JP2022094506A - Pet food recommendation device and pet food recommendation method - Google Patents

Abstract

To provide a pet food recommendation device and a pet food recommendation method, for recommending food suitable for a pet.SOLUTION: A pet food recommendation method includes a classification step for classifying which group a pet belongs to in multiple groups determined beforehand, based on attribute information of the pet, and on intestinal flora information of the pet, and a step for recommending food in consideration of which group the pet is classified into.SELECTED DRAWING: Figure 1

Description

The present invention relates to a pet food recommendation device and a pet food recommendation method for recommending food suitable for pets.

Today, various pet foods have been proposed (eg, Patent Document 1). However, it is not easy to make a pet food that is suitable for all pets.

Japanese Unexamined Patent Publication No. 2017-2053

An object of the present invention is to provide a pet food recommendation device and a pet food recommendation method for recommending food suitable for pets.

According to one aspect, a classification step of classifying the pet into one of a plurality of predetermined groups based on the attribute information of the pet and the information of the intestinal flora of the pet, and classification into any group. A pet food recommendation method is provided that comprises a step of recommending food, taking into account what has been done.

In the recommending step, the veterinarian may recommend food suitable for the pet.

In the classification step, based on the information of the intestinal flora, classification may be performed according to whether or not the specific bacterium possessed by the pet exceeds the reference value.

The reference value may correspond to the attribute information of the pet.

According to one aspect, a step of scoring the pet and a step of recommending a food according to the score given to the pet based on the attribute information of the pet and the information of the intestinal flora of the pet. And, a pet food recommendation method is provided.

A hood to be recommended for each score is preset, and in the recommendation step, the hood set to the score attached to the pet may be recommended.
In the recommending step, the veterinarian may recommend food suitable for the pet.

In the step of scoring, the pet may be scored in consideration of the intestinal age of the pet.

In the step of scoring, the pet may be scored based on the information of the intestinal flora, considering the number and / or the type of the specific bacteria contained in the intestine of the pet.

According to one aspect, the step of letting the artificial intelligence device learn in advance the relationship between the intestinal flora and the food to be recommended, and the artificial intelligence device based on the intestinal flora information of the target pet. A pet food recommendation method is provided that includes a step of recommending food to the pet.

In the step of learning, the artificial intelligence device is made to learn in advance the relationship between the attributes of the pet, the information of the intestinal flora, and the food to be recommended, and in the step of making the recommendation, the object is targeted by the artificial intelligence device. Food may be recommended to the pet based on the attribute information of the pet and the information of the intestinal flora.

In the step of learning, the artificial intelligence device is trained to create a model showing the relationship between the explanatory variable including the information of the intestinal flora and the information of the hood, and the objective variable, and in the step of recommending, the creation is performed. Based on the model, a hood that maximizes the objective variable may be recommended for the information of the intestinal flora.

The objective variable may include user satisfaction with each food or the health of the pet when ingesting each food.

In the learning step, the artificial intelligence device may be trained to create a plurality of models according to the information of the intestinal flora.

The explanatory variables may include pet attributes.

In the step of learning, supervised learning or reinforcement learning may be performed based on the explanatory variables and the objective variables prepared in advance.

The learning step may include unsupervised learning so as to dimensionally compress the explanatory variables to generate explanatory variables input to the model.

According to one aspect, a pet food recommendation method is provided in which a veterinarian recommends a food suitable for the pet based on the attribute information of the pet and the information of the intestinal flora of the pet.

According to one aspect, based on a step of defining the relationship between the pet's attributes and intestinal flora and the food to be recommended as a rule, the target pet's attribute information and intestinal flora information, and the above-mentioned rule. A pet food recommendation method comprising a step of recommending food to the pet is provided.

It may be provided with a step of inspecting the pet's stool to obtain information on the pet's intestinal flora.

A step of estimating the information of the intestinal flora of the pet from the attribute information of the pet and the information of the food ingested by the pet may be provided.

According to one embodiment, a means for classifying the pet into one of a plurality of predetermined groups based on the attribute information of the pet and the information on the intestinal flora of the pet.
A pet food recommendation device is provided that comprises a means of recommending food, taking into account which group it was classified into.

According to one aspect, a means for scoring the pet and a means for recommending food according to the score given to the pet based on the attribute information of the pet and the information on the intestinal flora of the pet. And, a pet food recommendation device is provided.

According to one aspect, an artificial intelligence device that has learned in advance the relationship between the intestinal flora and the food to be recommended and recommends the food to the pet based on the intestinal flora information of the target pet. A pet food recommendation device is provided.

According to one embodiment, the pet is subjected to the rule based on the rule defining the relationship between the pet's attributes and intestinal flora and the food to be recommended, and the attribute information and intestinal flora information of the target pet. A pet food recommendation device is provided with a means of recommending food.

The block diagram which shows the schematic structure of the pet food recommendation system which concerns on 1st Embodiment. The figure which shows the screen example of the Web page for inputting the attribute information of a pet. The figure which specifically explains the group classification by the classification means 3a. The figure which shows an example of the recommendation screen for the pet classified into Young-F type. The figure which shows an example of the recommendation screen for the pet classified into Young-LC type. The figure which shows an example of the recommendation screen for the pet classified into Young-N type. The figure which shows an example of the recommendation screen for the pet classified into YOUNG-C type. The figure which shows an example of the recommendation screen for the pet classified into Young-N type. Hood manufacturing process diagram. The block diagram which shows the schematic structure of the pet food recommendation system which concerns on 2nd Embodiment. The block diagram which shows the schematic structure of the pet food recommendation system which concerns on 3rd Embodiment. The block diagram which shows the schematic structure of the pet food recommendation system which concerns on 4th Embodiment. The block diagram which shows the schematic structure of the pet food recommendation system which concerns on 5th Embodiment. The figure explaining the artificial intelligence device 3c. The block diagram which shows the schematic structure of the pet food recommendation system which concerns on 6th Embodiment.

(First Embodiment)
Hereinafter, embodiments according to the present invention will be specifically described with reference to the drawings. Although dogs are illustrated below as pets, they can also be applied to other pets such as cats and birds.

FIG. 1 is a block diagram showing a schematic configuration of a pet food recommendation system according to the first embodiment. The pet food recommendation system includes an attribute information acquisition means 1, an intestinal flora information acquisition means 2, a recommendation means 3, a recommendation result presentation means 4, and a food ordering means 5. These may be composed of one device or may be distributed to a plurality of devices. As an example, the attribute information acquisition means 1, the recommendation result presentation means 4, and the food ordering means 5 are realized on the Web server. The intestinal flora information acquisition means 2 is provided in a predetermined inspection institution. The recommendation means 3 is provided on the server of the administrator of this system (of course, this server may be integrated with the above Web server).

The attribute information acquisition means 1 acquires the attribute information of the pet. More specifically, the attribute information acquisition means 1 acquires the attribute information of the pet manually input by the owner from a predetermined form on the website. The pet attribute information here preferably includes information on the breed, weight, age, gender and living environment (indoor / outdoor), but it may be a part of the information or other information. There are no particular restrictions as long as it is related to pet food.

As a specific example, the attribute information acquisition means 1 displays a Web page as shown in FIG. 2 on a display of a terminal device (not shown) of the owner. Then, the owner accepts input of attribute information such as a pet's name and a dog breed. Then, when the "apply" button is selected, the input attribute information is transmitted to the recommendation means 3. In addition, a kit for stool test will be delivered to the owner.

Returning to FIG. 1, the intestinal flora information acquisition means 2 acquires information on the intestinal flora of a pet. In the following, two examples of specific acquisition methods will be described.

As a first example, the intestinal flora information acquisition means 2 may inspect the stool of a pet to acquire information on the intestinal flora.

That is, the intestinal flora information acquisition means 2 performs a stool test of a pet. More specifically, the owner receives a stool test kit and collects and returns the pet's stool. Then, the intestinal flora information acquisition means 2 inspects and analyzes the stool, and obtains the intestinal flora information and the like as the inspection result. The intestinal flora is the ecology of bacteria that inhabit the intestines of pets. And the information on the intestinal flora includes, for example, the diversity of intestinal bacteria possessed by pets, the number of lactic acid bacteria, the number of butyrate-producing bacteria (both Ficaribacterium and Clostridium), and Firmicutes. The ratio of Cutes to Bacteroides, which may include, but is not limited to, the FB ratio indicating the ease of gaining weight. For example, Fermicutes, Bacteroides, Proteobacterium, Radial Bacterial, Fusobacteria, Ficalibacterium, Clostridium, Lactobacillus, Bifidus, Osirospira, Blautia, SMB53, It may include one or more of the genera Fusobacterium, Prevotera, Luminococcus, Turicibacter, Streptococcus, 02d06, Stetterella, Rosebria, Drea and eubacteria.

According to this method, it is possible to obtain appropriate supplements because it is possible to obtain information on the intestinal flora including objective and quantitative data such as the type and number of bacteria contained in the intestine from the stool test results. It becomes.

As a second example, the intestinal flora information acquisition means 2 estimates the intestinal flora information from the pet attribute information acquired by the attribute information acquisition means 1 and the food information ingested by the pet. May be good.

Specifically, the intestinal flora information acquisition means 2 acquires information on the food ingested by the pet. For example, a Web page for inputting the food that the pet has ingested (preferably most recently) is displayed on the owner's terminal device, and the input from the owner is accepted.

Then, the intestinal flora information acquisition means 6 estimates the information of the pet's intestinal flora from the attribute information of the pet and the information of the food ingested by the pet. As a specific example, the intestinal flora information acquisition means 6 holds a database showing the relationship between the pet attribute information and the food ingested by the pet and the intestinal flora, and by referring to the database, the intestine is used. Information on the inner flora can be estimated.

With this technique, it is not necessary to inspect the pet's stool, so it is easier to recommend the right food for the pet.

The recommendation means 3 recommends a food suitable for a pet based on the attribute information of the pet and the information of the intestinal flora of the pet. The recommended food is not limited to one type, and a plurality of foods having different tastes may be recommended. Specific recommendation methods and recommended foods will be described later.

The recommendation result presentation means 4 presents information on the recommended food to the owner. As a specific example, the recommendation result presenting means 4 posts information on the recommended food and preferably a pet test result on a Web page accessible to the owner. In addition, the recommendation result presenting means 4 may send information on the recommendation food and a medical certificate showing the test result to the owner.

The food ordering means 5 accepts an order for recommended food presented to the owner from the owner. As a specific example, the food ordering means 5 provides an icon for purchasing the food on a Web page on which the recommended food is posted. Depending on the selection of the icon, the food ordering means 5 performs the necessary payment processing. As a result, the recommended food is delivered to the owner. The amount may be, for example, an amount consumed in a certain period (one month, etc.) in consideration of the weight of the pet.

Subsequently, a specific recommendation method by the recommendation means 3 will be illustrated.

The recommendation means 3 has a classification means 3a and classifies the pet into one of a plurality of predetermined groups according to the information of the intestinal flora. Recommended foods are preset for each group. That is, the recommendation means 3 classifies the pets into a specific group and recommends the food set in the group.

FIG. 3 is a diagram specifically explaining group classification by the classification means 3a. The recommendation means 3 performs group classification based on the diversity of intestinal bacteria and group classification based on lactic acid bacteria / butyrate-producing bacteria and FB ratio.

First, the classification means 3a classifies into a plurality of groups (here, four examples are shown) based on the diversity of intestinal bacteria in the information of the intestinal flora. In general, the younger the age, the more diverse the intestinal bacteria are, so for convenience, the group names are Young, Adult, Senior, and High Senior in the order of abundant diversity. The diversity of intestinal bacteria is determined, for example, by whether or not the type of bacteria contained in the intestine exceeds the reference value.

Subsequently, the classification means 3a further classifies each group classified according to the diversity into one of two groups according to the amount of lactic acid bacteria. As a specific example, it can be classified according to whether or not the number of lactic acid bacteria is larger than the standard value.

Then, the classification means 3a further classifies the group having a large amount of lactic acid bacteria into one of the two groups according to the level of the FB ratio. As a specific example, it can be classified according to whether or not the FB ratio is higher than the reference value. The group with a high FB ratio is defined as F type. On the other hand, the group with a low FB ratio is further classified into two groups according to the amount of butyrate-producing bacteria. As a specific example, it can be classified according to whether or not the number of butyrate-producing bacteria is larger than the standard value. The group with a large number of butyrate-producing bacteria is referred to as LC type, and the group with a small number of butyrate-producing bacteria is referred to as C type.

Further, the classification means 3a further classifies the group in which the number of lactic acid bacteria is low into one of the two groups according to the amount of butyrate-producing bacteria. As a specific example, it can be classified according to whether or not the number of butyrate-producing bacteria is larger than the standard value. The group with a large number of butyrate-producing bacteria is referred to as type C, and the group with a small number of butyrate-producing bacteria is referred to as type N.

F type has many lactic acid bacteria but high FB ratio, LC type has many lactic acid bacteria and butyrate-producing bacteria, L type has many lactic acid bacteria but few butyrate-producing bacteria, and C type has many butyrate-producing bacteria but few lactic acid bacteria. Group, N type can be said to be a group with few both lactic acid bacteria and butyrate-producing bacteria.

As described above, by classifying into one of the four groups according to the diversity of intestinal bacteria and into one of the five groups based on the lactic acid bacteria / butyrate-producing bacteria and the FB ratio, any of the 20 groups in total. Crab pets are classified. For example, if the classification according to diversity is Young and the classification based on lactic acid bacteria / butyrate-producing bacteria and FB ratio is F type, the classification of the pet is Young-F type.

In addition, it is desirable that each reference value when classifying into a group is a value according to the pet attribute information such as the dog breed, age, and weight of the pet. In addition, classification by the amount of lactic acid bacteria and butyrate-producing bacteria shall include classification according to the presence or absence.

As described above, the classification means 3a classifies the pet into groups in consideration of the number of specific bacteria contained in the intestine and the type of the bacteria based on the information of the intestinal flora. It is desirable to consider pet attribute information when classifying.

The food recommended by the recommendation means 3 can be a base food mixed with a few percent of the necessary useful bacterial food. The base food may be irrelevant to the information on the intestinal flora, but it is desirable that it is based on the attribute information of the pet, specifically, the type suitable for the actual age of the pet and the weight suitable for the weight of the pet. It is good to be. On the other hand, the useful bacteria correspond to the group classified by the information of the intestinal flora, and it is desirable that the useful bacteria include the deficient bacteria and the substance that activates (supports) the bacteria.

It is preferable that the F-type group has lactic acid bacteria, but the FB ratio is high and it is easy to gain weight. Therefore, in addition to the base food, foods containing water-soluble dietary fiber and oligosaccharides for enhancing diversity are recommended as useful bacterial foods.

The LC type is an ideal flora with many lactic acid bacteria and butyrate-producing bacteria. Therefore, no useful fungal food is required and the base food is recommended.

The L type is a flora with many lactic acid bacteria but few butyrate-producing bacteria. Therefore, in addition to the base food, a food containing butyrate-producing bacteria and water-soluble dietary fiber that activates the butyrate-producing bacteria is recommended as a useful bacterial food.

Type C is a flora with many butyrate-producing bacteria but few lactic acid bacteria. Therefore, in addition to the base food, a food containing lactic acid bacteria and oligosaccharides that activate them is recommended as a useful bacterial food.

Type N is a flora with few butyrate-producing bacteria and butyrate-producing bacteria. Therefore, in addition to the base food, foods containing lactic acid bacteria and butyrate-producing bacteria are recommended as useful bacterial foods.

Since Senior and High Senior have a low diversity of intestinal bacteria, useful bacteria further containing water-soluble dietary fiber and oligosaccharides not only in F type but also in LC type, L type, C type and N type. The hood may be recommended.

Moreover, when comparing the Young-F type and the Adult-F type, it is common that a useful bacterial food containing water-soluble dietary fiber and oligosaccharide is recommended. However, due to the less diversity of gut microbiota in the Adult-F type, useful bacterial foods containing more water-soluble dietary fiber and oligosaccharides are recommended. As described above, even if the F type is the same, there are many water-soluble dietary fibers and oligosaccharides as useful bacterial foods in the order of Young, Adult, Senior, and High Senior. This point is the same for LC type, L type, C type and N type.
Further, even if it is a specific type, it is desirable that the food corresponds to the attribute information of the pet.

For example, one breed of dog, Chihuahua, has calcium (because of its thin bones), oligosaccharides and dietary fiber (because of its delicate gastrointestinal tract), and omega-3 fatty acid (because of maintaining joint health). Need more than. Therefore, for example, for Chihuahuas classified as Young-F type, we recommend foods containing more calcium, oligosaccharides, dietary fiber, and omega 3 fatty acids than other dog breeds also classified as Young-F type. It is desirable to do.

As another example, older dogs consume less energy and are more prone to constipation due to lack of exercise and reduced bowel function. Therefore, for example, for old dogs classified as Young-F type (for example, 8 years old or older), puppies (for example, up to 1 year old) and adult dogs (for example, 1 to 8 years old) also classified as Young-F type. ), It is desirable to recommend a small amount of high-calorie food that contains more dietary fiber.

As another example, indoor dogs are prone to obesity. Therefore, for example, indoor dogs classified as Young-F type have higher protein and higher protein than outdoor dogs classified as Young-F type in order to maintain muscle mass and increase basal metabolism. It is advisable to recommend low-calorie foods.

As described above, the recommended food for the pet is determined. Then, the recommendation result presenting means 4 displays the following Web page (hereinafter referred to as a recommendation screen).

4A-4E are diagrams showing an example of a recommendation screen for pets classified into Young-F type, Young-LC type, Young-N type, Young-C type and Young-N type, respectively. As shown, the recommendation screen contains a description of the classified types and their characteristics. Then, a specific hood is displayed. Furthermore, it is desirable that the recommendation screen numerically shows the diversity of intestinal bacteria used for group classification, FB ratio, lactic acid bacteria, and butyrate-producing bacteria, and displays simple comments and precautions.

Further, the recommendation screen may include the "intestinal age" calculated by the recommendation means 3. The intestinal age is a parameter indicating the age of the intestinal flora, and is calculated based on the information of the intestinal flora and the attribute information. The intestinal age may be calculated according to the type and sex of the pet.

By the way, as described above, the recommended food is composed of a base food and a useful fungus food. By doing so, the manufacturing process can be made more efficient.

FIG. 5 is a manufacturing process diagram of the hood. The base hood is manufactured by the following process. First, raw material pretreatment such as crushing each material is performed (step S1a). Subsequently, the raw material is stirred (step S2a) and extruded while being heated by an extruder (step S3a). Then, it is dried at a high temperature (step S4a). Then, the deformed particles and the like are removed (step S5a), and the base hood is completed.

On the other hand, the useful fungus food is produced by the following process. First, raw material preparation such as baking chicken, drying and crushing, and minced meat is performed (step S1b). Subsequently, useful bacteria are added to stir the raw material (step S2b), and the raw material is extruded without heating with an extruder (step S3b). Then, it is dried at a low temperature such that useful bacteria do not die (step S4b). Then, the deformed particles and the like are removed (step S5b), and the base hood is completed.

Then, the base hood and the useful fungus hood are weighed in desired amounts (steps S6a and 6b), packed in a bag, and labeled (step S7) to complete the hood.

Thus, in the first embodiment, information on the intestinal flora (bacterial diversity, FB ratio, butyrate-producing bacteria, lactic acid bacteria, etc.) and pet attribute information (dog breed, weight, age, gender, living environment, etc.) ) And can recommend foods that are appropriate for pets. In particular, in the present embodiment, by inspecting the stool, quantitative measured values such as the diversity of intestinal bacteria, the number of each bacterium such as lactic acid bacteria and butyrate-producing bacteria, and the FB ratio can be obtained. Therefore, pets can be classified objectively and appropriate foods can be recommended.

Hereinafter, some modifications will be described, but common reference numerals are given to the functional parts common to the first embodiment, and the description thereof will be omitted.

(Second embodiment)
FIG. 6 is a block diagram showing a schematic configuration of the pet food recommendation system according to the second embodiment. The difference from the first embodiment (FIG. 1) is that the veterinarian makes the recommendation instead of the recommendation means 3.

That is, it is the same as the first embodiment until the pet is classified into one of a plurality of predetermined groups. Then, the classification result is transmitted to the veterinarian. Veterinarians recommend at least foods suitable for the pet (eg, foods that improve the intestinal flora, and so on), depending on which group the pet was classified into. In addition, the veterinarian may make recommendations with further consideration of the pet's attribute information. The food recommended by the veterinarian is input to the recommended result presenting means 4, and the recommended result presenting means 4 presents the information of the recommended food to the owner.

(Third embodiment)
FIG. 7 is a block diagram showing a schematic configuration of a pet food recommendation system according to a third embodiment. As a difference from the first embodiment (FIG. 1), the score evaluation means 3b is provided in the recommendation means 3 instead of the classification means 3a.

The score evaluation means 3b scores the pet based on the information of the intestinal flora. The score evaluation means 3b may give a score based on the diversity (type) of intestinal bacteria, the number of specific bacteria (lactic acid bacteria, butyrate-producing bacteria, etc.), and / or the FB ratio, etc., based on the intestinal flora. can. For example, the higher the score, the better the condition. In this case, the more diverse the intestinal bacteria, the higher the score, the larger the number of specific bacteria, the higher the score, and the higher the FB ratio, the higher the score.

In addition, the score evaluation means 3b may give a score to the pet in consideration of the attribute information of the pet and / or the intestinal age. For example, the younger the pet's age, the higher the score, and the younger the intestinal age, the higher the score.

The recommended hood is preset for each score. The recommendation means 3 recommends food according to the score attached to the pet.

(Fourth Embodiment)
FIG. 8 is a block diagram showing a schematic configuration of a pet food recommendation system according to a fourth embodiment. This embodiment is a modification of the third embodiment (FIG. 7), and the veterinarian makes the recommendation instead of the recommendation means 3.

That is, it is the same as the third embodiment until the pet is scored. Then, the score of the pet is transmitted to the veterinarian. The veterinarian will at least recommend foods suitable for the pet, depending on the pet's score. The veterinarian may also consider the pet's attribute information further. The food recommended by the veterinarian is input to the recommended result presenting means 4, and the recommended result presenting means 4 presents the information of the recommended food to the owner.

In the first to fourth embodiments described above, the relationship between the pet attribute and the intestinal flora and the recommended pet food is defined in advance as a rule. Then, it can be said that the recommendation means 3 recommends pet food based on the attributes of the pet, the intestinal flora, and the predetermined rules.

(Fifth Embodiment)
The fifth embodiment described below is for recommending pet food without performing group classification or score evaluation.

FIG. 9A is a block diagram showing a schematic configuration of a pet food recommendation system according to a fifth embodiment. As a difference from the first embodiment (FIG. 1), for example, a deep learning type artificial intelligence device 3c is provided in the recommendation means 3 instead of the classification means 3a.

The artificial intelligence device 3c has learned in advance the relationship between the pet's attributes and the intestinal flora and the food to be recommended. For this learning, we used statistical data on what kind of food was given to pets with what attributes and intestinal flora, and what kind of result was obtained (favorable result or not). can do.

Then, the artificial intelligence device 3c recommends the most probable food based on the attribute information of the target pet and the information of the intestinal flora. With such a technique, there is no need to predetermine rules that indicate the relationship between the pet's attributes and intestinal flora and the recommended food, and for each target pet (not by group or score). You can recommend pet food suitable for.

Hereinafter, a specific example of the artificial intelligence device 3c will be described. The artificial intelligence device 3c aims to recommend the optimum food for each pet according to various data obtained (for example, information on the intestinal flora such as the distribution of the intestinal flora before feeding the food).

FIG. 9B is a diagram illustrating an artificial intelligence device 3c. The artificial intelligence device 3c is learned to create a model showing the relationship between the explanatory variable and the objective variable.

Explanatory variables include information on the intestinal flora, such as intestinal flora distribution data prior to feeding. The explanatory variables also include food information such as the type of food. In the case of food type, the type of base food and the type and amount of useful kibble added at the time of compounding can be used as explanatory variables. Based on the information provided by the user and the delivery frequency information of the food, the feeding amount in one feeding may be estimated and used as an explanatory variable. In addition, the explanatory variables may include pet attributes such as pet age, gender, medical history, medication information, and breeding environment. Further, the explanatory variables may be these time series information.

The objective variable is a numerical value of the satisfaction level of the user (pet owner, etc.), the health level of the pet, or the behavior data by a questionnaire or the like. The degree of health may be a subjective evaluation or an objective evaluation (daily activity amount, voice data of crying, facial expression by camera, activity record, image / video data showing the amount / color / shape of stool, etc. It may be quantified). The larger the value of the objective variable, the more desirable the state.

Then, the artificial intelligence device 3c is trained to create a model (a model for converting the explanatory variable into the objective variable) showing the relationship between the explanatory variable and the objective variable based on the explanatory variable and the objective variable prepared in advance (for example, described later). Supervised learning or reinforcement learning). This models the relationship between the intestinal flora information and food information (and, in some cases, attribute information) in the explanatory variables and the objective variable.

By performing the above learning, it becomes possible to recommend food. At the time of food recommendation, information on the intestinal flora (if necessary, attribute information, the same applies hereinafter) such as intestinal bacterial flora distribution data of the target pet is input to the artificial intelligence device 3c. Then, the artificial intelligence device 3c outputs a hood that maximizes the objective variable for the input information of the intestinal flora based on the created model. The output hood is recommended to the user. User satisfaction with this food, pet health or behavioral data may be fed back and used for learning.

Here, the learning at the time of model creation may be supervised learning. Supervised learning collectively handles explanatory variables and objective variables for a predetermined period. In supervised learning, a model is created that outputs a hood in which the objective variable is maximized for a sample having certain gut microbiota distribution data. As the analysis method, for example, logistic regression analysis, decision tree, random forest, support vector machine and the like can be applied. One model may be created, or a plurality of models may be created according to the information of the intestinal flora. For example, a model for a pet having lactic acid bacteria and a model for pets without lactic acid bacteria may be created separately.

The learning at the time of model creation may be reinforcement learning. Reinforcement learning deals with explanatory variables and objective variables in chronological order. In reinforcement learning, a model is created that outputs a hood in which the objective variable is maximized for a sample having intestinal flora distribution data at a certain point in time based on continuous time series data. The objective variable in this case may be the latest value in the time series data or a cumulative value. In creating the model, Q-learning, Sarsa, Monte Carlo method, etc. may be applied with the information of the intestinal flora at a certain point in time as the state s, the information of the hood as the action a, and the objective variable as the reward r. This models the action a that maximizes the reward r for the state s.

In addition, if there are too many explanatory variables, unsupervised learning may be further performed so that the raw data of the explanatory variables is dimensionally compressed (the number of variables is reduced) to generate the explanatory variables to be input to the model. .. In unsupervised learning, the relationships between explanatory variables or the features of the intestinal flora distribution data in all the obtained samples are analyzed by k-means method, hierarchical clustering, k-nearest neighbor method, latent class analysis, etc. May be extracted or classified into groups, and the feature quantities and the classified groups may be used as explanatory variables to be input to the model. It is advisable to carry out analysis as appropriate according to the addition of explanatory variables, update the logic of group classification, and grasp the affiliation ratio and feature amount of each classified group in chronological order.

The intestinal bacterial flora distribution data in the explanatory variables and the intestinal bacterial flora distribution data input to the artificial intelligence device 3c at the time of recommendation may be general ones represented by continuous values of 0 to 1, and each of them may be used. It may be the presence or absence of bacteria or the relative amount. Further, the intestinal flora distribution data may be a characteristic amount or a tendency value of the entire intestinal flora distribution data extracted by principal component analysis or association (correlation between bacteria). .. Alternatively, the intestinal flora distribution data is based on the class probability or the most probable cluster for a pre-classified group by analysis such as k-means method, hierarchical clustering, k-nearest neighbor method, latent class analysis, etc. There may be.

In addition, the intestinal flora distribution data used for learning is expected to be used mainly at the genus level, but in the case of animal species where it is difficult to obtain sufficient intestinal flora distribution data at the genus level. May use gut microbiota distribution data at the family level.

(Sixth Embodiment)
FIG. 10 is a block diagram showing a schematic configuration of a pet food recommendation system according to a sixth embodiment. In this embodiment, the veterinarian recommends the appropriate food based on the pet's attributes and intestinal flora. With such a technique, there is no need to predefine rules that indicate the relationship between pet attributes and intestinal flora and recommended pet food, and it is targeted (not by group or by score). You can recommend pet food suitable for each pet.

(7th Embodiment)
Alternatively, a uniform pet food may be recommended for any pet without considering the pet's attributes or intestinal flora. The recommended pet food should be one that is expected to best improve the pet's intestinal flora.
According to each of the above-described embodiments, pet food suitable for pets can be recommended.

In each embodiment, both the intestinal flora information and the pet attribute are used, but only one of the information may be used. In addition, the food recommended based on the information of the intestinal flora may be corrected by the attribute.

Further, the present invention is applicable not only to pets but also to humans. In this case, what is recommended may be foods such as supplements.

The above-described embodiments have been described for the purpose of allowing a person having ordinary knowledge in the technical field to which the present invention belongs to carry out the present invention. Various modifications of the above embodiment can be naturally made by those skilled in the art, and the technical idea of the present invention can be applied to other embodiments. Accordingly, the invention is not limited to the described embodiments and should be the broadest scope in accordance with the technical ideas defined by the claims.

1 Attribute information acquisition means 2 Intestinal flora information acquisition means 3 Recommendation means 3a Classification means 3b Score evaluation means 3c Artificial intelligence device 4 Recommendation result presentation means 5 Food ordering means

Claims (25)

A classification step that classifies the pet into one of a plurality of predetermined groups based on the pet's attribute information and the pet's intestinal flora information.
A pet food recommendation method that includes steps to recommend food, taking into account which group it was classified into. The pet food recommendation method according to claim 1, wherein in the recommendation step, the veterinarian recommends a food suitable for the pet. The pet food recommendation method according to claim 1 or 2, wherein in the classification step, classification is performed according to whether or not a specific bacterium possessed by the pet exceeds a reference value based on the information of the intestinal flora. The pet food recommendation method according to claim 3, wherein the reference value corresponds to the attribute information of the pet. A step of scoring the pet based on the pet's attribute information and the pet's intestinal flora information.
A pet food recommendation method comprising a step of recommending food according to the score attached to the pet. The hood that is recommended for each score is preset,
The pet food recommendation method according to claim 5, wherein in the recommendation step, the food set to the score attached to the pet is recommended. The pet food recommendation method according to claim 5, wherein in the recommendation step, the veterinarian recommends a food suitable for the pet. The pet food recommendation method according to any one of claims 5 to 7, wherein in the step of scoring, the pet is scored in consideration of the intestinal age of the pet. 5. The step of scoring is to score the pet based on the information of the intestinal flora and considering the number and / or the type of the specific bacteria contained in the intestine of the pet. The pet food recommendation method according to any one of 8 to 8. A step to let the artificial intelligence device learn in advance the relationship between the information on the intestinal flora and the food to be recommended,
A pet food recommendation method comprising a step of recommending food to the pet based on information on the intestinal flora of the target pet by the artificial intelligence device. In the step of learning, the artificial intelligence device is made to learn in advance the relationship between the pet's attributes, the information on the intestinal flora, and the food to be recommended.
The pet food recommendation method according to claim 10, wherein in the recommendation step, the artificial intelligence device recommends food to the pet based on the attribute information of the target pet and the information of the intestinal flora. In the learning step, the artificial intelligence device is trained to create a model showing the relationship between the explanatory variables including the intestinal flora information and the hood information and the objective variable.
The pet food recommendation method according to claim 11, wherein in the recommendation step, a food that maximizes the objective variable with respect to the information of the intestinal flora is recommended based on the created model. The pet food recommendation method according to claim 12, wherein the objective variable includes user satisfaction with each food or pet health when each food is ingested. The pet food recommendation method according to claim 12 or 13, wherein in the learning step, the artificial intelligence device is trained to create a plurality of models according to the information of the intestinal flora. The pet food recommendation method according to any one of claims 12 to 14, wherein the explanatory variable includes a pet attribute. The pet food recommendation method according to any one of claims 12 to 15, wherein in the learning step, supervised learning or reinforcement learning is performed based on the explanatory variables and the objective variables prepared in advance. The pet food recommendation method according to any one of claims 12 to 16, wherein the learning step includes unsupervised learning so as to dimensionally compress the explanatory variables to generate explanatory variables input to the model. .. A pet food recommendation method in which a veterinarian recommends a food suitable for the pet based on the attribute information of the pet and the information of the intestinal flora of the pet. Steps to determine the relationship between pet attributes and intestinal flora and recommended foods as a rule,
A pet food recommendation method comprising a step of recommending food to the pet based on the attribute information and the intestinal flora information of the target pet and the above rule. The pet food recommendation method according to any one of claims 1 to 19, further comprising a step of inspecting the pet's stool and acquiring information on the pet's intestinal flora. The pet food recommendation method according to any one of claims 1 to 19, further comprising a step of estimating information on the intestinal flora of the pet from the attribute information of the pet and the information on the food consumed by the pet. A means for classifying the pet into one of a plurality of predetermined groups based on the pet attribute information and the pet's intestinal flora information.
A pet food recommendation device comprising a means of recommending food, taking into account which group it was classified into. A means for scoring the pet based on the pet's attribute information and the pet's intestinal flora information.
A pet food recommendation device comprising a means for recommending food according to a score attached to the pet. A pet food recommendation device provided with an artificial intelligence device that has learned in advance the relationship between the intestinal flora and the food to be recommended, and recommends the food to the pet based on the intestinal flora information of the target pet. A means for recommending food to the pet based on the rules defining the relationship between the pet's attributes and intestinal flora and the food to be recommended, and the target pet's attribute information and intestinal flora information. Equipped with pet food recommendation device.

Images