collection time period to obtain the Fluctuation curve of blood glucose monitoring data within a collection time period.

4. The calibration method of a non-invasive blood glucose monitor based on joint optimization according to claim 2, characterized in that: The specific method of obtaining the possibility that each IMF component belongs to high-frequency noise includes: For any IMF component, a frequency spectrum of the IMF component is obtained by Fourier transform; a peak detection method is used to obtain the frequency corresponding to each peak in the frequency spectrum, and the frequency is recorded as the peak frequency; The cumulative sum of the absolute values ​​of the differences between the mean of all peak frequencies of the IMF component and the mean of all peak frequencies of all other IMF components is taken as the possibility that the IMF component belongs to high-frequency noise.

5. The calibration method of a non-invasive blood glucose monitor based on joint optimization according to claim 1, characterized in that: The acquisition All similar data sequences at the collection moment, including the specific methods: Preset a similarity threshold parameter , if The blood glucose monitoring data sequence at the collection moment is The cosine similarity between the blood glucose monitoring data sequences at the target position in the reference band is greater than or equal to the similarity threshold parameter , will The blood glucose monitoring data sequence at the target position in the reference band is used as the first Similar data sequences at the same collection moment.

6. The calibration method of a non-invasive blood glucose monitor based on joint optimization according to claim 1, characterized in that: The specific method for obtaining the difference degree of the blood glucose monitoring data sequence at each collection moment in each cycle based on the difference in the fluctuation difference of the blood glucose monitoring data sequence between each collection moment and other collection moments is as follows: In the During the collection period, The sum of the absolute values ​​of the differences between the fluctuation difference of the blood glucose monitoring data sequence at the first collection moment and the fluctuation difference of the blood glucose monitoring data sequence at other collection moments is taken as the first The difference of blood glucose monitoring data series at each collection moment.

7. The calibration method of a non-invasive blood glucose monitor based on joint optimization according to claim 1, characterized in that: The specific method for obtaining the abnormality degree at each collection moment in each cycle according to the difference between each collection time period and the adjacent collection time periods is as follows: Preset a time period neighborhood parameter , will Before the collection period The collection period and Each collection time period is taken as The reference collection time period of each collection time period; The first The first The first The difference between the blood glucose monitoring data sequence at the left adjacent collection time and the The first The first The absolute value of the difference between the difference of the blood glucose monitoring data sequence at the right adjacent collection time is recorded as The adjacent abnormal difference of the reference collection time period; The cumulative sum of the adjacent abnormal differences of all reference collection time periods in the collection time period is recorded as Anomaly correction factor for each acquisition period; The anomaly correction factor is combined with the The product of the differences between the blood glucose monitoring data sequences at the time of collection is used as the The abnormality degree of the blood glucose monitoring data sequence at each collection moment.

8. The calibration method of a non-invasive blood glucose monitor based on joint optimization according to claim 1, characterized in that: The method of obtaining the correction amplitude of the blood glucose concentration at each sampling moment in each cycle according to the difference in the degree of abnormality between each sampling moment and surrounding sampling moments includes the following specific methods: Preset a collection time neighborhood parameter , in During the collection period, before the collection time The collection time and The time sequence composed of the collection moments is recorded as Neighborhood reference time sequence of each acquisition moment; The first The mean of the patient blood glucose concentration data at all collection moments in the neighborhood reference time sequence of the collection moment is recorded as the first mean; During the collection period The absolute value of the difference between the patient's blood glucose concentration data at each collection moment and the first mean value is recorded as the correction factor; The first The normalized value of the product of the abnormality degree of the blood glucose monitoring data sequence at the time of collection and the correction factor is used as the first The correction range of blood glucose concentration at each collection moment.

9. The calibration method of a non-invasive blood glucose monitor based on joint optimization according to claim 8, characterized in that: The method of obtaining the optimized correction factor of the non-invasive blood glucose monitor according to the correction amplitude of the blood glucose concentration at all sampling moments in all cycles in all sampling time periods includes the following specific methods: The first During the collection period The patient's blood glucose concentration data at the time of collection is During the collection period The difference between the mean values ​​of the patient's blood glucose concentration data at all collection moments in the neighborhood reference time sequence of the collection moment is recorded as the first difference; The first difference and the The product of the correction amplitude of the blood glucose concentration at the time of collection is used as the Correction value at each acquisition moment; If the first difference is greater than or equal to 0, The difference between the patient's blood glucose concentration data at the time of collection and the corrected value is taken as the Corrected patient blood glucose concentration data at each collection moment; If the first difference is less than 0, the The sum of the patient's blood glucose concentration data at the time of collection and the correction value is taken as the Corrected patient blood glucose concentration data at each collection moment; The ratio between the mean of the corrected patient blood glucose concentration data at all collection moments in all cycles of all collection time periods and the mean of the patient blood glucose concentration data at all collection moments in all cycles of all collection time periods is used as the optimized correction factor of the non-invasive blood glucose monitor.

Images