Insurance fee calculation system and insurance fee calculation method

Abstract

To provide an insurance fee calculation system and an insurance fee calculation method which calculate an insurance fee suitable for an animal.SOLUTION: An insurance fee calculation system includes an acquisition part for receiving intestinal bacterial flora data of an insurance subscription target animal, and a calculation part for calculating an insurance fee of the insurance subscription target animal, using the intestinal bacterial flora data of the insurance subscription target animal, wherein the calculation part classifies the insurance subscription target animal into a predetermined class, on the basis of the intestinal bacterial flora data of the insurance subscription target animal, by a program distributing the animal to a learned model of a relation between the intestinal bacterial flora data of the animal and a classification result using β diversity of the intestinal bacterial flora or a class for the animal set on the basis of the β diversity of the intestinal bacterial flora from the intestinal bacterial flora data of the animal, and calculates an insurance fee using the classification information of the class.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to an insurance premium calculation system and an insurance premium calculation method, and more particularly to an insurance premium calculation system and an insurance premium calculation method, and a disease risk prediction system, that calculate insurance premiums appropriate for animals using data on the intestinal flora of the animals. [Background technology]

[0002] Pets such as dogs, cats, and rabbits, as well as livestock such as cows and pigs, are irreplaceable to humans. In recent years, while the average lifespan of animals kept by humans has increased significantly, the number of animals suffering from some kind of injury or illness during their lives has increased, and the soaring medical expenses borne by owners have become a problem.

[0003] In order to maintain the health of animals, it is important to manage their physical condition through daily diet and exercise, and to respond quickly to any illnesses, but since animals cannot express their physical discomfort in their own words, the reality is that caretakers only realize that their animals are suffering from illness when symptoms progress and some externally observable signs appear. Also, when the musculoskeletal system is weakened due to calcium or vitamin deficiency, there are rarely any external signs, and in reality, the caretakers only notice that something is wrong with the animal when it suffers an injury such as a fracture.

[0004] Pet insurance has become popular as a way to deal with the rising medical costs for animals. It is desirable for the premiums for pet insurance to be calculated based on the prediction of future medical expenses for the insured pet, but since it is difficult to predict the specific medical expenses for each individual pet, the premiums are often calculated based on basic information such as the breed, age, and sex of the pet.

[0005] However, even for pets of the same breed, age, and sex, the medical expenses incurred over the course of a pet's life vary greatly from one individual to another due to differences in living environment, genetic background, etc. Therefore, depending on the individual, the premiums may end up being higher than expected under the previous premium calculation method.

[0006] Therefore, there is a need for an insurance premium calculation system or method that can easily calculate insurance premiums appropriate for individual animals that are the subject of pet insurance.

[0007] Patent Document 1 discloses an intestinal flora adjusting or improving composition that has the effect of effectively adjusting or improving the intestinal flora by increasing the number of Bacteroidetes bacteria and decreasing the number of Firmicutes bacteria in the intestinal flora, but does not disclose a method for predicting whether an animal will suffer from an injury or illness based on data regarding the animal's intestinal flora, or a method for calculating insurance premiums using intestinal flora data. [Prior art documents] [Patent documents]

[0008] [Patent Document 1] International Publication No. 2017 / 094892 Brochure Summary of the Invention [Problem to be solved by the invention]

[0009] Therefore, an object of the present invention is to provide an insurance premium calculation system and an insurance premium calculation method that can easily calculate an insurance premium appropriate for an individual animal that is the subject of pet insurance. [Means for solving the problem]

[0010] The inventors analyzed and examined data on the intestinal flora of animals that have pet insurance and a huge amount of data on whether or not the animals have filed insurance claims, i.e., whether or not they have been injured or ill. As a result, they discovered that data on the intestinal flora of animals can be used to calculate appropriate insurance premiums for each animal, and to predict the financial burden (medical expenses) and risk of contracting disease that the user will bear, and thus completed the present invention.

[0011] That is, the present invention relates to the following [1] to

[22] . [1] An insurance premium calculation system comprising: an acquisition unit that accepts intestinal flora data of an insured animal; and a calculation unit that calculates an insurance premium for the insured animal using the intestinal flora data of the insured animal, wherein the calculation unit classifies the insured animal into a predetermined class based on the intestinal flora data of the insured animal using a trained model that has learned the relationship between the intestinal flora data of the animal and a classification result using the beta diversity of the intestinal flora, or a program that assigns the animal from its intestinal flora data to one of the classes set based on the beta diversity of the intestinal flora, and calculates the insurance premium using the class classification information. [2] The insurance premium calculation system of [1], wherein the class in the classification result using the beta diversity of the intestinal bacterial flora, or the class set based on the beta diversity of the intestinal bacterial flora, is obtained by classifying each individual of the animal population for creating the classification based on the beta diversity among each individual of the animal population for creating the classification. [3] The insurance premium calculation system of [1] or [2], wherein the class in the classification result or the class set based on the beta diversity of the intestinal bacterial flora is a clustering result obtained by cluster analysis. [4] The insurance premium calculation system of [2], wherein the beta diversity includes a Jaccard coefficient or a Hamming distance. [5] The insurance premium calculation system according to [4], wherein the class in the classification result using the beta diversity of the intestinal flora, or the class set based on the beta diversity of the intestinal flora, is obtained by preparing data on the intestinal flora of multiple animal individuals for use in creating the classification, calculating the Jaccard coefficient for each of the multiple animal individuals with the intestinal flora of other animal individuals, and performing classification based on the Jaccard coefficient. [6] The insurance premium calculation system according to [5], wherein the class in the classification result using the beta diversity of the intestinal flora, or the class set based on the beta diversity of the intestinal flora, is obtained by preparing data on the intestinal flora of multiple animal individuals for use in creating the classification, calculating the Jaccard coefficient for each of the multiple animal individuals with the intestinal flora of other animal individuals, and performing cluster analysis on the data including the Jaccard coefficients for the multiple individuals obtained. [7] The insurance premium calculation system of [6], wherein the class in the classification result using the beta diversity of the intestinal flora, or the class set based on the beta diversity of the intestinal flora, is obtained by preparing data on the intestinal flora of multiple animal individuals for use in creating the classification, calculating the Jaccard coefficient for each of the multiple animal individuals with the intestinal flora of other animal individuals, performing a dimensionality reduction process on the data including the Jaccard coefficients for the multiple individuals obtained, plotting the data obtained by the dimensionality reduction process for each individual on a two-dimensional graph, and performing cluster analysis to perform classification. [8] The insurance premium calculation system of [6], wherein the cluster analysis is spectral clustering. [9] The insurance premium calculation system according to [2], wherein the trained model has been trained using, as training data, data on the intestinal flora of each animal in an animal population used to create a classification, and data on which class an animal has been classified into.

[10] A method for calculating insurance premiums comprising the steps of: a computer receiving intestinal flora data of an insured animal; and a computer calculating an insurance premium for the insured animal using the intestinal flora data of the insured animal, wherein the step of calculating the insurance premium for the insured animal is a step in which the computer classifies the insured animal into a predetermined class based on the intestinal flora data of the insured animal using a trained model that has learned the relationship between the intestinal flora data of the animal and a classification result using the beta diversity of the intestinal flora, or a program that assigns the animal from its intestinal flora data to one of the classes set based on the beta diversity of the intestinal flora, and calculates the insurance premium using the class classification information.

[11] An insurance premium calculation system comprising: an acquisition unit that accepts intestinal microbiota data of an insured animal; and a calculation unit that uses the intestinal microbiota data of the insured animal to calculate an insurance premium for the insured animal, wherein the calculation unit determines whether the intestinal microbiota data of the insured animal has characteristics possessed by intestinal microbiota data of an individual animal belonging to a specified cluster or is similar to the intestinal microbiota data of an individual animal belonging to a specified cluster, and calculates a loss rate or an insurance premium based on the determination, wherein the specified cluster is a cluster obtained by performing cluster analysis based on the beta diversity of the intestinal microbiota of a plurality of individual animals and classifying the plurality of individual animals into a plurality of clusters.

[12] A disease risk prediction system comprising an acquisition unit that receives intestinal flora data of a target animal, and a prediction unit that calculates a disease risk of the target animal using the intestinal flora data of the target animal, wherein the prediction unit classifies the target animal into a predetermined class based on the intestinal flora data of the target animal using a trained model that has learned the relationship between the intestinal flora data of the animal and a classification result using the beta diversity of the intestinal flora, or a program that assigns the animal from its intestinal flora data to one of the classes set based on the beta diversity of the intestinal flora, and predicts the disease risk using the class classification information.

[13] A disease risk prediction system according to

[12] , wherein the class in the classification result using the β diversity of the intestinal microbiota, or the class set based on the β diversity of the intestinal microbiota, is obtained by classifying each individual of the animal population for creating the classification based on the β diversity among each individual of the animal population for creating the classification.

[14] A disease risk prediction system according to

[12] or

[13] , wherein the classes in the classification result or the classes set based on the beta diversity of the intestinal microbiota are clustering results obtained by cluster analysis.

[15] The disease risk prediction system of

[12] , wherein the beta diversity includes a Jaccard coefficient or a Hamming distance.

[16] A disease risk prediction system according to

[15] , wherein the class in the classification result using the beta diversity of the intestinal microbiota, or the class set based on the beta diversity of the intestinal microbiota, is obtained by preparing data on the intestinal microbiota of multiple animal individuals for creating a classification, calculating the Jaccard coefficient for each of the multiple animal individuals with the intestinal microbiota of other animal individuals, and performing classification based on the Jaccard coefficient.

[17] A disease risk prediction system according to

[16] , wherein the classes in the classification result using the beta diversity of the intestinal microbiota, or the classes set based on the beta diversity of the intestinal microbiota, are obtained by preparing data on the intestinal microbiota of multiple animal individuals for use in creating a classification, calculating the Jaccard coefficient for each of the multiple animal individuals with the intestinal microbiota of other animal individuals, and performing cluster analysis on the data including the Jaccard coefficients for the multiple individuals obtained.

[18] The disease risk prediction system according to

[17] , wherein the classes in the classification result using the beta diversity of the intestinal microbiota, or the classes set based on the beta diversity of the intestinal microbiota, are obtained by preparing data on the intestinal microbiota of multiple animal individuals for use in creating a classification, calculating the Jaccard coefficient for each of the multiple animal individuals with the intestinal microbiota of the other animal individuals, performing a dimensionality reduction process on the data including the Jaccard coefficients for the multiple individuals obtained, plotting the data obtained by the dimensionality reduction process for each individual on a two-dimensional graph, and performing cluster analysis to perform classification.

[19] The disease risk prediction system of

[18] , wherein the cluster analysis is spectral clustering.

[20] A disease risk prediction system according to

[13] , in which the trained model is trained using, as training data, data on the intestinal flora of each individual animal in the animal population being used to create the classification, and data on which class an individual animal is classified into.

[21] A device comprising: an acquisition unit that receives intestinal flora data of a target animal; and a prediction unit that calculates a disease risk of the target animal using the intestinal flora data of the target animal, wherein the prediction unit classifies the target animal into a predetermined class based on the intestinal flora data of the target animal using a trained model that has learned the relationship between the intestinal flora data of the animal and a classification result using the β diversity of the intestinal flora, or a program that assigns the animal to one of the classes set based on the β diversity of the intestinal flora from the intestinal flora data of the animal, and predicts the disease risk using classification information of the class, and the prediction unit predicts the disease risk using the classification information of the class, wherein the prediction unit classifies the target animal into a predetermined class based on the intestinal flora data of the target animal using a trained model that has learned the relationship between the intestinal flora data of the animal and a classification result using the β diversity of the intestinal flora, or a program that assigns the animal to one of the classes set based on the β diversity of the intestinal flora from the intestinal flora data of the animal. the determined classes are obtained by preparing data on the intestinal flora of a plurality of animal individuals for use in creating a classification, calculating the Jaccard coefficient or Hamming distance for each of the plurality of animal individuals with the intestinal flora of the other animal individuals, performing a dimensionality reduction process on the data including the Jaccard coefficient or Hamming distance for the obtained plurality of individuals, plotting the data obtained by the dimensionality reduction process for each individual on a two-dimensional graph, and performing cluster analysis to perform classification; and a predetermined disease risk value is assigned to each class in the classification result using the β diversity of the intestinal flora, or to each class set based on the β diversity of the intestinal flora.

[22] The disease risk prediction system of

[21] , wherein a predetermined disease risk value based on the average Shannon coefficient of the intestinal microbiota data of individual animals belonging to each class is assigned to each class in the classification result using the beta diversity of the intestinal microbiota, or to each class set based on the beta diversity of the intestinal microbiota. Effect of the Invention

[0012] The present invention makes it possible to provide an insurance premium calculation system, an insurance premium calculation method, and a disease risk prediction system that can simply calculate insurance premiums appropriate for individual animals that are the subject of pet insurance. [Brief description of the drawings]

[0013] [Figure 1] FIG. 1 is a schematic diagram of an insurance premium calculation system. [Diagram 2] FIG. 13 is a diagram illustrating an insurance premium calculation method. [Diagram 3] FIG. 1 is a graph according to an embodiment. [Figure 4] FIG. 1 is a graph according to an embodiment. [Diagram 5] FIG. 1 is a graph according to an embodiment. [Figure 6] FIG. 1 is a graph according to an embodiment. [Figure 7] FIG. 1 is a schematic diagram of a disease risk prediction system. [Figure 8] FIG. 1 is a diagram illustrating a method for predicting disease risk. [Figure 9A] FIG. 1 is a graph according to an embodiment. [Figure 9B] FIG. 1 is a graph according to an embodiment. [Figure 9C] FIG. 1 is a graph according to an embodiment. [Figure 9D] FIG. 1 is a graph according to an embodiment. [Figure 10A] FIG. 1 is a graph according to an embodiment. [Figure 10B] FIG. 1 is a graph according to an embodiment. [Figure 10C] FIG. 1 is a graph according to an embodiment. [Figure 10D] FIG. 1 is a graph according to an embodiment. [Figure 11A] FIG. 1 is a graph according to an embodiment. [Figure 11B] FIG. 1 is a graph according to an embodiment. [Figure 11C] FIG. 1 is a graph according to an embodiment. [Figure 11D] FIG. 1 is a graph according to an embodiment. [Figure 12A] FIG. 1 is a graph according to an embodiment. [Figure 12B] FIG. 1 is a graph according to an embodiment. [Figure 12C] FIG. 1 is a graph according to an embodiment. [Figure 12D] FIG. 1 is a graph according to an embodiment. [Figure 13A] FIG. 1 is a graph according to an embodiment. [Figure 13B] FIG. 1 is a graph according to an embodiment. [Figure 13C] FIG. 1 is a graph according to an embodiment. [Figure 13D] FIG. 1 is a graph according to an embodiment. [Figure 14A] FIG. 1 is a graph according to an embodiment. [Figure 14B] FIG. 1 is a graph according to an embodiment. [Figure 14C] FIG. 1 is a graph according to an embodiment. [Figure 14D] FIG. 1 is a graph according to an embodiment. [Figure 14E] FIG. 1 is a graph according to an embodiment. [Figure 15] FIG. 1 is a graph according to an embodiment. [Figure 16] FIG. 1 is a graph according to an embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0014] [Insurance premium calculation system] The insurance premium calculation system of the present invention has an acquisition unit that accepts intestinal flora data of the insured animal, and a calculation unit that calculates the insurance premium for the insured animal using the intestinal flora data of the insured animal, wherein the calculation unit classifies the insured animal into a predetermined class based on the intestinal flora data of the insured animal using a trained model that has learned the relationship between the intestinal flora data of the animal and the classification result using the beta diversity of the intestinal flora, or a program that assigns the animal from its intestinal flora data to one of the classes set based on the beta diversity of the intestinal flora, and calculates the insurance premium using the class classification information.

[0015] [Acquisition Department] The acquisition unit acquires data on the intestinal flora of the animal. The configuration of the acquisition unit is not particularly limited, and any known configuration for accepting or acquiring data can be adopted. For example, the acquisition unit may be configured to accept transmission or input of data on the intestinal flora or diversity data from an external terminal or computer via a network such as the Internet, or may be an interface.

[0016] [Animals eligible for insurance] The insured animal is an animal for which an insurance for animals, such as pet insurance, is about to be applied for. The insured animal is preferably a pet animal, more preferably a dog, cat, rabbit, bird, reptile, or amphibian, and particularly preferably a dog, cat, or rabbit.

[0017] [Intestinal microbiota data] The data on the intestinal microbiota is data on the bacteria present in the intestines of an animal, and includes data on the type, presence or absence, ratio, distribution, etc. of bacteria. For example, data on the presence or absence, ratio, etc. of bacteria classified into a specific family, phylum, or genus can be included. The data of the intestinal microbiota can be measured by known metagenomic analysis methods and microbiota analysis methods, such as amplicon sequencing and shotgun sequencing using a sequencer such as NGS. For example, a method of collecting a sample such as feces from an animal and analyzing the DNA and RNA sequence information of all organisms contained in the sample using a next-generation sequencer to identify the organisms contained in the sample is exemplified. Preferably, a method of amplifying all or a part of the 16SrRNA gene contained in the sample as necessary, sequencing, and analyzing the obtained sequence using software to obtain bacterial composition data in the sample is exemplified.

[0018] An example of amplicon analysis (meta-16S analysis) of 16SrRNA genes using NGS (next-generation sequencer) will be specifically described. First, DNA is extracted from a sample using a DNA extraction reagent, and the 16SrRNA gene is amplified from the extracted DNA by PCR. Then, the amplified DNA fragments are comprehensively sequenced using NGS, low-quality reads and chimeric sequences are removed, and the sequences are clustered to perform OTU (Operational Taxonomic Unit) analysis. An OTU is an operational taxonomic unit that treats sequences with a certain level of similarity (e.g., 96-97% or more homology) as a single bacterial species. Therefore, the number of OTUs represents the number of bacterial species that make up the bacterial flora, and the number of reads belonging to the same OTU represents the relative abundance of that species. In addition, a representative sequence can be selected from the number of reads belonging to each OTU, and the family name and genus / species name can be identified by database search. In this way, the number of bacterial species belonging to a specific family or phylum can be measured. Analysis by ASV (Amplicon Sequence Variant) is also possible. Since ASV is created after removing erroneous sequences generated during PCR and sequencing, it is possible to distinguish sequence mutations of one base unit and to perform more precise identification. The insurance premium calculation system of the present invention may use data of intestinal flora measured in advance, or may receive a fecal sample from a user to measure data of the intestinal flora and use the data. When using data of intestinal flora measured in advance, the user, for example, sends a fecal sample to an intestinal bacteria analysis company, requests the measurement of the intestinal flora, and receives the data of the intestinal flora from the requested analysis company. The user can transmit the data of the intestinal flora received in this way to the insurance premium calculation system of the present invention through a user terminal. In addition, the requested analysis company may transmit the data of the intestinal flora to the insurance premium calculation system of the present invention on behalf of the user.

[0019] The data on the intestinal microbiota may be linked to data related to the bacterial diversity (alpha diversity) of the intestinal microbiota of an animal. A high diversity of the intestinal microbiota means that the intestinal microbiota contains a wide variety of bacteria evenly. There are several types of diversity indicators, so-called diversity indices, and any known indices may be used in the present invention. Examples of diversity indices include the Shannon-Weiner diversity index (hereinafter sometimes abbreviated as "Shannon index"), the Simpson index, the number of unique sequences detected by a sequencer (Amplicon Sequence Variant: ASV), the number of OTUs (Operational Taxonomic Units), Faith's PD, Pielou's evenness, etc.

[0020] [Calculation section] The calculation unit of the insurance premium calculation system of the present invention uses the intestinal flora data of the insured animal to calculate the insurance premium for the insured animal, and specifically, classifies the insured animal into a predetermined class based on the intestinal flora data of the insured animal using a trained model that has learned the relationship between the intestinal flora data of the animal and the classification result using the β diversity of the intestinal flora, or a program that assigns the animal from the intestinal flora data of the animal to one of the classes set based on the β diversity of the intestinal flora, and calculates the insurance premium using the classification information of the class. The calculation unit may calculate a loss ratio based on the intestinal flora data, and calculate the insurance premium based on the loss ratio. The calculation unit is composed of, for example, a processor or an arithmetic processing device such as a CPU or a GPU, and calculates the insurance premium using a trained model or software or a program including the trained model. The trained model or software or a program including the trained model may be stored in a separate storage device. In addition, the software or program having the function of classifying the insured animal into a predetermined class using the trained model and the software or program for calculating the insurance premium using the class classification information may be configured separately or may be integrated. The program for classifying the animal into one of the classes set based on the beta diversity of the intestinal flora from the intestinal flora data of the animal may or may not include a trained model. Such a program is, for example, a program having the function of classifying the animal into one of the predetermined classes according to the specific bacterial composition or characteristics in the intestinal flora data, the presence or absence or content ratio of a predetermined bacterium, etc.

[0021] [Trained model] The trained model used in the present invention is preferably artificial intelligence (AI). Artificial intelligence (AI) is software or a system that imitates the intellectual work performed by the human brain using a computer, specifically, a computer program that understands natural language used by humans, performs logical inference, and learns from experience. The artificial intelligence may be either a general-purpose type or a specialized type, specifically, any of the well-known types such as logistic regression, decision tree, k-means, multilayer perceptron, recurrent neural network, deep neural network, convolutional neural network, etc., and publicly available software can be used.

[0022] An example of learning to generate a trained model is machine learning.

[0023] The learning method for generating a trained model is not particularly limited, and publicly available software can be used. For example, learning may be performed by a known support vector machine method published in "Introduction to Support Vector Machines" (Kyoritsu Shuppan) or the like. Machine learning can be either unsupervised learning or supervised learning, but supervised learning is preferred. The method of supervised learning is not particularly limited, and examples include decision trees, ensemble learning, gradient boosting, and the like. Examples of published machine learning algorithms include XGBoost, CatBoost, and LightGBM.

[0024] [Classification results] The trained model of the present invention has learned the relationship between an animal's intestinal microbiota data and classification results using the beta diversity of the intestinal microbiota. β diversity represents the degree of difference in diversity between two samples. For example, it is an index expressed as the distance between two points, and the greater the distance, the more different the composition of the two samples is. Examples of β diversity include the Jaccard coefficient (Jaccard distance), Dice coefficient, Simpson coefficient, Bray-Curtis distance, Hamming distance, Unifrac distance, Unweighted Unifrac distance, and Weighted Unifrac distance, and the Jaccard coefficient is preferred. Note that α diversity represents the diversity in one sample. The class in the classification result using the β diversity of the intestinal microbiota refers to, for example, the result of dividing each individual into a predetermined number of classes using the β diversity calculated from the data of the intestinal microbiota of each of a plurality of animals prepared in advance, or each class obtained thereby, and the classification result using the β diversity of the intestinal microbiota of the target animal refers to which class the target animal is assigned to. Since β diversity represents the degree of difference in diversity between two samples, for example, if individuals with similar β diversity are classified into one class, it is possible to form one class with individuals with similar compositions of intestinal microbiota. The present inventors have found that such classification can be used to calculate insurance premiums. After classification using β diversity, other parameters, for example, α diversity, may be used to combine classes, reduce the number of classes, or further subdivide classes. As a specific example, data on the intestinal flora of multiple animal individuals, for example 10,000 individuals, is prepared, and β diversity such as the Jaccard coefficient is calculated between each individual. Then, using the β diversity, the multiple individuals, for example 10,000 individuals, are classified into a predetermined number of classes, for example five classes A to E. The classification criteria are not particularly limited, and classification may be performed based on the average size of β diversity or by performing cluster analysis as described below. The number of individuals is not particularly limited, but is preferably 5,000 or more, more preferably 10,000 or more, and even more preferably 30,000 or more.

[0025] In the insurance premium calculation system of the present invention, it is preferable that the classes in the classification result using the β diversity of the intestinal flora are obtained by preparing data on the intestinal flora of multiple animal individuals for creating the classification, calculating the Jaccard coefficient for each of the multiple animal individuals with the intestinal flora of other animal individuals, and performing classification based on the Jaccard coefficient.

[0026] In the insurance premium calculation system of the present invention, it is more preferable that the classes in the classification result using the β diversity of the intestinal flora are obtained by preparing data on the intestinal flora of multiple animal individuals for creating the classification, calculating the Jaccard coefficient for each of the multiple animal individuals with the intestinal flora of the other animal individuals, performing a dimensionality reduction process (dimensionality compression process) on the data including the Jaccard coefficients for the multiple individuals obtained, plotting the data obtained by the dimensionality reduction process for each individual on a two-dimensional graph, and performing cluster analysis to perform classification. Examples of the cluster analysis include non-hierarchical cluster analysis and spectral clustering, and spectral clustering is preferred. The method of dimensionality reduction is not particularly limited, and examples include UMAP, PCA, t-SNE, PCoA, etc.

[0027] The number of classes in the classification is not particularly limited, and may be any of 3 to 100, preferably any of 3 to 70, and more preferably any of 3 to 50. The more classes there are, the more detailed the insurance premium setting becomes, but if there are too many classes, the calculation becomes complicated.

[0028] In the present invention, by using a trained model that has learned the relationship between the classification result of which class an individual has been classified into and the intestinal flora data of that individual, it is possible to classify the insured animal and calculate the insurance premium without having to calculate the beta diversity with the intestinal flora data of other individuals one by one. Such a trained model can be generated by performing training using, for example, data on the intestinal flora of each individual animal in the animal population for creating classifications and data on which class an individual animal has been classified into as training data. In the above specific example, data on the intestinal flora of some or all of the 10,000 animals for creating classifications and the classification results indicating which class each individual has been classified into (A to E) can be used as training data.

[0029] The calculation unit of the present invention may use multiple trained models for each type or breed of animal. For example, when multiple trained models are used for each breed of animal, the trained model used is a trained model corresponding to a specific breed. Such a trained model is a trained model that has learned the relationship between the intestinal flora data for a specific breed of animal and the classification result of the animal.

[0030] As the trained model corresponding to the breed, it is preferable to use data on the intestinal flora of animals related to one or more specific breeds and labels and data on the classification of the animals as teacher data. For example, it is a trained model trained using data on the intestinal flora of a group of animals of only a specific breed and labels and data on the classification of the animals. In this case, the group of animals of only a specific breed may be a group of only one breed, for example, a toy poodle, or a group including multiple breeds, for example, a toy poodle, a pomeranian, and a miniature dachshund. In addition, in the case of multiple breeds, the breeds may be classified into a trained model for large breeds trained using data on the intestinal flora of large breeds, a trained model for medium breeds trained using data on the intestinal flora of medium breeds, a trained model for small breeds trained using data on the intestinal flora of small breeds, and a trained model for extra-small breeds trained using data on the intestinal flora of extra-small breeds, and the trained model may be configured to have trained models corresponding to each classification. For example, in the case of dogs, breeds with an average adult weight of less than 4 kg are called extra small dogs, those with an average adult weight of 4-10 kg are called small dogs, those with an average adult weight of 10-25 kg are called medium dogs, and those with an average adult weight of 25 kg or more are called large dogs. Breeds may also be classified by height, length, genetic relationship, etc.

[0031] When the calculation unit of the present invention uses a plurality of trained models for each type or breed of animal, the system of the present invention is preferably configured such that, for example, the acquisition unit acquires information on the type or breed of the animal as basic information of the animal. Alternatively, the system of the present invention may be configured to separately include a breed determination unit. The breed determination unit is a means for determining the breed of an animal from, for example, an image of the animal input to the acquisition unit, using a trained model for breed determination. The trained model for breed determination is, for example, a trained model that has learned the relationship between the image of the animal and the breed of the animal. Preferably, the trained model includes a trained model that uses an image of the animal and the breed of the animal as training data for learning, and has the image of the animal as the input and the determination of the breed of the animal as the output.

[0032] In addition, the calculation unit of the present invention may use multiple trained models for each age or age group of the animal. The age group is a classification for each age range, and specific examples include classifications such as young (under 1 year old), adult (1 year to under 7 years old), and old (7 years or older). The method of classification is not particularly limited, and classifications such as 0 to under 3 years old, 3 to under 7 years old, and 7 years old or older may be used. When the calculation unit of the present invention uses multiple trained models for each age or age group of the animal, the trained model used is a trained model corresponding to a specific age or age group. Such a trained model is a trained model that has learned the relationship between data on the intestinal flora of an individual or group of animals belonging to a specific age or age group and the classification result of the animal. When the calculation unit of the present invention uses multiple trained models for each age or age group of the animal, the system of the present invention is preferably configured such that, for example, the acquisition unit acquires information on the age of the animal as basic information of the animal. Alternatively, the system of the present invention may be configured to separately include an age determination unit. The age determination unit is a means for determining the age of an animal from, for example, an image of the animal input to the acquisition unit, using a trained model for age determination. The trained model for age determination is, for example, a trained model that has learned the relationship between the image of the animal and the age of the animal. Preferably, the trained model includes a trained model that uses the image of the animal and the age of the animal as teacher data for learning, and has the image of the animal as input and the determination of the age of the animal as output.

[0033] [Calculating insurance premiums using class classification information] The insurance premium calculation system of the present invention calculates the insurance premium using classification information on the class into which the insured animal is classified. The calculation method is not particularly limited, and for example, a correspondence table between each class and insurance premium may be created in advance, and the corresponding insurance premium may be calculated according to the class into which the animal is classified. The insurance premium may be calculated by using the class into which the animal is classified as an addition or subtraction factor, and taking into account other basic information such as breed, age, sex, and medical history. The insurance premium may be calculated according to the class into which the animal is classified based on the correspondence table, and the final insurance premium may be calculated by adding or subtracting the insurance premium according to the basic information. The insurance premium calculated using the class classification information may be corrected based on information on the food that the insured animal normally takes in and information on the lifestyle habits, and the final insurance premium may be calculated. The loss ratio or its average may be calculated for each class, and the insurance premium may be calculated according to the loss ratio.

[0034] In another embodiment of the calculation unit of the present invention, the calculation unit may determine whether the intestinal flora data of the insured animal has characteristics of the intestinal flora data of an animal individual belonging to a predetermined cluster or is similar to the intestinal flora data of an animal individual belonging to a predetermined cluster, and calculate the loss ratio or insurance premium based on the determination. In this case, the predetermined cluster is preferably a cluster obtained by performing cluster analysis based on the β diversity of the intestinal flora of a plurality of animal individuals and classifying the plurality of animal individuals into a plurality of clusters. For example, data on the intestinal flora of 10,000 animals is prepared, and cluster analysis is performed on the data based on β diversity to classify the data into a predetermined number of clusters. As a result, it is assumed that a certain cluster (called cluster X) contains many animal individuals having bacteria a, bacteria b, and bacteria c as intestinal bacteria, and when the average loss ratio of the animal individuals included in cluster X is calculated, it is found that the average loss ratio is N. On the other hand, if the data on the intestinal flora of the insured individual animal is examined and it is found to have the same characteristics as those of cluster X, i.e., bacteria a, b, and c, then the insured individual animal is expected to have an average loss rate N similar to that of cluster X. Then, the insurance premium can be calculated based on the expected value of the loss rate.

[0035] [output] The calculation unit calculates the insurance premium appropriate for the insured animal based on the data on the intestinal flora of the insured animal. The format of the output of the calculation result is not particularly limited. The insurance premium calculation system of the present invention may also have a separate output unit that receives the calculation results from the calculation unit and outputs the calculation results, and a transmission unit that transmits the judgment results to a personal computer or mobile terminal used by the user via a communication line.

[0036] [Insurance premium calculation method] The insurance premium calculation method of the present invention includes a step of a computer receiving intestinal flora data of an insured animal, and a step of a computer calculating an insurance premium for the insured animal using the intestinal flora data of the insured animal, the step of calculating the insurance premium for the insured animal being characterized in that the computer classifies the insured animal into a predetermined class based on the intestinal flora data of the insured animal using a trained model that has learned the relationship between the intestinal flora data of the animal and the classification result using the β diversity of the intestinal flora, or a program that assigns the animal from the intestinal flora data of the animal to one of the classes set based on the β diversity of the intestinal flora, and calculates the insurance premium using the classification information of the class. The step of classifying the insured animal into a predetermined class based on the intestinal flora data of the insured animal using the trained model and the step of calculating the insurance premium using the classification information of the class may be the same step or may be executed as separate steps. The meanings of specific terms in each step are the same as those in the above-mentioned insurance premium calculation system.

[0037] [System Overview] Fig. 1 is a diagram for explaining an overview of an insurance premium calculation system according to one embodiment of the present invention. As shown in Fig. 1, the insurance premium calculation system according to this embodiment includes a server 1 and a user terminal 2. The server 1 and the terminal 2 are connected via a network. The server 1 includes a processing operation unit (CPU) 10, a storage unit 20, and an interface unit 30. It should be noted that users in the present invention include owners of insured animals, as well as agents, breeders, insurance companies that provide insurance for animals such as pet insurance, and the like.

[0038] Further, the processing operation unit (CPU) 10 includes a calculation unit 11, the storage unit 20 includes at least a setting unit 21, and the interface unit 30 includes an acquisition unit 31 and an output unit 32. The setting unit 21 may be configured to store a trained model that has learned the relationship between the data of the intestinal microbiota and the classification result using β diversity, or software including the trained model. In such a configuration, the calculation unit 11 can call up the trained model stored in the setting unit 21 or software including the trained model based on the basic information of the animal, and calculate the insurance risk or insurance premium based on the diversity data of the animal. In this embodiment, the trained model is stored in the setting unit 21, but the trained model may be stored in the storage unit instead of the setting unit 21.

[0039] Here, the calculation unit 11 determines into which class the insured animal is classified based on a trained model that has learned the relationship between data on the animal's intestinal flora and the classification results using β diversity, and calculates the insurance premium using the classification information for that class. Furthermore, the acquisition unit 31 may acquire intestinal bacterial flora diversity data, and preferably further basic information about the pet animal, and the output unit 32 presents information such as insurance premiums calculated by the calculation unit 11 to the user.

[0040] [Insurance premium calculation method] FIG. 2 is a diagram for explaining an outline of an insurance premium calculation method according to one embodiment of the present invention. As shown in Figure 2, the insurance premium calculation method according to this embodiment sequentially includes step S1 of acquiring data on the intestinal flora of the insured animal, step S2 of classifying the insured animal into a predetermined class based on the data on the intestinal flora of the insured animal, and step S3 of calculating the insurance premium using the classification information of the class.

[0041] Here, it is preferable to further include a step of acquiring basic information of the insured animal, and it is preferable to correct the insurance risk and the insurance premium based on the basic information of the insured animal. Alternatively, it is also preferable to calculate a provisional insurance premium once based on the basic information, and then correct the provisional insurance premium using the insurance risk calculated by the trained model to calculate the final insurance premium. By using the basic information of the insured animal, a more suitable insurance premium can be calculated. In this way, when correcting the provisional insurance premium based on the basic information by adding the insurance risk derived using the class classification information, it is preferable to correct the provisional insurance premium between 0.5 and 2.0 times. In addition, the insurance premium may be calculated based on the medical expenses expected based on the basic information, and the medical expenses obtained by correcting the medical expenses according to the insurance risk derived based on the class classification information.

[0042] Furthermore, the method may include a step of predicting the future insurance risk of the insured animal based on information on the food that the insured animal normally consumes. Since the alpha diversity of the intestinal flora increases depending on the food that the insured animal eats, by taking into account the information on the food, it is expected that the calculation of the insurance premium will be more precise.

[0043] Generally, insurance premiums are calculated by multiplying the medical expenses of the animal by the insurance company's burden rate, and then adding the insurance company's commission, operating expenses, etc. Therefore, it is possible to predict the financial burden (medical expenses) that the user will bear in the same way that the insurance premium calculation system calculates insurance premiums.

[0044] [Disease risk prediction system] The disease risk prediction system of the present invention comprises an acquisition unit that receives intestinal flora data of a target animal, and a prediction unit that predicts the disease risk of the target animal using the intestinal flora data of the target animal, and the prediction unit classifies the target animal into a predetermined class based on the intestinal flora data of the target animal using a trained model that has learned the relationship between the intestinal flora data of the animal and the classification result using the beta diversity of the intestinal flora, or a program that assigns the animal from its intestinal flora data to one of the classes set based on the beta diversity of the intestinal flora, and predicts the disease risk using the class classification information.

[0045] [disease] The type of disease that is the subject of the present invention is not particularly limited, and examples include skin diseases, otological diseases, musculoskeletal diseases, ophthalmic diseases, digestive diseases, systemic diseases, urinary diseases, hepatic, biliary and pancreatic diseases, circulatory diseases, nervous system diseases, respiratory diseases, dental and oral diseases, endocrine diseases, reproductive diseases, and blood and hematopoietic diseases. Skin diseases include, for example, dermatitis, atopic dermatitis, and pyoderma. Examples of otological diseases include otitis externa and otitis media. Examples of musculoskeletal disorders include patellar luxation and herniated disc. Examples of ophthalmological diseases include conjunctivitis, eye discharge, keratitis, corneal ulcer / erosion, epiphora, cataracts, and glaucoma. Digestive system diseases include, for example, gastritis and enteritis. Systemic illnesses include, for example, lethargy and collapse. Examples of urinary system diseases include cystitis and urolithiasis. Examples of diseases of the hepatic, biliary and pancreatic systems include cholesteatosis and chronic renal failure. Examples of circulatory system diseases include valvular disease and cardiomyopathy. Nervous system disorders include, for example, epilepsy and seizures. Respiratory diseases include, for example, coughing, rhinitis, tracheal collapse, and bronchial stenosis. Dental and oral diseases include, for example, periodontal disease and stomatitis. Endocrine diseases include, for example, hypothyroidism and diabetes. Reproductive system diseases include, for example, mammary tumors and balanitis. Blood and hematopoietic system diseases include, for example, tumors of lymphatic tissues and thrombocytopenia.

[0046] The acquisition unit of the disease risk prediction system is similar to the acquisition unit of the insurance premium calculation system described above.

[0047] [Target animals] The target animal is an animal that is the subject of disease risk prediction. For example, it is a pet or livestock kept or managed by a user of the disease risk prediction system or a representative thereof. Preferably, it is a pet, more preferably a dog, cat, rabbit, bird, reptile, or amphibian, and particularly preferably a dog, cat, or rabbit.

[0048] [Intestinal microbiota data] The data on the intestinal flora is the same as the data on the intestinal flora in the insurance premium calculation system.

[0049] [Prediction section] The prediction unit of the disease risk prediction system of the present invention predicts the disease risk of a target animal using intestinal flora data of the target animal, and specifically, classifies the target animal into a predetermined class based on the intestinal flora data of the target animal using a trained model that has learned the relationship between the intestinal flora data of the animal and a classification result using the β diversity of the intestinal flora, or a program that classifies the animal from the intestinal flora data of the animal into one of the classes set based on the β diversity of the intestinal flora, and predicts the disease risk using the classification information of the class. The prediction unit may calculate a damage rate based on the intestinal flora data, and calculate the disease risk based on the damage rate. The prediction unit is composed of, for example, a processor or an arithmetic processing device such as a CPU or a GPU, and predicts the disease risk using a trained model or software or a program including the trained model. The trained model or software or a program including the trained model may be stored in a separate storage device. In addition, the software or program having the function of classifying a target animal into a predetermined class using a trained model and the software or program for predicting a disease risk using class classification information may be configured separately or may be integrated. The program for classifying an animal into one of the classes set based on the beta diversity of the intestinal flora from the intestinal flora data of the animal preferably does not include a trained model. Such a program is, for example, a program having the function of classifying the animal into one of the predetermined classes according to the specific bacterial composition or characteristics in the intestinal flora data, the presence or absence or content ratio of a predetermined bacterium, etc.

[0050] [Trained model] The trained model is the same as the trained model in the above-mentioned insurance premium calculation system.

[0051] [Classification results] The trained model in the disease risk prediction system of the present invention has learned the relationship between an animal's intestinal microbiota data and classification results using the beta diversity of the intestinal microbiota. β diversity represents the degree of difference in diversity between two samples. For example, it is an index expressed as the distance between two points, and the greater the distance, the more different the composition of the two samples is. Examples of β diversity include the Jaccard coefficient (Jaccard distance), Dice coefficient, Simpson coefficient, Bray-Curtis distance, Hamming distance, Unifrac distance, Unweighted Unifrac distance, and Weighted Unifrac distance, and the Jaccard coefficient is preferred. Note that α diversity represents the diversity in one sample. The class in the classification result using the β diversity of the intestinal microbiota refers to, for example, the result of dividing each individual into a predetermined number of classes using the β diversity calculated from the data of the intestinal microbiota of each of a plurality of animals prepared in advance, or each class obtained thereby, and the classification result using the β diversity of the intestinal microbiota of the target animal refers to which class the target animal is assigned to. Since β diversity represents the degree of difference in diversity between two samples, for example, if individuals with similar β diversity are classified into one class, it is possible to form one class with individuals with similar compositions of the intestinal microbiota. The present inventors have found that such classification can be used to predict disease risk. After classification using β diversity, other parameters, such as α diversity, may be used to combine classes to reduce the number of classes or to further subdivide the classes. As a specific example, data on the intestinal flora of multiple animal individuals, for example 10,000 individuals, is prepared, and β diversity such as Jaccard coefficient or Hamming distance between each individual is calculated. Then, using the β diversity, the multiple individuals, for example 10,000 individuals, are classified into a predetermined number of classes, for example, five classes A to E. The classification criteria are not particularly limited, and classification may be performed based on the average size of β diversity or by performing cluster analysis as described below. The number of individuals is not particularly limited, but is preferably 5,000 individuals or more, more preferably 10,000 individuals or more, and even more preferably 30,000 individuals or more.

[0052] In the disease risk prediction system of the present invention, it is preferable that the classes in the classification result using the β diversity of the intestinal microbiota are obtained by preparing data on the intestinal microbiota of multiple animal individuals for creating the classification, calculating the Jaccard coefficient for each of the multiple animal individuals with the intestinal microbiota of other animal individuals, and performing classification based on the Jaccard coefficient.

[0053] In the disease risk prediction system of the present invention, it is more preferable that the classes in the classification result using the β diversity of the intestinal flora are obtained by preparing data on the intestinal flora of multiple animal individuals for creating a classification, calculating the Jaccard coefficient for each of the multiple animal individuals with the intestinal flora of the other animal individuals, performing a dimensionality reduction process (dimensionality compression process) on the data including the Jaccard coefficient for the multiple individuals obtained, plotting the data obtained by the dimensionality reduction process for each individual on a two-dimensional graph, and performing cluster analysis to perform classification. Examples of the cluster analysis include non-hierarchical cluster analysis and spectral clustering, and spectral clustering is preferred. The method of dimensionality reduction is not particularly limited, and examples include UMAP, PCA, t-SNE, PCoA, etc.

[0054] The number of classes in the classification is not particularly limited, and may be any of 3 to 100, preferably any of 3 to 70, and more preferably any of 3 to 50. The more classes there are, the more detailed the prediction of disease risk becomes possible, but if the number of classes is too large, the calculation becomes complicated.

[0055] In the disease risk prediction system of the present invention, a trained model is used that has learned the relationship between the classification result of which class an individual is classified into and the intestinal flora data of that individual, making it possible to classify the target animal and predict disease risk without having to calculate the beta diversity with the intestinal flora data of other individuals one by one. Such a trained model can be generated by performing training using, for example, data on the intestinal flora of each individual animal in the animal population for creating classifications and data on which class an individual animal has been classified into as training data. In the above specific example, data on the intestinal flora of some or all of the 10,000 animals for creating classifications and the classification results indicating which class each individual has been classified into (A to E) can be used as training data.

[0056] The prediction unit of the present invention may use multiple trained models for each type or breed of animal. For example, when multiple trained models are used for each breed of animal, the trained model used is a trained model corresponding to a specific breed. Such a trained model is a trained model that has learned the relationship between the data on the intestinal flora of an animal of a specific breed and the classification result of the animal.

[0057] As the trained model corresponding to the breed, it is preferable to use data on the intestinal flora of animals related to one or more specific breeds and labels and data on the classification of the animals as teacher data. For example, it is a trained model trained using data on the intestinal flora of a group of animals of only a specific breed and labels and data on the classification of the animals. In this case, the group of animals of only a specific breed may be a group of only one breed, for example, a toy poodle, or a group including multiple breeds, for example, a toy poodle, a pomeranian, and a miniature dachshund. In addition, in the case of multiple breeds, the breeds may be classified into a trained model for large breeds trained using data on the intestinal flora of large breeds, a trained model for medium breeds trained using data on the intestinal flora of medium breeds, a trained model for small breeds trained using data on the intestinal flora of small breeds, and a trained model for extra-small breeds trained using data on the intestinal flora of extra-small breeds, and the trained model may be configured to have trained models corresponding to each classification. For example, in the case of dogs, breeds with an average adult weight of less than 4 kg are called extra small dogs, those with an average adult weight of 4-10 kg are called small dogs, those with an average adult weight of 10-25 kg are called medium dogs, and those with an average adult weight of 25 kg or more are called large dogs. Breeds may also be classified by height, length, genetic relationship, etc.

[0058] When the prediction unit of the present invention uses a plurality of trained models for each type or breed of animal, the disease risk prediction system of the present invention is preferably configured such that, for example, the acquisition unit acquires information on the type or breed of the animal as basic information on the animal. Alternatively, the disease risk prediction system of the present invention may be configured to separately include a breed determination unit. The breed determination unit is a means for determining the breed of an animal from, for example, an image of the animal input to the acquisition unit, using a trained model for breed determination. The trained model for breed determination is, for example, a trained model that has learned the relationship between the image of the animal and the breed of the animal. Preferably, the trained model includes a trained model that uses an image of the animal and the breed of the animal as training data for learning, and has the image of the animal as input and the determination of the breed of the animal as output.

[0059] In addition, the prediction unit of the present invention may use multiple trained models for each age or age group of the animal. The age group is a classification for each age range, and specific examples include classifications such as young (under 1 year old), adult (1 year to under 7 years old), and old (7 years or older). The method of classification is not particularly limited, and classifications such as 0 to under 3 years old, 3 to under 7 years old, and 7 years old or older may be used. When the prediction unit of the present invention uses multiple trained models for each age or age group of the animal, the trained model used is a trained model corresponding to a specific age or age group. Such a trained model is a trained model that has learned the relationship between data on the intestinal flora of an individual or group of animals belonging to a specific age or age group and the classification result of the animal. When the prediction unit of the present invention uses multiple trained models for each age or age group of the animal, the disease risk prediction system of the present invention is preferably configured such that, for example, the prediction unit acquires information on the age of the animal as basic information of the animal. Alternatively, the disease risk prediction system of the present invention may be configured to separately include an age determination unit. The age determination unit is a means for determining the age of an animal from, for example, an image of the animal input to the acquisition unit, using a trained model for age determination. The trained model for age determination is, for example, a trained model that has learned the relationship between the image of the animal and the age of the animal. Preferably, the trained model includes a trained model that uses the image of the animal and the age of the animal as teacher data for learning, and has the image of the animal as input and the determination of the age of the animal as output.

[0060] [Disease risk prediction using class classification information] The disease risk prediction system of the present invention predicts the disease risk of a target animal using class classification information indicating which class the target animal is classified into. The prediction method is not particularly limited, and for example, a correspondence table of each class and the numerical value or level of disease susceptibility may be created in advance, and the presence or absence, numerical value (for example, a numerical value divided into stages according to the magnitude of risk), probability (%), and level of the corresponding disease risk may be determined or predicted according to which class the target animal is classified into. In addition, the class into which the target animal is classified may be used as an additive or subtractive factor for the disease risk, and the disease risk may be predicted after taking into account other basic information such as breed, age, sex, and medical history. In addition, the presence or absence, probability, and level of the corresponding disease risk may be calculated based on the correspondence table, and the presence or absence, probability, or level of the disease risk may be added or subtracted according to the basic information to calculate the final presence or absence, probability, or level of the disease risk. In addition, the disease risk calculated using the class classification information may be corrected based on information about the food the target animal normally takes in and information about its lifestyle habits to calculate the final disease risk. Also, a configuration may be adopted in which a damage ratio or its average is calculated for each class, and the disease risk is calculated according to the damage ratio.

[0061] In another embodiment of the prediction unit of the present invention, the prediction unit may determine whether the intestinal flora data of the target animal has characteristics of the intestinal flora data of an animal individual belonging to a predetermined cluster or is similar to the intestinal flora data of an animal individual belonging to a predetermined cluster, and calculate the damage risk or disease risk based on the determination. In this case, the predetermined cluster is preferably a cluster obtained by performing cluster analysis based on the β diversity of the intestinal flora of a plurality of animal individuals and classifying the plurality of animal individuals into a plurality of clusters. For example, data on the intestinal flora of 10,000 animals is prepared, and cluster analysis is performed on the data based on the β diversity to classify the data into a predetermined number of clusters. As a result, it is assumed that a certain cluster (referred to as cluster X) contains many animal individuals having bacteria a, bacteria b, and bacteria c as intestinal bacteria, and when the average damage rate of the animal individuals included in cluster X is calculated, it is found that the average damage rate is N. On the other hand, if the data on the intestinal flora of a target animal is examined and it is found to have the same characteristics as those of cluster X, i.e., bacteria a, b, and c, then the target animal is expected to have the same average damage rate N and disease risk as cluster X. Then, the disease risk can be calculated based on the expected value of the damage rate.

[0062] [output] The prediction unit calculates a disease risk appropriate for the target animal based on the data on the intestinal flora of the target animal. The format of the output of the prediction result is not particularly limited. The disease risk prediction system of the present invention may also have a separate output unit that receives the prediction result from the prediction unit and outputs the prediction result, and a transmission unit that transmits the judgment result to a personal computer or mobile terminal used by the user via a communication line.

[0063] A disease risk prediction system according to another aspect of the present invention includes an acquisition unit that receives intestinal flora data of a target animal, and a prediction unit that calculates a disease risk of the target animal using the intestinal flora data of the target animal, wherein the prediction unit classifies the target animal into a predetermined class based on the intestinal flora data of the target animal using a trained model that has learned the relationship between the intestinal flora data of the animal and a classification result using the β diversity of the intestinal flora, or a program that assigns the animal to one of the classes set based on the β diversity of the intestinal flora from the intestinal flora data of the animal, and predicts the disease risk using classification information of the class, and The classes set based on the diversity of intestinal microbiota are obtained by preparing data on the intestinal microbiota of multiple animal individuals for use in creating a classification, calculating the Jaccard coefficient or Hamming distance for each of the multiple animal individuals from the intestinal microbiota of the other animal individuals, performing a dimensionality reduction process on the data including the Jaccard coefficient or Hamming distance for the multiple individuals obtained, plotting the data obtained by the dimensionality reduction process for each individual on a two-dimensional graph, and performing cluster analysis to perform classification, and are characterized in that a predetermined disease risk value is assigned to each class in the classification result using the β diversity of the intestinal microbiota or the class set based on the β diversity of the intestinal microbiota. In the disease risk prediction system of this aspect, it is preferable that a predetermined disease risk value based on the average of the Shannon coefficients of the intestinal microbiota data of individual animals belonging to each class is assigned to each class in the classification result using the β diversity of the intestinal microbiota, or to each class set based on the β diversity of the intestinal microbiota. The disease risk prediction system of this aspect enables prediction of disease risk without linking the intestinal microbiota data to data such as insurance payment data or loss ratio.

[0064] [Overview of the disease risk prediction system] Fig. 7 is a diagram for explaining an overview of a disease risk prediction system according to one embodiment of the present invention. As shown in Fig. 7, the disease risk system according to this embodiment includes a server 1 and a user terminal 2. The server 1 and the terminal 2 are connected via a network. The server 1 includes a processing operation unit (CPU) 10, a storage unit 20, and an interface unit 30. Note that users in the present invention include owners of target animals, as well as agents, breeders, insurance companies that provide insurance for animals such as pet insurance, and the like.

[0065] Further, the processing operation unit (CPU) 10 includes a prediction unit 12, the storage unit 20 includes at least a setting unit 22, and the interface unit 30 includes an acquisition unit 31 and an output unit 32. The setting unit 22 may be configured to store a trained model that has learned the relationship between the data of the intestinal microbiota and the classification result using β diversity, or software including the trained model. In such a configuration, the prediction unit 12 can call up the trained model stored in the setting unit 22 or software or a program including the trained model based on the basic information of the animal, and predict the disease risk based on the intestinal microbiota data of the animal. In this embodiment, the trained model is stored in the setting unit 22, but the trained model may be stored in the storage unit instead of the setting unit 22.

[0066] Here, the prediction unit 12 determines into which class the target animal is classified based on a trained model that has learned the relationship between data on the animal's intestinal flora and the classification results using β diversity, and predicts the disease risk using the classification information of that class. Furthermore, the acquisition unit 31 may acquire intestinal microbiota diversity data, and preferably also basic information about the pet, and the output unit 32 presents information such as disease risk calculated by the prediction unit 12 to the user.

[0067] [Disease risk prediction method] FIG. 8 is a diagram for explaining an outline of a disease risk prediction method according to one embodiment of the present invention. As shown in Figure 8, the disease risk prediction method of this embodiment sequentially includes step S11 of acquiring data on the intestinal flora of a target animal, step S21 of classifying the target animal into a predetermined class based on the data on the intestinal flora of the target animal, and step S31 of calculating the disease risk using the classification information of the class.

[0068] Here, it is preferable to further include a step of acquiring basic information of the target animal, and it is preferable to correct the disease risk based on the basic information of the target animal. Alternatively, it is also preferable to calculate a provisional disease risk once based on the basic information, and then use the disease risk calculated by the trained model to correct the provisional disease risk and calculate the final disease risk. By using the basic information of the target animal, a more accurate disease risk can be calculated. In this way, when correcting the provisional disease risk based on the basic information by taking into account the disease risk derived using the class classification information, it is preferable to correct the value of the provisional disease risk between 0.5 and 2.0 times.

[0069] Furthermore, the method may include a step of predicting the future disease risk of the target animal based on information on the food that the target animal normally consumes. Since the alpha diversity of the intestinal flora increases depending on the food that the target animal eats, by taking into account the information on the food, it is expected that a more precise calculation of the disease risk can be made. EXAMPLES

[0070] Examples of the present invention will be described below, but the present invention is not limited to the following examples. [Example 1]

[0071] (Selection of dog) Gut microbiota data were obtained from fecal samples of 47,578 aged 0-year-old dogs as described below.

[0072] (DNA extraction from fecal samples) Fecal samples were collected from each dog and DNA was extracted as follows. Dog owners collected fecal samples from their dogs using a fecal collection kit, which was then suspended in fixative (10% EtOH, 1.07% NH4Cl, 5 mM EDTA, 0.09% NaN3). Next, 200 uL of the fecal suspension and 810 uL of lysis buffer (containing 224 ug / mL Proteinase K) were added to the bead tube, and the beads were crushed using a bead homogenizer (6,000 rpm, crushing for 20 seconds, interval for 30 seconds, crushing for 20 seconds). The sample was then placed on a 70°C heat block for 10 minutes to treat with Proteinase K, and then placed on a 95°C heat block for 5 minutes to inactivate Proteinase K. DNA was automatically extracted from the lysed sample using a chemagic 360 (PerkinElmer) using the chemagic kit stool protocol, and 100 uL of DNA extract was obtained.

[0073] (Meta 16S RNA gene sequence analysis) Meta 16S sequencing analysis was performed using a modified version of Illumina 16S Metagenomic Sequencing Library Preparation (version 15044223 B). First, a 460 bp region including the variable region V3-V4 of the 16S rRNA gene was amplified by PCR using universal primers (Illumina_16S_341F and Illumina_16S_805RPCR). The PCR reaction mixture was prepared by mixing 10 uL of DNA extract, 0.05 uL of each primer (100 uM), 12.5 uL of 2x KAPA HiFi Hot-Start ReadyMix (F. Hoffmann-La Roche, Switzerland), and 2.4 uL of PCR grade water. PCR was performed by thermal denaturation at 95°C for 3 minutes, followed by 30 cycles of 95°C for 30 seconds, 55°C for 30 seconds, and 72°C for 30 seconds, followed by a final extension reaction at 72°C for 5 minutes. The amplified products were purified using magnetic beads and eluted with 50 uL of Buffer EB (QIAGEN, Germany). The purified amplified products were PCR-indexed using the Nextera XT Index Kit v2 (Illumina, CA, US). The PCR reaction solution was prepared by mixing 2.5 uL of the amplified products, 2.5 uL of each primer, 12.5 uL of 2x KAPA HiFi Hot-Start ReadyMix, and 5 uL of PCR grade water. PCR was performed by thermal denaturation at 95°C for 3 minutes, followed by 12 cycles of 95°C for 30 seconds, 55°C for 30 seconds, and 72°C for 30 seconds, followed by a final extension reaction at 72°C for 5 minutes. The indexed amplified products were purified using magnetic beads and eluted with 80-105 uL of Buffer EB. The concentration of each amplified product was measured using a NanoPhotometer (Implen, CA, US) and adjusted to 1.4 nM. Equal amounts of each amplified product were mixed together to prepare the sequencing library. The DNA concentration and size of the amplified products in the sequencing library were confirmed by electrophoresis and analyzed using MiSeq. For analysis, 2 × 300 bp paired-end sequencing was performed using MiSeq Reagent Kit V3.The obtained sequences were analyzed using analysis software called QIIME2 to obtain bacterial composition data. The sequence of the universal primer used above is as follows: This universal primer can be purchased commercially. Illumina_16S_341F 5′-TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGCCTACGGGNGGCWGCAG- 3' llumina_16S_805R 5′-GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGGACTACHVGGGTATCTAATCC- 3'

[0074] Next, we calculated the Jaccard coefficient for each of the 47,578 individuals with the bacterial composition data of the gut microbiota of the other individuals, thereby obtaining the Jaccard coefficient for each of the 47,578 individuals with the other 47,577 individuals. After that, the data including the Jaccard coefficient for each of the 47,578 individuals was subjected to dimensionality reduction processing using UMAP to obtain two-dimensional data, which were then plotted on a two-dimensional graph. The results are shown in Figure 3. Then, 47,578 individuals were divided into 50 classes using spectral clustering. The classification results are shown in Figure 4. In Figure 4, the value written in the box showing each class indicates the average damage rate of individuals in that class.

[0075] As a comparative example, the Shannon index (a type of alpha diversity) was calculated for each of the 47,578 individuals, and they were sorted in descending order of Shannon index and divided into 50 classes from top to bottom (the classes were divided so that the number of individuals in each class was approximately equal). For convenience, the 50 classes were assigned labels from 0 to 49.

[0076] The average loss ratio of individuals included in each class of the examples and comparative examples was calculated. The results of the class with the highest average loss ratio (class 0), the class with the lowest loss ratio (class 49), and the class with the average loss ratio between class 0 and class 49 (class 24) for reference are shown in FIG. 5. As a result, it was found that the difference in loss ratio between class 0 and class 49 was more significant in the examples. This shows that by performing the classification of the present invention, it is possible to more clearly classify individuals into high-risk and low-risk individuals from the data of the intestinal microflora. The loss ratio is the ratio of insurance money paid to the income premium in non-life insurance, and is calculated by dividing the insurance money (Loss) by the premium (Premium).

[0077] [Example 2]

[0078] (Selection of dog) As in Example 1 above, gut microbiota data was obtained from fecal samples for 46,508 0-year-old dogs, 34,012 1-year-old dogs, 42,614 2- to 3-year-old dogs, 35,484 4- to 6-year-old dogs, and 24,768 dogs aged 7 years or older.

[0079] Next, for each of the 46,508 0-year-old dogs, we calculated the Jaccard coefficient with the bacterial composition data of the intestinal microbiota of the other dogs, thereby obtaining the Jaccard coefficient for each of the 46,508 dogs with the other 46,507 dogs. Then, the data including the Jaccard coefficient for each of the 46,508 individuals was subjected to dimensionality reduction processing using UMAP to convert the data into two-dimensional data, which was then plotted on a two-dimensional graph. Then, the 46508 individuals were divided into 50 classes by spectral clustering. The classification is shown in Fig. 9A to Fig. 9D. In Fig. 9, the numerical value written in the box showing each class indicates the insurance claim ratio of individuals included in that class. The insurance claim ratio was calculated using insurance claim data of pet insurance, and was calculated by dividing the number of individuals belonging to each class who made an insurance claim due to the disease in a predetermined period (180 days) by the number of individuals belonging to each class. Fig. 9A shows the insurance claim ratio for skin disease, Fig. 9B shows the insurance claim ratio for diabetes, Fig. 9C shows the insurance claim ratio for periodontal disease, and Fig. 9D shows the insurance claim ratio for valvular heart disease by class. It can be seen from Fig. 9A to Fig. 9D that the insurance claim ratio for each disease is significantly different between classes. This makes it possible to determine and predict whether an individual animal is at high or low risk of developing a certain disease, depending on which class it is assigned to based on its intestinal flora data.

[0080] Similarly, for 34,012 1-year-old dogs, 42,614 2-3-year-old dogs, 35,484 4-6-year-old dogs, and 24,768 7-year-old dogs or older, we calculated the Jaccard coefficient, performed dimensionality reduction processing using UMAP, plotted the data on a graph, and performed spectral clustering for the populations belonging to each age category. Figures 10 to 13 show the classification for each age category.

[0081] 10A to 10D show the insurance claim rate for a specific disease for each class of 34,012 1-year-old dogs classified according to the classification, with FIG. 10A showing the insurance claim rate for skin disease, FIG. 10B showing the insurance claim rate for diabetes, FIG. 10C showing the insurance claim rate for periodontal disease, and FIG. 10D showing the insurance claim rate for valvular heart disease. From FIG. 10A to 10D, it can be seen that the insurance claim rate for each disease is significantly different between the classes. This makes it possible to determine and predict whether an individual animal is at high risk of developing a certain disease, depending on which class the individual animal is classified into based on the data on the intestinal flora.

[0082] 11A to 11D show the insurance claim rate for a specific disease for each class of 42,614 dogs aged 2 to 3 years old, which were classified. FIG. 11A shows the insurance claim rate for skin disease, FIG. 11B shows the insurance claim rate for diabetes, FIG. 11C shows the insurance claim rate for periodontal disease, and FIG. 11D shows the insurance claim rate for valvular heart disease. From FIG. 11, it can be seen that the insurance claim rate for each disease is significantly different between classes. This makes it possible to determine and predict whether an individual animal is at high risk of developing a certain disease depending on which class the individual animal is classified into based on the data on the intestinal flora.

[0083] 12A to 12D show the insurance claim rate for a specific disease for each class of 35,484 dogs aged 4 to 6 years old, which were classified. FIG. 12A shows the insurance claim rate for skin disease, FIG. 12B shows the insurance claim rate for diabetes, FIG. 12C shows the insurance claim rate for periodontal disease, and FIG. 12D shows the insurance claim rate for valvular heart disease. From FIG. 12A to 12D, it can be seen that the insurance claim rate for each disease is significantly different between the classes. This makes it possible to determine and predict whether an animal has a high or low risk of developing a certain disease depending on which class the animal is classified into based on the data on the intestinal microflora.

[0084] 13A to 13D show the insurance claim rate for a specific disease for each class of 24,768 dogs aged 7 years or older, which were classified. FIG. 13A shows the insurance claim rate for skin disease, FIG. 13B shows the insurance claim rate for diabetes, FIG. 13C shows the insurance claim rate for periodontal disease, and FIG. 13D shows the insurance claim rate for valvular heart disease. From FIG. 13A to 13D, it can be seen that the insurance claim rate for each disease is significantly different between the classes. This makes it possible to determine and predict whether an animal has a high risk of developing a certain disease depending on which class the animal is classified into based on the data on the intestinal flora.

[0085] [Example 3]

[0086] (Cat Selection) As in Example 1 above, gut microbiota data was obtained from fecal samples for 16,729 0-year-old cats, 12,575 1-year-old cats, 13,938 2-3 year-old cats, 10,509 4-6 year-old cats, and 7,174 cats aged 7 years or older.

[0087] Next, for each of the 16,729 0-year-old cats, we calculated the Jaccard coefficient with the bacterial composition data of the intestinal microbiota of the other cats, thereby obtaining the Jaccard coefficient for each of the 16,729 cats with the other 16,728 cats. Then, the data including the Jaccard coefficient for each of the 16,729 individuals was subjected to dimensionality reduction processing using UMAP to convert the data into two-dimensional data, which was then plotted on a two-dimensional graph. Then, the 16,729 individuals were divided into 50 classes by spectral clustering. The classification is shown in FIG. 14A. In FIG. 14, the numerical value written in the box showing each class indicates the proportion of individuals in that class who claimed insurance due to kidney disease. The insurance claim proportion was calculated using pet insurance claim data, and was calculated for each disease that was the reason for the insurance claim by dividing the number of individuals belonging to each class who claimed insurance due to that disease within a specified period (180 days) by the number of individuals belonging to each class. In addition, for 12575 1-year-old cats, 13938 2-3-year-old cats, 10509 4-6-year-old cats, and 7174 7-year-old or older cats, we calculated the Jaccard coefficient, performed dimensionality reduction processing using UMAP, plotted on a graph, and performed spectral clustering for the populations belonging to each age category. Figures 14B to 14E show the classification for each age category. Figure 14A shows the insurance claim proportion for kidney disease by class for 16,729 cats aged 0 years, Figure 14B shows the insurance claim proportion for kidney disease by class for 12,575 cats aged 1 year, Figure 14C shows the insurance claim proportion for kidney disease by class for 13,938 cats aged 2 to 3 years, Figure 14D shows the insurance claim proportion for kidney disease by class for 10,509 cats aged 4 to 6 years, and Figure 14E shows the insurance claim proportion for kidney disease by class for 7,174 cats aged 7 years or older.

[0088] [Reference example 1] For the 7,174 cats aged 7 years or older for which intestinal flora data was collected in Example 3 above, (1) the 7,174 cats were divided into three classes by dividing them into thirds in descending order of Shannon index (Figure 15(A)), and (2) the cats were divided into 50 classes by calculating the Jaccard coefficient, performing dimensionality reduction processing using UMAP, plotting on a graph, and performing spectral clustering (Figure 15(B)). A comparison was made of the proportion of insurance claims for kidney disease. Regarding (1), the low Shannon index group included individuals with a Shannon index of 1.64 to 4.78, the intermediate group included individuals with a Shannon index of 4.78 to 5.29, and the high Shannon index group included individuals with a Shannon index of 5.29 to 6.40. The figures shown in each graph are the proportion of insurance claims within each class. Regarding (2), after dividing into 50 classes, each class was consolidated into three classes based on the insurance claim ratio. The high insurance claim ratio class was 4.0-7.2%, the medium class was 2.75-4.0%, and the low class was 1.9-2.75%. The figures shown in each graph are the average insurance claim ratio within each class after the consolidation. Comparing Figures 15(A) and (B), we can see that the difference in the insurance claim rate between the classes is larger when cats are classified based on β diversity than when they are classified into three classes based on the Shannon index. This shows that classification based on β diversity allows for more accurate prediction of disease risk than classification using the Shannon index.

[0089] [Reference example 2] For the 7,174 cats aged 7 years or older for which intestinal flora data was collected in Example 3 above, (1) the 7,174 cats were divided into three classes by dividing them into thirds in descending order of Shannon index (Figure 16(A)), and (2) the cats were divided into 20 classes by calculating the Jaccard coefficient, performing dimensionality reduction processing using UMAP, plotting on a graph, and performing spectral clustering (Figure 16(B)). A comparison was made of the proportion of insurance claims for kidney disease. Regarding (1), the low Shannon index group included individuals with a Shannon index of 1.64 to 4.78, the intermediate group included individuals with a Shannon index of 4.78 to 5.29, and the high Shannon index group included individuals with a Shannon index of 5.29 to 6.40. The figures shown in each graph are the proportion of insurance claims within each class. Regarding (2), after dividing into 20 classes, each class was consolidated into three classes based on the insurance claim ratio. The high insurance claim ratio class was 4.0-7.2%, the medium class was 2.75-4.0%, and the low class was 1.9-2.75%. The figures shown in each graph are the average insurance claim ratio within each class after the consolidation. Comparing Figure 16 (A) and (B), it can be seen that the difference in the insurance claim rate between the classes is larger when cats are classified based on β diversity than when they are classified into three classes based on the Shannon index. This shows that classification based on β diversity allows for more accurate prediction of disease risk than classification using the Shannon index.

Claims

1. an acquisition unit for receiving intestinal microbiota data of an insured animal; a calculation unit that calculates an insurance premium for the insured animal using the intestinal flora data of the insured animal, The calculation unit classifies the insured animal into a predetermined class based on the intestinal flora data of the insured animal using a trained model that has learned the relationship between the intestinal flora data of the animal and the classification result using the β diversity of the intestinal flora, or a program that assigns the animal to one of the classes set based on the β diversity of the intestinal flora from the intestinal flora data of the animal, and calculates the insurance premium using the classification information of the class. Insurance premium calculation system.

2. The insurance premium calculation system of claim 1, wherein the class in the classification result using the β diversity of the intestinal bacterial flora, or the class set based on the β diversity of the intestinal bacterial flora, is obtained by classifying each individual of the animal population for creating the classification based on the β diversity between each individual of the animal population for creating the classification.

3. 3. The insurance premium calculation system according to claim 1, wherein the classes in the classification results or the classes set based on the beta diversity of the intestinal bacterial flora are clustering results obtained by cluster analysis.

4. 3. The insurance premium calculation system according to claim 2, wherein the beta diversity includes a Jaccard coefficient or a Hamming distance.

5. The insurance premium calculation system of claim 4, wherein the class in the classification result using the beta diversity of the intestinal flora, or the class set based on the beta diversity of the intestinal flora, is obtained by preparing data on the intestinal flora of multiple animal individuals for creating a classification, calculating the Jaccard coefficient for each of the multiple animal individuals with the intestinal flora of other animal individuals, and performing classification based on the Jaccard coefficient.

6. The insurance premium calculation system of claim 5, wherein the class in the classification result using the beta diversity of the intestinal flora, or the class set based on the beta diversity of the intestinal flora, is obtained by preparing data on the intestinal flora of multiple animal individuals for creating a classification, calculating the Jaccard coefficient for each of the multiple animal individuals with the intestinal flora of other animal individuals, and performing cluster analysis on the data including the Jaccard coefficients for the multiple individuals obtained.

7. The insurance premium calculation system of claim 6, wherein the classes in the classification result using the beta diversity of the intestinal flora, or the classes set based on the beta diversity of the intestinal flora, are obtained by preparing data on the intestinal flora of multiple animal individuals for creating a classification, calculating the Jaccard coefficient for each of the multiple animal individuals with the intestinal flora of other animal individuals, performing a dimensionality reduction process on the data including the Jaccard coefficients obtained for the multiple individuals, plotting the data obtained by the dimensionality reduction process for each individual on a two-dimensional graph, and performing cluster analysis to perform classification.

8. 7. The insurance premium calculation system according to claim 6, wherein the cluster analysis is spectral clustering.

9. The insurance premium calculation system of claim 2, wherein the trained model is trained using, as training data, data on the intestinal flora of each individual animal in an animal population used to create classifications, and data on which class an individual animal is classified into.

10. receiving gut microbiota data of an insured animal by a computer; and a step of calculating an insurance premium for the insured animal by using the intestinal flora data of the insured animal by a computer, The step of calculating the insurance premium for the insured animal is a step of classifying the insured animal into a predetermined class based on the intestinal flora data of the insured animal using a trained model that has learned the relationship between the intestinal flora data of the animal and the classification result using the β diversity of the intestinal flora, or a program that classifies the animal based on the intestinal flora data of the animal into one of the classes set based on the β diversity of the intestinal flora, and calculating the insurance premium using the classification information of the class. Insurance premium calculation method.

11. an acquisition unit for receiving intestinal microbiota data of an insured animal; a calculation unit that calculates an insurance premium for the insured animal using the intestinal flora data of the insured animal, the calculation unit determines whether the intestinal flora data of the insured animal has characteristics of the intestinal flora data of an animal individual belonging to a specified cluster or is similar to the intestinal flora data of an animal individual belonging to a specified cluster, and calculates a loss ratio or an insurance premium based on the determination; The predetermined cluster is a cluster obtained by performing a cluster analysis based on the β diversity of the intestinal microbiota of a plurality of animal individuals and classifying the plurality of animal individuals into a plurality of clusters. Insurance premium calculation system.

12. an acquisition unit for receiving gut microbiota data of a target animal; A prediction unit that calculates a disease risk of the target animal using the intestinal flora data of the target animal, The prediction unit classifies the target animal into a predetermined class based on the intestinal flora data of the target animal using a trained model that has learned the relationship between the intestinal flora data of the animal and a classification result using the β diversity of the intestinal flora, or a program that classifies the animal into one of the classes set based on the β diversity of the intestinal flora from the intestinal flora data of the animal, and predicts the disease risk using the classification information of the class. Disease risk prediction system.

13. The disease risk prediction system of claim 12, wherein the class in the classification result using the β diversity of the intestinal microbiota, or the class set based on the β diversity of the intestinal microbiota, is obtained by classifying each individual of the animal population for creating the classification based on the β diversity between each individual of the animal population for creating the classification.

14. The disease risk prediction system according to claim 12 or 13, wherein the classes in the classification results or the classes set based on the β diversity of the intestinal bacterial flora are clustering results obtained by cluster analysis.

15. The disease risk prediction system according to claim 12, wherein the beta diversity comprises a Jaccard coefficient or a Hamming distance.

16. The disease risk prediction system of claim 15, wherein the class in the classification result using the β diversity of the intestinal microbiota, or the class set based on the β diversity of the intestinal microbiota, is obtained by preparing data on the intestinal microbiota of multiple animal individuals for creating a classification, calculating the Jaccard coefficient for each of the multiple animal individuals with the intestinal microbiota of other animal individuals, and performing classification based on the Jaccard coefficient.

17. The disease risk prediction system of claim 16, wherein the class in the classification result using the beta diversity of the intestinal microbiota, or the class set based on the beta diversity of the intestinal microbiota, is obtained by preparing data on the intestinal microbiota of multiple animal individuals for creating a classification, calculating the Jaccard coefficient for each of the multiple animal individuals with the intestinal microbiota of other animal individuals, and performing cluster analysis on the data including the Jaccard coefficients for the multiple individuals obtained.

18. The disease risk prediction system of claim 17, wherein the class in the classification result using the β diversity of the intestinal microbiota, or the class set based on the β diversity of the intestinal microbiota, is obtained by preparing data on the intestinal microbiota of multiple animal individuals for creating a classification, calculating the Jaccard coefficient for each of the multiple animal individuals with the intestinal microbiota of other animal individuals, performing a dimensionality reduction process on the data including the Jaccard coefficients obtained for the multiple individuals, plotting the data obtained by the dimensionality reduction process for each individual on a two-dimensional graph, and performing cluster analysis to perform classification.

19. The disease risk prediction system according to claim 18, wherein the cluster analysis is spectral clustering.

20. The disease risk prediction system of claim 13, wherein the trained model is trained using, as teacher data, data on the intestinal flora of each individual animal in an animal population used for creating classifications, and data on which class an individual animal is classified into.

21. an acquisition unit for receiving gut microbiota data of a target animal; A prediction unit that calculates a disease risk of the target animal using the intestinal flora data of the target animal, the prediction unit classifies the target animal into a predetermined class based on the intestinal flora data of the target animal using a trained model that has trained a relationship between the intestinal flora data of the animal and a classification result using the β diversity of the intestinal flora, or a program that classifies the animal into one of classes set based on the β diversity of the intestinal flora from the intestinal flora data of the animal, and predicts a disease risk using classification information of the class; the classes in the classification result using the β diversity of the intestinal microbiota, or the classes set based on the β diversity of the intestinal microbiota, are obtained by preparing data on the intestinal microbiota of a plurality of animal individuals for creating a classification, calculating the Jaccard coefficient or Hamming distance between each of the plurality of animal individuals and the intestinal microbiota of the other animal individuals, performing a dimensionality reduction process on the data including the Jaccard coefficient or Hamming distance obtained for the plurality of individuals, plotting the data obtained by the dimensionality reduction process for each individual on a two-dimensional graph, and performing a cluster analysis to perform classification; A predetermined disease risk value is assigned to each class in the classification result using the β diversity of the intestinal microbiota or the class set based on the β diversity of the intestinal microbiota. Disease risk prediction system.

22. A predetermined disease risk value based on the average of the Shannon coefficients of the intestinal microbiota data of individual animals belonging to each class is assigned to each class in the classification result using the β diversity of the intestinal microbiota or the class set based on the β diversity of the intestinal microbiota. The disease risk prediction system according to claim 21.

Images