Genetic testing method to realize skin counseling

Abstract

Since skin condition is significantly influenced by environmental factors associated with lifestyle habits, it has been difficult to forecast skin condition by means of SNP analysis. The purpose of the present invention is to determine one's current skin condition, which has been difficult with the prior art. The present invention provides a method for determining one's current and future skin type by determining skin condition from genetic and environmental factors.

Description

[Technical field]

[0001] The present invention relates to a skin evaluation method, a skin evaluation system, and a control program for a skin condition evaluation system based on genetic testing. Skin counseling is made possible based on the evaluation method, system, and program of the present invention. [Background technology]

[0002] In recent years, skin tests based on gene SNP analysis have been proposed, and the relationship between SNP and skin quality has been studied, particularly for genes that code for proteins directly involved in skin quality (Patent Documents 1 to 4). As a result, methods for predicting skin characteristics and skin troubles based on SNP analysis results have been provided. In these studies, SNPs are examined for genes that are known to be directly involved in skin quality, such as wrinkles and age spots. Examples of such target genes include MC1R, MMP1, SOD2, GPX1, and ASIP. On the other hand, since there are a wide variety of genes that affect the skin characteristics of the skin, the relationship between SNPs and skin characteristics in many more genes has also been examined (Patent Document 1: International Publication No. 2018 / 101449). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] International Publication No. 2018 / 101449 Summary of the Invention [Problem to be solved by the invention]

[0004] Although skin tests based on SNP analysis show the skin constitution that one was born with, it is known that the skin condition is greatly influenced by environmental factors due to lifestyle habits. Therefore, it has been difficult to determine the current skin condition only based on the skin constitution evaluated by SNP analysis. In addition, since the current skin condition is closely related to the future skin condition, it has been found that the accuracy of predicting the future skin condition based only on genetic factors is low. Therefore, the present invention has been made to solve such problems, and aims to provide a method for determining the current skin condition based on genetic factors and environmental factors, and a method for predicting the future skin condition. [Means for solving the problem]

[0005] The inventors have conducted research and development into a method for determining current skin condition and a method for predicting future skin condition by combining genetic factors determined by SNP analysis, etc. with environmental factors such as lifestyle habits, and have established a method for evaluating current or future skin condition.

[0006] Therefore, the present invention provides a method for evaluating skin using information on an individual's SNPs and information on at least one environmental factor selected from the group consisting of age, height, weight, BMI, information on exposure to ultraviolet light, and smoking information, comprising: The skin condition is determined by weighting the contribution of the information on SNPs and information on environmental factors to the skin condition. The present invention also relates to a skin evaluation system that executes the skin evaluation method, and a program and a control method for controlling the skin evaluation system. Effect of the Invention

[0007] By combining information about SNPs with information about environmental factors, it is possible to more accurately evaluate current or future skin conditions compared to conventional methods of determining skin constitution based solely on genetic information. [Brief description of the drawings]

[0008] [Figure 1] Fig. 1A is a diagram showing a schematic diagram of an information processing device including an input unit, a storage unit, a processing unit, a learning unit, and an output unit. Fig. 1B is a diagram showing a schematic diagram of a system in which the learning unit exists in a learning device external to the information processing device. [Diagram 2] FIG. 2 is a schematic diagram showing a flow for creating a learning model for determining a skin condition score. [Diagram 3] FIG. 3 shows a graph in which subjects belonging to a certain percentage of the high and low value sides of the measured values ​​of skin condition (wrinkles) are classified into a high value group and a low value group. [Figure 4] FIG. 4 is a schematic diagram showing a flow for creating a learning model for determining skin type. [Figure 5A] Figure 5A is a schematic diagram showing the process of creating a learning model that determines a skin condition score, dividing a dataset into a learning dataset and a validation dataset in a processing unit, constructing a learning model in a learning unit based on the divided learning dataset, and validating the constructed learning model. [Figure 5B] FIG. 5B is a schematic diagram showing a process for creating a learning model for determining skin type, in which a processing unit divides a data set into a learning data set and a validation data set, a learning unit constructs a learning model based on the divided learning data set, and a learning unit validates the constructed learning model. [Figure 6A-B] Fig. 6A is a graph of an ROC curve showing the relationship between sensitivity and specificity, which was created during validation of a learning model for determining skin type. Fig. 6B is a graph for creating a coefficient of determination R2, which was created during validation of a learning model for determining a skin condition score. [Figure 6C] FIG. 6C is a graph showing SNP occurrence frequency generated during validation of the learning model. [Figure 7] FIG. 7 is a schematic diagram showing a flow for predicting a future skin condition using a learning model that determines a skin condition score. [Figure 8] FIG. 8 is a schematic diagram showing a flow for determining skin type using a learning model for determining skin type. [Figure 9A]FIG. 9A shows a graph of future skin condition score evolution using a learned model for determining the skin condition score. [Figure 9B] FIG. 9B shows the display on the application, where predicted changes in skin condition score are displayed by selecting slide bars for age, sun protection measures, and smoking habits. [Figure 10] Fig. 10A is a graph showing skin types determined using a learning model for determining each skin type, and Fig. 10B is a fan graph showing skin types evaluated from both the susceptibility and difficulty of developing the skin type. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0009] The present invention relates to a method for evaluating skin using information on an individual's SNPs (hereinafter also referred to as genetic information) and information on environmental factors (hereinafter also referred to as environmental information). Combining the genetic information and the environmental information allows for a more accurate evaluation of the current or future skin condition. Specifically, the skin evaluation method includes a step of determining the skin condition by weighting the genetic information and the environmental information so as to contribute to the skin condition. Specifically, the weighting that contributes to the skin condition is performed by subjecting the relationship between the genetic information and the environmental information and a previously measured value of the skin condition to machine learning, and inputting the genetic information and the environmental information into a learning model that has been weighted to contribute to a specific skin condition.

[0010] The method of the present invention can be realized by a skin evaluation system including an input unit, a learning unit, a processing unit, and an output unit. Another aspect of the present invention relates to the skin evaluation system. Specifically, the skin evaluation system includes the following: an input section for inputting information on an individual's SNPs and information on environmental factors selected from the group consisting of age, height, weight, BMI, information on exposure to ultraviolet light, and smoking information; A learning unit that determines information about the skin condition using a learning model that is trained based on the relationship between information about SNPs and information about environmental factors and the skin condition and is weighted to contribute to the skin condition; and A processing unit that performs processing on the skin condition determined by the learning unit. An output section that outputs the processed skin condition Yet another aspect of the present invention relates to a program or method for controlling the skin evaluation system of the present invention.

[0011] The program for controlling the skin condition evaluation system of the present invention is a control program for a skin evaluation system comprising a learning unit that determines information about the skin condition from input information about SNPs and information about environmental factors through a learning model in which weights are assigned to contribute to the skin condition, an input unit, an output unit, and a processing unit, and the control program includes the following command: The information on the individual's SNPs input from the input unit and information on environmental factors selected from the group consisting of age, height, weight, BMI, information on exposure to ultraviolet light, and smoking information are input to a learning unit; A learning unit determines information on skin condition determined through a learning model in which the input information on SNPs and information on environmental factors are weighted to contribute to the skin condition; causing a processing unit to process the information regarding the determined skin condition; The processed skin condition is output to the output unit. The present invention relates to the program.

[0012] FIG. 1A is a diagram showing a schematic diagram of an information processing device 10 / system 20 including an input unit 12, a storage unit 11, a processing unit 14, a learning unit 13, and an output unit 15. The learning unit 13 exists in the information processing device 10, but the learning unit 13 may exist in an external learning device 30 and be connected to the information processing device 10 via a communication unit to form a system 20 as a whole (FIG. 1B). The learning unit 13 including the created learning model can also be transferred to another information processing device. The connection between the external learning device 30 and the information processing device 10 may be wired or wireless. As an example, they may be connected via an intranet or the Internet. In the present invention, the system may include a device.

[0013] 2 and 4 are schematic diagrams showing an example of a flow of processing performed in the input unit 12, the storage unit 11, the processing unit 14, and the learning unit 13 when creating a learning model. A data set including genetic information, environmental information, and previously measured values ​​of skin condition is input from the input unit 12, stored in the storage unit 11, and then processed in the processing unit 14 and the learning unit 13. In the processing unit 14, the data set may be divided into a learning data set and a validation data set, but in another embodiment, a validation data set may be provided separately. In the learning unit 13, a learning model is created in which the input skin condition measurement value or preprocessed data on the measurement value is used as the objective variable, and the environmental information and genetic information are used as explanatory variables, and weighted to obtain the objective variable. In addition, prior to creating the learning model, variable selection may be performed.

[0014] Generally, SNPs are classified into recessive homozygote, heterozygote, and dominant homozygote according to genotype, but one SNP may be affected by interaction with other SNPs. Therefore, four genetic models consisting of recessive model, dominant model, additive model, and synergistic model are used for each SNP as explanatory variables to perform analysis, and the optimal model for each SNP can be selected. This analysis can be performed by regularizing with Elastic Net. It is expected that the accuracy of determining skin type by SNPs can be improved by determining the genetic model prior to creating the learning model according to the present invention.

[0015] In one embodiment, the measured value of the skin condition is used as the objective variable, and information on 79 types of SNPs and information on at least one environmental factor selected from the group consisting of age, height, weight, BMI, exposure to ultraviolet light, and smoking information are used as explanatory variables, and weighting is performed for each explanatory variable (FIG. 2). This weighting can be performed by multiple regression analysis. When an individual's genetic information and environmental information are input to a learning unit including a learning model created in this way, the learning unit can output a skin condition score. Therefore, the learning unit that outputs the skin condition score can be called a learning unit that determines the skin condition score. At this time, by varying and inputting the age of the environmental information, the learning unit outputs a future skin condition score. In this way, the processing unit inputs the genetic information and the environmental information with age as a variable to the learning unit 13, and based on the skin condition score output from the learning unit 13, a graph can be created that displays the change in the skin condition with age as a variable. Furthermore, by varying and inputting behavioral factors such as smoking and ultraviolet light measures from the environmental information to the learning unit, the learning unit outputs a skin condition score when such behavioral factors are reflected, thereby making it possible to predict the future skin condition. Therefore, the learning section that determines the skin condition score can also be considered a learning model that predicts changes in the skin condition.

[0016] In another embodiment, the preprocessed data of the measured value of the subject's skin condition can be used as the objective variable. As an example of preprocessing, a certain percentage of the high value side or low value side is set as a high value group or a low value group based on the measured value of the skin condition, and the subject is classified into a high value group and a non-high value group or a low value group and a non-low value group for each skin condition (FIG. 3). Then, whether or not the subject belongs to the high value group or the low value group can be used as the objective variable (FIG. 3). Classification into a high value group and a non-high value group or a low value group and a non-low value group can be performed by the normal processing unit 14. The high value group and the low value group can be determined according to age, for example, with the age range being 10 years, 5 years, and 2 years. The high value group and the low value group can be, for example, an X% high value group and an X% low value group, where X% can be any value selected, for example, from the group consisting of 5%, 10%, 15%, 20%, 25%, and 30%. Depending on the skin condition selected, the age range and X may vary.

[0017] After this pre-processing, the learning unit 13 can determine factors contributing to the high value group from information on the SNPs of the subject and / or information on environmental factors by logistic analysis in the high value group and non-high value group. As an example, a learning model is created by defining subjects in the high value group as 1 for a certain skin condition and subjects in the non-high value group as 0, and using these as the objective variables, and weighting is applied to explanatory variables consisting of information on 79 types of SNPs and information on environmental factors. Similarly, the learning unit 13 can determine factors contributing to the low value group from information on the SNPs of the subject and / or information on environmental factors by logistic analysis in the low value group and non-low value group. As an example, a learning model is created by defining subjects in the low value group as 1 for a certain skin condition and subjects in the non-low value group as 0, and using these as the objective variables, and weighting is applied to explanatory variables consisting of information on 79 types of SNPs and / or information on environmental factors. Logistic regression can be performed to construct these learning models. In this way, a learning model for determining whether the skin condition belongs to the high value group or the low value group is constructed for each skin condition, and the learning unit 13 holds these learning models. By inputting information on the SNP of the subject and / or information on environmental factors to the learning unit 13 including the learning model, the learning unit 13 outputs whether the subject belongs to the high value group or the low value group for a certain skin condition. As an example, when a subject in the high value group for a certain skin condition is defined as 1 and a subject in the non-high value group is defined as 0 and used as a response variable, the possibility that the subject belongs to the high value group is output as a numerical value in the range of 0 to 1. Similarly, when a subject in the low value group for a certain skin condition is defined as 1 and a subject in the non-low value group is defined as 0 and used as a response variable, the possibility that the subject belongs to the low value group is output as a numerical value in the range of 0 to 1. In this way, by creating a learning model for determining the high value group / low value group for each of the high value group and the low value group for a certain skin condition, the skin constitution can be determined from both the possibility of belonging to the high value group and the possibility of belonging to the low value group for a certain skin condition.

[0018] Prior to learning in the learning unit 13, a dataset including information on the SNPs and / or environmental factors of the subject and information on the quantified skin condition of the subject may be divided into a learning dataset and a validation dataset. By dividing the dataset into the learning dataset and the validation dataset, any number x of cross-validation datasets can be constructed (FIG. 5). This preprocessing can be performed in the processing unit 14. By performing this preprocessing, it becomes possible to perform multiple learning / validation processes based on the dataset N, and it becomes possible to construct a more accurate model even with a limited dataset.

[0019] The method for constructing a learning model, including pre-processing in which a data set is divided into a training data set and a validation data set, includes the following steps: (i) selecting a training dataset n; (ii) dividing the data set into a training data set n and a validation data set Nn; (iii) constructing a learning model for the learning data set n; (iv) A step of validating the constructed model using a validation data set Nn. The desired number of times, X, the training data set (n1...n X ) and repeat steps (i) to (iv). This construction method can be performed by the information processing device 10 / the system 20.

[0020] Specifically, the information processing device / system includes: A memory unit for storing a data set N including explanatory variables and a response variable; A processing unit that selects a part of the stored data set as a learning data set and inputs it to a learning unit, and validates the learning model based on the objective variables of the validation data set and the output values ​​of the validation data set input to a learning unit including the constructed learning model. A learning unit that constructs a learning model that is trained to output a dependent variable when an explanatory variable is input, from an explanatory variable and a dependent variable of an input learning data set, (i) a step in which the processing unit selects a training dataset n from a dataset N stored in the storage unit, divides the training dataset n into a training dataset n and a validation dataset Nn, and inputs the training dataset n to a learning unit; (ii) constructing a learning model for the learning data set n by the learning unit, the learning model being trained to output a target variable when an explanatory variable is input; (iii) a step in which the processing unit inputs a validation data set Nn to a learning unit including the constructed learning model. (iv) A step of verifying the constructed model using the values ​​output by the learning unit and the objective variables of the validation data set Nn. The desired number of times, X, the training data set (n1...n X ) and repeat steps (i) to (iv); The process includes:

[0021] Verification can be performed by inputting validation data into the constructed learning model and performing statistical processing on the output value and the objective variable of the validation data. Verification can be performed by creating an ROC curve or calculating the coefficient of determination R2 or correlation coefficient r. Since the data set is divided and there are x cross-validation data sets, up to x statistical processing times are obtained for up to x learning models created (Figures 6A and B). In addition, the occurrence frequency of for each SNP in up to x learning models created is shown (Figure 6C). Based on this information, and further taking into consideration biochemical knowledge, an optimal learning model can be selected.

[0022] In one embodiment, when dividing a data set N into a learning data set n and a validation data set Nn, the age group, information on exposure to ultraviolet light, smoking information, and skin measurement test facilities included in the explanatory variables of the multivariate analysis can be distributed as evenly as possible. This allows the construction of a more accurate model. Also, if the number of data items under each condition is less than two, the data is distributed as learning data. This can solve problems caused by imbalance between the learning data set and the validation data set in the process of constructing a model based on the multivariate analysis and the process of verifying the constructed model (the regression coefficient of the multivariate analysis cannot be estimated, making it impossible to construct a model, or overfitting to the learning data set, resulting in insufficient generalization ability of the model).

[0023] FIG. 7 shows a flow in which the evaluation system of the present invention inputs genetic information and environmental information to predict future skin conditions. The method, evaluation system, and program of the present invention utilize a learning model that predicts changes in the constructed skin condition. The method, evaluation system, and program of the present invention can input environmental information and genetic information via the input unit 12. The input environmental information and genetic information are sent to the learning unit 13, and the learning unit 13 outputs a skin condition score. In the processing unit, the age and behavioral factors of the environmental information are varied and input to the learning unit, and the learning unit outputs the skin condition score in each case. In the processing unit 14, by plotting the skin condition score output against the age, it is possible to create a graph of the change in the skin condition score when the age is changed, and further, a graph of the change in the skin condition score when the behavioral factor is changed can be created. The graph created in the processing unit 14 is sent to the output unit 15, and the graph is output in the output unit 15. The results of the processing of each step may be stored in the memory unit 11.

[0024] FIG. 8 shows the flow of the evaluation system of the present invention inputting genetic information and environmental information to determine skin constitution. The method, evaluation system, and program of the present invention utilize a learning unit for determining skin constitution that has been constructed. The learning unit for determining skin constitution has a learning model for high value group determination that outputs the possibility of belonging to the high value group, and a learning model for low value group determination that outputs the possibility of belonging to the low value group. The method, evaluation system, and program of the present invention can input environmental information and genetic information via the input unit 12. The input environmental information and genetic information are sent to the learning unit 13, and the learning unit 13 outputs the possibility of belonging to the high value group by the learning model for high value group determination of the learning unit 13, and outputs the possibility of belonging to the low value group by the learning model for low value group determination. The processing unit 14 creates a graph showing the output possibility of belonging to the high value group and the possibility of belonging to the low value group. The graph created in the processing unit 14 is sent to the output unit, and the graph is output in the output unit 15. The results of the processing of each step may be stored in the memory unit 11.

[0025] When using a learning model for predicting changes in skin condition, the learning unit 13 can also output a formula for outputting the skin condition score. In this formula, the processing unit 14 can also create a predicted graph of skin condition transition in which age is shown on the X-axis and the skin condition score is shown on the Y-axis. Furthermore, a predicted graph of skin condition transition when behavioral factors, such as smoking habits and UV protection, are changed can be created (FIG. 9). FIG. 9A is a predicted graph of skin condition (wrinkles) transition with age as a variable when wrinkles are selected as the skin condition as an example, and further shows changes due to sunburn protection. FIG. 9B is an example of a display presented to a customer, and by selecting the future (several years later), changes in skin condition (wrinkle) score can be visually displayed. Furthermore, by inputting behavioral factors such as UV protection and smoking information, future changes in skin condition score can be displayed. These graphs may be displayed on a terminal at the storefront or on the Web at the customer's terminal. They may be printed on paper and sent as a report of the analysis results. Regarding UV protection, the system can display the transition of skin condition in cases where sunburn protection measures, such as the use of sunscreen, are avoided as much as possible, or where sunscreen is actively implemented to the same extent as in the past.

[0026] When the learning unit 13 for determining skin constitution is used, the learning unit 13 outputs the possibility of the skin constitution belonging to the high value group and / or the low value group. As an example, when age spots are selected as the skin condition, the high value group of age spot measurement values ​​(Visia Age Spot Index) is an indicator that age spots are likely to occur, and the low value group of age spot measurement values ​​(Visia Age Spot Index) is an indicator that age spots are unlikely to occur. Therefore, when the environmental information and the genetic information are inputted into the learning unit 13, it is possible to determine the skin constitution for age spots from both the perspective of how likely it is that age spots will occur (possibility of belonging to the high value group) and how unlikely it is that age spots will occur (possibility of belonging to the low value group). The processing unit 14 may process the possibility of belonging to the high value group and the possibility of belonging to the low value group determined by the learning unit 13 to generate a graph showing the ease of production (the possibility of belonging to the high value group) and the difficulty of production (the possibility of belonging to the low value group) (Figure 10A), or it may generate a fan-shaped graph with the ease of production (the possibility of belonging to the high value group) and the difficulty of production (the possibility of belonging to the low value group) as orthogonal axes (Figure 10B).

[0027] Counseling can be performed based on the skin type or future skin condition output from the output unit 15. In the counseling, beauty ingredients and their concentrations suitable for the skin type, or cosmetics suitable for the skin type can be presented. In addition to cosmetics suitable for the skin type, information on the frequency and method of use can also be provided. The relationship between the skin type and the beauty ingredients and their concentrations, or cosmetics may be stored in advance in the storage unit 11 as a table, and information on the beauty ingredients and their concentrations, or cosmetics can be provided together with the output of the skin type.

[0028] The information on environmental factors is input as a result of a questionnaire survey. As an example, the questionnaire includes at least one piece of information selected from the group consisting of age, height, weight, ultraviolet exposure information, and smoking information. The ultraviolet exposure history information and smoking information are surveyed by a questionnaire. The ultraviolet exposure information is surveyed by a questionnaire about current sunburn and a questionnaire about past sunburn. As an example, the questionnaire about current sunburn can be classified into four levels, and the questionnaire about past sunburn can be classified into three levels. The smoking information can be surveyed about whether or not the person currently smokes or has smoked in the past, the number of cigarettes smoked per day if they have smoked, and the smoking period. As an example, the survey can be conducted by dividing the person into three groups: no smoking history, a smoking history of 20 or more cigarettes per day in the past, and a smoking history of less than 20 cigarettes per day in the past. The information on current sunburn and the current smoking information among the ultraviolet exposure information can be used as behavioral factors to predict future changes in skin condition.

[0029] The skin condition measurements are measured by selecting an appropriate measuring means according to the skin condition to be evaluated. Specifically, measurements of skin conditions selected from the group consisting of age spots, wrinkles and firmness, moisture, texture, skin color, and sebum are used. These skin conditions can be further classified into subgroups. For example, age spots can be classified into the number of age spots and the area of ​​age spots measured by a skin image measuring device in addition to the measurements by Visia (Visia age spot index). Wrinkles and firmness can be classified into the wrinkle measurements by Visia (Visia wrinkle index), replica analysis (wrinkles at the corner of the eyes), and skin elasticity value. Moisture content can be classified into stratum corneum moisture content and transepidermal water loss (barrier function) using a dedicated commercially available device. Skin color can be classified into cheek skin color (melanin amount, skin brightness, yellowness) and upper arm skin color measured by a color difference meter. These measurements are measured using any method known in the art according to the skin condition to be measured. As one example, age spots, wrinkles, and texture can be quantified and measured using a skin image measuring device such as VISIA (registered trademark), etc. Skin moisture and sebum can be quantified and measured using a skin measuring device such as a Corneometer, Vapometer, or Sebumeter.

[0030] As the skin condition, at least one selected from the group consisting of spots, wrinkles, firmness, skin moisture, texture, skin color, and sebum can be determined. SNPs can be selected according to the skin condition to be determined. The information on SNPs used in the present invention is information on the following 79 SNPs: rs1800629、rs2108622、rs1047781、rs12203592、rs16891982、rs3760776、rs9340799、rs12913832、rs17822931、rs964184、rs1801133、rs2228479、rs10515552、rs10741657、rs10882272、rs11057830、rs11234027、rs12272004、rs12377462、rs12785878、rs12931267、rs1540771、rs1667255、rs1993116、rs2060793、rs2227564、rs2282679、rs2298585、rs3829251、rs41281112、rs4654748、rs492602、rs602662、rs1030868、rs1126643、rs1256062、rs12934922、rs1485766、rs1501299、rs17577、rs1799724、rs1800012、rs1800414、rs2010963、rs2234693、rs2241145、rs2285053、rs2287074、rs2287076、rs2987983、rs3918242、rs4065、rs4252125、rs6152、rs7201、rs74653330、rs7501331、rs8110862、rs8326、rs833061、rs1050565、rs1799750、rs4880、rs1050450、rs6058017、rs1061622、rs2046571、rs2232228、rs3785079、rs2246416、rs1107946、rs11568737、rs41303970、rs12051272、rs182052、rs3865188、rs6810075、rs7799039、rs1137101。

[0031] As an example, when the skin condition is a blemish, a learning model can be created for each of the blemish number and blemish area measured by a skin image measuring device in addition to the blemish measurement value (Visia blemish index) by Visia. The SNPs used for evaluating these skin conditions are as follows. One or more SNPs selected from the group consisting of SNPs listed in the table below can be used to create the learning model and to evaluate the skin condition (blemish) using the learning model. [Table 1] [Table 2] [Table 3]

[0032] As an example, when the skin condition is skin color, learning models can be created for the amount of melanin measured by a color difference meter, the brightness of the skin, and yellowness. The SNPs used to evaluate these skin conditions are as follows. One or more SNPs selected from the group consisting of the SNPs listed in the table below can be used to create the learning model and to evaluate the skin condition (skin color 1 to 4) using the learning model. [Table 4]

[0033] [Table 5]

[0034] [Table 6]

[0035] [Table 7]

[0036] As an example, when the skin condition is wrinkles, two types of populations can be created using the wrinkle measurement value (Visia wrinkle index) and a learning model can be created for each. The SNPs used to evaluate these skin conditions are as follows. One or more SNPs selected from the group consisting of SNPs listed in the table below can be used to create the learning model and to evaluate the skin condition (wrinkles 1 to 2) using the learning model.

[0037] [Table 8]

[0038] [Table 9]

[0039] As an example, when the skin condition is texture, a learning model can be created for each of the texture measurements by Skinvisiom (texture score by Skinvisiom). The SNPs used for evaluating these skin conditions are as follows. One or more SNPs selected from the group consisting of SNPs listed in the table below can be used to create the learning model and to evaluate the skin condition (texture) using the learning model. [Table 10]

[0040] As an example, when the skin condition is elasticity, a learning model can be created for each elasticity measured by Cutometer. The SNPs used for evaluating these skin conditions are as follows. One or more SNPs selected from the group of SNPs listed in the table below can be used to create a learning model and to evaluate the skin condition (elasticity) using the learning model. [Table 11]

[0041] As an example, when the skin condition is sebum amount, a learning model can be created for each sebum amount measured by Sebumeter. The SNPs used for evaluating these skin conditions are as follows. One or more SNPs selected from the group consisting of SNPs listed in the table below can be used to create the learning model and to evaluate the skin condition (sebum amount) using the learning model. [Table 12]

[0042] As an example, when the skin condition (cheek, inner forearm) is moisture, a learning model can be created for each of the moisture measurement value by the Corneometer and the transepidermal water loss by the Vapometer. The SNPs used to evaluate these skin conditions are as follows. One or more SNPs selected from the group consisting of SNPs listed in the table below can be used to create the learning model and to evaluate the skin conditions (moisture 1 and 2) using the learning model. [Table 13]

[0043] [Table 14]

[0044] As an example, when the skin condition (cheek, inner forearm) is a barrier function, a learning model can be created for each of the transepidermal water loss (barrier function) measured by a Vapometer. The SNPs used to evaluate these skin conditions are as follows. One or more SNPs selected from the group of SNPs listed in the table below can be used to create the learning model and to evaluate the skin conditions (barrier 1 and 2) using the learning model. [Table 15]

[0045] [Table 16]

[0046] As an example, in the case of wrinkle area, wrinkle volume, and maximum wrinkle depth of wrinkles at the corners of the eyes, a learning model can be created for each measurement value obtained by replica analysis. The SNPs used to evaluate these skin conditions are as follows. One or more SNPs selected from the group of SNPs listed in the table below can be used to create a learning model and to evaluate the skin condition (wrinkle area, wrinkle volume, maximum wrinkle depth) using the learning model. [Table 17]

[0047] [Table 18]

[0048] [Table 19]

[0049] In further embodiments of the present invention, the following are contemplated: rs1800629, rs2108622, rs1047781, rs12203592, rs16891982, rs3760776, rs9340799, rs12913832, rs17822931, rs964184, rs1801133, rs2228479, rs10515552, rs10741657, rs10882272, rs11057830, rs11234027, rs12272004, rs12377462, rs1278587 8, rs12931267, rs1540771, rs1667255, rs1993116, rs2060793, rs2227564, rs2282679, rs2298585, rs3829251, rs41281112, rs4654748 , rs492602, rs602662, rs1030868, rs1126643, rs1256062, rs12934922, rs1485766, rs1501299, rs17577, rs1799724, rs1800012, rs180 0414, rs2010963, rs2234693, rs2241145, rs2285053, rs2287074, rs2287076, rs2987983, rs3918242, rs4065, rs4252125, rs6152, rs72 01, rs74653330, rs7501331, rs8110862, rs8326, rs833061, rs1050565, rs1799750, rs4880, rs1050450, rs6058017, rs1061622, rs2046 571, rs2232228, rs3785079, rs2246416, rs1107946, rs11568737, rs41303970, rs12051272, rs182052, rs3865188, rs6810075, rs7799039, rs1137101, and information on at least one environmental factor selected from the group consisting of age, height, weight, BMI, exposure to ultraviolet light, and smoking information, Skin condition is determined by weighting the contribution of SNP information and environmental factor information to skin condition. The present invention also relates to a method for evaluating skin, including the above. The SNP to be used can be appropriately selected according to the skin condition to be evaluated. More specifically, the relationship between the skin condition to be evaluated and the SNP to be used is as shown in Tables 1 to 19.

[0050] In the present invention, SNPs are represented by rs numbers (Reference SNP ID numbers). Detailed information on SNPs (chromosomal locations and mutations) corresponding to each rs number is managed by the National Center for Biotechnology Information (NCBI) and is available on the NCBI homepage ( http: / / www.ncbi.nlm.nih.gov / ) for reference.

[0051] The detection of SNP may be performed by any known method. As an example, the subject's nucleic acid is purified from a biological sample of the subject, such as saliva, blood, mucous membrane, tissue fragment, hair, etc., based on a standard method, and SNP is detected in the purified nucleic acid. Therefore, the method for determining skin characteristics of the present invention may further include a sample preparation step and a nucleic acid purification step. The nucleic acid may be DNA or RNA, and in the case of RNA, it is preferable to perform reverse transcription after purification to prepare DNA. In the SNP detection step, any method can be used as long as it allows the detection of SNP in purified nucleic acid. SNP can be detected by determining the sequence around the position of the target SNP, or by using a PCR-based method, a method using a DNA probe, or a method using mass spectrometry. Examples of PCR-based methods include SNP typing, TaqMan PCR, single-base extension, Pyrosequencing, and Exonuclease Cycling Assay. An example of a method using a DNA probe is the Invader method (Comprehensive Gene Polymorphism Analysis (SNP), Japanese Journal of Pharmacology, 125, 148-152, 2005). SNPs may be accompanied by other SNPs in linkage disequilibrium. The target SNP can be detected by detecting an SNP in linkage disequilibrium with the target SNP. Therefore, in the present invention, detection of an SNP is not limited to direct detection of the target SNP, but also includes detection of the target SNP by detecting an SNP in linkage disequilibrium.

[0052] The presence of SNPs can also be detected based on an individual's base sequence that has already been sequenced. In this case, the presence of SNPs can be detected by examining the sequence at the position where the SNP is present in the data of the already determined base sequence. Therefore, the detection of SNPs in this case may be performed by inputting base sequence data into a computer that stores SNP information. In a more preferred embodiment, the input is made from a terminal via the Internet, SNPs are detected on a server, and the detection results can be output to the terminal via the Internet.

[0053] The input unit 12 includes an interface. The interface may be connected to an operation unit such as a keyboard or a mouse, a communication unit, or an external storage device such as a CD-ROM, a DVD-ROM, a BD-ROM, or a memory stick. Information about SNPs and information about environmental factors may be input from the input unit and stored in the storage unit. Instead of information about SNPs, genetic sequence information of the target may be input from the input unit. The input sequence information may be temporarily stored in the storage unit or may be sent directly to the processing unit. The genetic sequence information of the target may be the entire genome sequence of the target, or may be only a partial sequence of interest, or only SNP information. An instruction for processing in the processing unit can be given from the input unit via the operation unit.

[0054] The output unit 15 includes an interface. The interface outputs the skin type and skin condition output by the learning device. In the learning device, information on SNPs related to skin type can also be output together with the skin type and skin condition. The output unit may be a display device such as a liquid crystal display that directly displays the results of the learning system, an output means such as a printer, or a communication unit for outputting to an external storage device or via a network.

[0055] The communication unit relates to a communication interface such as a LAN or a port for connecting the information processing device to a network, and can function as an input unit and / or an output unit.

[0056] The storage unit 11 includes a memory device such as a RAM, a ROM, or a flash memory, a fixed disk device such as a hard disk drive, or a portable storage device such as a flexible disk or an optical disk. The storage unit stores data and instructions input from the input unit, processing results in the learning system, and other data, as well as programs, databases, output forms, and other data used in various computer processes. The computer program may be installed from a computer-readable recording medium such as a CD-ROM or a DVD-ROM, or via the Internet. The computer program is installed in the storage unit using a known setup program or the like.

[0057] The processing unit 14 controls the information processing device so that the SNP information and the environmental factor information input from the input unit are input to the learning system via the input acquisition unit, and the skin constitution and skin condition output from the learning system are acquired via the communication unit. The processing unit executes various arithmetic processing according to the program stored in the storage unit. The arithmetic processing is performed by a processor included in the processing unit. This processor includes functional modules that control the input unit, the storage unit, the communication unit, and the output unit, and can perform various controls. Each of these units may be composed of an independent integrated circuit, a microprocessor, firmware, or the like. Information about the presence of SNPs detected by the processing unit is temporarily stored in the storage unit. The processing unit can also operate to detect the presence of SNPs in the genetic sequence information when genetic sequence information of the target is input instead of the SNP information. More specifically, the processing unit can detect the presence of SNPs in the sequence information of the target from the sequence information of SNPs stored in advance in the storage unit and the sequence information of the target input from the input unit. The presence of an SNP can be detected by comparing the sequence information of the SNP stored in the memory unit with the sequence information of the subject input from the input unit, or by searching for the sequence information of the SNP stored in the memory unit in the sequence information of the subject input from the input unit.

[0058] The learning unit 13 uses known statistical analysis techniques and machine learning techniques to learn the relationship between the environmental information and genetic information about SNPs and the actual skin condition data measured in advance. In the present invention, variables are selected using an elastic net, and a learning model is constructed using generalized linear regression (logistic regression, multiple regression).

[0059] In yet another aspect of the invention, the following are provided: rs1800629, rs2108622, rs1047781, rs12203592, rs16891982, rs3760776, rs9340799, rs12913832, rs17822931, rs964184, rs1801133, rs2228479, rs10515552, rs10741657, rs10882272, rs11057830, rs11234027, rs12272004, rs12 377462, rs12785878, rs12931267, rs1540771, rs1667255, rs1993116, rs2060793, rs2227564, rs2282679, rs2298585, rs3829 251, rs41281112, rs4654748, rs492602, rs602662, rs1030868, rs1126643, rs1256062, rs12934922, rs1485766, rs1501299, rs 17577, rs1799724, rs1800012, rs1800414, rs2010963, rs2234693, rs2241145, rs2285053, rs2287074, rs2287076, rs2987983 , rs3918242, rs4065, rs4252125, rs6152, rs7201, rs74653330, rs7501331, rs8110862, rs8326, rs833061, rs1050565, rs17997 The present invention relates to a method and computer for determining skin characteristics based on the presence of at least one SNP selected from the group consisting of rs11568737, rs41303970, rs12051272, rs182052, rs3865188, rs6810075, rs7799039, and rs1137101, as well as a program or method for controlling the computer.Among the listed SNPs, at least one selected from the group consisting of rs1047781, rs1050450, rs1137101, rs12203592, rs12934922, rs182052, rs2287076, rs3918242, rs41303970, rs4252125, rs602662, rs6152, and rs6810075 may be used.

[0060] More specifically, the method for determining skin characteristics includes the following: detecting at least one SNP as described above; A process for determining skin characteristics based on the detected SNPs Includes.

[0061] The computer that determines the skin characteristics more specifically is a storage unit that stores information about at least one SNP and information about skin characteristics associated with the SNP; An input section for inputting the user's DNA information; a processing unit for detecting the presence of the stored SNPs in the input DNA information; An output section that outputs information on skin characteristics related to the detected SNPs Includes.

[0062] More specifically, the program or method for controlling a computer to determine skin characteristics comprises: storing information about the at least one SNP and information about skin characteristics associated with the SNP in a storage unit; The user's DNA information is entered into the input section. causing a processing unit to detect the presence of the stored SNPs in the input DNA information; Outputting information on skin characteristics related to the detected SNPs to an output unit. A further aspect of the present invention relates to a storage medium storing such a control program.

[0063] The computer according to the present invention may be a skin characteristic determination device. The evaluation system and computer according to the present invention may constitute a system including a server and a terminal device. When the server and the terminal device are constituted, the input unit and the output unit are connected to a network via an interface unit or a communication unit, respectively. In this case, the genetic information and / or environmental information of the user is input from an individual terminal device connected to the network, processed in the server, and information on the skin condition and skin characteristics is output to the same or another terminal device. The genetic information of the user may provide genome information or partial sequence information of the genome, or only information on the identified SNPs. When all or part of the genome information is provided, the presence of SNPs is detected based on the SNPs stored in the memory unit in the processing unit of the server. In addition, in the system, a plurality of servers may be distributed and arranged on the network so that the service can be provided in the form of cloud computing, and each server may cooperate to share the processing in the processing unit and the learning unit. This allows the evaluation system to execute processing more efficiently.

[0064] Here, the skin characteristics may be selected from the group consisting of skin shape characteristics, skin type, and skin color characteristics. The skin shape characteristics may further be selected from the group consisting of wrinkles, texture, and firmness. The skin type may further be selected from the group consisting of sebum amount, moisture amount, and skin barrier function. The skin color characteristics may further be selected from the group consisting of skin background color, skin color of ultraviolet-exposed areas, spots, brightness, and yellowness. Cosmetics, supplements, and / or beauty methods can be provided based on the skin characteristics determined by the method for determining skin characteristics of the present invention or the computer that determines skin characteristics.

[0065] Although skin characteristics can be determined by detecting a single SNP, it is preferable to detect multiple SNPs to determine skin characteristics in terms of achieving higher sensitivity.In order to determine skin characteristics based on the detection of multiple SNPs, for example, the SNP is determined by determining whether the SNP has a positive or negative effect on the skin characteristics, and by assigning points according to the presence of the SNP, the SNP score is calculated.By determining the relationship between the SNP score and the skin characteristics in advance, the skin characteristics can be determined from the SNP score.

[0066] The skin characteristics can be determined for at least one selected from the group consisting of skin shape characteristics, skin type, and skin color characteristics. The relationship between the SNP and these skin characteristics may be such that the presence of the SNP in a heterozygous or homozygous state has a negative or positive effect on these skin characteristics. The SNP score can be calculated based on the number of SNPs that have a negative effect and the number of SNPs that have a positive effect. As an example, the SNP score can be calculated by subtracting the number of SNPs that have a negative effect from the number of SNPs that have a positive effect. As another example, the SNP score may be calculated in the reverse manner, that is, by subtracting the number of SNPs that have a positive effect from the number of SNPs that have a negative effect. By previously determining the relationship between the skin characteristics and the SNP score, the skin characteristics of the subject can be determined based on the calculated SNP score.

[0067] When skin characteristics are determined by the skin characteristic determination method of the present invention, counseling can be performed based on information about such skin characteristics. Therefore, one aspect of the present invention also relates to a counseling method. In such counseling, advice can be provided on lifestyle habits and the selection of cosmetics and supplements to be used. As an example, among skin color characteristics, the amount of melanin in the upper arm is related to the original skin color of the person, and the amount of melanin in the face is related to the skin color affected by ultraviolet rays. Therefore, it can be determined that a subject who is prone to low melanin amounts in the upper arm and high melanin amounts in the face through SNP analysis is a subject who is fair-skinned but is susceptible to the effects of ultraviolet rays and therefore requires more care against ultraviolet rays. For such subjects, it is possible to propose lifestyle habits that avoid ultraviolet rays and to propose the use of cosmetics containing stronger ultraviolet protection agents.

[0068] In the case where it is judged that wrinkles are likely to occur, cosmetics or supplements containing ingredients capable of enhancing the production of collagen fibers such as collagen and elastin and elastic fibers and / or inhibiting the decomposition of elastic fibers, such as matrix metalloproteinase inhibitors, can be provided. Such ingredients include extracellular matrix decomposition inhibitors, such as matrix metalloproteinase inhibitors, and more specifically collagenase inhibitors. Specific ingredients include retinol, winged bean extract, mangosteen extract, etc. (Usefulness of cosmetics: Progress and future prospects of evaluation technology, Yakuji Nipposha, (2001) Section 7, Wrinkle-resistant cosmetics, pp. 162-177).

[0069] If it is determined that the skin is prone to problems related to skin color characteristics such as age spots, cosmetics or supplements containing ingredients that have UV protection, melanin production inhibition, and / or keratin turnover promotion effects can be provided. Examples of such ingredients include UV protection agents, vitamin C derivatives, hydroquinone, tranexamic acid, potassium 4-methoxysalicylic acid, arbutin, kojic acid, rucinol, vitamin C derivatives, etc. (Usefulness of Cosmetics: Progress and Future Prospects of Evaluation Technology, Yakuji Nipposha, (2001), Section 6, Whitening Cosmetics, pp. 144-161). Also, appropriate makeup bases, cosmetics, and makeup methods can be proposed.

[0070] When skin functions such as texture, firmness, sebum, moisture content, and skin barrier function are judged to be poor, cosmetics and supplements containing ingredients that can improve skin functions can be provided. Skin texture improvers, sebum reducers, moisture content increase promoters, and skin barrier function improvers are known. Examples of such ingredients include moisturizers, tranexamic acid, and winged bean extract.

[0071] Another aspect of the present invention relates to a kit for determining skin characteristics, comprising the following: rs1800629, rs2108622, rs1047781, rs12203592, rs16891982, rs3760776, rs9340799, rs12913832, rs17822931, rs964184, rs1801133, rs2228479, rs10515552, rs10741657, rs10882272, rs11057830, rs11234027, rs12272004, rs12377462 ... s12785878, rs12931267, rs1540771, rs1667255, rs1993116, rs2060793, rs2227564, rs2282679, rs2298585, rs3829251, rs412 81112, rs4654748, rs492602, rs602662, rs1030868, rs1126643, rs1256062, rs12934922, rs1485766, rs1501299, rs17577, rs17 99724, rs1800012, rs1800414, rs2010963, rs2234693, rs2241145, rs2285053, rs2287074, rs2287076, rs2987983, rs3918242, rs4065, rs4252125, rs6152, rs7201, rs74653330, rs7501331, rs8110862, rs8326, rs833061, rs1050565, rs1799750, rs4880, rs The kit includes a nucleic acid reagent for detecting at least one SNP selected from the group consisting of rs1050450, rs6058017, rs1061622, rs2046571, rs2232228, rs3785079, rs2246416, rs1107946, rs11568737, rs41303970, rs12051272, rs182052, rs3865188, rs6810075, rs7799039, and rs1137101, or a SNP in linkage disequilibrium therewith. Such a kit may further include a nucleic acid extraction reagent and a purification reagent. Also, the kit may include an instruction manual for explaining the relationship between the detected SNP and the skin condition.Among the listed SNPs, at least one selected from the group consisting of rs1047781, rs1050450, rs1137101, rs12203592, rs12934922, rs182052, rs2287076, rs3918242, rs41303970, rs4252125, rs602662, rs6152, and rs6810075 may be used.

[0072] An example of such a nucleic acid reagent includes a primer capable of selectively amplifying an SNP, a DNA polymerase, dNTPs, and optionally an amplification buffer. The primer capable of amplifying an SNP can be arbitrarily designed for each SNP.

[0073] When a nucleic acid probe is used as the nucleic acid reagent, the kit may further include a probe detection reagent. Such a nucleic acid probe may be an unfixed probe, or may be fixed on a DNA chip, so that multiple SNPs can be detected simultaneously. Such a nucleic acid probe may be labeled so that a specific SNP can be detected.

[0074] All documents mentioned herein are incorporated by reference in their entirety.

[0075] The following examples of the present invention are for illustrative purposes only and do not limit the technical scope of the present invention. The technical scope of the present invention is limited only by the claims. The present invention can be modified, for example, by adding, deleting, and replacing the constituent elements of the present invention, without departing from the spirit of the present invention. EXAMPLES

[0076] Example 1: Analysis subjects The subjects were female volunteers aged 20-79 years. Prior to the analysis, a questionnaire survey was conducted on age, height, weight, BMI (calculated from height and weight according to the standard method), UV exposure information, and smoking information. Regarding UV exposure information, current awareness of sunburn was classified into four stages (two stages of sunburn orientation, two stages of non-sunburn orientation), and further, in order to grasp the response to UV rays in detail, the survey was conducted by classifying the sunburn prevention measures taken from birth to the present (1-14 years old, 15-19 years old, 20-24 years old, 25-29 years old, and thereafter in 10-year increments) into three stages (actively taking no UV protection measures, taking UV protection measures against strong sunlight, and taking UV protection measures against weak sunlight). Regarding smoking information, the survey was conducted by dividing the subjects into three groups: no smoking history, smoking history of 20 or more cigarettes per day in the past, and smoking history of less than 20 cigarettes per day in the past.

[0077] Example 2: Acquisition of actual skin measurement data of analysis subjects Skin characteristics were measured for female volunteers aged 20 to 79 years old, including skin wrinkles, skin pigmentation (cheek melanin amount, skin melanin amount (inside upper arm), brightness, and yolk), and skin condition (barrier function, sebum amount, firmness, and texture). For each measurement, a Visia Evolution (Canfield Scientific) was used to measure wrinkles and pigmentation, and the index values ​​for wrinkles and pigmentation were calculated from the images taken using a dedicated analysis method. For pigmentation, the number of pigmentations and the area of ​​pigmentation were calculated using a skin image measuring device and used for analysis. Skin color was analyzed using a spectrophotometer CM-700d (Konica Minolta), and the melanin amount, brightness (L* colorimetric value), and yellowness (b* colorimetric value) of the cheeks were measured. In addition, the skin color of the inner upper arm, which is not exposed to sunlight, was also measured, and the amount of melanin was calculated. The skin condition was measured using specialized commercially available equipment to measure the stratum corneum moisture content (Corneometer), transepidermal water loss (barrier function) (Vapometer), sebum content (Sebumeter), and firmness (Cutometer), while the texture condition was measured using an original measuring device, Skinvisiom, and specialized analysis software to measure the degree of texture irregularity. For wrinkles around the eyes, a replica of the skin surface was made using silicon, and measurements were taken in 3D. The data was then imported into a computer, and the images were processed using a proprietary analysis system to calculate the depth, volume, and area of ​​the wrinkles.

[0078] Example 3: Determination of SNPs in the Analyte The following 79 SNPs were selected for analysis because they are expected to affect skin characteristics based on previous dermatological findings: rs1800629, rs2108622, rs1047781, rs12203592, rs16891982, rs3760776, rs9340799, rs12913832, rs17822931, rs96 4184, rs1801133, rs2228479, rs10515552, rs10741657, rs10882272, rs11057830, rs11234027, rs12272004, rs12377 462, rs12785878, rs12931267, rs1540771, rs1667255, rs1993116, rs2060793, rs2227564, rs2282679, rs2298585, rs 3829251, rs41281112, rs4654748, rs492602, rs602662, rs1030868, rs1126643, rs1256062, rs12934922, rs1485766, r s1501299, rs17577, rs1799724, rs1800012, rs1800414, rs2010963, rs2234693, rs2241145, rs2285053, rs2287074, r s2287076, rs2987983, rs3918242, rs4065, rs4252125, rs6152, rs7201, rs74653330, rs7501331, rs8110862, rs8326, r s833061, rs1050565, rs1799750, rs4880, rs1050450, rs6058017, rs1061622, rs2046571, rs2232228, rs3785079, rs2 246416, rs1107946, rs11568737, rs41303970, rs12051272, rs182052, rs3865188, rs6810075, rs7799039, rs1137101. The selection criteria were factors that function in the epidermis, basement membrane, dermis, subcutaneous adipose tissue, and the whole body (hormones, vitamins, etc.), enzymes, extracellular matrix proteins, etc., and from these, genes with SNPs were selected as candidates. These SNPs may affect various skin traits. Next, for each of the above 79 SNPs, four types of genetic models, including recessive model, dominant model, additive model, and synergistic model, were set, and the SNPs that considered a total of 316 types of genetic models for the 79 SNPs were used as genetic information. Using this genetic information and environmental information as explanatory variables, variable selection was performed using Elastic Net to determine the optimal genetic model for each SNP. It is expected that the accuracy of determining skin type by SNPs will be improved by determining the genetic model prior to creating the learning model of the present invention.

[0079] In determining the SNPs of the analysis subjects, saliva was used as a sample. The saliva was collected using Oragene (registered trademark) DNA OG-500 (DNA Genotek Inc.), and DNA was stabilized. DNA was purified from the saliva, and SNPs were determined using a DNA array to obtain information on the above 79 preselected SNPs.

[0080] Example 4: Creation of a learning model for determining a skin condition score using skin condition measurements as a response variable A learning model for determining a skin condition score was created by performing multiple regression analysis using skin condition measurements selected from the group consisting of wrinkles, age spots, cheek melanin amount, scalp melanin amount (upper inner arm), brightness, yolk, barrier function, sebum amount, firmness, texture, wrinkle area, wrinkle volume, and maximum wrinkle depth as objective variables, and genetic information about SNPs and information about environmental factors such as age, height, weight, BMI, UV exposure information, and smoking information as explanatory variables (Figure 2). The number of SNPs used in creating the learning model was n. n = 1 to 4 and 5 to 10 were created, respectively. In this learning model for determining a skin condition score, a constant term is determined for each explanatory variable, and the skin condition, which is the objective variable, can be output by inputting the explanatory variables. At this time, by changing one or more of the explanatory variables (e.g., age, UV exposure information, smoking information, etc.), it is possible to predict future changes in skin condition.

[0081] Example 5: Creation of a learning model for classifying skin type into a high value group and a non-high value group (or a low value group and a non-low value group) and using the classification result as a response variable Based on the skin condition measurements selected from the group consisting of wrinkles, age spots, cheek melanin amount, scalp melanin amount (inner upper arm), brightness, yolk, barrier function, sebum amount, firmness, and texture, high and low value groups were determined for each age group from the distribution of each measurement value. Specifically, in the high value group model that judges the susceptibility of skin type (for example, the susceptibility to age spots if the target skin type is age spots), the skin measurement value obtained by skin test is defined as 1 if it is in the top 25% of all data included in each age group (5-year age range), and the other cases (non-high value group) are defined as 0 (Figure 3). In addition, in the low value group model that judges the difficulty of skin type (for example, the difficulty of age spots if the target skin type is age spots), the skin measurement value obtained by skin test is defined as 1 if it is in the bottom 25% of all data included in each age group (5-year age range), and the other cases are defined as 0 (Figure 3).

[0082] In the case of the high value group judgment model, the high value group was set to 1 and the non-high value group to 0 as the objective variable, and information on 79 types of SNPs and information on environmental factors based on the questionnaire results were input as explanatory variables, and a logistic regression analysis was performed to create a learning model for determining skin type. In creating the learning model, the number of SNPs used was n. n = 1 to 4 and 5 to 10 were created, respectively. In the learning model for determining skin type, a constant term was determined for each explanatory variable, and by inputting the explanatory variables, it is possible to output the objective variable, which is the likelihood of having a skin type (the possibility of belonging to the high value group of skin conditions) and / or the difficulty of having a skin type (the possibility of belonging to the low value group of skin conditions).

[0083] Example 6: Validation of the learning model In creating the learning model, 80% of the dataset to be analyzed was divided into a learning dataset and 20% was divided into a validation dataset. In this way, 100 sets of cross-validation datasets consisting of the learning dataset and the validation dataset were created. Based on the operations of Examples 4 and 5, a learning model for determining a skin condition score and a learning model for determining a skin constitution were created, and as a result, 100 sets of learning models were created for each of the skin conditions. The SNPs used in each of the learning models created for each skin condition were as follows. In the table below, the SNPs used in the learning model for determining 100 skin condition scores are listed in the "linear" column. In addition, the SNPs used in the learning model for determining 100 skin constitutions that determine the susceptibility of the skin constitution (possibility of belonging to the high skin condition group) are listed in the "upper" column. In addition, the SNPs used in the learning model for determining 100 skin constitutions that determine the difficulty of the skin constitution (possibility of belonging to the low skin condition group) are listed in the "lower" column.

[0084] [Table 20] [Table 21] [Table 22] [Table 23]

[0085] [Table 24]

[0086] [Table 25]

[0087] [Table 26]

Table 27

[0088]

Table 28

Table 29

[0089]

Table 30

[0090]

Table 31

[0091]

Table 32

[0092]

Table 33

[0093]

Table 34

[0094]

Table 35

Table 36

[0095]

Table 37

[0096] [Table 38]

[0097] For the created learning model for determining skin type, the explanatory variables of the validation data were input to obtain the skin type susceptibility (or difficulty) output, and based on the classification of the skin condition of the validation data into a high value group (or a low value group), an ROC curve was obtained, and the area under the ROC curve (AUC) and sensitivity were calculated (Figure 6A). For the sensitivity, the optimal cutoff value obtained from the ROC curve (such as the value of the point at which the distance from the upper left corner of the ROC curve (the point where both the specificity and specificity are 1) is the smallest) or a value set independently of the ROC curve (for example, since the skin type score obtained from the prediction model is a probability value, the cutoff value is set to 0.5, etc.) was used. In the learning model for determining 100 types of skin types created in this way, the occurrence frequency of SNPs was investigated (Figure 6C). Based on the occurrence frequency of SNPs and biochemical knowledge, SNPs to be used in the learning model were selected, and from the created learning models, a learning model with a high AUC using such SNPs was selected (Figure 5B).

[0098] The skin condition score output by inputting the explanatory variables of the validation data into the learning model that determines the created skin condition score and the coefficient of determination (R 2 ) was obtained (Figure 6B). The frequency of occurrence of SNPs was examined in the 100 learning models thus created (Figure 6C). Based on the frequency of occurrence of SNPs and biochemical knowledge, SNPs to be used in the learning models were selected, and from the created learning models, R 2 We selected the model with the highest learning rate (Figure 5A).

[0099] Through the above operations, SNPs related to skin condition were selected. SNPs selected when the measured value of skin condition was used as the objective variable were displayed as Linear. SNPs selected when the classification result into the high value group was used as the objective variable were displayed as Upper. SNPs selected when the classification result into the low value group was used as the objective variable were displayed as Lower. SNPs used when the number of SNPs used was 1 to 4 and when it was 5 to 10 are both listed. The skin conditions of Skin01 to 16 and Skin21 to 23, displayed as codes, are as shown in the table below. [Table 39]

[0100] Example 7: Method for evaluating skin condition using a learning model for determining a created skin condition score The skin condition of the subject was evaluated using a learning unit including a learning model for determining a skin condition score created in Example 4 and verified in Example 6. Specifically, genetic information related to SNP and information on environmental factors selected from the group consisting of age, height, weight, BMI, information on exposure to ultraviolet light, and smoking information were input as explanatory variables to a learning unit created to obtain a skin condition score selected from the group consisting of age spots 1 to 3, skin color 1 to 4, wrinkles 1 to 2, texture, elasticity, sebum amount, moisture amount 1 to 2, barrier function 1 to 2, wrinkle area, wrinkle volume, and maximum wrinkle depth. The learning unit inputs genetic information and environmental information as explanatory variables into the formula of the learning model to output a score of the current skin condition. Furthermore, by changing age and behavioral factors in the formula of the learning model, future changes in the skin condition were predicted. By selecting sunburn measures and smoking behavior as behavioral factors, it became possible to predict the score of the skin condition relative to age changes (FIGS. 9A and B).

[0101] Example 8: Determination of skin type using a created learning model for determining skin type The skin condition of the subject was evaluated using a learning unit including a learning model for determining skin constitution created in Example 5 and verified in Example 6. Specifically, genetic information related to SNP and information on environmental factors selected from the group consisting of age, height, weight, BMI, information on exposure to ultraviolet rays, and smoking information were input to a learning unit created to obtain a skin condition score selected from the group consisting of age spots 1 to 3, skin color 1 to 4, wrinkles 1 to 2, texture, elasticity, sebum amount, moisture amount 1 to 2, barrier function 1 to 2, wrinkle area, wrinkle volume, and maximum wrinkle depth. The learning unit inputs explanatory variables related to genetic information and environmental information into the mathematical formula of the learning model to output the likelihood of developing a skin constitution (possibility of belonging to a high value group of skin conditions) and / or the difficulty of developing a skin constitution (possibility of belonging to a low value group of skin conditions), which are the objective variables. The outputted likelihood of developing a skin type (likelihood of belonging to the high skin condition group) and / or unlikelihood of developing a skin type (likelihood of belonging to the low skin condition group) were displayed as a plot diagram and a fan graph, respectively (Figures 10A and 10B).

[0102] Explanation of symbols 10. Information processing device 11 Storage section 12 Input section 13 Learning Department 14 Processing section 15 Output section 20 Rating System 30 Learning Device

Claims

1. A skin evaluation method executed by a computer having a learning unit that determines a skin condition through a learning model that is trained based on the relationship between information on SNPs and information on environmental factors and the skin condition and is weighted to contribute to the skin condition, and an input unit, inputting the subject's SNP information and environmental factor information input from the input unit into a learning unit; causing the learning unit to determine a skin condition of the subject; The method comprising:

2. The method described in claim 1, wherein the learning unit includes a learning model in which information about SNPs and information about environmental factors are used as explanatory variables, and the measured value of skin condition is used as a dependent variable, and each explanatory variable is weighted against the dependent variable by multiple regression analysis.

3. The computer further includes a processing unit and an output unit, causing the learning unit to determine a skin condition as information about the skin condition and to determine a formula using age as a variable; causing the processing unit to create a graph of skin condition over time based on a mathematical formula; and The method according to claim 2 , further comprising a step of causing the output unit to output a graph of the progression of skin condition with respect to age as the processed skin condition.

4. The computer further includes a processing unit and an output unit, and causes the learning unit to determine a skin condition as information about the skin condition and to output a formula with age and behavioral factors as variables; causing the processing unit to create a graph of changes in skin condition score versus age when behavioral factors are changed based on a mathematical formula; and The method according to claim 1 , further comprising instructions for causing the output unit to output a graph of skin condition progression with respect to age as the processed skin condition.

5. The method according to claim 4, wherein the behavioral factors are protection against ultraviolet light and smoking habits.

6. The method according to claim 1, wherein the learning unit divides the skin into a high value group and a non-high value group based on the measured skin condition, and the classification result into the high value group and the non-high value group is used as the objective variable, and is weighted by logistic regression using information about the subject's SNPs and information about environmental factors.

7. A method according to claim 1 or 6, wherein the learning unit divides the skin into a low value group and a non-low value group based on the measured skin condition, and the classification result into the low value group and the non-low value group is used as the objective variable, and is weighted by logistic regression using information about the subject's SNPs and information about environmental factors.

8. A skin evaluation system including a processing unit, an input unit, a learning unit, and an output unit, an input section for inputting information on an individual's SNPs and information on environmental factors selected from the group consisting of age, height, weight, BMI, information on exposure to ultraviolet light, and smoking information; a learning unit that determines information about the skin condition using a learning model that is trained based on the relationship between the information about SNPs and the information about environmental factors and the skin condition and that is weighted to contribute to the skin condition; A processing unit that performs processing on the skin condition determined by the learning unit; and An output section that outputs the processed skin condition a learning unit that inputs information about the individual's SNPs and information about the environmental factors input from an input unit, the learning unit determines information about the skin condition, the processing unit processes the information about the determined skin condition, and the output unit outputs the processed skin condition.

9. The evaluation system according to claim 8, wherein the learning unit includes a learning model in which information about SNPs and information about environmental factors are used as explanatory variables, and skin condition measurements are used as objective variables, and each explanatory variable is weighted relative to the objective variable by multiple regression analysis.

10. The learning unit determines a skin condition as information about the skin condition and determines a formula using age as a variable; The processing unit creates a graph of the skin condition over time based on a mathematical formula, The output unit outputs a graph of the change in skin condition with respect to age as the processed skin condition. The evaluation system of claim 9 .

11. The learning unit determines a skin condition as information about the skin condition, and outputs a formula with age and behavioral factors as variables; the processing unit creates a graph showing a change in skin condition score versus age when a behavioral factor is changed based on a mathematical formula; The output unit outputs a graph of the change in skin condition with respect to age as the processed skin condition. The evaluation system of claim 9 .

12. The evaluation system according to claim 11 , wherein the behavioral factors are measures against ultraviolet rays and smoking habits.

13. The evaluation system according to claim 8, wherein the learning unit divides the skin into a high value group and a non-high value group determined based on the measured values ​​of the skin condition, and the classification results into the high value group and the non-high value group are used as the objective variable, and are weighted by logistic regression using information about the subject's SNPs and information about environmental factors.

14. The evaluation system according to claim 8 or 13, wherein the learning unit divides the skin into a low value group and a non-low value group determined based on the measured values ​​of the skin condition, and the classification result into the low value group and the non-low value group is used as the objective variable, and is weighted by logistic regression based on information about the subject's SNPs and information about environmental factors.

15. A program for controlling a skin evaluation system including a learning unit that determines a skin condition through a learning model that is learned based on a relationship between information on SNPs and information on environmental factors and the skin condition and that is weighted to contribute to the skin condition, an input unit, an output unit, and a processing unit, The following directive: inputting information on the individual's SNPs input from the input unit and information on environmental factors selected from the group consisting of age, height, weight, BMI, information on exposure to ultraviolet light, and smoking information into a learning unit; A learning unit determines information on a skin condition determined through a learning model in which the input information on SNPs and information on environmental factors are weighted to contribute to skin constitution; causing a processing unit to process the information regarding the determined skin condition; The processed skin condition is output to the output unit. The program comprising:

16. The program according to claim 15, wherein the learning unit includes a learning model in which information about SNPs and information about environmental factors are used as explanatory variables, and skin condition measurements are used as objective variables, and each explanatory variable is weighted relative to the objective variable by multiple regression analysis.

17. causing the learning unit to determine a skin condition as information about the skin condition and to determine a formula using age as a variable; causing the processing unit to create a graph of skin condition over time based on a mathematical formula; and The program according to claim 15, further comprising instructions for causing the output unit to output a graph of a change in skin condition with respect to age as the processed skin condition.

18. causing the learning unit to determine a skin condition as information about the skin condition and output a formula with age and behavioral factors as variables; causing the processing unit to create a graph of changes in skin condition score versus age when behavioral factors are changed based on a mathematical formula; and The program according to claim 15, further comprising instructions for causing the output unit to output a graph of a change in skin condition with respect to age as the processed skin condition.

19. 20. The program of claim 18, wherein the behavioral factors are protection from ultraviolet light and smoking habits.

20. The program according to claim 15, wherein the learning unit divides the skin into a high value group and a non-high value group determined based on the skin condition measurement values, and the classification results into the high value group and the non-high value group are used as the objective variable, and are weighted by logistic regression using information about the subject's SNPs and information about environmental factors.

21. The program according to claim 15 or 20, wherein the learning unit divides the skin into a low value group and a non-low value group based on the measured values ​​of the skin condition, and the classification result into the low value group and the non-low value group is used as the objective variable, and is weighted by logistic regression using information about the subject's SNPs and information about environmental factors.

22. An information processing device for performing skin evaluation, comprising a processing unit, an input unit, a learning unit, and an output unit, comprising: the input unit to which information on an individual's SNPs and information on environmental factors selected from the group consisting of age, height, weight, BMI, information on exposure to ultraviolet light, and smoking information are input; a learning unit that determines information about the skin condition using a learning model that is trained based on the relationship between the information about SNPs and the information about environmental factors and the skin condition and that is weighted to contribute to the skin condition; A processing unit that performs processing on the skin condition determined by the learning unit; and An output section that outputs the processed skin condition The information processing device includes: information about the individual's SNPs and information about the environmental factors input from an input unit to a learning unit; information about the skin condition is determined in the learning unit; the information about the determined skin condition is processed in a processing unit; and the processed skin condition is output from an output unit.

Images