Portable noninvasive glucometer utilizing singular coefficient analysis technology

Abstract

The invention provides a portable noninvasive glucometer utilizing a singular coefficient analysis technology. The portable noninvasive glucometer comprises a quantum cascade laser system, an attenuated total reflection window, a detector, an analog-to-digital converter and a main control unit integrating a singular coefficient calculator and an advanced machine learning regression model processor. In the using process, the laser system accurately emits time-domain monochromatic laser pulses with adjustable wave numbers to the attenuated total reflection prism in contact with the skin of a patient, and then the time-domain monochromatic laser pulses are captured by the detector and processed into time-domain monochromatic absorbance pulse signals through the analog-to-digital converter. The main control unit carries out data mining on the time-domain monochromatic absorbance pulse signal by utilizing a singular coefficient analysis technology and extracts a key singular coefficient, and further calculates the blood glucose concentration by utilizing a machine learning regression model. The accuracy of the noninvasive glucometer provided by the invention reaches the standard of replacing a fingertip glucometer specified by the American Food and Drug Administration, and the noninvasive glucometer shows excellent performance and reliability.

Description

Technical Field

[0001] The present invention relates to the field of blood sugar concentration analysis, and in particular to a portable non-invasive blood sugar meter based on quantum cascade laser and artificial intelligence technology. Background Art

[0002] The need for blood sugar testing is extremely urgent. Given that there are more than 500 million diagnosed diabetics, they are advised to monitor their blood sugar levels daily. In addition, the number of diabetics continues to rise, and the initial age of onset of high blood sugar continues to decline, now down to 35 years old. Although diabetes is difficult to cure, monitoring blood sugar concentrations can help to more effectively manage diabetes and reduce the risk of fatal diabetes complications. According to the latest statistics released by the World Health Organization, non-communicable diseases kill 4.1 million people each year, accounting for 74% of global deaths. The four most dangerous diseases are cardiovascular disease (17.9 million cases), cancer (9.3 million cases), chronic respiratory diseases (4.1 million cases) and diabetes (2 million cases). Diabetes medical care urgently needs innovative blood sugar meters.

[0003] At present, blood glucose testing generally uses fingertip blood glucose meters, which are achieved by testing the conductivity of blood glucose test strips. During each test, the patient needs to draw blood from the fingertips and drip it onto the test strip. The blood glucose on the test strip will be oxidized into conductive gluconic acid under the action of the pre-loaded glucose-specific enzyme. Although there are microneedle arrays on the market that can alleviate the inconvenience and slow process of fingertip pricking, its invasiveness and related health risks still exist. From a theoretical perspective, by penetrating the patient's subcutaneous microcirculation network to the interstitial fluid, and then returning, the molecular characteristics of glucose can be quickly detected in a non-invasive manner.

[0004] Indeed, research and development of noninvasive glucose meters using infrared, Raman, and acousto-optic spectroscopy has been active for decades. However, intensive efforts around the world have yielded few results. For example, the U.S. Food and Drug Administration (FDA) stipulates that the error of such noninvasive glucose meters must not exceed 15% relative to the test results of a standard fingertip glucose meter, and 95% of the measurements must fall within this error range. In 2019, Chen et al. (published in Talanta, Vol. 197, pp. 211-217) pointed out that all previous noninvasive spectrometers had not met the FDA's accuracy benchmark until the pressure between the patient and the attenuated total reflection prism (ATR) window of the Fourier transform infrared spectrometer (FTIR) was increased to 20N / cm 2 , blood glucose testing meets FDA accuracy benchmarks. Figure 1The associated pressure application (shown in (a)) effectively shortens the optical path length from the ATR window to the patient's subcutaneous microcirculatory network; therefore, the spectral signal-to-noise ratio is exponentially improved.

[0005] Recently, it was found (Song et al., Spectrochimica Acta Part A #124738) that the accuracy of the above spectral analysis method is affected by the presence of other components in the interstitial fluid at 1080 cm -1 By accurately selecting the spectral data with the least interference for in-depth exploration, it was confirmed that the performance of the spectrum can be improved by improving the data management software without any hardware changes. Figure 1 (a) The accuracy of the non-invasive blood glucose meter configuration shown. More specifically, for 1000-1040cm -1 The spectral intensity of the range is sampled and the principal component analysis method is applied, such as Figure 1 However, despite the progress, there is still a lack of a portable non-invasive blood glucose meter that can meet FDA accuracy standards without causing inconvenience or discomfort to patients (such as requiring patients to press the sensor-patient interface with a pressure of 20N / cm2). Summary of the invention

[0006] In order to improve the technical problems of existing fingertip blood glucose meters and make up for the lack of portable non-invasive blood glucose meters that meet FDA accuracy standards on the market, the present invention provides a portable non-invasive blood glucose meter (referred to as a singular coefficient blood glucose meter) using singular coefficient analysis technology to achieve accurate measurement of blood glucose concentration.

[0007] A portable non-invasive blood glucose meter using singular coefficient analysis technology, the blood glucose meter includes a main control unit (MCU) composed of an integrated circuit chip with a built-in singular coefficient calculator and a machine learning regression model processor, and a quantum cascade laser system (QCL), an infrared transparent attenuated total reflection prism (ATR), a detector (DETECTOR), and an analog-to-digital converter (ADC) connected thereto.

[0008] In one embodiment of the present application, the main control unit (MCU) signal generator composed of an integrated circuit chip with a built-in singular coefficient calculator and a machine learning regression model processor is connected to a quantum cascade laser system to control the shape and emission time of a comb pulse signal with adjustable wave number emitted by the quantum laser system;

[0009] In one embodiment of the present application, the attenuated total reflection prism is connected to the output end of the quantum cascade laser. In a typical detection process, the patient presses the skin of the hypothenar or other parts on the attenuated total reflection prism of the non-invasive blood glucose meter to carry out data collection. The multiple attenuated total reflection prism (MATR) has a 45° end face to produce 3-15 total reflections on the upper surface. As the most suitable, this attenuated total reflection prism forms 7 total reflections on the upper surface.

[0010] In one embodiment of the present application, the detector is disposed at the end of the attenuated total reflection prism, and is used to capture the optical signal output by the multiple attenuated total reflection prism;

[0011] In one embodiment of the present application, the analog-to-digital converter is connected to the detector and is used to convert the optical signal captured by the detector into a digital comb-shaped time-domain absorbance signal containing blood glucose information;

[0012] In one embodiment of the present application, the singular coefficient calculator processes the comb time-domain absorbance signal of the analog-to-digital converter, uses the multifractal detrended fluctuation analysis (MFDFA) method or other singular coefficient analysis methods similar to MFDFA to derive the singular cardinality and singular spectrum, and extracts the key singular coefficients, and then uses common machine learning regression methods to perform data mining on the key singular coefficients to obtain the blood glucose concentration.

[0013] In one embodiment of the present application, the key singular strength is a first-level key singular strength extracted from a singular spectrum and composed of the singular strengths of the leftmost point, the rightmost point and the maximum point, and multiple second-level key singular strengths are derived from the first-level key singular strength; the multiple second-level key singular strengths include a second-level key singular strength defined as the width of the leftmost point and the rightmost point, and if necessary, also include other derivative values ​​of the first-level key singular strength; the singular cardinality, the first-level key singular strength and the second-level key singular strength constitute a set of key singular coefficients.

[0014] In one embodiment of the present application, the commonly used machine learning regression method is a support vector machine regression method.

[0015] In one embodiment of the present application, the quantum cascade laser system includes a laser pulse having a frequency of 10 kHz to 1 MHz, 100 kHz to 500 kHz, or 150 kHz to 200 kHz, which is directly or indirectly connected to the main control unit to receive and track the control signal from the main control unit, and then emits a laser pulse having a frequency of 10 kHz to 1 MHz, 100 kHz to 500 kHz, or 150 kHz to 200 kHz in the time domain; the quantum cascade laser system includes a laser pulse having a frequency of 20 to 300 cm -1 , from 40 to 200cm -1 or 200cm -1 Wide wavenumber range and central wavenumber at 1000-1100 cm -1The quantum cascade laser system comprises a grating whose rotation and filtering wave number are controlled by a micro-electromechanical actuator or a grating whose length and filtering wave number are controlled by voltages applied to two end electrodes of the grating plate, and is used to filter broadband laser pulses into monochromatic laser pulses with adjustable wave number, and the width of the filtered laser pulses is set to 50-500ns, 50-200ns, 50-100ns or 100ns; the quantum cascade laser system comprises a plurality of apertures along the optical path of the laser system, which are used to filter the laser pulses and set the spectral resolution to 0.05-8cm -1 、1-8cm -1 , 4cm -1 or 1.6cm -1 .

[0016] In one embodiment of the present application, the detector includes an InAsSb infrared detector or an MCT infrared detector.

[0017] In one embodiment of the present application, the analog-to-digital converter (ADC) includes an analog-to-digital converter (ADC) operating at a sampling frequency of 10-100 MHz or 20-60 MHz.

[0018] In one embodiment of the present application, a quantum cascade laser system, an attenuated total reflection prism, a detector, an analog-to-digital converter (ADC), and a main control unit (MCU) are packaged on a plastic printed circuit board or a ceramic printed circuit board using surface mount technology, or are highly integrated into a single-chip non-invasive blood glucose meter using chip integration technology.

[0019] Compared with the prior art, the beneficial effects of adopting the above technical solution are as follows: the non-invasive blood glucose meter proposed by the present invention can integrate and miniaturize the non-invasive blood glucose detection equipment, saving computing costs while ensuring the detection accuracy. At the same time, this non-invasive blood glucose meter can accurately measure blood glucose concentration without the need for a pressure sensor, and meets the standards that must be met by fingertip blood glucose meters specified by the FDA. This standard means that the blood glucose concentration measured by the non-invasive blood glucose meter has a probability of more than 95% that the difference with the blood glucose concentration measured by the standard fingertip blood glucose meter under the same conditions is no more than 15%. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 (a) is the schematic diagram of the ATR-FTIR non-invasive blood glucose meter with pressure driver; Figure 1 (b) is to use Figure 1 Clarke error grid obtained by the non-invasive blood glucose meter in (a) using the hypothenar instead of the finger to test blood glucose. The spectral range of the data acquisition is 1000-1040cm -1 .

[0021] Figure 2This is a schematic diagram of the singular coefficient blood glucose meter proposed by the present invention.

[0022] Figure 3 (a) is a schematic diagram of the singular coefficient blood glucose meter. Figure 3 (b) Figure 3 (a) Simplified schematic diagram of hypothenar blood glucose measurement.

[0023] Figure 4 Schematic diagram of the conversion route from the original time-domain absorbance signal to the key singular coefficient database generated by singular coefficient analysis and then obtaining the blood glucose concentration through statistical analysis.

[0024] Figure 5 A schematic diagram of a set of original time-domain absorbance signals of a singular coefficient blood glucose meter.

[0025] Figure 6 This is a set of 41 singular spectra of concentrations obtained by testing a subject using a singular coefficient blood glucose meter.

[0026] Figure 7 This is a schematic diagram of the Clarke error grid obtained by comparing the singular coefficient blood glucose meter and the standard fingertip blood glucose meter after testing the blood glucose of the subject.

[0027] FIG8(a) is a schematic diagram of the conversion process of the time-domain absorbance signal of the singular coefficient blood glucose meter to the molecular optical spectrum and then the standard concentration calculation. FIG8(b) is the optical spectrum obtained through the conversion process of FIG8(a).

[0028] Fig. 9 The Clarke error grid of blood glucose concentration is calculated by converting the molecular optical spectrum of the time-domain absorbance signal measured by the singular coefficient blood glucose meter.

[0029] Fig.10 For a Figure 1 (a) A comprehensive analysis of the size and weight of the ATR-FTIR blood glucose meter shown and the singular coefficient blood glucose meter provided by the present invention.

[0030] Fig.11 For the general Figure 3 Schematic diagram comparing the concentrations measured by the new device and the standard fingerstick glucose meter after replacing the ADC.

[0031] Fig.12 Schematic diagram of the integrated laser system of a blood glucose meter upgraded for miniaturization.

[0032] Fig.13 Schematic diagram of packaging a device for accurately measuring molecular concentration using singular coefficient analysis technology on a printed circuit board in one embodiment of the present invention.

[0033] Fig.14Schematic diagram of an apparatus for realizing accurate measurement of molecular concentration using singular coefficient analysis technology integrated on a single chip in one embodiment of the present invention.

[0034] Figure numerals: 100 - monochromatic laser pulse generation system with adjustable wave number, 200 - attenuated total reflector, 300 - detector and ADC, 400 - main control unit with integrated singular coefficient calculator and machine learning regression model processor, 101 - QCL laser, 102 - grating plate. DETAILED DESCRIPTION

[0035] Embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar modules or modules with the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application, and cannot be construed as limitations on the present application. On the contrary, the embodiments of the present application include all changes, modifications and equivalents that fall within the spirit and connotation of the appended claims.

[0036] Please refer to Figure 2 , Figure 3 (a), the present invention discloses a portable non-invasive blood glucose meter using singular coefficient analysis technology, comprising a quantum cascade laser (QCL) with a wavelength range of 8.9 to 10.7 μm, a multiple attenuated total reflection prism (MATR) with 7 total reflections on the upper surface, a detector (DETECTOR), an analog-to-digital converter (ADC) and a main control unit (MCU). When measuring blood glucose, the patient's hypothenar is pressed on the MATR, please refer to Figure 3 (b). During the operation of this portable non-invasive blood glucose meter using the singular coefficient analysis technology, data mining is directly performed on the original comb-shaped time-domain absorbance signal collected by the detector to obtain the blood glucose concentration, without the need to convert the original comb-shaped time-domain absorbance signal into an absorption spectrum, and then calculate the blood glucose concentration based on the absorption spectrum. The complex process of detecting molecular concentrations with traditional light sources is avoided. The singular coefficient analysis technology in the present invention effectively avoids the inefficiency and inevitable information loss in traditional spectra when using light sources to measure molecular concentrations. It has been verified in practice that the singular coefficient blood glucose meter disclosed in the present invention can achieve fast and accurate non-invasive blood glucose measurement, minimize the inconvenience and discomfort of patients, and the measurement passes the FDA's accuracy benchmark. The blood glucose meter can replace the common fingertip blood glucose meter and has a compact blood glucose meter configuration. The portable non-invasive blood glucose meter using the singular coefficient analysis technology proposed in the present invention is specifically introduced below.

[0037] Figure 3(a) and (b) show an embodiment of the present invention, which discloses a portable non-invasive blood glucose meter using singular coefficient analysis technology. The blood glucose meter includes a main control unit (MCU) composed of an integrated circuit chip with a built-in singular coefficient calculator and a machine learning regression model processor, and a quantum cascade laser system (QCL), an infrared transparent multi-attenuation total reflection prism (MATR), a detector (DETECTOR), and an analog-to-digital converter (ADC) connected thereto. The output end of the quantum cascade laser is coupled to the MATR, and the upper end face of the MATR is cut into a 45° angle surface to form multiple reflection channels. In the conventional data acquisition process, the finger or hypothenar of the test subject is placed on the MATR surface of the blood glucose meter. The infrared detector is connected to the MATR to capture the light signal output by the MATR, and digitizes it into a time-domain absorbance signal containing blood glucose information through the ADC connected to the detector. The singular coefficient calculator receives the time-domain absorbance signal from the ADC, and uses the singular coefficient analysis method to calculate the singular cardinality and singular spectrum, and extracts the key singular coefficients; then the common machine learning regression method in the machine learning regression model processor is used to convert these key singular coefficients into blood glucose concentrations.

[0038] Specifically, in this embodiment, a broadband (8.9 to 10.7 μm) pulsed QCL component (QCL14890-09, Hamamatsu Photonics) with an average full pulse power of 600 mW, a MATR, an infrared detector (C12494-011LH, Hamamatsu Photonics) and an MCU are used. The wavelength range of the QCL is equivalent to 935 to 1128 cm -1 The wavelength range fully covers the critical spectral region of 1000-1040cm-1 for detecting blood glucose, with minimal spectral interference. The dimensions of the QCL laser assembly are 82mm long, 81mm wide, and 88mm high. In this assembly, a function generator (T340, Highland Technology) outputs a 180,050Hz square wave pulse sequence to help the QCL emit pulses of this frequency; another 1800Hz sine wave sequence is used to drive the built-in microelectromechanical system (MEMS) grating plate to narrow the laser pulse width and improve the spectral resolution performance of each laser pulse.

[0039] The QCL pulse signal is coupled to a ZnSe MATR prism, which is 72.4 mm long and 5.7 mm high, and the end face is cut into a 45° face angle to facilitate 7 total reflections on the upper surface. In a typical data acquisition process, the patient's hypothenar is pressed against the MATR window of the singular coefficient blood glucose meter. The optical signal is captured by an MCU-controlled InAsSb infrared detector (C12494-011LH, Hamamatsu Photonics), with dimensions of 65 mm long, 20 mm wide, and 50 mm high. The detector fully covers the spectral response range from 2 to 11 μm. The analog signal from the detector is digitized and converted to digital by a compact and fast ADC connected to a common USB. Finally, the acquired data is transferred to the singular coefficient calculator for data processing.

[0040] Figure 4 The original time-domain absorbance signal of the infrared detector ( Figure 5 ) is a new path to directly convert it into blood glucose concentration. Specifically, the original comb-shaped time-domain absorbance signal is applied to singular coefficient analysis to obtain the singular spectrum and extract the key singular coefficients. With the help of common machine learning regression methods (such as support vector machine regression method), the blood glucose concentration is calculated based on the key singular coefficient database. Figure 6 41 singular spectra from patients are shown as a typical example of singular coefficient blood glucose meter operation. From each singular spectrum, three first-order key singular intensities and one second-order key singular intensity are expected to be obtained, which covers the left end point α +∞ , α0 at the peak, α at the right end point -∞ And Δα=α -∞ -α +∞ .. Machine learning regression method uses Figure 6 Statistical calculations are performed on the key singular intensities and related singular cardinal numbers of 41 singular spectra in the dataset to calculate the corresponding blood glucose concentration.

[0041] Specifically, in some embodiments, the singular coefficient analysis includes the following calculation steps:

[0042] (1) Take the absorbance x of the i-th time unit in the time domain absorbance signal i , calculate its average value when the total duration is N time units <x>, and construct the filtered absorbance signal profile Y(i),

[0043]

[0044] (2) The signal Y(i) is divided into M non-overlapping segments of equal time length s from i = 1 to i = N, and then divided into another group of M non-overlapping segments of time length s from i = N to i = 1. Then, the local trend y of the vth segment is calculated using the least squares fitting of the polynomial function. v (i) and determine the variance function Y of the vth segment var (v,s),

[0045]

[0046] (3) Calculate the qth order wave function F(q,s),

[0047]

[0048] (4) Calculate the singular cardinality β, the singular intensity α, and the multifractal dimension f(α) to construct the singular spectrum f(α), where h(q) is calculated by d(log(F)) / d(log(s), and h′(q) is calculated by d(h) / d(q).

[0049]

[0050] α=h(q)+qh′(q)#(7)

[0051] f(α)=q 2 h′(q)+1#(8)

[0052] (5) Extract three first-order key singular intensities from the singular spectrum, the leftmost point α +∞ , the rightmost point α –∞ , the peak point of α, thereby constructing a set of first-level key singular intensities. Multiple second-level key singular intensities can be derived from the first-level key singular intensities, wherein the multiple second-level key singular intensities include a second-level key singular intensities defined as the width of the leftmost point and the rightmost point, and other derivative values ​​of the first-level key singular intensities when necessary; the singular cardinality, the first-level key singular intensities and the second-level key singular intensities constitute a set of key singular coefficients. .

[0053] In some embodiments, the machine learning regression model comprises a standard support vector machine regression model.

[0054] Clarke error grid method is conveniently used to evaluate the singular coefficient of blood glucose meter because it is the ISO15197 blood glucose clinical measurement standard ( Figure 3 ) and FTIR-based blood glucose meters ( Figure 1 (a)) Correlation with the measurement results of a standard fingertip blood glucose meter. In the Clarke error grid, paired measurement data obtained by two different methods are plotted in the Clarke error grid diagram, and the regions are classified according to the correlation error. Therefore, on the Clarke error grid diagram, the error area is used as a reference point to visualize the location of the plotted data, which is beneficial to evaluate the correlation between the two methods. In this work, the paired data (blood glucose reference values ​​of the standard fingertip blood glucose meter and blood glucose measurement values ​​of the non-invasive blood glucose meter) are plotted in the Clarke error grid diagram, and the data in area A of the grid (with a correlation error of less than 20%) are regarded as a convenient preliminary correlation benchmark. In addition, a gray grid is added to each Clarke error grid diagram, in which the correlation error is less than 15%, in order to meet the FDA's requirement that 95% of the relevant data be within the 15% error grid.

[0055] The accuracy of the singular coefficient blood glucose meter disclosed in this embodiment is evaluated and verified by blood glucose measurement data of a group of patients with different blood glucose concentrations for cross-checking the diabetic status of the patients, all of whom meet the "Oral Glucose Tolerance Test" criteria recognized by the American Diabetes Association. The accuracy of the blood glucose meter is compared with the accuracy of a standard fingertip blood glucose meter. The following experimental procedures can be performed to match the recent evaluation and verification of other non-invasive blood glucose measurement methods:

[0056] 1) Specifically, four healthy individuals without diabetes are selected as subjects, including two men and two women. The subjects are required to fast for 10 to 12 hours before 8 am every day. After the fasting period, the actual blood glucose value of the subject is first detected and recorded using a standard fingertip blood glucose meter. Then, the singular coefficient blood glucose meter provided by the present invention is used to measure the blood glucose of the subject, and three QCL signal scans are performed for each subject. The measured blood glucose concentration is expected to be close to the normal level of 3.9mmol / L in healthy people.

[0057] 2) Next, each subject was asked to take 75 grams of glucose orally, and then undergo a series of blood glucose tests to track blood glucose changes. The test sequence covered 12 tests, with a total log time of 120 minutes, and the time interval between two consecutive tests was set to 10 minutes. The data measured blood glucose concentrations exceeding 7.9mmol / l were discarded, and 41 values ​​with a concentration range of 3.9 to 7.9mmol / l and increments of 0.1mmol / l were measured and recorded.

[0058] 3) For each set of data, each patient is required to perform blood glucose measurement using a rapid singular coefficient blood glucose meter within 9 minutes after the blood sampling process using a standard fingertip blood glucose meter is completed. The reason for setting the delay time of 9 minutes is that the fingertip blood sampling test measures glucose in the blood, while the singular coefficient blood glucose meter detects glucose in the interstitial fluid, and the average time delay for glucose to diffuse from the blood to the interstitial fluid is approximately 8-10 minutes. It should be noted that the test performed using the singular coefficient blood glucose meter of the present invention can be completed at a rate of about 1 second per time, so for such a short time difference, the changes in blood glucose over time can be ignored.

[0059] 4) Each non-invasive blood glucose test is performed by requiring each patient to press the hypothenar on the MATR window of the blood glucose meter in a relaxed state. Similarly, each test requires 3 QCL signal scans. In the present invention, 5 repeated "oral glucose tolerance test" processes were performed on 2 men and 2 women (all non-diabetic patients).

[0060] In this embodiment, the data management method for the test data collected by the singular coefficient blood glucose meter is as follows:

[0061] For each set of original comb-shaped time-domain absorbance signals of each blood glucose measurement, the singular coefficient analysis method is directly used to process them, and the blood glucose concentration statistics are generated with the support vector machine regression method. Figure 6 As shown, 41 singular spectra from patients are covered.

[0062] The non-invasive blood glucose test results in this embodiment show a high correlation with the test results of the standard fingertip blood glucose meter. Figure 7 As presented. Obviously, all relevant data are in the gray area with an error of less than 15%. Therefore, the singular coefficient blood glucose meter and its operation disclosed in this embodiment is an excellent non-invasive blood glucose measurement design, and its value proposition is completely capable of replacing the standard fingertip blood glucose meter. The typical test time of the operation of the singular coefficient blood glucose meter is 1 second, while the test time of the FTIR-based blood glucose meter is about 10 minutes.

[0063] It should be noted that, in the description of the embodiments of the present invention, unless otherwise clearly specified and limited, the terms "setting" and "connection" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a direct connection, or an indirect connection through an intermediate medium. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances; the drawings in the embodiments are used to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings herein can be arranged and designed in various different configurations.

[0064] Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limitations on the present application. Ordinary technicians in the field can change, modify, replace and modify the above embodiments within the scope of the present application.

[0065] Non-invasive blood glucose testing example

[0066] Example 1: Performance comparison of singular coefficient blood glucose meter based on singular coefficient analysis route in data processing and based on comb-like time-domain absorbance signal to optical spectrum route

[0067] In order to realize non-invasive blood glucose measurement using a singular coefficient blood glucose meter, in addition to using MFDFA or a singular coefficient analysis method similar to MFDFA to directly perform data mining operations on the comb-shaped time-domain absorbance signal combined with regression analysis, the comb-shaped time-domain absorbance signal of the blood glucose meter can also be converted into an optical spectrum, and then the concentration is calculated by regression analysis of the optical spectrum data to complete the blood glucose measurement. Figure 8(a) summarizes the process of converting the comb-shaped time-domain absorbance signal to the optical spectrum, and Figure 8(b) shows 41 optical spectra. These spectra are consistent with Figure 6 The 41 sets of comb-shaped time-domain absorbance signals in the same Figure 7 and Fig. 9 Clarke error grid diagram in . Obviously, if Figure 7 As shown in Figure 1, all relevant data obtained through the singular coefficient analysis path of direct comb time-domain absorbance signal data mining are within the gray grid area, and the relevant error is less than 15%. Fig. 9 As shown in the figure, only about 78% of the relevant data obtained from the comb-time absorbance signal conversion to the optical spectrum path in data processing is located in the gray grid area. Therefore, the singular coefficient blood glucose meter using the singular coefficient analysis path of direct comb-time absorbance signal data mining has reached the FDA's accuracy standard and can replace the fingertip blood glucose meter, while the accuracy of the blood glucose meter using the comb-time absorbance signal to optical spectrum path in data processing is much lower and does not meet the FDA standard. In fact, the speed of obtaining blood glucose measurement results by the singular coefficient analysis path of direct comb-time absorbance signal data mining can also be about twice as fast as the comb-time absorbance signal to optical spectrum path in data processing.

[0068] Example 2: Performance comparison of a singular coefficient blood glucose meter using a singular coefficient analysis data processing route and a FTIR blood glucose meter using a standard optical spectroscopy data processing route

[0069] In addition to using a singular coefficient blood glucose meter for non-invasive blood glucose measurement, FTIR can also be used for non-invasive blood glucose measurement ( Figure 1 (a)). The blood glucose meter includes a pressure driver for applying 20N / cm 2 The pressure was adjusted to improve the measurement accuracy. The blood glucose concentration was calculated by principal component analysis (PCA) of the spectral data, using the standard with the least spectral interference. Figure 1 (b) shows a Clarke error grid plot of the results of blood glucose measurements using an FTIR blood glucose meter. Clearly, 95.5% of the relevant data are within the gray grid area. Although the FTIR blood glucose meter equipped with a pressure driver meets the FDA's accuracy standards and can replace blood fingerstick blood glucose meters, its accuracy is inferior to the singular coefficient blood glucose meter using the singular coefficient analysis data management approach. In fact, the singular coefficient blood glucose meter using the singular coefficient analysis data processing approach is more than 60 times faster than the FTIR blood glucose meter using the optical spectroscopy data processing approach. Clearly, the FTIR blood glucose meter is far inferior to the singular coefficient blood glucose meter in terms of size and weight. For a summary of the comparison, see Fig.10 .

[0070] Example 3

[0071] In this example, in order to further reduce the size and weight of the apparatus of the present invention, and to measure the practicality of integrating all hardware components onto a single printed circuit board, Figure 3 The hardware design of the specific device described in (a) was re-examined. For example, the 100MHz ADC was changed to a 20MHz ADC. The original ADC size was 15cm*12cm*2.5cm and weighed 0.5kg. The physical size of the replaced 20MHz ADC was 5cm*2.5cm*1cm and weighed 10g, which was 50 times lighter than the original ADC. After replacing the ADC, Fig.11 As shown, more than 97% of the measurements are still within the error grid area with an error of no more than 15%.

[0072] In addition, other hardware modifications are investigated in this case study. Consider replacing the Figure 3 (a) QCL system. Specifically, Figure 3 The QCL system in (a) has a size of 11.2cm*8.2cm*8.8cm and weighs 1.20kg. It operates at 180kHz and has a bandwidth of 193cm -1 , with a rated power of more than 600mW. A QCL that can replace it has a physical size of 4mm*4mm*2mm and weighs less than 10g, such as Fig.12 shown.

[0073] Overall, Figure 3 (a) The overall physical size of the hardware design is more than 10 times smaller than the state-of-the-art FTIR spectrometer, and the weight is reduced from 70kg to 5kg. The new version of the hardware in this case is 10-100 times smaller and 10-100 times lighter than the original. Fig.13 The display uses surface mount technology to integrate the miniaturized upgraded laser system, simplified and miniaturized MATR, simplified and miniaturized detector and ADC components, and miniaturized upgraded MCU on a ceramic printed circuit board, achieving better hardware miniaturization effect. Fig.14 The best hardware design is to integrate all these components into a single chip. Fig.13 , Fig.14 In the figure, reference numeral 100 denotes a monochromatic laser pulse generating system with adjustable wave number, reference numeral 200 denotes an attenuated total reflector, reference numeral 300 denotes a detector and ADC, and reference numeral 400 denotes a main control unit integrating a singular coefficient calculator and a machine learning regression model processor.

[0074] The present invention proposes a portable non-invasive blood glucose meter using singular coefficient analysis technology. The hardware of the blood glucose meter includes a main control unit composed of an integrated circuit chip with a built-in singular coefficient calculator and a machine learning regression model processor, and a quantum cascade laser system, an infrared transparent attenuated total reflection prism, an infrared detector, and an analog-to-digital converter connected thereto. Compared with the design of a blood glucose meter based on FTIR, the size and weight of the singular coefficient blood glucose meter are more suitable for portable applications. The comb-shaped time domain pulse signal emitted by the quantum cascade laser system directly uses singular coefficient analysis to generate a set of singular spectra after passing through an attenuated total reflection prism in contact with the patient's body surface. The key singular coefficients are extracted from the singular spectrum and converted into concentration data through advanced statistical analysis involving machine learning. This advanced non-invasive blood glucose meter is fast and convenient, and its accuracy has exceeded the FDA benchmark, and it can replace the outdated and inconvenient fingertip blood glucose meters on the market.< / x>

Claims

1. A portable non-invasive blood glucose meter using singular coefficient analysis technology, characterized in that: The portable non-invasive blood glucose meter includes a main control unit composed of an integrated circuit chip with a built-in singular coefficient calculator and a machine learning regression model processor, and a quantum cascade laser system, an infrared transparent attenuated total reflection prism, a detector and an analog-to-digital converter connected thereto; the non-invasive blood glucose meter is instructed by the main control unit to transmit a time-domain monochromatic pulse light whose wave number continuously changes with the time domain, and the light enters the detector and the analog-to-digital converter after passing through the outer surface of the attenuated total reflection prism in contact with the patient's skin; the singular coefficient calculator in the main control unit performs data mining on the time-domain absorbance signal, extracts a set of key singular coefficients, and submits them to the machine learning regression model processor for further analysis to deduce the blood glucose concentration.

2. The portable non-invasive blood glucose meter according to claim 1, characterized in that: The portable non-invasive blood glucose meter is more than 10 times or 100 times smaller in size and weight than a typical FTIR spectrometer.

3. The portable non-invasive blood glucose meter according to claim 1, characterized in that: The singular coefficient calculator includes software and hardware matched by the integrated circuit chip of the main control unit. The singular coefficient calculator takes the absorbance x of the i-th time unit from the time domain absorbance signal. i , calculate its average value within a total duration of N time units <x> , and construct the filtered absorbance signal profile Y(i):< / x> The signal Y(i) is divided into M non-overlapping segments of equal time length s from i=1 to i=N, and then divided into another group of M non-overlapping segments of time length s from i=N to i=1. Then the local trend y of the vth segment is calculated by the least squares fitting of the polynomial function. v (i) and determine the variance function Y of the vth segment var (v,s): Then from Y var (v,s) calculates a series of q-order wave functions F(q,s): Then, the singular cardinality β, singular intensity α and multifractal dimension f(α) are calculated from F(q,s) to construct the singular spectrum f(α), where h(q) is calculated by d(log(F)) / d(log(s)) and h′(q) is calculated by d(h) / d(q). α=h(q)+qh′(q)# f(α)=q 2 h′(q)+1# Finally, the first-order key singular intensity consisting of the singular intensities of the leftmost point, the rightmost point and the maximum point is extracted from the singular spectrum, and multiple second-order key singular intensities are derived from the first-order key singular intensity; The multiple secondary key singular strengths include a secondary key singular strength defined as the width of the leftmost point and the rightmost point; the singular cardinality, the primary key singular strength and the secondary key singular strength constitute a set of key singular coefficients.

4. The portable non-invasive blood glucose meter according to claim 1, characterized in that: The machine learning regression model processor includes software for setting up and applying a support vector machine regression model to data mine blood glucose concentration from a set of key singular coefficients output by a singular coefficient calculator.

5. The portable non-invasive blood glucose meter according to claim 1, characterized in that: The quantum cascade laser system includes a control unit directly or indirectly connected to the main control unit to receive and track control signals from the main control unit, thereby emitting laser pulses with a frequency of 10kHz to 1MHz, 100kHz to 500kHz or 150kHz to 200kHz in the time domain.

6. The portable non-invasive blood glucose meter according to claim 1, characterized in that: The quantum cascade laser system includes from 20 to 300 cm -1 , from 40 to 200cm -1 or 200cm -1 Wide wavenumber range and central wavenumber at 1000-1100 cm -1 The quantum cascade laser system comprises a grating whose rotation and filtering wave number are controlled by a micro-electromechanical actuator or a grating whose length and filtering wave number are controlled by voltages applied to two end electrodes of the grating plate, and is used to filter broadband laser pulses into monochromatic laser pulses with adjustable wave number, and the width of the filtered laser pulses is set to 50-500ns, 50-200ns, 50-100ns or 100ns; the quantum cascade laser system comprises a plurality of apertures along the optical path of the laser system, which are used to filter the laser pulses and set the spectral resolution to 0.05-8cm -1 、1-8cm -1 , 4cm -1 or 1.6cm -1 .

7. The portable non-invasive blood glucose meter according to claim 1, characterized in that: The detector comprises an InAsSb infrared detector or an MCT infrared detector.

8. The portable non-invasive blood glucose meter according to claim 1, characterized in that: The analog-to-digital converter includes an analog-to-digital converter operating at a sampling frequency of 10-100 MHz or 20-60 MHz.

9. The portable non-invasive blood glucose meter according to claim 1, characterized in that: The analog-to-digital converter comprises an n-bit analog-to-digital converter with n ranging from 10 to 24.

10. The non-invasive blood glucose meter according to claim 1, characterized in that: The quantum cascade laser system, attenuated total reflection prism, detector, analog-to-digital converter and main control unit are packaged on a plastic printed circuit board or a ceramic printed circuit board using surface mounting technology, or are highly integrated into a single-chip non-invasive blood glucose meter using chip integration technology.

Images