HIV infection prediction model and construction method thereof

Abstract

The invention discloses an HIV infection prediction model and a construction method thereof, and relates to the technical field of HIV virus detection. The formula of the HIV infection prediction model disclosed by the invention is as follows: logit (P) =-2.54 + 1.47 * A-0. 72 * B + 0.76 * C + 7.42 * D, wherein the letter meanings in the formula are as follows: A is a sex value, male is 0, and female is 1; b is an educational degree value which is 0 above the high school and 1 below the high school; c represents whether the submitted crowd is a VCT value or not, VCT is 0, and non-VCT is 1; d is a recheck result value of the preliminary screening test, the reaction is 0 in two times, and the reaction is 1 in only one time. The prediction model provided by the invention predicts an immunoblotting HIV-1 antibody confirmation result before confirmation experiments, and makes an optimal choice in two confirmation experiments, thereby solving the technical problems of long detection time and high detection cost in existing HIV infection confirmation.

Description

Technical Field

[0001] The present invention relates to the technical field of HIV virus detection, and particularly relates to an HIV infection prediction model and a method for constructing the same. Background Art

[0002] AIDS is a disease that seriously endangers human health and even life, and is a threat to social public health. How to effectively detect AIDS virus infected persons is an issue that cannot be ignored in the prevention and control of AIDS. Regarding the detection problem of AIDS infection, the recommended AIDS detection strategy is to first conduct an HIV screening test, and if there is a reaction in the screening, a confirmation test is supplemented. The immunoblotting method is the most widely used HIV antibody confirmation test at present because of its strong specificity and high sensitivity. However, due to the influence of multiple factors, it is prone to the results of HIV-1 antibody indeterminacy and false negatives. Such results cannot make a clear judgment on whether a person is infected with HIV. In view of this, HIV-1 nucleic acid detection is added as a confirmation test. This method has higher sensitivity and has a supplementary value for the results of immunoblotting that are indeterminate or false negative. However, due to factors such as relatively harsh experimental conditions and high costs, it cannot be widely carried out at the grass-roots level. In actual situations, if each sample to be confirmed is first subjected to the immunoblotting method and then further subjected to nucleic acid detection, it will increase the time for HIV infection diagnosis and the detection cost, which is also not conducive to the progress of AIDS prevention and control work. Summary of the Invention

[0003] The present invention provides an HIV infection prediction model. Based on the initial screening test results and basic information such as epidemiology, this prediction model predicts the results of the HIV-1 antibody confirmation test by immunoblotting before the confirmation experiment. Doctors can make the best choice between the two confirmation experiments according to the results predicted by the model: if the model predicts that the immunoblotting result of a certain patient is positive, the immunoblotting method can be directly selected, and a positive diagnosis can be directly obtained with a high probability. If a positive diagnosis is not obtained, then the HIV-1 nucleic acid detection is selected for HIV confirmation; if the model predicts that the immunoblotting result of a certain patient is negative or indeterminate, the HIV-1 nucleic acid detection can be directly selected for HIV confirmation, thus solving the technical problems of long detection time and high detection cost in the existing HIV infection diagnosis.

[0004] The technical solution adopted by the present invention is as follows:

[0005] An HIV infection prediction model, and the formula of the HIV infection prediction model is:

[0006] logit(P)=-2.54 + 1.47×A - 0.72×B + 0.76×C + 7.42×D;

[0007] Among them, the meanings of the letters in the formula are as follows: A is the gender value, 0 for male and 1 for female; B is the education level value, 0 for above high school and 1 for high school and below; C is the value indicating whether the tested population is VCT (Voluntary Counseling and Testing for AIDS), 0 for VCT and 1 for non-VCT; D is the value of the initial and retest results, 0 for having a reaction in both tests and 1 for having a reaction only once; the diagnostic cut-off value of logit(P) is 0.285, that is, when logit(P) is greater than 0.285, the model prediction result is "HIV-1 antibody negative or indeterminate".

[0008] The method for constructing the HIV infection prediction model includes the following steps:

[0009] (1) Collect information of no less than 1000 HIV presumptive reactive patients to be confirmed, including basic and epidemiological information and test information. Epidemiological information includes gender, age, occupation, ethnicity, and marital status. Test information includes methodology, reagent test type, and initial test result. Those with incomplete information above are excluded;

[0010] (2) Use the immunoblotting method to conduct HIV antibody confirmation tests on the samples retained in step (1) to obtain three confirmation test results: positive, indeterminate, and negative;

[0011] (3) Randomly divide the data of all the above test results and patient information into a training set and a validation set at a ratio of 7:3. The training set data is used to construct the model, and after the model is constructed, it is verified in the validation set;

[0012] (4) Construct the model: Use the training set data, with the immunoblotting results (positive, negative / indeterminate) as independent variables and the basic and epidemiological information and test information as dependent variables. Use univariate Logistic regression analysis and multivariate Logistic regression analysis. Finally, determine that gender, education level, whether the tested population is VCT, and the retest result of the initial screening test are the influencing factors for HIV-1 antibody negative and indeterminate. Then, incorporate these four predictors of gender, education level, whether the tested population is VCT, and the retest result of the initial screening test into the prediction model to obtain the HIV infection prediction model formula;

[0013] (5) Model verification: Use the validation set data to verify the feasibility of the HIV infection prediction model formula obtained in step (4).

[0014] Furthermore, in step (1), information of 2468 HIV presumptive reactive patients to be confirmed was collected.

[0015] Furthermore, in step (3), the data of the test results and patient information are randomly divided into a training set and a validation set at a ratio of 7∶3.

[0016] In summary, compared with the prior art, the present invention has the following advantages and beneficial effects:

[0017] 1. The HIV infection prediction model provided by the present invention has good calibration (χ² = 12.094, P = 0.098) and diagnostic value (AUC value is 0.874, sensitivity is 0.99, and specificity is 0.63).

[0018] 2. The HIV infection prediction model provided by the present invention can predict the Western blot HIV-1 antibody confirmation result before the confirmation experiment only based on the primary screening test results and basic information such as epidemiology. Doctors can make the best choice in the two confirmation experiments according to the results predicted by the model: if the model predicts that a patient's Western blot result is positive, the Western blot can be directly selected, and a positive diagnosis can be obtained with a high probability. If a positive diagnosis is not obtained, the HIV-1 nucleic acid test can be selected for HIV confirmation; if the model predicts that a patient's Western blot result is negative or indeterminate, the HIV-1 nucleic acid test can be directly selected for HIV confirmation. For this part of the patients who choose the HIV-1 nucleic acid test method through the prediction model, the original two-step confirmation experiment method is reduced to one step, which simplifies the detection process, shortens the time for HIV detection, reduces the detection cost, and is of great significance for the early diagnosis and treatment of patients in the acute HIV infection period. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is the model nomogram drawn during the model construction process of Example 1;

[0020] Figure 2 It is the fitting diagram of the standard curve, prediction curve and ideal curve drawn during the model verification process of Example 1;

[0021] Figure 3 It is the ROC curve diagram drawn using logit(P) during the model verification process of Example 1. DETAILED DESCRIPTION OF THE INVENTION

[0022] The present invention will be specifically described below in combination with the specific embodiments and examples, and the advantages and various effects of the present invention will be presented more clearly therefrom. Those skilled in the art should understand that these specific embodiments and examples are used to illustrate the present invention, rather than to limit the present invention.

[0023] Throughout the specification, unless otherwise specifically stated, the terms used herein should be understood as having the meanings commonly used in the art. Therefore, unless otherwise defined, all technical and scientific terms used herein have the same meaning as the general understanding of those skilled in the art to which the present invention belongs. In case of conflict, this specification shall prevail.

[0024] Unless otherwise specified, various raw materials, reagents, instruments, and equipment used in the present invention can be obtained through market purchases or can be prepared by existing methods.

[0025] Example 1

[0026] In this example, an HIV infection prediction model was constructed according to the following steps:

[0027] (1) Basic and epidemiological information (including gender, age, occupation, ethnicity, marital status, etc.) and test information of the initial screening test (including methodology, reagent test type, initial screening test results, etc.) of 2468 patients awaiting confirmation with reactive initial HIV screening were collected. Those with incomplete information were excluded.

[0028] (2) The above samples were subjected to HIV antibody confirmation using the immunoblotting method. Positive results can directly diagnose HIV infection, and negative and indeterminate results were further subjected to HIV nucleic acid testing; the confirmation test results were combined with the initial information for further statistical analysis.

[0029] (3) All the above data were randomly divided into a training set and a validation set at a ratio of 7:3. The training set data were used to construct the model, and the constructed model was verified in the validation set.

[0030] (4) Model construction: Using the immunoblotting results (positive, indeterminate / negative) as independent variables and the basic and epidemiological information and test information of the initial screening test as dependent variables, univariate Logistic regression analysis and multivariate Logistic regression analysis were used. Finally, gender, education level, the tested population (VCT), and retest results were determined as the influencing factors for negative and indeterminate HIV-1 antibodies (P<0.05). These four predictors were incorporated into the prediction model to obtain the regression equation logit(P)= -2.54 + 1.47×gender (male = 0, female = 1) - 0.72×education level (above high school = 0, high school and below = 1) + 0.76×tested population (VCT = 0, non-VCT = 1) + 7.42×retest results (both reactions = 0, only one reaction = 1). The corresponding model nomogram was drawn as shown in Figure 1 ;

[0031] (5) Model verification: The model constructed with the training set was brought into the validation set data for verification. The Hosmer Lemeshow test showed that χ² = 12.094 and P = 0.098, indicating a high goodness of fit of the model; the calibration curve showed that the prediction curve and the calibration curve were close to the ideal curve, indicating a high consistency between the predicted probability and the actual probability, showing that the model has good calibration, as shown in Figure 2; The ROC curve was plotted using logit(P), with an AUC value of 0.874 (95% CI: 0.826 - 0.922). The diagnostic cut-off value was 0.285 (i.e., logit(P) greater than 0.285 indicates a "negative or indeterminate result"). The sensitivity was 0.99 and the specificity was 0.63, showing that the model has good discrimination, as shown in Figure 3 .

[0032] Finally, it should also be noted that the term "comprising", "including" or any other variation is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements but also other elements not expressly listed, or elements inherent to such process, method, article or device. Although the preferred embodiments of the present invention have been described, those skilled in the art can make additional changes and modifications once they learn the basic creative concept. Therefore, the appended claims are intended to be construed as including the preferred embodiments and all changes and modifications falling within the scope of the present invention.

[0033] The above-described embodiments merely represent the specific implementation manners of the present application, and the description thereof is relatively specific and detailed, but it should not be construed as a limitation on the protection scope of the present application. It should be noted that for those of ordinary skill in the art, without departing from the concept of the technical solution of the present application, several changes and improvements can still be made, and these all fall within the protection scope of the present application.

Claims

1. An HIV infection prediction model, characterized in that, The formula of the HIV infection prediction model is as follows: logit(P) = -2.54 + 1.47×A - 0.72×B + 0.76×C + 7.42×D; Among them, the meanings of the letters in the formula are as follows: A is the gender value, 0 for male and 1 for female; B is the education level value, 0 for above high school and 1 for high school and below; C is the value indicating whether the tested population is VCT, 0 for VCT and 1 for non-VCT; D is the retest result value of the preliminary screening test, 0 for positive reactions in both tests and 1 for positive reaction in only one test; the diagnostic cut-off value of logit(P) is 0.285, that is, when logit(P) is greater than 0.285, the model prediction result is "HIV-1 antibody negative or uncertain".

2. The method for constructing the HIV infection prediction model according to claim 1, wherein, It includes the following steps: (1) Collect the information of no less than 1000 patients to be confirmed with positive HIV preliminary screening, including basic and epidemiological information and test information. The epidemiological information includes gender, age, occupation, ethnicity, and marital status. The test information includes methodology, reagent test type, and preliminary screening test results. Those with incomplete above information are excluded; (2) Use the immunoblotting method to conduct the HIV antibody confirmation test on the samples retained in step (1) to obtain three confirmation test results: positive, uncertain, and negative; (3) Randomly divide the data of all the test results and patient information obtained above into a training set and a validation set according to a certain ratio. The training set data is used to construct the model, and after the model is constructed, it is verified in the validation set; (4) Construct the model: Use the training set data, with the immunoblotting results (positive, negative / uncertain) as independent variables and the basic and epidemiological information and test information as dependent variables. Use univariate Logistic regression analysis and multivariate Logistic regression analysis. Finally, determine that gender, education level, whether the tested population is VCT, and the retest result of the preliminary screening test are the influencing factors for HIV-1 antibody negative and uncertain. Then incorporate these four predictors of gender, education level, whether the tested population is VCT, and the retest result of the preliminary screening test into the prediction model to obtain the HIV infection prediction model formula; (5) Model verification: Use the validation set data to verify the feasibility of the HIV infection prediction model formula obtained in step (4).

3. The method for constructing an HIV infection prediction model according to claim 2, wherein, In step (1), the information of 2468 patients to be confirmed with positive HIV preliminary screening was collected.

4. The method for constructing an HIV infection prediction model according to claim 2, wherein In step (3), the data of all the test results and patient information obtained above were randomly divided into a training set and a validation set according to a ratio of 7:3.

Images