WO2022022663A1 - Method for constructing alcohol tolerance prediction model - Google Patents

Abstract

Disclosed is a method for constructing an alcohol tolerance prediction model, the method comprising: S1, acquiring relationships between alcohol consumption capacities and alcohol consumption amounts of respective samples, dividing the alcohol consumption amounts into a first preset number of alcohol consumption tiers, and establishing a first database according to the alcohol consumption tiers and the relationships between the alcohol consumption capacities and the alcohol consumption amounts of the samples; S2, acquiring genetic data of the samples and formatting the genetic data; and S3, selecting, according to the formatted genetic data of the samples and the first database, a machine learning model, and constructing a first alcohol consumption amount prediction model. The beneficial effects are as follows: an alcohol tolerance prediction model is constructed by means of machine learning, so as to provide a standard for determining the alcohol consumption amount and to quantify the alcohol consumption capacity of an individual, thereby facilitating effective prediction of alcohol tolerance levels of users.

Description

A method of constructing an alcohol volume prediction model technical field

The invention relates to the technical field of biological genes, in particular to a method for constructing an alcohol quantity prediction model.

Background technique

After alcohol enters the human body, it enters the blood circulation directly through the biofilm through the oral cavity, esophagus, stomach, intestine and other organs, and is quickly transported to various tissues and organs of the body for metabolism and utilization. There are two enzymes in the human body for alcohol metabolism: ethanol is oxidized to acetaldehyde under the catalysis of alcohol dehydrogenase; acetaldehyde is converted to acetic acid by acetaldehyde dehydrogenase. Alcohol metabolism is mainly completed by two enzymes (alcohol dehydrogenase and aldehyde dehydrogenase), and the difference in drinking ability (alcohol amount) between individuals is mainly determined by the activities of these two enzymes, and the activity of the enzymes is determined by the gene In the final analysis, the amount of alcohol a person can drink is determined by genetics.

As an important drink in some people's lives, wine has derived various wine cultures and has become an indispensable existence in specific occasions. However, studies have shown that drinking is not suitable for everyone, and excessive drinking is extremely harmful to the body; and the drinking ability of different people is also quite different. It is very important to have a correct understanding of one's ability to metabolize alcohol and have a healthy drinking standard. However, there is a lack of drinking standards in the prior art, so it cannot effectively predict the drinking amount of a user.

SUMMARY OF THE INVENTION

The present invention aims to solve one of the technical problems in the above technologies at least to a certain extent. To this end, the purpose of the present invention is to propose a method for constructing an alcohol consumption prediction model, constructing an alcohol consumption prediction model through machine learning, and by collecting the relationship between the drinking ability and the drinking quantity of the sample, the relationship between the drinking ability and the drinking quantity of the sample is calculated. 、Drinking rank to establish the first database to improve the accuracy of the data, collect the genetic data of the sample and format the genetic data, match the genetic data of the sample with the data in the first database, and construct the first drinking through the machine learning model. The quantity prediction model provides drinking quantity judgment criteria, quantifies the drinking ability of individuals, and is conducive to effective alcohol consumption prediction for users.

In order to achieve the above purpose, the embodiment of the present invention proposes a method for constructing an alcohol volume prediction model, including:

S1. Obtain the relationship between the drinking ability and the drinking amount of the sample, divide the sample into a first preset number of drinking rank according to the drinking amount, and establish a first database according to the relationship between the drinking ability and the drinking amount of the sample, and the drinking rank;

S2. Obtain the genetic data of the sample and format the genetic data;

S3. Select a machine learning model to construct a first alcohol consumption prediction model according to the gene data of the formatted sample and the first database.

According to the construction method of the alcohol consumption prediction model proposed by the present invention, the drinking ability and the drinking quantity of the sample are obtained through the questionnaire survey method, and the data is analyzed to obtain the corresponding relationship between the drinking ability and the drinking quantity, and the drinking quantity is divided into the first Presetting a number of drinking grades, establishing a first database, classifying drinking levels into drinking levels, and specifically quantifying drinking ability is conducive to giving more valuable drinking suggestions. Obtain the genetic data of the sample and format the genetic data. According to the formatted sample genetic data and the first database, a machine learning model is used to construct the first drinking volume prediction model, which provides the drinking volume judgment standard and quantifies the drinking ability of the individual, which is beneficial to Effective alcohol prediction for users.

According to some embodiments of the present invention, the acquiring genetic data of the sample and formatting the genetic data includes:

S21, collecting saliva from the sample;

S22, performing DNA extraction according to the saliva of the sample, and performing gene sequencing on the extracted DNA;

S23. Process the gene data after gene sequencing to obtain the genotype of the locus related to the alcohol consumption of each sample;

S24. Format the gene loci into numbers according to the genotype.

According to some embodiments of the present invention, genetic locus screening is performed on the genetic data of the formatted sample, including:

S241, calculate the characteristic value of each gene locus respectively and each data subset obtained after the described first database is divided and the purity promotion value or the uncertainty reduction value of the data set before division;

S242. Select the gene locus N with the maximum purity improvement value or the maximum uncertainty reduction value and the characteristic value n of the gene locus N, wherein the gene locus N is used as a node, according to the gene locus N The grouping of the eigenvalue n splits the first database into two sub-data sets;

S243. In the two sub-data sets in turn, calculate the purity improvement value or the uncertainty reduction value of the characteristic value of each gene locus in the sub-data set; select the gene locus M and the uncertainty reduction value with the largest purity improvement value or the largest uncertainty reduction value. The eigenvalue m of the gene locus M, wherein the gene locus M is used as a sub-node, and the sub-data set is split again according to the grouping of the eigenvalue m of the gene locus M;

S244. When it is determined that the purity of the divided sub-data set is greater than the preset purity threshold or the uncertainty value is less than the preset uncertainty threshold, stop the splitting, and finally obtain the gene locus related to the alcohol consumption and the gene locus and the gene locus The relationship between drinking grades.

According to some embodiments of the present invention, the loci related to alcohol consumption include the rs1229984 locus and the rs671 locus, wherein the rs1229984 locus is located on the ADH1B gene, and when the result of the rs1229984 locus is TT type, The activity of alcohol dehydrogenase is strong, and the metabolism of ethanol is fast; the result is CT type, the activity of alcohol dehydrogenase is moderate, and the rate of alcohol metabolism is moderate; the result is that the activity of alcohol dehydrogenase is weak, and the rate of alcohol metabolism is slow; the rs671 gene locus is located in In the ALDH2 gene, when the rs671 gene locus was GG type, the activity of acetaldehyde dehydrogenase was strong, and the acetaldehyde metabolism was fast; when the result was GA\AA type, the activity of acetaldehyde dehydrogenase was weak, and the metabolism of acetaldehyde was slow.

According to some embodiments of the present invention, after constructing the first alcohol consumption prediction model, the method further includes:

S4. Re-divide the drinking amount into a second preset number of drinking grades, and establish a second database according to the relationship between the drinking ability of the sample and the drinking amount, and the second preset number of drinking grades;

S5. Select a machine learning model to reconstruct a drinking amount prediction model according to the loci of the sample related to the drinking amount and the second database, and obtain a second drinking amount prediction model.

According to some embodiments of the present invention, DNA extraction is performed according to the saliva of the sample, including:

S421. Put 0.2 ml of saliva in a centrifuge tube, add 600 μL of 0.01 mol/L PBS solution, and centrifuge at high speed at 10000*g for 3 min in a centrifuge;

S422, get the precipitate and add 50ul, 5mol/l potassium iodide solution, 75ul, 0.9% sodium chloride solution, 120ul phenol:chloroform (20:13) solution, shake and shake for 1min, and centrifuge at high speed at 10000*g for 3min by centrifuge;

S423, take 80 ul of the supernatant, add 80 ul of isopropanol, shake for 30 s, and centrifuge at a high speed of 10000*g for 3 min in a centrifuge;

S424, taking the precipitate, adding 500 μl of absolute ethanol to wash, and centrifuging at 10000*g for 3 min in a centrifuge;

S425, take the precipitate to dry at room temperature, and dissolve it with TE buffer.

According to some embodiments of the present invention, it also includes:

S71. Acquire second information that affects the drinking amount of the user, where the second information includes: disease history, drinking type, drinking degree, and drinking frequency;

S72. Modify the second alcohol consumption prediction model according to the second information to obtain a third alcohol consumption prediction model.

Other features and advantages of the present invention will be set forth in the description which follows, and in part will be apparent from the description, or may be learned by practice of the invention. The objectives and other advantages of the invention may be realized and attained by the structure particularly pointed out in the written description and drawings.

The technical solutions of the present invention will be further described in detail below through the accompanying drawings and embodiments.

Description of drawings

The accompanying drawings are used to provide a further understanding of the present invention, and constitute a part of the specification, and are used to explain the present invention together with the embodiments of the present invention, and do not constitute a limitation to the present invention. In the attached image:

Fig. 1 is the flow chart of the construction method of a kind of alcohol quantity prediction model according to an embodiment of the present invention;

FIG. 2 is a flowchart of processing genetic data of a sample according to an embodiment of the present invention;

FIG. 3 is a flowchart of the screening of genetic loci related to alcohol consumption according to an embodiment of the present invention;

4 is a flowchart of establishing a second alcohol consumption prediction model according to an embodiment of the present invention;

5 is a flow chart of DNA extraction from saliva according to an embodiment of the present invention;

FIG. 6 is a schematic diagram of a decision tree of drinking quantity-related genes and drinking rank according to an embodiment of the present invention.

detailed description

The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, but not to limit the present invention.

The following describes a method for constructing an alcohol volume prediction model proposed by an embodiment of the present invention with reference to FIG. 1 to FIG. 6 .

Fig. 1 is the flow chart of the construction method of a kind of alcohol quantity prediction model according to an embodiment of the present invention; As shown in Fig. 1, the embodiment of the present invention proposes a kind of construction method of alcohol quantity forecast model, including:

S1. Obtain the relationship between the drinking ability and the drinking amount of the sample, divide the sample into a first preset number of drinking rank according to the drinking amount, and establish a first database according to the relationship between the drinking ability and the drinking amount of the sample, and the drinking rank;

S2. Obtain the genetic data of the sample and format the genetic data;

S3. Select a machine learning model to construct a first alcohol consumption prediction model according to the gene data of the formatted sample and the first database.

According to the construction method of the alcohol consumption prediction model proposed by the present invention, the drinking ability and the drinking quantity of the sample are obtained through the questionnaire survey method, and the data is analyzed to obtain the corresponding relationship between the drinking ability and the drinking quantity, and the drinking quantity is divided into the first Presetting a number of drinking grades, establishing a first database, classifying drinking levels into drinking levels, and specifically quantifying drinking ability is conducive to giving more valuable drinking suggestions. Obtain the genetic data of the sample and format the genetic data, select a machine learning model to construct the first alcohol consumption prediction model according to the formatted sample genetic data and the first database, and construct an alcohol consumption prediction model through the decision tree model in the machine learning, It provides drinking quantity judgment criteria and quantifies the drinking ability of individuals, which is conducive to effective alcohol consumption prediction for users. The machine learning model includes: at least one of linear classification, linear regression, support vector machine (SVM), decision tree classification model, naive Bayes, random forest, and neural network model. Among them, the decision tree classification model is easy to understand and implement, and the training results are easy to deduce the corresponding logical expressions, which has good readability.

FIG. 2 is a flowchart of processing genetic data of a sample according to an embodiment of the present invention; as shown in FIG. 2 , the acquiring and formatting the genetic data of the sample includes:

S21, collecting saliva from the sample;

S22, performing DNA extraction according to the saliva of the sample, and performing gene sequencing on the extracted DNA;

S23. Process the gene data after gene sequencing to obtain the genotype of the locus related to the alcohol consumption of each sample;

S24. Format the gene loci into numbers according to the genotype.

The working principle and beneficial effects of the above technical solutions: DNA extraction, gene sequencing, and genotyping are performed on the saliva of the sample to obtain the genetic data of the sample; the gene sequencing methods include: chip sequencing, second-generation sequencing, third-generation sequencing, PCR sequencing, At least one of panel sequencing. Finally, the genotypes of the loci related to the alcohol consumption of each sample are obtained. In order to effectively calculate the influence of the loci on the alcohol consumption, the loci are formatted into numbers according to the genotype. Illustratively, wild type is 0, heterozygous mutant is 1, and homozygous mutant is 2. For example, at the rs1229984 locus, CC is a homozygous mutant, formatted as a number 2; TT is a wild type, formatted as a number 0; CT is a heterozygous mutant, formatted as a number 1; such as at the rs671 locus On the dot, AA is a homozygous mutant, formatted as a number 2; GG is a wild type, formatted as a number 0; AG is a heterozygous mutant, formatted as a number 1.

In one embodiment, gene locus screening is performed on the genetic data of the formatted sample, including:

S241, respectively calculating the characteristic value of each gene locus and dividing the first database to obtain the purity improvement value or uncertainty reduction value of each data subset and the data set before division;

S242. Select the gene locus N with the maximum purity improvement value or the maximum uncertainty reduction value and the characteristic value n of the gene locus N, wherein the gene locus N is used as a node, according to the gene locus N The grouping of the eigenvalue n splits the first database into two sub-data sets;

S243. In the two sub-data sets in turn, calculate the purity improvement value or the uncertainty reduction value of the characteristic value of each gene locus in the sub-data set; select the gene locus M and the uncertainty reduction value with the largest purity improvement value or the largest uncertainty reduction value. The eigenvalue m of the gene locus M, wherein the gene locus M is used as a sub-node, and the sub-data set is split again according to the grouping of the eigenvalue m of the gene locus M;

S244. When it is determined that the purity of the divided sub-data set is greater than the preset purity threshold or the uncertainty value is less than the preset uncertainty threshold, stop the splitting, and finally obtain the gene locus related to the alcohol consumption and the gene locus and the gene locus The relationship between drinking grades.

The working principle and beneficial effect of the above technical solution: the method for measuring the purity and uncertainty of the data set before and after dividing the data set includes calculating at least one parameter in the information gain, the information gain rate, and the Gini coefficient. Specifically, the method is determined according to the Gini coefficient. In the method of purity and uncertainty, the larger the Gini coefficient, the higher the uncertainty of the data, and the lower the sample purity, indicating that the proportion of the target sample in the data set to the total sample is smaller; the smaller the Gini coefficient, the uncertainty of the data. The lower the stability, the higher the sample purity, which means that the proportion of the target sample in the total sample is higher; when the Gini coefficient is less than the preset value, it means that the purity of the divided sub-data set is greater than the preset purity threshold or uncertainty When the value is less than the preset uncertainty threshold, the splitting is stopped, and the gene locus related to the drinking amount and the relationship between the locus and the drinking rank are finally obtained. Exemplarily, when the Gini coefficient is equal to 0, all samples in the dataset are of the same class.

In one embodiment, as shown in FIG. 6 , it is determined whether the result of the rs671 gene locus of the sample is GG, that is, it is determined whether the format of the rs671 gene locus of the sample is 0, and according to whether the rs671 gene locus is 0, the first The database is divided into two data sets, the first data set and the second data set, wherein the rs671 gene locus of the samples in the first data set is GG, and the rs671 gene locus results of the samples in the second data set are AA and AG; Calculate the Gini coefficient of the eigenvalues of each gene locus in the first data set and the second data set, and select the gene locus A with the smallest calculated Gini coefficient and the eigenvalue a of the gene locus A, wherein the all The gene locus A is used as a child node, and the data set is split again according to the grouping of the characteristic value a of the gene locus A; for example, in the first data set, it is judged whether the rs1229984 gene locus of the sample is CC or CT , when the judgment is False, the sample rs1229984 locus is TT, that is, the sample rs671 locus in this grouping is GG, and the sample rs1229984 locus is TT. As shown in Table 1, the drinking segment is 8 segments. When it is determined that each group is a sample of the same type, that is, the Gini coefficient is 0, the splitting is stopped, and the gene loci related to the amount of drinking and the relationship between the locus and the drinking rank are finally obtained. The genes related to alcohol consumption are divided into corresponding drinking grades according to their gene types, which is convenient for memory, and can accurately reflect the corresponding relationship between gene types and alcohol consumption, which is clear at a glance and improves user experience.

According to some embodiments of the present invention, the loci related to alcohol consumption include the rs1229984 locus and the rs671 locus, wherein the rs1229984 locus is located on the ADH1B gene, and when the result of the rs1229984 locus is TT type, The activity of alcohol dehydrogenase is strong, and the metabolism of ethanol is fast; the result is CT type, the activity of alcohol dehydrogenase is moderate, and the rate of alcohol metabolism is moderate; the result is that the activity of alcohol dehydrogenase is weak, and the rate of alcohol metabolism is slow; the rs671 gene locus is located in In the ALDH2 gene, when the rs671 gene locus was GG type, the activity of acetaldehyde dehydrogenase was strong, and the acetaldehyde metabolism was fast; when the result was GA\AA type, the activity of acetaldehyde dehydrogenase was weak, and the metabolism of acetaldehyde was slow.

In one embodiment, the genetic loci directly related to alcohol consumption further include: rs6413413, rs698, rs2298755 and other loci; alcohol dependence-related loci include: rs2066702, rs55768019, rs1789891 and other genetic loci; alcohol use disorder related genes The loci include: rs4975012, rs7078436, rs3114045 and other loci; alcoholism-related loci include: rs9556711, rs2140418, rs8040009 and other loci; alcohol sensitivity-related loci include: rs112834343, rs75536499, rs146298733 alleles The loci related to drinking response include rs143894582, rs200848948, rs397813807 and other loci; the loci related to alcohol-dependent impulsive behavior include: rs34997829 locus; Gene loci, according to the different permutations and combinations of the genotypes of the related loci, to establish a drinking quantity prediction model. It is within the protection scope of the present invention that a person skilled in the art establishes a drinking quantity prediction model by different arrangement and combination of genotypes of related loci.

In one embodiment, a decision tree model is used to construct a drinking quantity prediction model;

Algorithms include:

Use Python to program and call the DecisionTreeClassifier module of Sklearn to perform data mining and build a drinking volume prediction model;

DecisionTreeClassifier module main parameter settings:

criterion='gini': select the Gini coefficient as the metric for the quality of node division;

splitter='best': find the best split point among all features;

max_depth=None: Set the maximum depth of the decision tree, None means that the maximum depth of the decision tree is not constrained until the samples on each leaf node belong to the same class;

min_samples_split=2: When dividing an internal node, the minimum number of samples on the node is required to be 2;

min_samples_leaf=1: Set the minimum number of samples on the leaf node to 1;

Finally, the rs1229984 locus and the rs671 locus were found to be related to alcohol consumption.

Fig. 4 is a flow chart of establishing a second drinking amount prediction model according to an embodiment of the present invention; as shown in Fig. 4, after constructing the first drinking amount prediction model, it also includes:

S4. Re-divide the drinking amount into a second preset number of drinking grades, and establish a second database according to the relationship between the drinking ability of the sample and the drinking amount, and the second preset number of drinking grades;

S5. Select a machine learning model to reconstruct a drinking amount prediction model according to the loci of the sample related to the drinking amount and the second database, and obtain a second drinking amount prediction model.

The working principle and beneficial effects of the above technical solutions: optimizing the first drinking amount prediction model, converting the first preset number of drinking tiers into a second preset number of drinking tiers, and the second preset number may be 9, according to The relationship between the drinking ability of the sample and the amount of drinking, and the second preset number of drinking tiers are used to establish a second database. When rebuilding the drinking amount prediction model, only the loci related to the amount of drinking are selected, and the correlation between the amount of drinking and the amount of drinking is determined by the sample. The genetic locus and the second database select a machine learning model to reconstruct the drinking volume prediction model to obtain the second drinking volume prediction model, which can reduce the calculation amount and complexity, and improve the prediction accuracy of the second drinking volume prediction model. The relationship between the rs1229984 locus and the rs671 locus and alcohol consumption is shown in Table 1.

Table I

In an embodiment, the second preset number is 7, and the drinking rank is 7 ranks, and the relationship between the rs1229984 locus and the rs671 locus and the drinking amount is obtained as shown in Table 2.

Table II

The working principle and beneficial effects of the above technical solutions: when the drinking rank is 0, there are three situations: 1. The rs1229984 locus is CC, and the rs671 locus is AA; 2. The rs1229984 locus is TT and the rs671 locus 3. The rs1229984 locus is CT, and the rs671 locus is AA. The naming of drinking tiers is done in a discontinuous way. If 3 and 6 tiers are missing, the discontinuous naming can match the drinking tier with the specific amount of alcohol consumed. For example, when the drinking tier is 9, the user The amount of alcohol consumed is 9 taels or more.

After the second drinking amount prediction model is obtained, the user's drinking amount is predicted based on the second drinking amount prediction model according to the user's genetic data. It can provide the judgment standard of drinking quantity, quantify the drinking ability of the individual, and give more intuitive and valuable drinking quantity evaluation and drinking advice according to the user's physical condition, so as to improve the user's experience.

Figure 5 is a flow chart of DNA extraction from saliva according to an embodiment of the present invention; as shown in Figure 5, DNA extraction from saliva includes:

S421, put 0.2 ml of saliva in a centrifuge tube, add 600 μL of 0.01 mol/L PBS solution, and centrifuge at high speed at 10000*g for 3 min in a centrifuge;

S422, get the precipitate and add 50ul, 5mol/l potassium iodide solution, 75ul, 0.9% sodium chloride solution, 120ul phenol:chloroform (20:13) solution, shake and shake for 1min, and centrifuge at high speed at 10000*g for 3min by centrifuge;

S423, take 80 ul of the supernatant, add 80 ul of isopropanol, shake for 30 s, and centrifuge at a high speed of 10000*g for 3 min in a centrifuge;

S424, taking the precipitate, adding 500 μl of absolute ethanol to wash, and centrifuging at 10000*g for 3 min in a centrifuge;

S425, take the precipitate to dry at room temperature, and dissolve it with TE buffer.

The working principle and beneficial effects of the above technical scheme: saliva contains oral cavity exfoliated cells, and the cells contain genetic material DNA. Through the above scheme, the user's DNA can be extracted from the saliva for corresponding processing and analysis.

According to some embodiments of the present invention, the genetic data of the sample can also be obtained by collecting blood for DNA extraction. DNA extraction through blood has high sensitivity, and the extracted DNA data is more accurate.

In one embodiment, it also includes:

S71. Acquire second information that affects the drinking amount of the user, where the second information includes: disease history, drinking type, drinking degree, and drinking frequency;

S72. Modify the second alcohol consumption prediction model according to the second information to obtain a third alcohol consumption prediction model.

The working principle and beneficial effects of the above technical solutions: the prediction result output by the second drinking volume prediction model based on the user's genetic data does not consider the actual situation of the user, and the prediction of drinking volume needs to be revised according to the second information affecting the drinking volume of the user , the establishment of a third alcohol consumption prediction model requires the establishment of a correction mechanism for alcohol consumption based on the second information. For example, as shown in Table 2, the genotype of the rs1229984 locus is CC and the genotype of the rs671 locus is GG, and the user's alcohol consumption is predicted to be 7 segments, that is, the user can drink 7 Two or more wines (take 50° liquor as an example), but the user has recently suffered from stomach problems and cannot drink alcohol. Drinking alcohol can easily cause gastric perforation and seriously endanger health. Similarly, according to the user's drinking type, drinking degree, and drinking frequency, the prediction of the user's drinking amount will also be affected. According to the actual situation of the user, a more effective alcohol consumption prediction is made, so that the prediction result of the third alcohol consumption prediction model is more accurate.

In one embodiment, the method for revising the second alcohol consumption prediction model according to the second information includes:

Calculate the amount of ethanol in the first prediction result given by the second alcohol consumption prediction model:

V ₁ =A×c

Wherein, A is the drinking volume (ml) output by the second drinking volume prediction model based on the user's genetic data; c is the preset alcohol concentration (%vol) in the second drinking volume prediction model;

Calculate the amount of ethanol that the user can drink according to the second information:

V ₂ =V ₁ ×d×t×f

Among them, d is the correlation coefficient between the user's disease history and drinking amount; t is the correlation coefficient between drinking type and drinking amount; f is the correlation coefficient between drinking frequency and drinking amount;

Alcohol consumption of the second prediction result given by the third alcohol consumption prediction model:

Among them, _cu is the drinking degree input by the user.

The working principle and beneficial effects of the above technical solutions: when the user belongs to patients with stomach disease, liver disease, cardiovascular and cerebrovascular diseases, pregnant women, taking cold medicine, sleeping pills, and tranquilizers, the correlation coefficient d between the user's disease history and the amount of alcohol consumption is 0, That is, the user cannot drink alcohol; the correlation coefficient d between the disease history of other users and the amount of drinking is between 0 and 1; the value of the correlation coefficient t between the type of drinking and the amount of drinking is shown in Table 3; The value of the correlation coefficient f is shown in Table 4; through the above algorithm, the second drinking volume prediction model is modified according to the second information to obtain the third drinking volume prediction model, which can be used for more effective drinking according to the actual situation of the user. Prediction results are more accurate, giving users the most correct drinking advice and improving user experience.

Table 3

type of drinking	Correlation coefficient t
Liquor	1
beer	1.5
wine	1.8

Table 4

frequency of drinking	Correlation coefficient f
drink every day	0.3
Drink once every three days	0.6
Drink once in 7 days	0.8

It will be apparent to those skilled in the art that various modifications and variations can be made in the present invention without departing from the spirit and scope of the invention. Thus, provided that these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims (7)

1. A method for constructing an alcohol volume prediction model, comprising:

   S1. Obtain the relationship between the drinking ability and the drinking amount of the sample, divide the sample into a first preset number of drinking rank according to the drinking amount, and establish a first database according to the relationship between the drinking ability and the drinking amount of the sample, and the drinking rank;

   S2. Obtain the genetic data of the sample and format the genetic data;

   S3. Select a machine learning model to construct a first alcohol consumption prediction model according to the formatted gene data of the sample and the first database.
2. The method for constructing an alcohol volume prediction model according to claim 1, wherein the obtaining the genetic data of the sample and formatting the genetic data comprises:

   S21, collecting saliva from the sample;

   S22, performing DNA extraction according to the saliva of the sample, and performing gene sequencing on the extracted DNA;

   S23. Process the gene data after gene sequencing to obtain the genotype of the locus related to the alcohol consumption of each sample;

   S24. Format the gene locus into a number according to the genotype.
3. The construction method of alcohol quantity prediction model as claimed in claim 2, is characterized in that,

   Perform gene locus screening on the genetic data of the formatted sample, including:

   S241, respectively calculating the characteristic value of each gene locus and dividing the first database to obtain the purity improvement value or uncertainty reduction value of each data subset and the data set before division;

   S242. Select the gene locus N with the maximum purity improvement value or the maximum uncertainty reduction value and the characteristic value n of the gene locus N, wherein the gene locus N is used as a node, according to the gene locus N The grouping of the eigenvalue n splits the first database into two sub-data sets;

   S243. In the two sub-data sets in turn, calculate the purity improvement value or the uncertainty reduction value of the characteristic value of each gene locus in the sub-data set; select the gene locus M and the uncertainty reduction value with the largest purity improvement value or the largest uncertainty reduction value. The eigenvalue m of the gene locus M, wherein the gene locus M is used as a sub-node, and the sub-data set is split again according to the grouping of the eigenvalue m of the gene locus M;

   S244. When it is determined that the purity of the divided sub-data set is greater than the preset purity threshold or the uncertainty value is less than the preset uncertainty threshold, stop the splitting, and finally obtain the gene locus related to the alcohol consumption and the gene locus and the gene locus The relationship between drinking grades.
4. The method for constructing an alcohol consumption prediction model according to claim 3, wherein the genetic locus related to alcohol consumption comprises the rs1229984 genetic locus and the rs671 genetic locus, wherein the rs1229984 genetic locus is located on the ADH1B gene, When the rs1229984 gene locus is TT type, the activity of alcohol dehydrogenase is strong and the alcohol metabolism is fast; when the result is CT type, the activity of alcohol dehydrogenase is moderate, and the rate of alcohol metabolism is moderate; when the result is CC type, the activity of alcohol dehydrogenase is weak, Alcohol metabolism is slow; the rs671 gene locus is located on the ALDH2 gene. When the rs671 gene locus is GG type, the activity of acetaldehyde dehydrogenase is strong, and acetaldehyde metabolism is fast; when the result is GA\AA type, the activity of acetaldehyde dehydrogenase is weak. , Acetaldehyde metabolism is slow.
5. The method for constructing an alcohol consumption prediction model as claimed in claim 4, characterized in that, after constructing the first alcohol consumption prediction model, the method further comprises:

   S4. Re-divide the drinking amount into a second preset number of drinking grades, and establish a second database according to the relationship between the drinking ability of the sample and the drinking amount, and the second preset number of drinking grades;

   S5. Select a machine learning model to reconstruct the alcohol consumption prediction model according to the genetic locus related to the alcohol consumption of the sample and the second database, and obtain a second alcohol consumption prediction model.
6. The method for constructing an alcohol quantity prediction model according to claim 2, wherein DNA extraction is performed according to the saliva of the sample, comprising:

   S421, put 0.2ml of saliva in a centrifuge tube, add 600μL of 0.01mol/L PBS solution, and centrifuge at high speed at 10000*g for 3min through a centrifuge;

   S422, get the precipitate and add 50ul, 5mol/l potassium iodide solution, 75ul, 0.9% sodium chloride solution, 120ul phenol:chloroform (20:13) solution, shake and shake for 1min, and centrifuge at high speed at 10000*g for 3min by centrifuge;

   S423, take 80ul of the supernatant, add 80ul of isopropanol, shake and shake for 30s, and centrifuge at a high speed of 10000*g for 3min by a centrifuge;

   S424, taking the precipitate, adding 500 μl of absolute ethanol to wash, and centrifuging at 10000*g for 3 min in a centrifuge;

   S425, take the precipitate to dry at room temperature, and dissolve it with TE buffer.
7. The construction method of alcohol quantity prediction model as claimed in claim 5, is characterized in that, also comprises:

   S71. Acquire second information that affects the drinking amount of the user, where the second information includes: disease history, drinking type, drinking degree, and drinking frequency;

   S72. Amend the second alcohol consumption prediction model according to the second information to obtain a third alcohol consumption prediction model.

Images