Food product evaluation device, food product evaluation method, and food product evaluation program

Abstract

In the present invention, a protein to be evaluated is specified and the results of predicting digestibility of the protein to be evaluated are obtained using a machine learning model that uses the amino acid composition of the protein as an explanatory variable and the protein digestibility as an objective variable.

Description

Food evaluation device, food evaluation method, and food evaluation program

[0001] The present invention relates to a food evaluation device, a food evaluation method, and a food evaluation program.

[0002] Non-Patent Document 1 discloses a technology for evaluating the amount of digestible and usable amino acids using the Digestible Indispensable Amino Acid Score (DIAAS), which is an amino acid score proposed by the Food and Agriculture Organization of the United Nations (FAO) in 2013 to evaluate the balance of essential amino acids in protein materials and related products, and which is formed by the product of pig ileal digestibility data (Standard ileal digestibility or True ileal digestibility) obtained by conducting in vivo tests and the essential amino acid content and digestibility of the evaluation product.

[0003] Dietary protein quality evaluation in human nutrition. Report of an FAO Expert Consultation, FAO Food Nutrition Paper, 2013;92:1-66.

[0004] However, in the conventional inventions, the digestibility data of the evaluation items is obtained by conducting in vivo tests, which has been problematic in terms of animal ethics, test costs, and low high-throughput.

[0005] The present invention has been made in consideration of the above-mentioned problems, and aims to provide a food evaluation device, a food evaluation method, and a food evaluation program that can predict digestibility using an evaluation protein material, evaluate the amount of digestible amino acids by multiplying the content of various amino acids by digestibility, and provide a method for utilizing the evaluation results.

[0006] In order to solve the above-mentioned problems and achieve the objective, a food evaluation device is provided which comprises a memory unit and a control unit, wherein the memory unit comprises a protein memory means which stores a machine learning model which uses the amino acid composition of the protein as an explanatory variable and protein digestibility as a target variable, and the control unit comprises an evaluation target setting means which sets the protein to be evaluated, and a prediction result acquisition means which uses the machine learning model to acquire a digestibility prediction result of the protein digestibility of the evaluation target.

[0007] In the food evaluation device according to the present invention, the explanatory variables include the type of the protein.

[0008] In the food evaluation device according to the present invention, the type of protein is a taxonomic rank of animal or plant family, genus, or species.

[0009] In the food evaluation device according to the present invention, the explanatory variables include at least one of the type of cooking method, temperature, and heating time.

[0010] In addition, in the food evaluation device of the present invention, the explanatory variables are characterized by including formulation data including water content and / or hardness, physical properties including solubility and / or emulsifying power, nutrient content including dietary fiber content, fat content, sugar content and / or mineral content, plant and / or animal part data, and / or source data of the country, region and / or manufacturer.

[0011] In the food evaluation device according to the present invention, the protein is a protein used as a cooking ingredient or a protein contained in a single dish.

[0012] In addition, in the food evaluation device of the present invention, the prediction result acquisition means is further characterized in that it uses the digestibility prediction result to calculate the amount of digestible amino acids, which is the value obtained by multiplying the content of various amino acids by the protein digestibility.

[0013] In addition, in the food evaluation device of the present invention, the prediction result acquisition means is further characterized in that it uses the digestibility prediction result to calculate the amount of digestible essential amino acids, which is the value obtained by multiplying the content of various essential amino acids and the protein digestibility, the digestible essential amino acid standard ratio, which is the value obtained by dividing the amount of digestible essential amino acids by a standard value, or the digestible essential amino acid score, which is the smallest value of the digestible essential amino acid standard ratios.

[0014] In addition, in the food evaluation device of the present invention, the control unit is characterized by further comprising a correction means for correcting the digestibility prediction result of the protein digestibility of the protein contained in the single dish to be evaluated, based on changes in digestibility due to the combination of ingredients that make up the single dish.

[0015] In addition, in the food evaluation device of the present invention, the control unit is characterized by further comprising a correction means for correcting the digestibility prediction result of the protein digestibility of the protein contained in the single dish to be evaluated, based on changes in digestibility due to eating the single dish with other dishes that make up a menu including the single dish.

[0016] In addition, in the food evaluation device of the present invention, the control unit is characterized by further comprising a correction means for correcting the digestibility prediction result of the protein digestibility of the subject to be evaluated based on changes in digestibility due to exercise before and after ingestion of the subject to be evaluated.

[0017] In addition, in the food evaluation device of the present invention, the control unit is characterized by further comprising a correction means for correcting the digestibility prediction result of the protein digestibility of the evaluation object based on changes in digestibility due to the constitution of the person ingesting the evaluation object.

[0018] In addition, in the food evaluation device of the present invention, the prediction result acquisition means is further characterized in that it acquires meal content suggestion data based on the digestibility prediction result and nutrient data including dietary reference intakes.

[0019] In the food evaluation device according to the present invention, the prediction result acquisition means may further display the digestibility prediction result.

[0020] In the food evaluation device according to the present invention, the evaluation target is the protein ingested in a predetermined period of time in an amount of one meal or more.

[0021] In addition, in the food evaluation device of the present invention, the prediction result acquisition means is characterized in that it acquires the meal content proposal data based on the digestibility prediction result and the nutrient data including the dietary reference intakes according to the attributes and / or purpose of intake of the person consuming the evaluation object.

[0022] In addition, in the food evaluation device of the present invention, the prediction result acquisition means is further characterized in that it creates product development data for a product that reflects the protein digestibility of the evaluation target based on the digestibility prediction result.

[0023] In addition, in the food evaluation device according to the present invention, the machine learning model is a model consisting of at least one of random forest, neural network, XGBoost, LightGBM, support vector regression, and linear regression including ElasticNet.

[0024] In addition, the food evaluation device of the present invention is a food evaluation device comprising a memory unit and a control unit, wherein the memory unit comprises a protein memory means for storing a machine learning model that uses the amino acid composition of a protein as an explanatory variable and the amount of digestible amino acids, the amount of digestible essential amino acids, the digestible essential amino acid standard ratio or the digestible essential amino acid score as a target variable, and the control unit comprises an evaluation target setting means for setting the protein to be evaluated, and a prediction result acquisition means for using the machine learning model to acquire the amount of digestible amino acids, the amount of digestible essential amino acids, the digestible essential amino acid standard ratio or the digestible essential amino acid score of the evaluation target.

[0025] Further, the food evaluation device according to the present invention is a food evaluation device comprising a memory unit and a control unit, wherein the memory unit comprises protein storage means for storing a dataset in which, as training data, at least one of the following is used as explanatory variables: amino acid composition of protein, type of protein, formulation data including water content and / or hardness, physical and chemical properties including solubility and / or emulsifying power, nutrient content including dietary fiber amount, fat amount, sugar amount and / or mineral amount, plant and / or animal part data, source data of country, region and / or manufacturer, type of processing and cooking method, cooking temperature, cooking time, and in vitro analysis results regarding enzyme activity inhibition and / or digestibility, and protein digestibility in an in vitro process or an in vivo process is used as a response variable; and the control unit comprises model construction means for training a machine learning model using the training data.

[0026] Furthermore, the food evaluation method of the present invention is a food evaluation method to be executed by a food evaluation device having a memory unit and a control unit, wherein the memory unit comprises a protein memory means for storing a machine learning model in which the amino acid composition of the protein is an explanatory variable and protein digestibility is an objective variable, and the food evaluation method is characterized in that it includes an evaluation target setting step executed in the control unit for setting the protein to be evaluated, and a prediction result acquisition step for using the machine learning model to obtain a digestibility prediction result of the protein digestibility of the evaluation target.

[0027] The food evaluation method of the present invention is a food evaluation method to be executed by a food evaluation device having a memory unit and a control unit, wherein the memory unit is equipped with protein memory means for storing a dataset in which, as training data, at least one of the following is used as explanatory variables: amino acid composition of protein, type of protein, formulation data including water content and / or hardness, physical and chemical properties including solubility and / or emulsifying power, nutrient content including dietary fiber amount, fat amount, sugar amount and / or mineral amount, plant and / or animal part data, source data of country, region and / or manufacturer, type of processing and cooking method, cooking temperature, cooking time, and in vitro analysis results regarding enzyme activity inhibition and / or digestibility, and protein digestibility in an in vitro process or an in vivo process is used as a target variable, and the method includes a model construction step, executed by the control unit, in which a machine learning model learns the training data.

[0028] In addition, the food evaluation program of the present invention is a food evaluation program to be executed by a food evaluation device having a memory unit and a control unit, wherein the memory unit has a protein memory means for storing a machine learning model in which the amino acid composition of the protein is an explanatory variable and protein digestibility is an objective variable, and the control unit executes an evaluation target setting step for setting the protein to be evaluated, and a prediction result acquisition step for using the machine learning model to obtain a digestibility prediction result of the protein digestibility of the evaluation target.

[0029] The present invention has the effect of being able to predict digestibility by incorporating nutrient information other than ingredients and protein, such as ash and moisture content. Furthermore, the present invention has the effect of being applicable to solving health issues such as decreased digestibility due to elderly people, athletes, illness, and food combinations, by proposing menus that maximize the use of vegetable protein while taking DIAAS into consideration. Furthermore, the present invention has the effect of being able to estimate the DIAAS of ingredients (including vegetable proteins) that have been subjected to various processes such as cooking and processing, thereby contributing to the development of high-quality protein products. Furthermore, the present invention has the effect of being able to estimate DIAAS at the product level.

[0030] FIG. 1 is a block diagram showing an example of the configuration of a food evaluation device according to this embodiment. FIG. 2 is a flowchart showing an example of a food evaluation process according to this embodiment. FIG. 3 is a diagram showing an example of data collection according to this embodiment. FIG. 4 is a diagram showing an example of model construction according to this embodiment. FIG. 5 is a diagram showing an example of protein digestibility prediction according to this embodiment. FIG. 6 is a diagram showing an example of digestible essential amino acid standard ratio prediction according to this embodiment. FIG. 7 is a diagram showing an example of differences between evaluation methods according to this embodiment. FIG. 8 is a diagram showing an example of assumptions made in each evaluation method according to this embodiment. FIG. 9 is a diagram showing an example of biological species of digestibility data according to this embodiment. FIG. 10 is a diagram showing an example of major classifications of protein types according to this embodiment. FIG. 11 is a diagram showing an example of minor classifications of protein types according to this embodiment. FIG. 12 is a diagram showing an example of types of cooking and processing methods according to this embodiment. FIG. 13 is a diagram showing an example of protein digestibility prediction according to this embodiment. FIG. 14 is a diagram showing an example of the coefficient of determination according to this embodiment. FIG. 15 is a diagram showing an example of the coefficient of determination according to this embodiment. FIG. 16 is a diagram showing an example of the coefficient of determination according to this embodiment. FIG. 17 is a diagram showing an example of the coefficient of determination according to this embodiment. FIG. 18 is a diagram showing an example of a digestible essential amino acid standard ratio in this embodiment. FIG. 19 is a diagram showing an example of a prediction accuracy verification result in this embodiment. FIG. 20 is a diagram showing an example of a prediction accuracy verification result in this embodiment. FIG. 21 is a diagram showing an example of a prediction accuracy verification result in this embodiment. FIG. 22 is a diagram showing an example of a prediction accuracy verification result in this embodiment. FIG. 23 is a diagram showing an example of an explanatory variable importance analysis result in this embodiment. FIG. 24 is a diagram showing an example of an explanatory variable importance analysis result in this embodiment. FIG. 25 is a diagram showing an example of a correlation in this embodiment. FIG. 26 is a diagram showing an example of a correlation in this embodiment. FIG. 27 is a diagram showing an example of a correlation in this embodiment.

[0031] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described in detail with reference to the accompanying drawings. However, the present invention is not limited to this embodiment.

[0032] [1. Overview] First, an overview of the present invention will be described.

[0033] While there have traditionally been several amino acid scores, a new one, DIAAS, was established by the FAO in 2013. DIAAS is an amino acid score that takes into account the recommended daily intake of amino acids and the digestibility of proteins. Here, DIAAS is the minimum value among the digestible essential amino acid standard ratios of various essential amino acids. Here, the digestible essential amino acid standard ratio (DIAA) is calculated by the formula: "digestible essential amino acid amount / standard value." Here, the digestible essential amino acid amount is calculated by the formula: "content of various essential amino acids (mg / g protein) x digestibility (%)." Here, the standard value is defined based on a good amino acid balance (described in the FAO DIAAS guidelines). Furthermore, DIAAS changes depending on the type of food ingredient, processing (e.g., freezing and thawing), or cooking (e.g., baking). The amount of digestible essential amino acids is calculated by "(content of various essential amino acids (mg / g protein) x digestibility (%))", and the amount of digestible amino acids is calculated by "(content of various amino acids (mg / g protein) x digestibility (%))".

[0034] Here, conventionally, calculation of DIAAS requires three points: (1) the amount of amino acids in the protein, (2) the recommended intake amount of amino acids provided by the FAO, and (3) the in vivo digestibility of the protein, which needs to be measured in living organisms (such as pigs). However, evaluation is generally difficult due to the lack of data on (3). Therefore, conventionally, only the high DIAAS of animal proteins has been promoted based on publicly known DIAAS evaluation results.

[0035] Therefore, in this embodiment, a mechanism is provided that enables evaluation and prediction of DIAAS even under unknown conditions that cannot be evaluated in vivo by predicting protein digestibility using an in silico prediction model.

[0036] That is, this embodiment aims to provide a method for predicting digestibility using the type of protein material to be evaluated (e.g., the genus or family of plant protein), cooking / processing method (e.g., fermentation, boiling, or freezing) and quantitative data (e.g., boiling time, heating temperature, or grinding size)), and amino acid composition, and then evaluating the DIAAS of materials or products or assessing the amount of digestible essential amino acids in dietary records, etc., as well as a method for utilizing the evaluation results. The amino acid composition and digestibility data utilize measurement results for 18 amino acids (Arg, His, Ile, Leu, Lys, Met, Phe, Thr, Trp, Val, Ala, Asp, Cys, Glu, Gly, Ser, Tyr, and Pro). The digestibility data utilizes results from in vivo tests (humans, pigs, rodents, etc.) and artificial intestinal fluid tests (in vitro results).

[0037] Here, the prediction technology in this embodiment is characterized by constructing a prediction model by machine learning (random forest, neural network, XGBoost, LightGBM, support vector regression, and / or linear regression including ElasticNet, etc.) using digestibility data as the objective variable and amino acid composition, cooking / processing method, and protein type as explanatory variables. The method for evaluating the digestible amino acid content in the DIAAS evaluation of ingredients and products, or in dietary records, etc., compares the ratio of the predicted amino acid utilization amount to the recommended amino acid utilization amount proposed by the FAO with the ratio in a reference protein material based on the predicted digestibility data, and makes an evaluation.

[0038] [2. Configuration of the food evaluation device 100] The food evaluation device 100 according to this embodiment can be configured by functionally or physically distributing or integrating any unit (either as a stand-alone type or a system type). In this embodiment, an example of the configuration of the food evaluation device 100 will be described with reference to Fig. 1. Fig. 1 is a block diagram showing an example of the configuration of the food evaluation device 100 according to this embodiment.

[0039] 1, the food evaluation device 100 may be an information processing device such as a personal computer or a workstation. The food evaluation device 100 includes a control unit 102, a storage unit 106, and an input / output unit 112, and the units included in the food evaluation device 100 are communicably connected via any communication path. The food evaluation device 100 is communicably connected to other devices via a network 300.

[0040] The input / output unit 112 may have a function for inputting and outputting data (I / O). Here, the input / output unit 112 may be, for example, a key input unit, a touch panel, a control pad (e.g., a touch pad, a game pad, etc.), a mouse, a keyboard, a microphone, etc. The input / output unit 112 may also be a display unit (e.g., a display, monitor, and touch panel made of liquid crystal or organic electroluminescence, etc.) that displays (input / output) information of application software, etc. The input / output unit 112 may also be an audio output unit (e.g., a speaker, etc.) that outputs audio information as audio. The input / output unit 112 may also be an image input unit (e.g., a camera, etc.) that records images (still images and videos) captured by an imaging element such as a CCD image sensor or a CMOS image sensor as digital data. The input / output unit 112 may also be a fingerprint sensor, a camera (e.g., an infrared camera, etc.) that can be used for iris authentication or face authentication, etc., and / or a biometric sensor such as a vein sensor.

[0041] The storage unit 106 stores various databases, tables, and / or files. The storage unit 106 stores computer programs that cooperate with an operating system (OS) to issue commands to a central processing unit (CPU) to perform various processes. The storage unit 106 may be, for example, a random access memory (RAM), a read-only memory (ROM), a hard disk drive (HDD), and / or a solid state drive (SSD). The storage unit 106 may store image data recorded by the input / output unit 112, data received via the network 300, and / or input data input via the input / output unit 112. The storage unit 106 conceptually includes a protein database 106a.

[0042] The protein database 106a stores protein data. The protein database 106a may also store a machine learning model related to protein. The protein database 106a may also store a machine learning model using the amino acid composition of a protein as an explanatory variable and protein digestibility (protein digestibility) as a target variable. The explanatory variables may include the type of protein. The type of protein may also be the taxonomic rank of an animal or plant, such as a family, genus, or species. The explanatory variables may also include at least one of the type of cooking method, temperature, and heating time. The explanatory variables may also include formulation data including moisture content and / or hardness, physical and chemical properties including solubility and / or emulsifying power, nutrient content including dietary fiber content, fat content, sugar content, and / or mineral content, plant and / or animal part data, and / or source data for a country, region, and / or manufacturer. The protein may be a protein used as a cooking ingredient or a protein contained in a single dish. The machine learning model may be a model consisting of at least one of random forest, neural network, XGBoost, LightGBM, support vector regression, and linear regression including ElasticNet. The protein database 106a may store a dataset (variable data) that serves as explanatory variables and target variables for machine learning. The protein database 106a may store a machine learning model that uses the amino acid composition of the protein as an explanatory variable and the digestible amino acid amount, digestible essential amino acid amount, digestible essential amino acid standard ratio, or digestible essential amino acid score as a target variable.The protein database 106a may also store, as training data, a dataset in which at least one of the following is used as explanatory variables: amino acid composition of the protein, type of protein, formulation data including water content and / or hardness, physical properties including solubility and / or emulsifying power, nutrient content including dietary fiber amount, fat amount, sugar amount and / or mineral amount, plant and / or animal part data, source data of country, region and / or manufacturer, type of processing and cooking method, cooking temperature, cooking time, and in vitro analysis results regarding enzyme activity inhibition and / or digestibility; and protein digestibility in an in vitro process or an in vivo process is used as a target variable.

[0043] The control unit 102 is a CPU or the like that comprehensively controls the food evaluation device 100. The control unit 102 has an internal memory for storing control programs such as an OS, programs that define various processing procedures, required data, etc., and executes various information processing based on these stored programs. Functionally, the control unit 102 conceptually includes a model construction unit 102a, an evaluation target setting unit 102b, a prediction result acquisition unit 102c, and a correction unit 102d.

[0044] The model construction unit 102a constructs a machine learning model. Here, the model construction unit 102a may cause the machine learning model to learn training data. The model construction unit 102a may also register data sets that serve as explanatory variables and target variables for machine learning in the protein database 106a.

[0045] The evaluation target setting unit 102b sets a protein to be evaluated. Here, the evaluation target may be a protein ingested in a predetermined period of time, such as one or more servings.

[0046] The prediction result acquisition unit 102c acquires a digestibility prediction result for the protein digestibility of the protein. Here, the prediction result acquisition unit 102c may acquire the digestibility prediction result for the protein digestibility of the evaluation target using a machine learning model. Furthermore, the prediction result acquisition unit 102c may use the digestibility prediction result to calculate a digestible amino acid amount, which is a value obtained by multiplying the content of various amino acids by the protein digestibility. Furthermore, the prediction result acquisition unit 102c may use the digestibility prediction result to calculate a digestible essential amino acid amount, which is a value obtained by multiplying the content of various essential amino acids by the protein digestibility, a digestible essential amino acid standard ratio, which is a value obtained by dividing the digestible essential amino acid amount by a standard value, or a digestible essential amino acid score, which is the minimum value of the digestible essential amino acid standard ratios. Furthermore, the prediction result acquisition unit 102c may acquire meal content proposal data based on the digestibility prediction result and nutrient data including dietary reference intakes. Furthermore, the prediction result acquisition unit 102c may display the digestibility prediction result. The prediction result acquisition unit 102c may also acquire meal content proposal data based on the digestibility prediction result and nutrient data including dietary reference intakes according to the attributes and / or purpose of the person consuming the evaluation target. The prediction result acquisition unit 102c may also create product development data for a product that reflects the protein digestibility of the evaluation target based on the digestibility prediction result. The prediction result acquisition unit 102c may also acquire the digestible amino acid amount, digestible essential amino acid amount, digestible essential amino acid standard ratio, or digestible essential amino acid score of the evaluation target using a machine learning model.

[0047] The correction unit 102d corrects the digestibility prediction result. Here, the correction unit 102d may correct the digestibility prediction result of the protein digestibility of the protein contained in the a la carte dish to be evaluated based on the change in digestibility due to the combination of the a la carte dish with the constituent dishes that make up the menu including the a la carte dish. The correction unit 102d may also correct the digestibility prediction result of the protein digestibility of the evaluation target based on the change in digestibility due to exercise before and after ingestion of the evaluation target. The correction unit 102d may also correct the digestibility prediction result of the protein digestibility of the evaluation target based on the change in digestibility due to the constitution of the person ingesting the evaluation target. The correction unit 102d may also correct the digestibility prediction result of the protein digestibility of the protein contained in the a la carte dish to be evaluated based on the change in digestibility due to the combination of ingredients that make up the a la carte dish.

[0048] 3. Food Evaluation Processing An example of the food evaluation processing according to this embodiment will be described with reference to Fig. 2 to Fig. 10. Fig. 2 is a flowchart showing an example of the food evaluation processing according to this embodiment.

[0049] As shown in Figure 2, when a user inputs a dataset that will be the explanatory variables and target variables for machine learning via the input / output unit 112, the model construction unit 102a registers the dataset as training data in the protein database 106a (step SA-1).

[0050] Then, based on the variable data stored in the protein database 106a, the model construction unit 102a constructs a machine learning model that uses at least the amino acid composition of the protein as an explanatory variable and the protein digestibility as a target variable (step SA-2).

[0051] Then, when a protein is designated by the user via the input / output unit 112, the evaluation target setting unit 102b sets the protein as the evaluation target (step SA-3).

[0052] Then, the prediction result acquisition unit 102c uses the machine learning model stored in the protein database 106a to acquire the digestibility prediction result of the protein digestibility to be evaluated, and uses the digestibility prediction result to calculate the digestible amino acid amount, digestible essential amino acid amount, digestible essential amino acid standard ratio, and / or digestible essential amino acid score (step SA-4).

[0053] Then, the correction unit 102d corrects the predicted digestibility result of the protein digestibility of the evaluation object based on changes in digestibility due to exercise before and after ingestion of the evaluation object, and / or changes in digestibility due to the constitution of the person ingesting the evaluation object (step SA-5).

[0054] Then, the prediction result acquisition unit 102c displays the digestibility prediction result on the input / output unit 112 (step SA-6), and ends the process.

[0055] An example of a model construction process in this embodiment will now be described with reference to Figures 3 and 4. Figure 3 is a diagram showing an example of data collection in this embodiment. Figure 4 is a diagram showing an example of model construction in this embodiment.

[0056] As shown in FIG. 3, in this embodiment, the type of protein (species, genus), cooking / processing method, amino acid composition of the evaluation product, and protein digestibility data are collected from existing literature and compiled into a database (STEP 1).

[0057] 4, in this embodiment, a protein digestibility prediction model is constructed by performing machine learning using protein digestibility as the objective variable and other data as explanatory variables (STEP 2). Here, the protein digestibility prediction model may not require amino acid composition data as input data, and predictions may be made taking into account changes in digestibility due to food combinations, aging, exercise, etc.

[0058] An example of verification of prediction accuracy in this embodiment will be described with reference to Figures 5 and 6. Figure 5 is a diagram showing an example of protein digestibility prediction in this embodiment. Figure 6 is a diagram showing an example of digestible essential amino acid standard ratio prediction in this embodiment.

[0059] As shown in FIG. 5 , in this embodiment, as shown in the plot of the protein digestibility results (100 predicted values) of four amino acids for 25 types of proteins, the error from the actual measured value remains within 10% in 96 of the 100 conditions.

[0060] As shown in Figure 6, in this embodiment, the predictability of the digestible essential amino acid standard ratio is improved by utilizing the protein digestibility model, as shown by a comparison of the correlation between the digestible essential amino acid standard ratio (conventional) calculated using the protein digestibility correction coefficient (average value for all conditions (86.4%)) and the digestible essential amino acid standard ratio (present) calculated using the digestibility prediction results and the actual measured value.

[0061] An example of an evaluation method in this embodiment will be described with reference to Fig. 7 and Fig. 8. Fig. 7 is a diagram showing an example of a difference between the evaluation methods in this embodiment. Fig. 8 is a diagram showing an example of an assumption in each evaluation method in this embodiment.

[0062] As shown in Figure 7, in this embodiment, although there is a possibility of errors in prediction accuracy due to differences in cooking and processing methods and ingredients, protein digestibility in pigs is assumed to be equivalent to that in humans, and human protein digestibility is predicted using existing pig digestibility data, and various in vitro pig digestibility results are used to predict extrapolated results for humans.

[0063] Furthermore, as shown in FIG. 8 , in this embodiment, (i) in evaluations based on existing digestibility data, although the cooking and processing methods and ingredients are different, it is assumed that the digestibility evaluation results for pigs and the digestibility evaluation results for humans are equivalent; (ii) in evaluations based on digestibility prediction models that predict changes in digestibility due to heating, although the ingredients are different, it is assumed that the digestibility prediction results for pigs predicted from the digestibility evaluation results for pigs and the digestibility evaluation results for humans are equivalent; and (iii) in evaluations based on in vitro digestibility predictions that also predict ingredients not in the database, such as squid and cheese, it is assumed that the digestibility prediction results for pigs extrapolated from in vitro test results and the digestibility evaluation results for humans are equivalent.

[0064] In this embodiment, the explanatory variables of the protein digestibility prediction model may exclude data related to cooking and processing methods, and some of the types of amino acids added to the explanatory variables may be excluded. Furthermore, in this embodiment, in order to refine the digestibility prediction results, formulation data of the evaluation product (e.g., moisture content, hardness, etc.), physical and chemical properties (e.g., solubility, emulsifying power, etc.), nutrient content (e.g., dietary fiber, fat content, carbohydrates, minerals, etc.), protein material data (e.g., plant parts (beans, pods, stems, flowers, etc.), animal parts (ribs, fillets, peaches, etc.), source of acquisition (country, region, manufacturer, etc.)) may be added to the explanatory variables.

[0065] In addition, in this embodiment, the machine learning model may be used in a form in which the data set of explanatory variables is divided into multiple groups depending on the evaluation purpose and the contents of the learning data set, and multiple models, including a model for predicting digestibility and a model for correcting the digestibility prediction results, are linked together.

[0066] Furthermore, in this embodiment, in predicting digestibility data, the construction of a protein digestibility prediction model may be a method of obtaining predicted values ​​using convenient coefficients such as scaling factors based on evidence of digestibility changes analyzed in humans or animals, or a method of correcting digestibility prediction results obtained through machine learning based on evidence of digestibility changes. Here, the evidence used may be digestibility changes due to food combinations (e.g., the effects of digestive enzymes or digestive enzyme inhibitors derived from ingredients, etc.), digestibility changes due to exercise (e.g., the type, duration, and frequency of exercise), or digestibility changes due to constitution (e.g., changes in digestive enzyme secretion or gastrointestinal function due to aging or disease, the effects of amino acid availability in the body, etc.).

[0067] Furthermore, in the food evaluation method of this embodiment, the evaluation subject may include dietary records for various periods such as one meal, one day, one month, etc. Furthermore, in this embodiment, the recommended amino acid utilization amount serving as the reference value may take into account the attributes of the subject to be evaluated (e.g., athlete, disease carrier, etc.) and the purpose of evaluation (e.g., diet, beauty, etc.), and a value other than the reference value proposed by FAO may be used.

[0068] The food evaluation method of this embodiment may also have a function for suggesting meal contents such as recipes and menus based on nutrient-related data such as dietary reference intakes. In this embodiment, meal contents may also be suggested based on dieter data (e.g., psychological data (preferences, mood, etc.)), analysis results (physical measurement results (height, weight, energy expenditure, electroencephalogram, etc.), biochemical test items (blood pressure, metabolites, etc.), genetic analysis (gene sequence and expression, intestinal bacteria, etc.)), or external data (e.g., season, trends, etc.).

[0069] In addition, in the food evaluation method of this embodiment, the results obtained from the prediction of digestibility and / or the digestible amino acid amount, digestible essential amino acid amount, digestible essential amino acid standard ratio and / or digestible essential amino acid score evaluation may be visualized.

[0070] In addition, in this embodiment, it is possible to predict changes in digestibility when a hypothetical cooking / processing method is performed on the evaluation material, and based on the digestibility prediction results, it is possible to develop and improve cooking / processing methods that are intended to change the digestibility, digestible amino acid amount, digestible essential amino acid amount, digestible essential amino acid standard ratio, or digestible essential amino acid score of the evaluation material, as well as to develop products (e.g., food development, etc.).

[0071] Example 1 Example 1 will be described with reference to Figs. 9 to 18. Fig. 9 is a diagram showing an example of biological species of digestibility data in this embodiment. Fig. 10 is a diagram showing an example of major classification of protein types in this embodiment. Fig. 11 is a diagram showing an example of minor classification of protein types in this embodiment. Fig. 12 is a diagram showing an example of types of cooking and processing methods in this embodiment. Figs. 13 and 18 are diagrams showing an example of protein digestibility prediction in this embodiment. Figs. 14 to 17 are diagrams showing examples of coefficients of determination in this embodiment.

[0072] In this example, the accuracy of digestibility prediction was verified for 119 conditions of digestibility test results for pigs, rats, etc., which were described in previous literature.

[0073] (1) Digestibility Data Collection First, in this example, digestibility test results (119 conditions) for pigs, rats, etc. published in PubMed (http: / / www.ncbi.nlm.nih.gov / pubmed / ) were collected, and the DIAA and amino acid composition of the evaluation materials were classified according to the type of evaluation material, as shown in Figures 9 to 12.

[0074] (2) Data Preprocessing Some published literature describing the digestibility and amino acid composition used in this example does not include DIAA, and the amino acid composition is described using different evaluation methods or units depending on whether or not the water content in the evaluation sample is taken into account. Furthermore, in this example, missing values ​​for DIAA were calculated and filled in according to FAO guidance (35978-02317b979a686a57aa4593304ffc17f06.pdf (fao.org)). Furthermore, in this example, the amino acid composition was converted into a form that could be uniformly subjected to machine learning by calculations that aligned the data to units common to all literature.

[0075] (3) Digestibility Prediction Model Construction In this example, a digestibility prediction random forest model was constructed using the programming language "R." In this example, digestibility data (quantitative data) from pigs, rats, etc. were used as the objective variable, and the amino acid composition (quantitative data) of the evaluation material and qualitative data categorized according to the items shown in Figures 10 to 12 for the protein types of the evaluation material were used as explanatory variables. The digestibility analysis results in this example were obtained from the amino acid analysis values ​​of the undigested material in the pig ileum, and thus were obtained as individual values ​​for each amino acid. Furthermore, in this example, 119 conditions randomly selected from these datasets were used as training data, and 28 conditions were used as evaluation data. The training data was divided into 10 groups, and hyperparameter optimization was performed by cross-validation.

[0076] (4) Verification of the prediction accuracy of the digestibility prediction model As shown in FIG. 13, in this example, a comparison of predicted digestibility data (Predicted values) and measured values ​​(Measured values) based solely on amino acid composition is plotted. As shown in FIG. 13, in this example, under many conditions, the graph shows that accurate predictions were possible, i.e., the predicted values ​​were close to a straight line, i.e., the measured values ​​were close to each other. Furthermore, as shown in FIG. 14, in this example, the coefficient of determination, which indicates the goodness of fit of regression analysis, which is one indicator of prediction accuracy, was calculated for the digestibility data predicted solely from the amino acid composition and the measured values. Here, the coefficient of determination is a relative criterion, and a value of 0.5 or higher is generally considered to indicate somewhat good prediction accuracy. However, as shown in FIG. 14, in this example, the coefficients of determination for 13 amino acids other than threonine, glutamic acid, and serine exceeded 0.5, and in particular, the coefficients of determination for lysine and tyrosine exceeded 0.7, indicating good predictability.

[0077] (5) Refinement of the digestibility prediction model In this embodiment, when the type of protein (qualitative data) is added to the explanatory variables in "(3) Digestibility prediction model construction", the coefficient of determination shown in Figure 15 is obtained. When the type of protein (qualitative data) and the type of cooking method (qualitative data) are added to the explanatory variables in "(3) Digestibility prediction model construction", the coefficient of determination shown in Figure 16 is obtained. When the type of protein (qualitative data), the type of cooking method (qualitative data), time and temperature (quantitative data) are added to the explanatory variables in "(3) Digestibility prediction model construction", the coefficient of determination shown in Figure 17 is obtained. As a result, as shown in Figures 16 and 17, in this embodiment, by adding explanatory variables, the number of amino acids with a coefficient of determination below 0.5 is reduced to one type, serine, and by adding both data on the type of protein and the type of cooking method to the explanatory variables, the number of amino acids with a coefficient of determination above 0.7 is increased to four or more types. From the above, this example shows that more precise predictions can be made by adding not only the amino acid composition but also the type of protein and the type of cooking and processing method to the explanatory variables.

[0078] In this example, the digestibility predicted using the digestibility prediction model constructed in "(3) Digestibility Prediction Model Construction" (the only explanatory variable is amino acid composition) was used to evaluate the digestible essential amino acid standard ratios in existing literature according to the FAO DIAAS evaluation formula. As shown in Figure 18, in this example, the digestible essential amino acid standard ratio evaluation results based on the predicted digestibility values ​​(Predicted values) and the actual digestibility measured values ​​(Measured values) using living organisms are plotted to compare the digestible essential amino acid standard ratio evaluation results with the literature values ​​(Observed values). Thus, this example demonstrates that the digestibility prediction technology can be used to accurately evaluate the digestible essential amino acid standard ratios. Here, SAA and AAA shown in FIG. 18 represent sulfur-containing amino acids (methionine + cysteine) and aromatic amino acids (phenylalanine + tyrosine), respectively.

[0079] Example 2 Example 2 will be described with reference to Fig. 19 to Fig. 22. Fig. 19 to Fig. 22 are diagrams showing an example of a prediction accuracy verification result in this embodiment.

[0080] In this example, the prediction accuracy of digestibility is verified for multiple machine learning models.

[0081] Specifically, in this example, for various protein ingredients including animal protein and vegetable protein, the digestibility data (lysine) is used as the dependent variable, and the amino acid composition, cooking / processing method, and type of protein are used as explanatory variables to verify the prediction accuracy.

[0082] In this embodiment, as shown in FIG. 19, the results of prediction accuracy verification using random forest are obtained, as shown in FIG. 20, the results of prediction accuracy verification using LightGBM are obtained, and as shown in FIG. 21, the results of prediction accuracy verification using neural network are obtained.

[0083] As shown in FIG. 22, in this example, the coefficient of determination and the average value and standard error of the root mean square error (RMSE) when predicting the digestibility of His, Ile, Leu, Lys, Met, Phe, Thr, Trp, and Val were examined, and it was shown that digestibility prediction was possible with any of the models.

[0084] Example 3 Example 3 will be described with reference to Fig. 23 and Fig. 24. Fig. 23 and Fig. 24 are diagrams showing an example of the explanatory variable importance analysis results in this embodiment.

[0085] In this example, the nutritional data of the digestibility evaluation materials (Carbohydrate, Starch, Fiber, Ash, Moisture, Cholesterol, Energy, Fatty Acid, Lipid, and Organic Acids) were assigned to the data of the ingredients most similar to those included in the 2020 edition (8th revision) of the Standard Tables of Food Composition in Japan, and used as explanatory variables for the digestibility prediction model.

[0086] Furthermore, in this example, the amino acid composition of 14 types of the evaluation ingredients (Arg, His, Ile, Leu, Lys, Met, Phe, Thr, Trp, Val, Ala, Asp, Glu, and Ser) and the crude protein content of the evaluation ingredients were added as explanatory variables (note that cooking / processing methods and protein types were not added as explanatory variables).

[0087] In this example, a random forest model is constructed using the above explanatory variables to predict digestibility, and a feature importance analysis of the explanatory variables is performed.

[0088] As shown in Figures 23 and 24, this example indicates that, among the nutritional data other than the amino acid composition and crude protein content data, the carbohydrate, sugar, and dietary fiber amounts in the evaluation material (Figure 24) may particularly contribute to digestibility prediction.

[0089] The graphs in Figures 23 and 24 show the results of models using Val (first from the top), Trp (second from the top), Thr (third from the top), Phe (fourth from the top), Met (fifth from the top), Leu (sixth from the top), Lys (seventh from the top), Ile (eighth from the top), and His (ninth from the top) as objective variables.

[0090] From the above results, this example shows that nutritional data other than amino acid composition can also be used to explain the characteristics of the digestibility of the evaluated materials and can be used for prediction.

[0091] Example 4 Example 4 will be described with reference to Fig. 25 to Fig. 27. Fig. 25 to Fig. 27 are diagrams showing an example of correlation in this embodiment.

[0092] In this example, the correlations of various evaluated protein ingredients were analyzed, including the correlation between digestibility (lysine) and lysine content as shown in Figure 25, the correlation between digestibility (lysine) and crude protein content as shown in Figure 26, and the correlation between digestibility (lysine) and carbohydrate content as shown in Figure 27.

[0093] As a result, as shown in Figures 25 and 26, a positive correlation was observed between digestibility (lysine) and lysine content, and between digestibility (lysine) and crude protein content, while a negative correlation was observed between digestibility (lysine) and carbohydrate content.

[0094] As such, the higher the protein content, the easier it is for the protein food material to react with digestive enzymes, improving the digestibility of the protein food material. However, as the amount of certain contaminants such as carbohydrates increases, the accessibility of digestive enzymes to the protein food material decreases, and so it can be seen that the digestibility of the protein food material decreases.

[0095] [4. Other Embodiments] In addition to the above-described embodiments, the present invention may be implemented in various different embodiments within the scope of the technical concept described in the claims.

[0096] For example, among the processes described in the embodiments, all or part of the processes described as being performed automatically can be performed manually, or all or part of the processes described as being performed manually can be performed automatically using known methods.

[0097] Furthermore, the processing procedures, control procedures, specific names, information including parameters such as registered data and search conditions for each process, screen examples, and database configurations shown in this specification and drawings can be changed as desired unless otherwise specified.

[0098] Furthermore, with regard to the food evaluation device 100 and the like, the components shown in the drawings are functional concepts, and do not necessarily have to be physically configured as shown in the drawings.

[0099] For example, all or any part of the processing functions of the food evaluation device 100, particularly the processing functions performed by the control unit, may be implemented by a CPU and a program interpreted and executed by the CPU, or may be implemented as hardware using wired logic. The program is recorded on a non-transitory computer-readable recording medium containing programmed instructions for causing the information processing device to execute the processes described in this embodiment, and is mechanically read as needed. That is, a computer program for providing instructions to the CPU in cooperation with the OS and performing various processes is recorded in a storage unit such as a ROM or HDD (hard disk drive). This computer program is executed by being loaded into RAM and cooperates with the CPU to form the control unit.

[0100] In addition, this computer program may be stored in an application program server connected to the food evaluation device 100, etc. via any network 300, and it is also possible to download all or part of it as needed.

[0101] Furthermore, the program for executing the processes described in this embodiment may be stored in a non-transitory computer-readable recording medium, or may be configured as a program product. Here, this "recording medium" includes memory cards, USB (Universal Serial Bus) memories, SD (Secure Digital) cards, flexible disks, magneto-optical disks, ROMs, EPROMs (Erasable Programmable Read Only Memory), EEPROMs (registered trademark) (Electrically Erasable and Programmable Read Only Memory), CD-ROMs (Compact Disk Read Only Memory), MOs (Magneto-Optical disks), DVDs (Digital Versatile Disks), and more. This includes any "portable physical medium" such as a Blu-ray Disc, a DVD player, a DVD player, a Blu-ray Disc, etc.

[0102] Furthermore, a "program" is a data processing method written in any language or description method, and does not matter whether it is in the form of source code or binary code. Note that a "program" is not necessarily limited to a single structure, but also includes a structure that is distributed as multiple modules or libraries, or a structure that achieves its function by cooperating with a separate program, such as an OS. Note that the specific configuration and reading procedure for reading a recording medium in each device shown in this embodiment, as well as the installation procedure after reading, can use well-known configurations and procedures.

[0103] The various databases stored in the memory unit are storage means such as memory devices such as RAM and ROM, fixed disk devices such as hard disks, flexible disks, and optical disks, and store various programs, tables, databases, and web page files used for various processes and providing websites.

[0104] The food evaluation device 100 may be configured as an information processing device such as a known personal computer or workstation, or may be configured as an information processing device connected to any peripheral device. The food evaluation device 100 may also be realized by installing software (including programs, data, etc.) that causes the device to perform the processing described in this embodiment.

[0105] Furthermore, the specific form of distribution and integration of the devices is not limited to that shown in the drawings, and all or part of them can be functionally or physically distributed and integrated in any unit depending on various additions or functional loads. In other words, the above-mentioned embodiments can be implemented in any combination, or embodiments can be implemented selectively.

[0106] The present invention is useful in a variety of industries, including the food industry and the information processing industry.

[0107] REFERENCE SIGNS LIST 100 Food evaluation device 102 Control unit 102a Model construction unit 102b Evaluation target setting unit 102c Prediction result acquisition unit 102d Correction unit 106 Storage unit 106a Protein database 112 Input / output unit 300 Network

Claims

1. A food evaluation device comprising a memory unit and a control unit, wherein the memory unit comprises: a protein memory means for storing a machine learning model in which the amino acid composition of a protein is an explanatory variable and protein digestibility is an objective variable, and the control unit comprises: an evaluation target setting means for setting the protein to be evaluated, and a prediction result acquisition means for acquiring a digestibility prediction result of the protein digestibility of the evaluation target using the machine learning model.

2. The food evaluation device according to claim 1, characterized in that the explanatory variables include the type of protein.

3. The food evaluation device according to claim 2, characterized in that the type of protein is a taxonomic rank of animal or plant family, genus, or species.

4. A food evaluation device according to any one of claims 1 to 3, characterized in that the explanatory variables further include at least one of the type of cooking method, temperature, and heating time.

5. The food evaluation device of claim 3, characterized in that the explanatory variables further include formulation data including moisture content and / or hardness, physical properties including solubility and / or emulsifying power, nutrient content including dietary fiber content, fat content, sugar content and / or mineral content, plant and / or animal part data, and / or country, region and / or manufacturer source data.

6. The food evaluation device according to claim 3, characterized in that the protein is a protein contained in a dish as an ingredient or a protein contained in a single dish.

7. The food evaluation device described in claim 1, characterized in that the prediction result acquisition means further uses the digestibility prediction result to calculate the amount of digestible amino acids, which is the product of the content of various amino acids and the protein digestibility.

8. The food evaluation device described in claim 1, characterized in that the prediction result acquisition means further uses the digestibility prediction result to calculate a digestible essential amino acid amount, which is the product of the content of various essential amino acids and the protein digestibility, a digestible essential amino acid standard ratio, which is the value obtained by dividing the digestible essential amino acid amount by a standard value, or a digestible essential amino acid score, which is the smallest value of the digestible essential amino acid standard ratios.

9. The food evaluation device described in claim 6, characterized in that the control unit further comprises a correction means for correcting the digestibility prediction result of the protein digestibility of the protein contained in the single dish to be evaluated based on a change in digestibility due to the combination of ingredients that make up the single dish.

10. The food evaluation device according to claim 6, further comprising a correction means for correcting the predicted digestibility result of the protein digestibility of the protein contained in the single dish to be evaluated based on a change in digestibility due to eating the single dish with other dishes that make up a menu including the single dish.

11. The food evaluation device of claim 1 or 2, characterized in that the control unit further comprises a correction means for correcting the digestibility prediction result of the protein digestibility of the subject to be evaluated based on a change in digestibility due to exercise before and after ingestion of the subject to be evaluated.

12. The food evaluation device of claim 1 or 2, characterized in that the control unit further comprises a correction means for correcting the digestibility prediction result of the protein digestibility of the evaluation subject based on changes in digestibility due to the constitution of the person ingesting the evaluation subject.

13. The food evaluation device according to claim 1 or 2, characterized in that the prediction result acquisition means further acquires meal content suggestion data based on the digestibility prediction result and nutrient data including dietary reference intakes.

14. The food evaluation device according to claim 1 or 2, characterized in that the prediction result acquisition means further displays the digestibility prediction result.

15. The food evaluation device according to claim 6, characterized in that the evaluation target is the protein ingested over a specified period of time in an amount of one meal or more.

16. The food evaluation device described in claim 13, characterized in that the prediction result acquisition means acquires the meal content proposal data based on the digestibility prediction result and the nutrient data including the dietary intake standards according to the attributes and / or purpose of intake of the person ingesting the evaluation subject.

17. The food evaluation device according to claim 1, characterized in that the prediction result acquisition means further creates product development data for a product that reflects the protein digestibility of the evaluation target based on the digestibility prediction result.

18. The food evaluation device of claim 1, characterized in that the machine learning model is a model consisting of at least one of random forest, neural network, XGBoost, LightGBM, support vector regression, and linear regression including ElasticNet.

19. A food evaluation device comprising a memory unit and a control unit, wherein the memory unit comprises: a protein memory means for storing a machine learning model having an explanatory variable of the amino acid composition of a protein and an amount of digestible amino acids, an amount of digestible essential amino acids, a standard ratio of digestible essential amino acids, or a score of digestible essential amino acids as objective variables, and the control unit comprises: an evaluation target setting means for setting the protein to be evaluated, and a prediction result acquisition means for using the machine learning model to acquire the amount of digestible amino acids, the amount of digestible essential amino acids, the standard ratio of digestible essential amino acids, or the score of the evaluation target.

20. A food evaluation device comprising a memory unit and a control unit, wherein the memory unit comprises: protein storage means for storing, as teacher data, at least one of the following as explanatory variables: amino acid composition of protein, type of protein, formulation data including water content and / or hardness, physicochemical properties including solubility and / or emulsifying power, nutrient content including dietary fiber amount, fat amount, sugar amount and / or mineral amount, plant and / or animal part data, country, region and / or manufacturer source data, type of processing and cooking method, cooking temperature, cooking time, and in vitro analysis results regarding enzyme activity inhibition and / or digestibility, and the protein digestibility in an in vitro process or an in vivo process is the objective variable; and the control unit comprises: model construction means for training a machine learning model to learn the teacher data.

21. A food evaluation method to be executed by a food evaluation device having a memory unit and a control unit, wherein the memory unit comprises: a protein memory means for storing a machine learning model in which the amino acid composition of a protein is an explanatory variable and protein digestibility is an objective variable; and the food evaluation method includes an evaluation target setting step, executed in the control unit, for setting the protein to be evaluated, and a prediction result acquisition step for using the machine learning model to obtain a digestibility prediction result of the protein digestibility of the evaluation target.

22. A food evaluation method to be executed by a food evaluation device having a memory unit and a control unit, wherein the memory unit comprises: a protein memory means for storing, as teacher data, at least one of the following as explanatory variables: amino acid composition of protein, type of protein, formulation data including water content and / or hardness, physicochemical properties including solubility and / or emulsifying power, nutrient content including dietary fiber amount, fat amount, sugar amount and / or mineral amount, plant and / or animal part data, country, region and / or manufacturer source data, type of processing and cooking method, cooking temperature, cooking time, and in vitro analysis results regarding enzyme activity inhibition and / or digestibility, and the protein digestibility in an in vitro process or an in vivo process is the objective variable; and a model construction step, executed in the control unit, for training a machine learning model to learn the teacher data.

23. A food evaluation program to be executed by a food evaluation device having a memory unit and a control unit, wherein the memory unit comprises: a protein memory means for storing a machine learning model in which the amino acid composition of a protein is an explanatory variable and protein digestibility is an objective variable; and the control unit executes: an evaluation target setting step for setting the protein to be evaluated; and a prediction result acquisition step for obtaining a digestibility prediction result of the protein digestibility of the evaluation target using the machine learning model.

Images