Calibration model and method for determining the concentration of analyte

Abstract

The present invention relates to a calibration model and method for determining the concentration of a substance to be measured. A calibration model for determining the concentration of a substance to be measured includes an acquisition unit and a calibration curve function determination unit. The acquisition unit is configured to acquire absorbance values ​​and concentration values ​​of the substance to be measured, thereby forming a plurality of calibration points configured to determine a calibration curve function based on the acquired absorbance values ​​and concentration values ​​of the substance to be measured, thereby fitting a calibration curve of the absorbance values ​​and concentration values ​​of the substance to be measured. The present invention can accurately express the functional relationship between the optical signal and the concentration of the substance to be measured, and combine the detected optical signal to accurately determine the concentration value of the substance to be measured, and provide a calibration model and method selection that can accurately express the functional relationship between the optical signal and the concentration of the substance to be measured, which can produce a better fitting effect in some systems.

Description

Technical Field

[0001] The present invention relates to the field of biochemical detection technology, and in particular to a calibration model and method for determining the concentration of a substance to be detected. Background Art

[0002] Biochemical analyzers, immunoassay analyzers, chemiluminescence analyzers, and coagulation analyzers are tools for measuring the concentration (activity) of substances to be tested in human blood and are indispensable in clinical practice. The common principle of these instruments is to convert the concentration (activity) signal of the substance to be tested into a light signal. The concentration (activity) value of the substance to be tested is determined by detecting the light signal and based on the calibration function curve that describes the relationship between the light signal and the concentration (activity) signal. Therefore, whether the calibration curve function relationship can accurately express the relationship between the light signal and the concentration (activity) signal is very critical to the measurement result.

[0003] The calibration process actually uses the correspondence between the known optical signal and concentration (activity) of finite points to derive a functional relationship curve composed of infinite points, which makes it possible to generate a calibration function in many ways. Known calibration function models include linear modules, spline function models such as the Akima model, Logit-Log 3P model, logit-log 4P model, logit-log 5P model and exponential model. The difference between these known calibration function models is whether they can more accurately describe the relationship between the optical signal and the concentration. In one system, a certain function model may be better, while in another system, another function model may be better. In this case, it is of great significance to choose a calibration model and method that can accurately express the functional relationship between the optical signal and the concentration of the substance to be measured.

[0004] The above description of the background technology is only for facilitating an in-depth understanding of the technical solution of the present invention (such as the technical means used, the technical problems solved and the technical effects produced), and should not be regarded as an admission or any form of implication that the information constitutes prior art already known to those skilled in the art. Summary of the invention

[0005] The object of the present invention is to provide a calibration model and method for determining the concentration of a substance to be tested, which can accurately express the functional relationship between the optical signal and the concentration of the substance to be tested, and combine the detected optical signal to accurately determine the concentration value of the substance to be tested.

[0006] According to an embodiment of the present invention, a calibration model for determining the concentration of a substance to be tested is provided, which includes: an acquisition unit, which is configured to acquire absorbance values ​​and concentration values ​​of the substance to be tested, thereby forming a plurality of calibration points; a calibration curve function determination unit, which is configured to determine a calibration curve function based on the acquired absorbance values ​​and concentration values ​​of the substance to be tested, thereby fitting a calibration curve of the absorbance values ​​and concentration values ​​of the substance to be tested.

[0007] Furthermore, the calibration curve function determination unit includes: a parameter value determination unit configured to determine the parameter value so that the sum of squares of residuals between the calibration curve function based on the parameter value and the calibration points is minimized.

[0008] Furthermore, the parameter value determination unit includes: an initial value determination unit, which is configured to fit the calibration points using the least squares method to obtain a least squares fitting curve, and use the parameter values ​​of the obtained least squares fitting curve as the initial values ​​of the parameter values ​​of the calibration curve function; a linearization unit, which is configured to simplify the calibration curve function into a linear expression about the increment of the parameter value using Taylor expansion; and an adjustment unit, which is configured to solve the linear expression using the Levenberg-Maguire method, and iteratively correct the increment of the parameter value so that the sum of squared residuals is minimized.

[0009] Furthermore, the number of the plurality of calibration points ranges from four to ten.

[0010] Furthermore, the calibration curve function is y=a+bsin(cx+d), y=a+bcos(cx+d), or Among them, x is the concentration value, y is the absorbance value, and a, b, c, and d are parameter values.

[0011] According to an embodiment of the present invention, a calibration method for determining the concentration of a substance to be tested is provided, which comprises: collecting absorbance values ​​and concentration values ​​of the substance to be tested, thereby forming a plurality of calibration points; determining a calibration curve function based on the collected absorbance values ​​and concentration values ​​of the substance to be tested, thereby fitting a calibration curve of the absorbance values ​​and concentration values ​​of the substance to be tested.

[0012] Further, determining the calibration curve function based on the sine function includes: determining a parameter value so that the sum of squares of residuals between the calibration curve function based on the parameter value and the calibration point is minimized.

[0013] Furthermore, determining the parameter value includes: fitting the calibration points using the least squares method to obtain a least squares fitting curve, and using the parameter values ​​of the obtained least squares fitting curve as initial values ​​of the parameter values ​​of the calibration curve function; simplifying the calibration curve function into a linear expression about the increment of the parameter value using the Taylor expansion; solving the linear expression using the Levenberg-Maguire method, and iteratively correcting the increment of the parameter value to minimize the residual sum of squares.

[0014] Furthermore, the number of the plurality of calibration points ranges from four to ten.

[0015] Furthermore, the calibration curve function is y=a+bsin(cx+d), y=a+bcos(cx+d), or Among them, x is the concentration value, y is the absorbance value, and a, b, c, and d are parameter values.

[0016] The present invention adopts the above technical solution, which has the following beneficial effects: the calibration model and method according to the present invention can accurately express the functional relationship between the optical signal and the concentration of the substance to be measured, and combine the detected optical signal to accurately determine the concentration value of the substance to be measured. In addition to the known calibration function model, the present invention provides a selection of calibration models and methods that can accurately express the functional relationship between the optical signal and the concentration of the substance to be measured. The calibration model and method of the present invention can produce better fitting effects in some systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The exemplary embodiments of the present invention will be described in more detail below in conjunction with the accompanying drawings. For the sake of clarity, the same components in different drawings are shown with the same reference numerals. It should be noted that the drawings are only for illustration and are not necessarily drawn to scale. In these drawings:

[0018] Figure 1 is a block diagram showing the configuration of a calibration model for determining the concentration of a substance to be measured according to an embodiment of the present invention.

[0019] Figure 2A is a fitting curve after 1 iteration as an example, Figure 2B is a fitting curve showing, as an example, the termination of the iteration.

[0020] FIG. 3A to FIG. 3H is a graph showing the fitting of calibration points using different calibration models.

[0021] Figure 4 is a flow chart illustrating a method for determining a calibration model for the concentration of a substance to be measured according to an embodiment of the present invention. DETAILED DESCRIPTION

[0022] The implementation scheme of the present invention is described in detail below. This implementation scheme is implemented on the premise of the technical scheme of the present invention, and a detailed implementation method and specific operation process are given, but the protection scope of the present invention is not limited to the implementation scheme described below.

[0023] Figure 1 is a block diagram showing the configuration of a calibration model for determining the concentration of a substance to be measured according to an embodiment of the present invention. Figure 1 As shown, the calibration model for determining the concentration of the substance to be tested according to an embodiment of the present invention may include an acquisition unit and a calibration curve function determination unit, the acquisition unit being configured to acquire absorbance values ​​and concentration values ​​of the substance to be tested, thereby forming a plurality of calibration points, and the calibration curve function determination unit being configured to determine the calibration curve function based on the acquired absorbance values ​​and concentration values ​​of the substance to be tested, thereby fitting a calibration curve of the absorbance values ​​and concentration values ​​of the substance to be tested.

[0024] According to the embodiment of the present invention, the calibration curve function is y=a+bsin(cx+d), y=a+bcos(cx+d), or

[0025] Among them, x is the concentration value, y is the absorbance value, and a, b, c, and d are parameter values.

[0026] As described above, the embodiments of the present invention use a calibration curve function based on a sine function, a cosine function, an inverse function of a sine function, and an inverse function of a cosine function to express the relationship between the optical signal and the concentration of the substance to be measured.

[0027] In the following, the sine function is used as an example to describe the calibration curve function. The sine function is a relatively simple function form in the field of mathematics. Fu Libo proposed that any periodic function can be expressed as a sine function series under certain conditions, indicating that the sine function has excellent characteristics in describing the essential characteristics of nature. The sine function sinx is monotonic in the range, which meets the special requirements of this type of detection. The rate of change of the interval is small at first, gradually increases in the middle, and gradually decreases at the end, which just meets the variation characteristics between absorbance and concentration. The sine function is intuitive to express, and the calculation of the derivative values ​​of each order of its function value is also very simple. Therefore, the embodiment of the present invention selects a calibration curve function based on a sine function, a cosine function, an inverse function of a sine function, and an inverse function of a cosine function to express the relationship between the optical signal and the concentration of the substance to be measured.

[0028] Based on the above calibration curve function, it is necessary to further determine the optimal values ​​of a, b, c and d in the calibration curve function so that the calibration curve function can accurately express the functional relationship between the optical signal and the concentration of the substance to be measured. Accordingly, the calibration curve function determination unit includes a parameter value determination unit. The parameter value determination unit is used to determine the optimal parameter value. Specifically, the parameter value determination unit is configured to determine the parameter value so that the residual square sum of the calibration curve function based on the parameter value and the calibration point is minimized, but the parameter value determination unit is not limited thereto.

[0029] The parameter value determination unit may include: an initial value determination unit, a linearization unit, and an adjustment unit. According to an embodiment of the present invention, the number of the plurality of calibration points ranges from greater than four, preferably from four to ten, but the present invention is not limited thereto.

[0030] The following Table 1 is a set of data including six calibration points. Based on these calibration point data, taking the calibration curve function y=a+bsin(cx+d) as an example, each unit in the parameter value determination unit according to the implementation scheme of the present invention is described in detail.

[0031] Table 1

[0032] Serial number concentration Absorbance 1 0.001 -18 2 12 534 3 22 1219 4 44 3777 5 62 6416 6 94 10679

[0033] According to Table 1, the known calibration point data are (0.001, -18), (12, 534), (22, 1219), (44, 3777), (62, 6416), (94, 10679). In fact, the known calibration point data is a set of (x, y) values.

[0034] The initial value determination unit can be configured to fit the calibration points using the least square method to obtain a least square fitting curve, and use the parameter values ​​of the obtained least square fitting curve as the initial values ​​of the parameter values ​​of the calibration curve function. Based on the calibration point data shown in Table 1, the initial values ​​of the parameter values ​​can be obtained as a=5233.5243, b=5453.6916, c=0.0280, and d=-1.4418.

[0035] When the calibration curve function y=a+bsin(cx+d) is used as the target curve expression, the values ​​of unknown parameters a, b, c and d are calculated using multiple sets of known calibration point data.

[0036] The least squares method is a method for solving linear expressions of unknowns. Since the target curve expression y=f(x)=a+bsin(cx+d) is not a linear expression of unknowns, in order to utilize the least squares method, the nonlinear target curve expression is Taylor expanded to obtain the linear expression of the target curve expression.

[0037] Thus, the linearization unit according to the embodiment of the present invention can be configured to Taylor expand the calibration curve function and generate a linear expression. "Simplification" refers to the process of Taylor expanding a nonlinear expression to generate a linear expression.

[0038] Since Taylor expansion is an approximate expression, it is necessary to iterate and approximate continuously to obtain the optimal result.

[0039] Taylor's first-order expansion is:

[0040]

[0041] Among them, y0 is the y value in the calibration point data. Δa, Δb, Δc, Δd are the increments of the parameter values ​​a, b, c, d. After the initial values ​​of the parameter values ​​a, b, c, d are determined, the increments are continuously iterated to approximate the optimal values.

[0042] When the number of calibration points is n, substituting the calibration point data, the linear equations of Δa, Δb, Δc, and Δd can be obtained as follows:

[0043]

[0044] Written in the form of matrix AX = B, then:

[0045]

[0046]

[0047]

[0048] Among them, x i ,y i It is the calibration point data

[0049]

[0050]

[0051]

[0052]

[0053] Then, the expression of the least squares method is:

[0054] A T ·A·X=A T ·B

[0055] Among them, A T represents the transpose of A.

[0056] In order to accelerate convergence, the adjustment unit is configured to use the Levenberg-Marquardt (LM) method, whose algorithm expression is:

[0057] [A T ·A+λ·diag(A T ·A)]·X=A T ·B

[0058] Among them, λ is the acceleration factor, and the initial value is e -3 , and then by continuously zooming in or out 10 times, calculate whether the residual sum of squares corresponding to different λ is the smallest, and get the optimal λ.

[0059] And among them, diag(A T ·A) is A T A is a diagonal matrix.

[0060] The iterative correction process is:

[0061] a=a-Δa

[0062] b=b-Δb

[0063] c=c-Δc

[0064] d=d-Δd

[0065] In the process of calculating whether the residual sum of squares is the smallest, as an example, the difference between the current residual sum of squares and the previous residual sum of squares is within the control range, specifically, the difference between the two residual sums of squares is set to be less than 0.00001. At this point, it can be considered that the iteration has converged.

[0066] As another example, the residual sum of squares has been decreasing, but it starts to increase after a certain time, which means that the minimum value of the residual sum of squares has been found. Although it may be local, it can only stop. At this time, it depends on the actual fitting effect to decide whether the calculation result is acceptable. Table 2 is a table showing parameter values ​​a, b, c, d and residual sum of squares under multiple iterations as an example. As shown in Table 2 below, the 21st iteration is already the minimum value, the 22nd iteration residual sum of squares increases, and the 23rd iteration actually recovers to the 21st iteration result.

[0067] Table 2

[0068] Iterations a b c d Residual sum of squares 0 5233.5243 5453.6916 0.0280 -1.4418 1 5641.8007 5647.6083 0.0266 -1.4952 0.072717 2 5829.4349 5797.6211 0.0262 -1.5090 0.045735 3 5918.5122 5872.9654 0.0259 -1.5095 0.035095 4 5969.7807 5924.4798 0.0256 -1.5050 0.029506 5 6004.4580 5964.6278 0.0254 -1.4991 0.025989 6 6030.8905 5998.1098 0.0252 -1.4932 0.023464 7 6052.6240 6027.0369 0.0251 -1.4878 0.021488 8 6071.2919 6052.5155 0.0249 -1.4829 0.019857 9 6087.7234 6075.2121 0.0248 -1.4785 0.018467 10 6102.3895 6095.5776 0.0247 -1.4746 0.017261 11 6115.5902 6113.9445 0.0246 -1.4712 0.016200 12 6127.5370 6130.5731 0.0245 -1.4681 0.015261 13 6138.3905 6145.6742 0.0244 -1.4653 0.014423 14 6148.2795 6159.4236 0.0243 -1.4628 0.013729 15 6157.3109 6171.9702 0.0243 -1.4606 0.013497 16 6165.5752 6183.4414 0.0242 -1.4585 0.013289 17 6173.1505 6193.9474 0.0241 -1.4567 0.013111 18 6180.1044 6203.5842 0.0241 -1.4550 0.012940 19 6186.4965 6212.4358 0.0240 -1.4535 0.012785 20 6192.3790 6220.5762 0.0240 -1.4521 0.012643 21 6197.7984 6228.0711 0.0240 -1.4508 0.012605 22 6202.7960 6234.9786 0.0239 -1.4496 0.012646 23 6197.7984 6228.0711 0.0240 -1.4508 0.012605

[0069] As another example, the number of iterations reaches a given value. The given value may be 100 times, but is not limited thereto. When the number of iterations exceeds the given value, the iteration stops, and based on experience, it can be checked whether the iteration result is acceptable. If the iteration result is acceptable, the iteration is completed; if the iteration result is unacceptable, further iteration is required.

[0070] Thus, the final value of the parameter value calculated by the calibration model for determining the concentration of the substance to be measured according to an embodiment of the present invention is:

[0071] a=6197.7984, b=6228.0711, c=0.0240, d=-1.4508

[0072] Fitting curve expression: y=6197.7984+6228.0711sin(0.024x+-1.4508).

[0073] Figure 2A is a fitting curve after 1 iteration as an example, Figure 2B is a fitting curve after the iteration is terminated as an example. Figure 2A and Figure 2B , the fitting curve after the iteration is terminated is closer to the calibration point data.

[0074] FIG. 3A to FIG. 3H is a graph showing the calibration point fitting using different calibration models based on the calibration point data shown in Table 1. Figure 3A It is a graph showing the fitting of the calibration points using the broken line model. Figure 3B It is a curve graph in which calibration points are fitted using a calibration model for determining the concentration of a substance to be measured according to an embodiment of the present invention. FIG. 3C to FIG. 3E These are curve graphs that fit the calibration points using the Logit-Log 3P model, the logit-log 4P model, and the logit-log 5P model respectively. Figure 3F It is a curve graph that uses an exponential model to fit the calibration points. Figure 3G It is a curve graph that uses the cubic spline function model to fit the calibration points. Figure 3H It is a curve graph that uses the Akima spline function model to fit the calibration points.

[0075] Among them, cubic spline and Akima spline are a piecewise fitting mathematical model, that is, each segment (between two calibration points) of the curve is an independent function expression. This fitting model can achieve complete accuracy for each calibration point; while other models are overall fitting mathematical models, that is, one function expression corresponds to all calibration points, so it cannot guarantee the complete accuracy of each calibration point. However, the spline model only represents mathematical accuracy, not the accuracy of biochemical reactions. Therefore, non-spline overall fitting methods also have their value.

[0076] from FIG. 3A to FIG. 3H It can be seen that for this set of calibration data, Figure 3B The sine function model according to the embodiment of the present invention shown has a better fitting effect. FIG. 3A to FIG. 3H ,Apart from Figure 3G , Figure 3H In addition to the two piecewise spline models shown, Figure 3B As an overall fitting model, the performance is optimal at the given set of calibration points. Therefore, in addition to the known calibration function models, the embodiments of the present invention provide a selection of a calibration model that can accurately express the functional relationship between the optical signal and the concentration of the substance to be measured.

[0077] Figure 4 FIG. 4 is a flow chart showing a method for determining a calibration model for determining the concentration of a substance to be measured according to an embodiment of the present invention. Figure 4 As shown, the method for the calibration model for determining the concentration of the substance to be tested according to the present invention includes: S410: collecting the absorbance value and concentration value of the substance to be tested to form a plurality of calibration points, S420, determining the calibration curve function based on the collected absorbance value and concentration value of the substance to be tested, thereby fitting a calibration curve of the absorbance value and concentration value of the substance to be tested.

[0078] In S420, the calibration curve function is y=a+bsin(cx+d), y=a+bcos(cx+d), or Among them, x is the concentration value, y is the absorbance value, and a, b, c, and d are parameter values.

[0079] Determining the calibration curve function based on the sine function includes: determining the parameter value so that the residual sum of squares between the calibration curve function based on the parameter value and the calibration point is minimized. Specifically, determining the parameter value includes: fitting the calibration point using the least squares method to obtain the least squares fitting curve, and using the parameter value of the least squares fitting curve obtained as the initial value of the parameter value of the calibration curve function; simplifying the calibration curve function into a linear expression about the increment of the parameter value using the Taylor expansion; solving the linear expression using the Levenberg-Maguire (LM) method, and iteratively correcting the increment of the parameter value so that the residual sum of squares is minimized.

[0080] Preferably, the number of the plurality of calibration points ranges from four to ten.

[0081] According to an embodiment of the present invention, the calibration model and method for determining the concentration of the substance to be tested can accurately express the functional relationship between the optical signal and the concentration of the substance to be tested, and combine the detected optical signal to accurately determine the concentration value of the substance to be tested.

[0082] In addition to the known calibration function models, the embodiments of the present invention use calibration curve functions based on sine function, cosine function, inverse function of sine function, and inverse function of cosine function, providing a selection of calibration models and methods that can accurately express the functional relationship between the optical signal and concentration of the substance to be measured. The calibration model and method of the present invention can produce better fitting effects in some systems.

[0083] The various embodiments of the invention are not an exhaustive list of all possible combinations, but are intended to describe representative aspects of the invention, and what is described in various embodiments may be applied independently or in combinations of two or more.

[0084] The description presented in the above exemplary embodiments is only used to illustrate the technical solution of the present invention, and is not intended to be exhaustive, nor is it intended to limit the present invention to the precise form described. Obviously, it is possible for a person of ordinary skill in the art to make many changes and variations based on the above teachings. The exemplary embodiments are selected and described to explain the specific principles of the present invention and its practical application, so that other technicians in the field can easily understand, implement and use the various exemplary embodiments of the present invention and its various selected forms and modified forms. The scope of protection of the present invention is intended to be limited by the attached claims and their equivalent forms.

Claims

1. A calibration model for determining the concentration of a substance to be measured, characterized in that: include: A collection unit configured to collect absorbance values ​​and concentration values ​​of the substance to be tested, thereby forming a plurality of calibration points; A calibration curve function determination unit, configured to determine a calibration curve function based on the collected absorbance value and concentration value of the substance to be tested, thereby fitting a calibration curve with respect to the absorbance value and concentration value of the substance to be tested; The calibration curve function determination unit includes a parameter value determination unit, the parameter value determination unit is configured to determine the parameter value so that the residual square sum between the calibration curve function based on the parameter value and the calibration point is minimized, and the parameter value determination unit includes: An initial value determination unit is configured to fit the calibration points using the least square method to obtain a least square method fitting curve, and use the parameter values ​​of the obtained least square method fitting curve as initial values ​​of the parameter values ​​of the calibration curve function; A linearization unit configured to simplify the calibration curve function into a linear expression with respect to increments of parameter values ​​using a Taylor expansion; An adjustment unit is configured to solve the linear expression using a Levenberg-Maguire method and iteratively correct the increment of the parameter value so that the residual square sum is minimized; The calibration curve function is y=a+bsin(cx+d), y=a+bcos(cx+d), or x is the concentration value, y is the absorbance value, and a, b, c, and d are parameter values.

2. The calibration model for determining the concentration of a substance to be measured according to claim 1, characterized in that: The number of the plurality of calibration points ranges from four to ten.

3. A calibration method for determining the concentration of a substance to be measured, characterized in that: include: Collecting the absorbance value and concentration value of the substance to be tested, thereby forming multiple calibration points; Determine the calibration curve function based on the collected absorbance value and concentration value of the substance to be tested, so as to fit the calibration curve about the absorbance value and concentration value of the substance to be tested; Wherein, determining the calibration curve function based on the sine function includes: determining a parameter value so that the residual square sum between the calibration curve function based on the parameter value and the calibration point is minimized, and determining the parameter value includes: The calibration points are fitted using the least square method to obtain a least square fitting curve, and the parameter values ​​of the obtained least square fitting curve are used as initial values ​​of the parameter values ​​of the calibration curve function; The calibration curve function is simplified into a linear expression about the increment of parameter value by using Taylor expansion; The linear expression is solved using the Levenberg-Maguire method, and the increment of the parameter value is iteratively corrected to minimize the residual sum of squares; The calibration curve function is y=a+bsin(cx+d), y=a+bcos(cx+d), or x is the concentration value, y is the absorbance value, and a, b, c, and d are parameter values.

4. The calibration method for determining the concentration of a substance to be measured according to claim 3, characterized in that: The number of the plurality of calibration points ranges from four to ten.

Images