JP7819901B2 - Biomarkers of hikikomori and support methods for identifying hikikomori - Google Patents

Description

The present invention relates to a biomarker for hikikomori and a method for supporting the identification of hikikomori using the biomarker.

"Hikikomori" refers to a state in which, as a result of various factors, a person avoids social participation and remains at home for six months or more. The number of hikikomori in Japan exceeds one million, and is expected to increase further in the future. Furthermore, there is a trend toward an increase in the number of hikikomori worldwide, and global countermeasures are desired. It has been suggested that hikikomori is related to various mental and physical illnesses, but the details are largely unknown.

In recent years, the discovery of blood biomarkers related to mental illness has advanced our understanding of the pathophysiology of mental illness; for example, biomarkers for depression have been reported (Patent Citations 1 and 2). Given this situation, there has been a desire to discover a biomarker for hikikomori.

Although a report on biomarkers for hikikomori has been published in Non-Patent Document 1, this study only examined biochemical analysis items from clinical tests, and did not broadly examine blood metabolites.

International Publication No. 2011/019072 International Publication No. 2017/082103

Sci. Rep. 2018, 8:2884.

The objective of this invention is to provide a biomarker for hikikomori and a method for supporting the identification of hikikomori using the biomarker.

In order to solve the above problem, the inventors conducted extensive research and discovered that there were significant differences in multiple components in blood samples from hikikomori individuals compared to non-hikikomori individuals.

The present invention has been completed based on these findings and includes the following broad aspects.
[Section 1]
A biomarker for identifying social withdrawal, comprising at least three members selected from the group consisting of ornithine, acylcarnitine, bilirubin, arginine, arginase, uric acid, and cholesterol ester.
[Section 2]
A biomarker for identifying social withdrawal, comprising at least four species selected from the group consisting of ornithine, acylcarnitine, bilirubin, arginine, arginase, uric acid, and cholesterol ester.
[Section 3]
Item 3. The biomarker for identifying social withdrawal according to Item 1 or 2, wherein the acylcarnitine is a long-chain acylcarnitine.
[Section 4]
A method for using blood components containing at least two selected from the group consisting of ornithine, acylcarnitine, bilirubin, arginine, and arginase as an indicator for identifying whether or not a subject is in a state of social withdrawal.
[Section 5]
A method for determining the severity of a subject's social withdrawal state by using blood components containing at least two selected from the group consisting of ornithine, acylcarnitine, bilirubin, arginine, and arginase as an indicator.
[Section 6]
A method for using blood components containing at least one selected from the group consisting of uric acid and cholesterol esters as an indicator for stratifying subjects into non-hikikomori groups, hikikomori groups with depression tendencies, and other hikikomori groups.
[Section 7]
Obtaining the content of the hikikomori identification biomarker according to any one of Items 1 to 3 in blood samples of a plurality of hikikomori individuals and non-hikikomori individuals;
generating training data by associating the content with a discrimination result of whether or not the subject is a hikikomori;
a step of performing machine learning using the training data to generate a hikikomori identification support model that outputs whether or not the person is a hikikomori when the amount of content is input;
A method for generating a hikikomori identification support model having the above.
[Section 8]
Obtaining the content of the hikikomori identification biomarker according to any one of Items 1 to 3 in blood samples of a plurality of hikikomori individuals and non-hikikomori individuals;
A step of generating training data by correlating the content with the assessment result of the severity of social withdrawal;
A step of performing machine learning using the training data to generate a hikikomori severity prediction support model that outputs the severity of hikikomori when the content is input;
A method for generating a model for supporting prediction of the severity of social withdrawal.
[Section 9]
Item 9. The method according to Item 7 or 8, wherein the biomarker for identifying social withdrawal comprises at least two selected from the group consisting of ornithine, acylcarnitine, bilirubin, arginine, and arginase.
[Section 10]
Obtaining the content of the hikikomori identification biomarker according to any one of Items 1 to 3 in blood samples of a plurality of hikikomori individuals and non-hikikomori individuals;
a step of generating training data by associating the content with a determination result of a clustering classification of hikikomori;
a step of performing machine learning using the training data to generate a hikikomori stratification support model that outputs a clustering classification of hikikomori when the content is input;
A method for generating a model for supporting the stratification of hikikomori.
[Section 11]
Item 11. The method according to Item 10, wherein the hikikomori identifying biomarker comprises at least one selected from the group consisting of uric acid and cholesterol esters.
[Section 12]
Item 10. The generation method according to Item 7 or 9, wherein the machine learning is machine learning using a random forest.
[Section 13]
Item 10. The generation method according to Item 8 or 9, wherein the machine learning is machine learning using partial least squares regression.
[Section 14]
Obtaining the content of the hikikomori identifying biomarker according to any one of Items 1 to 3 in a blood sample from a subject;
A step of predicting whether the subject is a hikikomori by inputting the content of the subject into a hikikomori identification support model generated by the generation method described in any one of items 7, 9, and 12;
A hikikomori identification support method comprising:
[Section 15]
Obtaining the content of the hikikomori identifying biomarker according to any one of Items 1 to 3 in a blood sample from a subject;
A step of predicting the severity of social withdrawal of the subject by inputting the content of the subject into a social withdrawal severity prediction support model generated by the generation method according to any one of items 8, 9 and 13;
A method for predicting the severity of social withdrawal.
[Section 16]
Obtaining the content of the hikikomori identifying biomarker according to any one of Items 1 to 3 in a blood sample from a subject;
Item 12. A step of predicting a clustering classification of hikikomori of the subject by inputting the content of the subject into a hikikomori stratification support model generated by the generation method according to Item 10 or 11;
A method for supporting the stratification of hikikomori.

The present invention makes it possible to identify, assess the severity of, and stratify hikikomori, and provides a clinically useful biomarker. Furthermore, the present invention makes it possible to provide the identification, assessment of the severity, and stratify hikikomori using a hikikomori biomarker.

1 is a schematic diagram of the present invention. 1 is a block diagram showing a schematic configuration of a hikikomori identification support model generation system 1. [0023]FIG. 10 is a flowchart showing the processing steps of a method for generating a hikikomori identification support model. 1 is a block diagram showing a schematic configuration of a support device 5 that is a device for implementing a hikikomori identification support method. 10 is a flowchart showing a processing procedure of a hikikomori identification support method. 1 is a block diagram showing a schematic configuration of a hikikomori severity prediction support model generation system 1'. 10 is a flowchart showing the processing steps of a method for generating a hikikomori severity prediction support model. FIG. 1 is a block diagram showing a schematic configuration of a support device 5' that is a device for implementing a hikikomori severity prediction support method. 10 is a flowchart showing the processing steps of a hikikomori severity prediction support method. FIG. 1 is a block diagram showing a schematic configuration of a hikikomori stratification support model generation system 1″. 10 is a flowchart showing the processing steps of a method for generating a hikikomori stratification support model. FIG. 1 is a block diagram showing a schematic configuration of a support device 5″ that is a device for implementing the hikikomori stratification support method of this embodiment. 10 is a flowchart showing the processing steps of a hikikomori stratification support method. The graph shows the differences in blood components between hikikomori and non-hikikomori (HC). A: Scatter plot of blood components. The horizontal dotted line in the graph indicates p=0.03. The vertical dotted lines in the graph indicate -0.25 and 0.25, respectively. AC indicates acylcarnitine, and the number after the underscore indicates the number of carbon atoms in the fatty acid and the double bond 0.001. ****: p<0.0001. Machine learning model for identifying the presence or absence of hikikomori in Example 1. A: Method for creating a random forest model based on the results of 127 blood components. B: Performance evaluation results of the random forest model created in Example 1. C: ROC curve of the performance evaluation results of the random forest model created in Example 1. The AUC was 0.854 (95% CI = 0.648-1.00, p <0.01). D: Components with large contributions to the random forest model created in Example 1 are shown. Graphs related to blood arginase in Example 2. A: Shows the enzyme reaction of arginase. B: Shows the analytical results of blood arginase measurement. C: A graph comparing blood arginase levels in men between hikikomori and non-hikikomori (HC). *: p < 0.05. D: Shows the correlation between blood arginine and blood ornithine in men. The dotted line indicates the 95% confidence interval in linear regression. Machine learning model for predicting the severity of social withdrawal in Example 3. A: Partial least squares model for predicting severity based on HQ-25. Actual HQ-25 results are shown on the horizontal axis, and prediction results from the machine learning model are shown on the vertical axis. B: Components with a large contribution to the partial least squares model created in Example 3 are shown. C: Results of an analysis of the correlation between hexadecanoylcarnitine (AC_C16:0) and HQ-25 using Spearman's rank correlation coefficient. The dotted line indicates the 95% confidence interval for linear regression. D: Results of an analysis of the correlation between arginine and HQ-25 using Spearman's rank correlation coefficient. The dotted line indicates the 95% confidence interval for linear regression. This relates to the stratification of hikikomori in Example 4. A: Results of principal component analysis of hikikomori (Hikikomori) and non-hikikomori (HC) based on psychiatric test data. B: Results for each cluster of the psychiatric test data used in the principal component analysis of A. The results of the cluster analysis regarding the stratification of hikikomori in Example 4 show 25 components that are particularly correlated. The distribution of components that showed a strong correlation with the results of the cluster analysis regarding the stratification of hikikomori in Example 4 is shown for each cluster. UA: uric acid, CE: cholesterol ester.

In this invention, the term "hikikomori" means a state in which a person does not leave their home for more than four days a week and does not participate in social activities for more than six consecutive months, resulting in functional impairment and distress.

In addition, in the present invention, the "content" of a biomarker in a blood sample preferably refers to the absolute concentration of the biomarker in the blood sample, but is not limited to this as long as it is a value that correlates with the absolute concentration of the biomarker and allows for comparison of the absolute concentration of each individual. It may also be a relative concentration, weight per unit volume, or a plasma level obtained from raw data measured to determine the absolute concentration (for example, a value obtained by standardizing the peak area of a graph obtained by LC/MS measurement).

1. Biomarker for Identifying Hikikomori The present invention provides a biomarker for identifying hikikomori. The biomarker for identifying hikikomori of the present invention comprises at least three species selected from the group consisting of ornithine, acylcarnitine, bilirubin, arginine, arginase, uric acid, and cholesterol ester, and preferably comprises at least four species selected from the group consisting of ornithine, acylcarnitine, bilirubin, arginine, arginase, uric acid, and cholesterol ester. Furthermore, the biomarker of the present invention may further comprise substances other than ornithine, acylcarnitine, bilirubin, arginine, arginase, uric acid, and cholesterol ester.

The biomarkers of the present invention make it possible to easily identify a subject's state of social withdrawal.

The components contained in the biomarkers of the present invention are described below.
<Acylcarnitine>
Acylcarnitines have a structure in which an acyl group derived from a fatty acid is bonded to the hydroxyl group of carnitine. They have a variety of structures depending on the carbon chain length of the fatty acid portion, the presence and number of unsaturated bonds, and the substitution of hydrogen atoms bonded to the carbon chain with oxygen atoms or hydroxy groups, but in this specification they are collectively referred to as acylcarnitines.

The acylcarnitine of the present invention is preferably a long-chain acylcarnitine. In this specification, "long-chain acylcarnitine" refers to an acylcarnitine having 12 to 20 carbon atoms in the fatty acid. Examples of long-chain acylcarnitines include dodecanoylcarnitine (AC_C12:0), dodecenoylcarnitine (AC_C12:1), tetradecanoylcarnitine (AC_C14:0), tetradecenoylcarnitine (AC_C14:1), tetradecadienoylcarnitine (AC_C14:2), 3-hydroxytetradecanoylcarnitine (AC_C14OH), hexadecanoylcarnitine (AC_C16:0), hexadecenoylcarnitine (AC_C16:1), and 3-hydroxyhexadecanoylcarnitine. Examples of carnitine include octadecanoylcarnitine (AC_C16OH), 3-hydroxyhexadecenoylcarnitine (AC_C16:1OH), octadecanoylcarnitine (AC_C18:0), octadecenoylcarnitine (AC_C18:1), octadecadienoylcarnitine (AC_C18:2), 3-hydroxyoctadecanoylcarnitine (AC_C18OH), 3-hydroxyoctadecenoylcarnitine (AC_C18:1OH), and 3-hydroxyoctadecadienoylcarnitine (AC_C18:2OH). Among these, tetradecanoylcarnitine (AC_C14:0), tetradecenoylcarnitine (AC_C14:1), tetradecadienoylcarnitine (AC_C14:2), hexadecanoylcarnitine (AC_C16:0), hexadecenoylcarnitine (AC_C16:1), octadecenoylcarnitine (AC_C18:1), octadecadienoylcarnitine (AC_C18:2), 3-hydroxyoctadecenoylcarnitine (AC_C18:1OH), and 3-hydroxyoctadecadienoylcarnitine (AC_C18:2OH) are preferred. The acylcarnitines used as biomarkers of the present invention may be used alone or in combination of two or more (e.g., two or more, ten or more, twenty or more, or thirty or more). Furthermore, there is no particular upper limit to the number of acylcarnitines that can be used as biomarkers in the present invention, but examples include 50 or fewer types, or 40 or fewer types.

In this specification, acylcarnitines may be referred to by abbreviations. For example, AC_C2:0 indicates an acylcarnitine with two fatty acid carbon atoms and zero double bonds. For example, AC_C3:DC indicates an acylcarnitine with three fatty acid carbon atoms, zero double bonds, and a carboxyl group in the fatty acid moiety. For example, AC_C4:OH indicates an acylcarnitine with two fatty acid carbon atoms, zero double bonds, and a hydroxyl group in the fatty acid moiety. For example, AC_16:1OH indicates an acylcarnitine with 16 fatty acid carbon atoms, one double bond, and a hydroxyl group in the fatty acid moiety.

Since the content of acylcarnitines in the blood samples of subjects with hikikomori is higher than that of non-hikikomori, it is preferably used as a biomarker for identifying hikikomori in the present invention.

<Bilirubin>
In biological samples such as blood, there are three types of bilirubin: conjugated bilirubin, unconjugated bilirubin, and delta bilirubin. The sum of these is called total bilirubin. Conjugated bilirubin is bilirubin that has been conjugated with glucuronid and is also called direct bilirubin. Unconjugated bilirubin is bilirubin that has not been conjugated with glucuronid and is also called indirect bilirubin. The bilirubin used as the biomarker of the present invention is preferably total bilirubin (T-Bil), direct bilirubin (D-Bil), or the direct bilirubin/indirect bilirubin ratio (R-Bil). Regarding the bilirubin used as the biomarker of the present invention, total bilirubin (T-Bil), direct bilirubin (D-Bil), or the ratio of direct bilirubin to indirect bilirubin (R-Bil) may be used alone or in combination of two or more.

The total bilirubin content, direct bilirubin content, and direct bilirubin/indirect bilirubin ratio in the subject's blood sample are all lower in hikikomori compared to non-hikikomori, making them suitable for use as biomarkers for identifying hikikomori in the present invention. In particular, lower results are obtained in male hikikomori compared to non-hikikomori, making them useful as indicators for characterizing male hikikomori.

<Arginine, ornithine, arginase>
Arginase is an enzyme that controls the reaction that produces ornithine from arginine. The arginine content in the blood sample of a subject is lower in hikikomori compared to non-hikikomori. Furthermore, the ornithine content in the blood sample of a subject is higher in hikikomori compared to non-hikikomori. Furthermore, the arginase content in the blood sample of a subject is higher in hikikomori compared to non-hikikomori. Therefore, arginine, ornithine, and arginase are each preferably used as a biomarker for identifying hikikomori in the present invention. Note that arginase is particularly useful as an indicator for characterizing male hikikomori.

<Uric acid>
Uric acid is the final metabolic product of nucleic acids and purines, which are components of ATP. Uric acid levels in a subject's blood sample can be used as an indicator for identifying social withdrawal. In particular, it can be used as an indicator for clustering subjects based on the degree of depression.

<Cholesterol esters>
Cholesterol esters are fatty acid esters of cholesterol and have a variety of structures depending on the carbon chain length of the fatty acid moiety, the presence or absence and number of unsaturated bonds, etc., but in this specification they are collectively referred to as cholesterol esters.

The cholesterol ester used in the present invention is preferably a cholesterol ester having a long-chain fatty acid, more preferably a cholesterol ester having a fatty acid with 12 to 20 carbon atoms, and particularly preferably a cholesterol ester having a fatty acid with 16 to 18 carbon atoms.

Examples of cholesterol esters having long-chain fatty acids include heptadecanoic acid cholesterol ester (CE_17:0), octadecanoic acid cholesterol ester (CE_18:0), octadecenoic acid cholesterol ester (CE_18:1), octadecadienoic acid cholesterol ester (CE_18:2), octadecatrienoic acid cholesterol ester (CE_18:3), eicosanoic acid cholesterol ester (CE_20:0), eicosenoic acid cholesterol ester (CE_20:1), and eicosadienoic acid cholesterol ester (CE_20:2). Cholesterol esters used as biomarkers of the present invention may be used alone or in combination of two or more (e.g., two or more, three or more, or five or more). There is no particular upper limit to the number of cholesterol esters used as biomarkers of the present invention, but examples include 20 or fewer, or 10 or fewer.

Note that cholesterol esters are sometimes referred to by abbreviations throughout this specification. For example, CE_17:0 indicates a cholesterol ester with 17 fatty acid carbon atoms and 0 double bonds.

The content of cholesterol esters in a subject's blood sample can be used as an indicator for identifying hikikomori. In particular, it can be preferably used as an indicator when clustering subjects by the degree of depressive tendency. Octadecatrienoic acid cholesterol ester (CE_18:3), eicosenoic acid cholesterol ester (CE_20:1), and eicosadienoic acid cholesterol ester (CE_20:2) are more preferably used as indicators when clustering subjects by the degree of depressive tendency.

<Others>
Examples of components included in the biomarkers of the present invention other than ornithine, acylcarnitine, bilirubin, arginine, arginase, uric acid, and cholesterol esters include, but are not limited to, the following: The following components included in the biomarkers of the present invention other than ornithine, acylcarnitine, bilirubin, arginine, arginase, uric acid, and cholesterol esters may be used alone or in combination of two or more.
・Biochemical test items
HDL cholesterol (HDL.C), LDL cholesterol (LDL.C), total cholesterol (Total.C), fibrinogen (FIB), fibrin-fibrinogen degradation products (FDP), high-sensitivity C-reactive protein (hsCRP), etc.
・Amino acids, etc.: Alanine, asparagine, aspartic acid, citrulline, creatine, cystine, glutamine, glutamic acid, hydroxyproline, isoleucine, leucine, lysine, methionine, phenylalanine, proline, serine, taurine, threonine, tryptophan, tyrosine, valine, etc.
Lysophosphatidylcholine: Lysophosphatidylcholine in which the carbon chain length of the fatty acid constituting the lysophosphatidylcholine is 12 or more and 22 or less. Examples include palmitoyl lysophosphatidylcholine (LPC_16:0), oleoyl lysophosphatidylcholine (LPC_18:1), linoleyl lysophosphatidylcholine (LPC_18:2), etc. In this specification, lysophosphatidylcholine may be abbreviated, and for example, LPC_16:0 indicates lysophosphatidylcholine with 16 fatty acid carbon atoms and 0 double bonds.
Phosphatidylcholine Phosphatidylcholine includes those in which the carbon chain length of the fatty acid constituting the phosphatidylcholine is 12 or more and 22 or less. Examples include dipalmitoylphosphatidylcholine (PC_32:0), dioleoylphosphatidylcholine (PC_36:2), dilinoleylphosphatidylcholine (PC_36:4), etc. In this specification, phosphatidylcholine may be abbreviated, and for example, PC_36:2 indicates that the total number of fatty acid carbon atoms in the phosphatidylcholine is 36 and the number of double bonds is 2.
Phosphatidylethanolamines include those in which the carbon chain length of the fatty acid constituting the phosphatidylethanolamine is 12 or more and 22 or less. Examples include dioleoylphosphatidylethanolamine (PE_36:2) and dilinoleylphosphatidylethanolamine (PE_36:4). Note that phosphatidylethanolamines are sometimes abbreviated herein, and for example, PE_36:2 indicates that the total number of fatty acid carbon atoms in the phosphatidylethanolamine is 36 and the number of double bonds is 2.
Triglycerides: Triglycerides in which the carbon chain length of the fatty acids constituting the triglycerides is 12 or more and 22 or less. Examples include palmitoyl dioleoyl glycerol (TG_52:2) and palmitoyl stearoyl linoleoyl glycerol (TG_52:3). In this specification, triglycerides are sometimes referred to by abbreviations, and for example, TG_52:2 indicates that the total number of fatty acid carbon atoms in the triglyceride is 52 and the number of double bonds is 2.

2. Method for Identifying Hikikomori 2.1 Identifying the Presence or Absence of Hikikomori In another embodiment of the present invention, there is provided a method for using a blood component of a subject as an indicator for identifying the presence or absence of a hikikomori state in a subject. The blood component can be a biomarker for identifying hikikomori of the present invention, and examples of the blood component include blood components containing at least two selected from the group consisting of ornithine, acylcarnitine, bilirubin, arginine, and arginase.

In this embodiment, a method for identifying the presence or absence of a hikikomori state is provided, which includes the steps of acquiring the content of a hikikomori identification biomarker of the present invention in a blood sample from a subject, and comparing the acquired content with a preset cutoff value to determine the presence or absence of a hikikomori state.

According to the identification method of this embodiment, there is no limit to the age of the subject; for example, the subject may be a young subject, specifically a subject under the age of 30, or an elderly subject, specifically a subject over the age of 65. There is also no limit to the gender of the subject; the subject may be a male or a female subject. Furthermore, the subject may or may not be receiving treatment, such as medication, for a mental illness such as social withdrawal, although it is preferable that the subject is not receiving treatment.

<Method for measuring the content of biomarkers in blood samples>
The steps for determining the content of the hikikomori identifying biomarker of the present invention in a blood sample from a subject are described below.

The blood sample used in this embodiment includes not only blood collected from a subject, but also blood obtained by processing the blood. Examples of blood obtained by processing the blood include serum and plasma. Serum and plasma can be obtained, for example, by leaving the blood to stand or by centrifuging the blood. In this specification, blood, serum, and plasma are collectively referred to as "blood sample."

Blood samples may be used directly to measure the biomarker content, or may be pretreated as needed before being used to measure the biomarker content. Examples of pretreatment include stopping enzyme reactions in the blood sample, removing lipid-soluble substances, and removing proteins. These pretreatments can be performed using known methods. Blood samples may also be diluted or concentrated as appropriate before use.

As a method for measuring the content of a biomarker, an appropriate known method may be selected depending on the type of biomarker. For example, the content can be measured using liquid chromatography-mass spectrometry (LC/MS), high-performance liquid chromatography (HPLC), gas chromatography-mass spectrometry (GC/MS), capillary electrophoresis-mass spectrometry (CE-MS), or capillary electrophoresis-time-of-flight mass spectrometry (CE-TOFMS). The content can also be measured by a quantitative method appropriate for the marker to be measured, such as nuclear magnetic resonance (NMR) quantification, acid-alkali neutralization titration quantification, amino acid analyzer quantification, enzymatic quantification, quantification using aptamers such as nucleic acid aptamers or peptide aptamers, or colorimetric quantification. The content of a biomarker can be measured using one of these methods alone, or two or more methods can be combined for measurement.

An example of a measurement method suitable for the biomarkers of the present invention is liquid chromatography-mass spectrometry (LC/MS).

<Determining whether or not someone is in a state of social withdrawal>
The process of comparing the acquired content with a preset cutoff value to determine whether or not a person is in a hikikomori state will be described below.

The content of the biomarkers for identifying hikikomori is compared with the cutoff value for each item to determine whether or not a subject is in a hikikomori state for each item. These results are then combined, and if the majority of the items are determined to be highly likely to have a hikikomori state, the subject can be determined to be in a hikikomori state. Furthermore, if less than half of the items are determined to be highly likely to have a hikikomori state, the subject can be determined to not be in a hikikomori state. The determination of whether or not a subject is in a hikikomori state for each item is described below.

The biomarkers for identifying hikikomori used in this embodiment are those described in Section "1. Biomarkers for identifying hikikomori," and include at least two selected from the group consisting of ornithine, acylcarnitine, bilirubin, arginine, arginase, uric acid, and cholesterol esters, and preferably at least two selected from the group consisting of ornithine, acylcarnitine, bilirubin, arginine, and arginase.

If the ornithine content in a blood sample is equal to or above a predetermined cutoff value, it can be determined that the subject is highly likely to be in a hikikomori state.

If the acylcarnitine content in a blood sample is equal to or greater than a predetermined cutoff value, it can be determined that the individual is highly likely to be in a state of social withdrawal.

If the bilirubin content in a blood sample is below a preset cutoff value, it can be determined that the person is likely to be in a hikikomori state.

If the arginine content in a blood sample is below a predetermined cutoff value, it can be determined that the person is likely to be in a hikikomori state.

When the arginase content in a blood sample is equal to or exceeds a predetermined cutoff value, it can be determined that the subject is highly likely to be in a hikikomori state. Arginase is particularly useful as an indicator for characterizing male hikikomori.

The content of components other than ornithine, acylcarnitine, bilirubin, arginine, and arginase can also be compared with pre-set cutoff values for each component, and values above or below the cutoff value can be set as a high probability of a hikikomori state, and a judgment can be made based on that.

The predetermined cutoff value can be set appropriately by a person skilled in the art depending on various conditions, such as the gender and age of the subject, the type of blood sample, the measurement method, and the type of biomarker. There are no particular limitations on the method for setting the cutoff value, and it can be set according to known methods.

The cutoff value may be set based solely on the results of measuring biomarkers in hikikomori individuals, or solely on the results of measuring biomarkers in non-hikikomori individuals; however, it is preferable to measure biomarkers in both hikikomori and non-hikikomori individuals and set the cutoff value based on the results of both. Non-hikikomori individuals may or may not be suffering from a disease, etc., as long as they are not in a hikikomori state, but it is preferable that they are not suffering from a disease that correlates with the measured biomarker.

For example, when setting a cutoff value based solely on the results of measuring biomarkers in hikikomori individuals, the cutoff value may be set so that a certain percentage of hikikomori individuals are included in the distribution of biomarker measurement results for multiple hikikomori individuals. The above description can also be applied mutatis mutandis when setting a cutoff value based solely on the results of measuring biomarkers in non-hikikomori individuals.

Furthermore, when biomarkers are measured for both hikikomori and non-hikikomori individuals and set based on the results of both, the cutoff value may be determined, for example, so that a predetermined percentage of non-hikikomori individuals fall within the normal value range, and a predetermined percentage of hikikomori individuals fall within the abnormal value range. Specifically, for example, in the case of a biomarker where a measured value at or above the cutoff value indicates a high possibility of hikikomori, the cutoff value can be determined so that a predetermined percentage of hikikomori individuals fall above the cutoff value, and a predetermined percentage of non-hikikomori individuals fall below the cutoff value. Furthermore, in the case of a biomarker where a measured value at or below the cutoff value indicates a high possibility of hikikomori, the cutoff value can be determined so that a predetermined percentage of hikikomori individuals fall below the cutoff value, and a predetermined percentage of non-hikikomori individuals fall above the cutoff value.

In the above, the specified percentage is, for example, 70% or more, preferably 80% or more, more preferably 90% or more, even more preferably 95% or more, and particularly preferably 100%.

In addition, when assessing whether a subject is in a hikikomori state, the above-mentioned hikikomori assessment results may be combined with other hikikomori tests to assess whether or not the subject is in a hikikomori state. Examples of other hikikomori tests include the pathological hikikomori assessment based on a structured interview and the self-administered HQ-25 (25-item Hikikomori Questionnaire).

In another embodiment, a method for using the biomarkers of the present invention as indicators for identifying whether a subject is in a hikikomori state is provided. The method includes the steps of: acquiring the content of the biomarkers for hikikomori identification of the present invention in blood samples from a plurality of hikikomori individuals and non-hikikomori individuals; generating training data by correlating the content with a discrimination result of whether or not the subject is a hikikomori; and performing machine learning using the training data to generate a hikikomori identification support model that outputs whether or not the subject is a hikikomori when the content is input. The method further includes the steps of acquiring the content of the biomarkers of the present invention in a blood sample from the subject; and predicting whether or not the subject is a hikikomori by inputting the content of the biomarkers of the subject into a hikikomori identification support model generated by the method for generating a hikikomori identification support model.

The method of this embodiment makes it possible to easily and accurately assess whether a subject is in a state of social withdrawal.

According to the identification method of this embodiment, there is no limit to the age of the subject; for example, the subject may be a young subject, specifically a subject under the age of 30, or an elderly subject, specifically a subject over the age of 65. There is also no limit to the gender of the subject; the subject may be a male or a female subject. Furthermore, the subject may or may not be receiving treatment, such as medication, for a mental illness such as social withdrawal, although it is preferable that the subject is not receiving treatment.

<Hikikomori identification support model>
2 is a block diagram showing the schematic configuration of the hikikomori identification support model generation system 1 of this embodiment. The hikikomori identification support model generation system 1 is a system for implementing the hikikomori identification support model generation method of this embodiment, and includes a content measuring device 2 for a blood sample from a subject, a training data generation device 3, and a machine learning device 4.

The content measuring device 2 measures the content of biomarkers for identifying hikikomori in blood samples from multiple hikikomori individuals and non-hikikomori individuals. The method described above in the "<Method for Measuring Biomarker Content in a Blood Sample>" section is preferably used as the measurement method. In this embodiment, the biomarkers for identifying hikikomori are those described in the "1. Biomarkers for Identifying Hikikomori" section, and include at least two biomarkers selected from the group consisting of ornithine, acylcarnitine, bilirubin, arginine, arginase, uric acid, and cholesterol esters, preferably at least three biomarkers selected from the group consisting of ornithine, acylcarnitine, bilirubin, arginine, arginase, uric acid, and cholesterol esters, and more preferably at least two biomarkers selected from the group consisting of ornithine, acylcarnitine, bilirubin, arginine, and arginase. The biomarkers for identifying social withdrawal more preferably include at least three selected from the group consisting of ornithine, acylcarnitine, bilirubin, arginine, and arginase, and particularly preferably include at least three selected from the group consisting of ornithine and acylcarnitine, bilirubin, arginine, and arginase, and particularly preferably include ornithine, nonanoylcarnitine (AC_C9:0), 3-hydroxydecanoylcarnitine (AC_C10:OH), tetradecanoylcarnitine (AC_C14:0), tetradecanoylcarnitine (AC_C15:0), tetradecanoylcarnitine (AC_C16:0), tetradecanoylcarnitine (AC_C17:0), tetradecanoylcarnitine (AC_C18:0), tetradecanoylcarnitine (AC_C19:0), tetradecanoylcarnitine (AC_C20:0), tetradecanoylcarnitine (AC_C21:0), tetradecanoylcarnitine (AC_C22:0), tetradecanoylcarnitine (AC_C23:0), tetradecanoylcarnitine (AC_C24:0), tetradecanoylcarnitine (AC_C25:0), tetradecanoylcarnitine (AC_C26:0), tetradecanoylcarnitine (AC_C27:0), tetradecanoylcarnitine (AC_C28:0), tetradecanoylcarnitine (AC_C29:0), tetradecanoylcarnitine (AC_C30:0), tetradecanoylcarnitine (AC_C31:0), tetradecanoylcarnitine (AC_C32:0), tetradecanoylcarnitine (AC_C33:0), tetradecanoylcarnitine (AC_C34:0), tetradecanoylcarnitine (AC_C35:0), tetradecanoylcarnitine ( Most preferably, the biomarkers include senoylcarnitine (AC_C14:1), tetradecadienoylcarnitine (AC_C14:2), hexadecanoylcarnitine (AC_C16:0), hexadecenoylcarnitine (AC_C16:1), octadecenoylcarnitine (AC_C18:1), octadecadienoylcarnitine (AC_C18:2), phosphatidylethanolamine ester (PE_34:2), phosphatidylcholine ester (PC_36:5), alanine, serine, bilirubin, arginine, and arginase. Combining the measurement results of multiple biomarkers can improve the accuracy of identifying whether or not a subject is socially withdrawn. The content data in the measured blood sample is sent to the training data generation device 3 along with the subject's ID, etc.

The teacher data generation device 3 can be configured as a general-purpose computer, and its hardware configuration includes a processor such as a CPU or GPU, a main memory device such as DRAM or SRAM (not shown), and an auxiliary memory device 30 such as an HDD or SSD. The auxiliary memory device 30 stores a database D1 of the identification results of whether each measurement subject is a hikikomori or not, as determined by the content measuring device 2, and various programs for operating the teacher data generation device 3.

The teacher data generation device 3 has two functional blocks: an acquisition unit 31 and a teacher data generation unit 32. In this embodiment, these units are realized in software by the processor of the teacher data generation device 3 reading a predetermined program into the main memory and executing it.

The acquisition unit 31 is a functional block that acquires the content of hikikomori identification biomarkers in blood samples from multiple hikikomori and non-hikikomori individuals. In this embodiment, the acquisition unit 31 acquires the content in the blood sample from the content measuring device 2 via a communication line or recording medium, but the content in the blood sample may also be stored in advance in the auxiliary storage device 30 and acquired by reading the content from the auxiliary storage device 30.

The teacher data generation unit 32 is a functional block that generates teacher data D2 by associating the content with the identification result of whether or not the person is a hikikomori. Specifically, when the teacher data generation unit 32 receives the content from the acquisition unit 31, it reads the identification result of whether or not the person being measured for that content is a hikikomori from the identification result database D1 by referencing an ID, etc., and generates teacher data D2 by associating the content with the read identification result. Furthermore, the teacher data generation unit 32 stores the generated teacher data D2 in the auxiliary storage device 30.

Note that if the identification result database D1 is not stored in the auxiliary storage device 30 or if the identification results of the subject are not included in the identification result database D1, the training data generation unit 32 may separately obtain the identification results from an external source and associate them with the content.

The machine learning device 4 can be configured as a general-purpose computer, and like the teacher data generation device 3, it has a processor such as a CPU or GPU, a main memory device such as DRAM or SRAM (not shown), and an auxiliary memory device 40 such as an HDD or SSD. The auxiliary memory device 40 stores the teacher data D2 transferred from the teacher data generation device 3, various programs for operating the machine learning device 4, etc.

In this embodiment, the machine learning device 4 is installed on the cloud, but it may also be installed in the same location as the teacher data generation device 3, or the teacher data generation device 3 and the machine learning device 4 may be configured as an integrated unit.

The machine learning device 4 includes a learning unit 41 as a functional block. The learning unit 41 performs machine learning based on the training data D2 to generate a hikikomori identification support model M1 that outputs whether or not a person is a hikikomori when the content in a blood sample is input. The learning method used by the learning unit 41 for machine learning is not particularly limited, but in this embodiment, machine learning using random forests is used. The generated hikikomori identification support model M1 is stored in the auxiliary storage device 40 and is then used to support hikikomori identification.

Figure 3 is a flowchart showing the processing steps of the method for generating a hikikomori identification support model in this embodiment.

In step S1, the acquisition unit 31 of the teacher data generation device 3 acquires the content of hikikomori identification biomarkers in blood samples from multiple hikikomori individuals and non-hikikomori individuals from the content measuring device 2. In step S2, the teacher data generation unit 32 associates the content with the identification result of whether or not the individual is a hikikomori, thereby generating teacher data D2.

After that, steps S1 and S2 are repeated while changing the subject, and when a predetermined amount of training data D2 necessary and sufficient for machine learning has been accumulated (YES in step S3), the training data D2 is transferred to the machine learning device 4.

Next, in step S4, the learning unit 41 performs machine learning using the training data D2. As a result, the learning unit 41 generates a hikikomori identification support model M1 that outputs whether or not a person is a hikikomori when the content is input (step S5).

<Predicting whether or not someone is a hikikomori>
4 is a block diagram showing a schematic configuration of the support device 5 of this embodiment. The support device 5 is a device for implementing the hikikomori identification support method of this embodiment.

The support device 5 can be configured as a general-purpose computer, and its hardware configuration includes a processor such as a CPU or GPU, a main memory device such as DRAM or SRAM (not shown), and an auxiliary memory device 50 such as an HDD or SSD. The auxiliary memory device 50 stores the hikikomori identification support model M1 generated by the hikikomori identification support model generation method described above, as well as various programs for operating the support device 5.

The support device 5 has two functional blocks: an acquisition unit 51 and a prediction unit 52. In this embodiment, these units are implemented in software by the processor of the support device 5 loading a predetermined program into the main memory and executing it.

The acquisition unit 51 is a functional block that acquires the content of the hikikomori identification biomarker in the subject's blood sample. In this embodiment, the acquisition unit 51 acquires the content from the content measuring device 2 via a communication line or recording medium, but the content may also be stored in advance in the auxiliary storage device 50 and acquired by reading the content from the auxiliary storage device 50. The type of content acquired by the acquisition unit 51 is the same as the type of content acquired by the acquisition unit 51 shown in Figure 2 when generating the hikikomori identification support model M1.

The prediction unit 52 is a functional block that predicts whether a subject is a hikikomori by inputting the content of a substance in the subject's blood sample into the hikikomori identification support model M1. Specifically, when the prediction unit 52 receives the content from the acquisition unit 51, it inputs the content into the hikikomori identification support model M1 and receives the prediction result of whether or not the subject is a hikikomori, which is output from the hikikomori identification support model M1 in response.

Furthermore, the prediction unit 52 outputs the prediction results to the display device 6. Note that the prediction results may also be output by printing or other methods.

Figure 5 is a flowchart showing the processing steps of the hikikomori identification support method of this embodiment.

In step S6, the acquisition unit 51 of the support device 5 acquires the content of the hikikomori identification biomarker in the subject's blood sample from the content measuring device 2. In step S7, the prediction unit 52 inputs the content in the subject's blood sample into the hikikomori identification support model M1, thereby predicting whether the subject is a hikikomori. In step 8, the prediction unit 52 outputs the prediction result to the display device 6 or the like.

As described above, the hikikomori identification support model M1 can be used to predict whether or not a subject is in a hikikomori state.

When assessing whether a subject is in a hikikomori state, the prediction results of the hikikomori identification support model may be combined with other hikikomori tests to assess whether the subject is in a hikikomori state. Examples of other hikikomori tests include pathological hikikomori assessments based on structured interviews and the self-administered HQ-25 (25-item Hikikomori Questionnaire).

2.2 Determining the severity of hikikomori The severity of hikikomori can be assessed to some extent using testing methods such as the pathological hikikomori assessment based on a structured interview or the self-administered HQ-25 (25-item Hikikomori Questionnaire). However, as these methods rely on the patient's subjective complaints and attitudes, a more objective evaluation method is needed.

In another embodiment of the present invention, a method is provided in which a blood component of a subject is used as an index for determining the severity of the subject's hikikomori state. The blood component can be a hikikomori identification biomarker of the present invention, and can include blood components containing at least two selected from the group consisting of ornithine, acylcarnitine, bilirubin, arginine, and arginase.

In this embodiment, a method for determining the severity of hikikomori status is provided, which includes the steps of: acquiring the content of a hikikomori identification biomarker of the present invention in a blood sample from a subject; and comparing the acquired content with a preset cutoff value to determine the severity of the hikikomori status.

According to the identification method of this embodiment, there is no limit to the age of the subject; for example, the subject may be a young subject, specifically a subject under the age of 30, or an elderly subject, specifically a subject over the age of 65. There is also no limit to the gender of the subject; the subject may be a male or a female subject. Furthermore, the subject may or may not be receiving treatment, such as medication, for a mental illness such as social withdrawal, although it is preferable that the subject is not receiving treatment.

<Assessing the severity of hikikomori status>
In this embodiment, the method described in the above section "<Method for measuring the content of biomarker in blood sample>" is preferably used for the step of obtaining the content of the biomarker for identifying hikikomori of the present invention in the blood sample of the subject.

The process of comparing the obtained content with a pre-set cutoff value to determine the severity of the hikikomori state is described below.

The content of the hikikomori identification biomarkers is compared with the respective cutoff values to determine the severity of the hikikomori state for each item. The results are then combined, and if it is determined that the severity of the hikikomori state is likely to be high for the majority of items, the subject can be determined to have a high severity of hikikomori state.

Regarding the severity of the hikikomori state for each item, in the case of a biomarker that is determined to be "highly likely to have a hikikomori state" when its content in a blood sample is equal to or greater than the cutoff value, it can be determined that the greater the content of the biomarker in the blood sample is above the cutoff value, the greater the likelihood of the severity of the hikikomori state. Conversely, it can be determined that the closer the content of the biomarker in the blood sample is to the cutoff value, the less likely the severity of the hikikomori state. Furthermore, in the case of a biomarker that is determined to be "highly likely to have a hikikomori state" when its content in the blood sample is equal to or less than the cutoff value, it can be determined that the smaller the content of the biomarker in the blood sample is below the cutoff value, the greater the likelihood of the severity of the hikikomori state. Conversely, it can be determined that the closer the content of the biomarker in the blood sample is to the cutoff value, the less likely the severity of the hikikomori state.

Furthermore, the biomarkers for identifying hikikomori used in this embodiment are those described in Section "1. Biomarkers for identifying hikikomori," and include at least two types selected from the group consisting of ornithine, acylcarnitine, bilirubin, arginine, arginase, uric acid, and cholesterol esters, and preferably include at least two types selected from the group consisting of ornithine, acylcarnitine, bilirubin, arginine, and arginase.

The higher the ornithine content in a blood sample is, the higher the pre-set cutoff value, indicating a higher likelihood of the severity of the hikikomori state.

The higher the acylcarnitine content in a blood sample is, the higher the pre-set cutoff value, indicating a higher likelihood of the severity of the hikikomori state.

The lower the bilirubin content in a blood sample is below a pre-set cutoff value, the more likely the person is to be suffering from social withdrawal.

The lower the arginine content in a blood sample is below a preset cutoff value, the more likely the severity of the hikikomori state is.

Regarding the arginase content in a blood sample, it can be determined that the higher the pre-set cutoff value, the more likely the severity of the hikikomori state is. Arginase is particularly useful as an indicator for characterizing male hikikomori.

The content of components other than ornithine, acylcarnitine, bilirubin, arginine, and arginase can also be compared with pre-set cutoff values for each component, and values above or below the cutoff value can be set as a high probability of a hikikomori state, and a judgment can be made based on that.

The predetermined cutoff value can be set appropriately by a person skilled in the art depending on various conditions, such as the gender and age of the subject, the type of blood sample, the measurement method, and the type of biomarker. There are no particular limitations on the method for setting the cutoff value, and it can be set according to known methods.

The cutoff value may be set based solely on the results of measuring biomarkers in hikikomori individuals, or solely on the results of measuring biomarkers in non-hikikomori individuals; however, it is preferable to measure biomarkers in both hikikomori and non-hikikomori individuals and set the cutoff value based on the results of both. Non-hikikomori individuals may or may not be suffering from a disease, etc., as long as they are not in a hikikomori state, but it is preferable that they are not suffering from a disease that correlates with the measured biomarker.

For example, when setting a cutoff value based solely on the results of measuring biomarkers in hikikomori individuals, the cutoff value may be set so that a certain percentage of hikikomori individuals are included in the distribution of biomarker measurement results for multiple hikikomori individuals. The above description can also be applied mutatis mutandis when setting a cutoff value based solely on the results of measuring biomarkers in non-hikikomori individuals.

Furthermore, when biomarkers are measured for both hikikomori and non-hikikomori individuals and set based on the results of both, the cutoff value may be determined, for example, so that a predetermined percentage of non-hikikomori individuals fall within the normal value range, and a predetermined percentage of hikikomori individuals fall within the abnormal value range. Specifically, for example, in the case of a biomarker where a measured value at or above the cutoff value indicates a high possibility of hikikomori, the cutoff value can be determined so that a predetermined percentage of hikikomori individuals fall above the cutoff value, and a predetermined percentage of non-hikikomori individuals fall below the cutoff value. Furthermore, in the case of a biomarker where a measured value at or below the cutoff value indicates a high possibility of hikikomori, the cutoff value can be determined so that a predetermined percentage of hikikomori individuals fall below the cutoff value, and a predetermined percentage of non-hikikomori individuals fall above the cutoff value.

In the above, the specified percentage is, for example, 70% or more, preferably 80% or more, more preferably 90% or more, even more preferably 95% or more, and particularly preferably 100%.

In addition, when assessing the severity of a subject's hikikomori state, the above-mentioned assessment results may be combined with other hikikomori tests to assess the severity of the hikikomori state. Examples of other hikikomori tests include the pathological hikikomori assessment based on a structured interview and the self-administered HQ-25 (25-item Hikikomori Questionnaire).

In another embodiment, a method for using the biomarkers of the present invention as indicators for determining the severity of a hikikomori state in a subject is provided, comprising the steps of: acquiring the content of the hikikomori identification biomarkers of the present invention in blood samples from a plurality of hikikomori individuals and non-hikikomori individuals; generating training data by correlating the content with the hikikomori severity assessment results; and performing machine learning using the training data to generate a hikikomori severity prediction support model that outputs a hikikomori severity level when the content is input. Furthermore, a method for predicting the severity of hikikomori in a subject is provided, comprising the steps of acquiring the content of the biomarkers of the present invention in a blood sample from the subject; and predicting the severity of hikikomori in the subject by inputting the content of the biomarkers of the subject in the hikikomori severity prediction support model generated by the method for generating a hikikomori severity prediction support model.

The method of this embodiment makes it possible to easily and accurately evaluate the severity of a subject's hikikomori state.

According to the identification method of this embodiment, there is no limit to the age of the subject; for example, the subject may be a young subject, specifically a subject under the age of 30, or an elderly subject, specifically a subject over the age of 65. There is also no limit to the gender of the subject; the subject may be a male or a female subject. Furthermore, the subject may or may not be receiving treatment, such as medication, for a mental illness such as social withdrawal, although it is preferable that the subject is not receiving treatment.

<Hikikomori severity prediction support model>
Figure 6 is a block diagram showing the schematic configuration of a hikikomori severity prediction support model generation system 1' according to one embodiment of the present invention. The hikikomori severity prediction support model generation system 1' is a system for implementing the hikikomori severity prediction support model generation method according to one embodiment of the present invention, and includes a content measuring device 2, a teacher data generation device 3', and a machine learning device 4'. Note that in Figure 6, components having the same functions as those in Figure 2 are designated by the same reference numerals, and detailed description thereof will be omitted.

The content measuring device 2 measures the content of biomarkers for identifying hikikomori in blood samples from multiple hikikomori individuals and non-hikikomori individuals. The measurement method preferably used is the method described above in the "<Method for measuring the content of biomarkers in blood samples>" section. In this embodiment, the biomarkers for identifying hikikomori include at least two species selected from the group consisting of ornithine, acylcarnitine, bilirubin, arginine, arginase, uric acid, and cholesterol esters, preferably at least three species selected from the group consisting of ornithine, acylcarnitine, bilirubin, arginine, arginase, uric acid, and cholesterol esters, and more preferably at least two species selected from the group consisting of ornithine, acylcarnitine, bilirubin, arginine, and arginase. The biomarkers for identifying social withdrawal more preferably include at least three selected from the group consisting of ornithine, acylcarnitine, bilirubin, arginine, and arginase, and particularly preferably include at least three selected from the group consisting of ornithine and acylcarnitine, bilirubin, arginine, and arginase, and further preferably include ornithine, octanoylcarnitine (AC_C8:0), tetradecanoylcarnitine (AC_C14:0), hexadecanoylcarnitine (AC_C16:0), hexadecanoylcarnitine (AC_C18:0), octanoylcarnitine (AC_C19:0), octanoylcarnitine (AC_C20:0), octanoylcarnitine (AC_C21:0), octanoylcarnitine (AC_C22:0), octanoylcarnitine (AC_C23:0), octanoylcarnitine (AC_C24:0), octanoylcarnitine (AC_C25:0), octanoylcarnitine (AC_C26:0), octanoylcarnitine (AC_C27:0), octanoylcarnitine (AC_C28:0), octanoylcarnitine (AC_C29:0), octanoylcarnitine (AC_C30:0), octanoylcarnitine (AC_C31:0), octanoylcarnitine (AC_C32:0), octanoylcarnitine (AC_C33:0), octanoylcarnitine (AC_C34:0), octanoylcarnitine (AC_C35:0), octanoylcarnitine (AC_C36:0), octanoylcarnitine (AC_C37:0), octanoylcarnitine (AC_C38:0), octanoylcarnit Most preferably, the biomarkers include all of the following: lunitine (AC_C16:1), octadecenoylcarnitine (AC_C18:1), octadecadienoylcarnitine (AC_C18:2), 3-hydroxyoctadecenoylcarnitine (AC_C18:1OH), 3-hydroxyoctadecadienoylcarnitine (AC_C18:2OH), hexadecenoyllysophosphatidylcholine (LPC_16:1), total bilirubin, direct bilirubin, direct bilirubin/indirect bilirubin ratio, LDL cholesterol, arginine, and arginase. Combining the measurement results of multiple biomarkers can improve the accuracy of identifying whether or not a subject is in a social withdrawal state. The measured content data is sent to the training data generation device 3' along with the subject's ID, etc.

The hardware configuration of the teacher data generation device 3' is the same as that of the teacher data generation device 3 shown in Figure 2. The auxiliary storage device 30 stores a database D3 of the results of the hikikomori severity assessment for each subject measured by the content measuring device 2, as well as various programs for operating the teacher data generation device 3'. The results of the hikikomori severity assessment for each subject can be obtained using known methods, such as a pathological hikikomori assessment based on a structured interview or the self-administered HQ-25 (25-item Hikikomori Questionnaire).

The teacher data generation device 3' has two functional blocks: an acquisition unit 31 and a teacher data generation unit 32'. In this embodiment, these units are realized in software by the processor of the teacher data generation device 3' reading a predetermined program into the main memory and executing it.

The acquisition unit 31 acquires the content of hikikomori identification biomarkers in blood samples of multiple hikikomori individuals and non-hikikomori individuals from the content measuring device 2.

The teacher data generation unit 32' is a functional block that generates teacher data D4 by associating the content with the assessment result of the severity of social withdrawal. Specifically, when the teacher data generation unit 32' receives the content from the acquisition unit 31, it reads the assessment result of the severity of social withdrawal of the person being measured for that content from the severity assessment result database D3 by referencing the ID, etc., and generates teacher data D4 by associating the content with the read assessment result. Furthermore, the teacher data generation unit 32' stores the generated teacher data D4 in the auxiliary storage device 30.

Note that if the severity assessment result database D3 is not stored in the auxiliary storage device 30 or if the assessment result for the subject is not included in the severity assessment result database D3, the teacher data generation unit 32' may separately obtain the assessment result from an external source and associate it with the content.

The hardware configuration of the machine learning device 4' is the same as that of the machine learning device 4 shown in Figure 2. The auxiliary storage device 40 stores teacher data D4 transferred from the teacher data generation device 3', various programs for operating the machine learning device 4', and the like.

The machine learning device 4' includes a learning unit 41' as a functional block. The learning unit 41' performs machine learning based on the training data D4 to generate a hikikomori severity prediction support model M2 that outputs the severity of hikikomori when a content is input. The learning method used by the learning unit 41' is not particularly limited, but in this embodiment, machine learning using partial least squares regression is used. The generated hikikomori severity prediction support model M2 is stored in the auxiliary storage device 40 and is then used to support hikikomori severity prediction.

Figure 7 is a flowchart showing the processing steps of the method for generating a hikikomori severity prediction support model in this embodiment.

In step S11, the acquisition unit 31 of the teacher data generation device 3' acquires the content of hikikomori identification biomarkers in blood samples from multiple hikikomori individuals and non-hikikomori individuals from the content measuring device 2. In step S12, the teacher data generation unit 32' associates the content with the assessment result of the severity of hikikomori to generate teacher data D4.

After that, steps S11 and S12 are repeated while changing the subject, and when a predetermined amount of training data D4 necessary and sufficient for machine learning has been accumulated (YES in step S13), the training data D4 is transferred to the machine learning device 4'.

Next, in step S14, the learning unit 41' performs machine learning using the training data D4. As a result, the learning unit 41' generates a hikikomori severity prediction support model M2 that outputs the severity of hikikomori when the content is input (step S15).

<Predicting the severity of hikikomori>
8 is a block diagram showing a schematic configuration of a support device 5' of this embodiment. The support device 5' is a device for implementing the hikikomori severity prediction support method of this embodiment.

The hardware configuration of the support device 5' is the same as that of the support device 5 shown in Figure 4. The auxiliary storage device 50 stores the hikikomori severity prediction support model M2 generated by the hikikomori severity prediction support model generation method described above, as well as various programs for operating the support device 5'.

The support device 5' includes an acquisition unit 51 and a prediction unit 52'. In this embodiment, these units are implemented in software by the processor of the support device 5' loading a predetermined program into the main memory and executing it.

The acquisition unit 51 acquires the content of the hikikomori identification biomarker in the subject's blood sample from the content measuring device 2. The type of content acquired by the acquisition unit 51 is the same as the type of content acquired by the acquisition unit 31 shown in Figure 6 when generating the hikikomori severity prediction support model M2.

The prediction unit 52' is a functional block that predicts the severity of hikikomori of a subject by inputting the content of the substance in the subject's blood sample into the hikikomori severity prediction support model M2. Specifically, when the prediction unit 52' receives the content from the acquisition unit 51, it inputs the content into the hikikomori severity prediction support model M2 and receives the prediction result of the severity of hikikomori output from the hikikomori severity prediction support model M2 in response. Furthermore, the prediction unit 52' outputs the prediction result to the display device 6.

Figure 9 is a flowchart showing the processing steps of the hikikomori severity prediction support method of this embodiment.

In step S16, the acquisition unit 51 of the support device 5' acquires the content of the hikikomori identification biomarker in the subject's blood sample from the content measuring device 2. In step S17, the prediction unit 52' inputs the content in the subject's blood sample into the hikikomori severity prediction support model M2, thereby predicting the severity of hikikomori for the subject. In step S18, the prediction unit 52' outputs the prediction result to the display device 6 or the like.

As described above, the hikikomori severity prediction support model M2 can be used to predict the severity of a subject's hikikomori.

When assessing the severity of a subject's hikikomori state, the prediction results of the hikikomori severity prediction support model may be combined with other hikikomori tests to assess the severity of the hikikomori state. Examples of other hikikomori tests include the pathological hikikomori assessment based on a structured interview and the self-administered HQ-25 (25-item Hikikomori Questionnaire).

2.3 Stratification of Hikikomori Individuals Various factors, including environmental factors, personality, and mental illness, are thought to be causes of hikikomori, and it is possible to stratify hikikomori individuals based on their background. In this specification, "stratification" refers to classifying the hikikomori state into subtypes based on the background and defining a clustering classification of hikikomori individuals for each subtype. Examples of backgrounds for hikikomori include depression, social anxiety disorder, developmental disorders, and a tendency toward schizophrenia. When hikikomori individuals are stratified based on whether or not they have a tendency toward depression, it can be determined that treatment for depression is effective in resolving the hikikomori state for those with a tendency toward depression, and that treatment for depression is ineffective in resolving the hikikomori state for those without a tendency toward depression.

In another embodiment of the present invention, a method is provided in which blood components of a subject are used as indicators for stratifying the subject. The type of stratification is not particularly limited, but preferably includes stratification into non-hikikomori, hikikomori with depression tendencies, and other hikikomori. The blood components can be the hikikomori identification biomarkers of the present invention, and include blood components containing at least one selected from the group consisting of uric acid and cholesterol esters.

In this embodiment, a method for supporting stratification of hikikomori is provided, which includes the steps of acquiring the content of the hikikomori identification biomarker of the present invention in a blood sample from a subject, and comparing the acquired content with a preset reference value to determine the clustering classification of hikikomori.

According to the identification method of this embodiment, there is no limit to the age of the subject; for example, the subject may be a young subject, specifically a subject under the age of 30, or an elderly subject, specifically a subject over the age of 65. There is also no limit to the gender of the subject; the subject may be a male or a female subject. Furthermore, the subject may or may not be receiving treatment, such as medication, for a mental illness such as social withdrawal, although it is preferable that the subject is not receiving treatment.

<Determining clustering classification of hikikomori>
In this embodiment, the method described in the above section "<Method for measuring the content of biomarker in blood sample>" is preferably used for the step of obtaining the content of the biomarker for identifying hikikomori of the present invention in the blood sample of the subject.

The process of comparing the acquired content with a preset standard value and determining the clustering classification of hikikomori is described below.

The content of each biomarker for identifying hikikomori is compared with a reference value, and the hikikomori clustering classification for each item is determined. These results are then combined, and the clustering classification determined for the greatest number of items can be determined to be the hikikomori clustering classification for the subject.

The biomarkers for identifying hikikomori used in this embodiment are those described in Section "1. Biomarkers for identifying hikikomori," and include at least two selected from the group consisting of ornithine, acylcarnitine, bilirubin, arginine, arginase, uric acid, and cholesterol esters, and preferably at least one selected from the group consisting of uric acid and cholesterol esters.

If the uric acid content in a blood sample is above the standard value, it can be determined that the person is likely to be a hikikomori with depression-prone individuals.

If the cholesterol ester content in a blood sample is above the standard value, it can be determined that the person is likely to be a hikikomori with depression-prone individuals.

The predetermined reference values may be set appropriately by a person skilled in the art depending on various conditions, such as the gender and age of the subject, the type of blood sample, the measurement method, and the type of biomarker. There are no particular limitations on the method for setting the reference values, and they can be set according to known methods.

The reference value may be set based solely on the results of measuring the subject's biomarkers in one clustering classification, but it is preferable to measure the subject's biomarkers in multiple clustering classifications and set the reference value by combining the results.

For example, when setting a reference value based solely on the results of measuring a subject's biomarkers in one clustering classification, the reference value may be set so that the subject is included at a predetermined rate in the distribution of biomarker measurement results for multiple subjects.

Furthermore, when biomarkers of subjects are measured in multiple clustering classifications and the results are combined to set the reference value, the reference value may be determined, for example, so that a predetermined proportion of subjects fall into each clustering classification. Specifically, for example, in the case of a biomarker for which a measured value equal to or greater than the reference value is likely to correspond to a group of hikikomori individuals with depressive tendencies, the reference value can be determined so that a predetermined proportion of subjects falling into the group of hikikomori individuals with depressive tendencies are above the reference value, and a predetermined proportion of subjects falling into other groups are below the reference value. Furthermore, for example, in the case of a biomarker for which a measured value equal to or less than the reference value is likely to correspond to a group of hikikomori individuals with depressive tendencies, the reference value can be determined so that a predetermined proportion of subjects falling into the group of hikikomori individuals with depressive tendencies are below the reference value, and a predetermined proportion of subjects falling into other groups are above the reference value.

In the above, the specified percentage is, for example, 70% or more, preferably 80% or more, more preferably 90% or more, even more preferably 95% or more, and particularly preferably 100%.

In addition, when determining a subject's hikikomori clustering classification, the clustering classification results described above may be combined with other hikikomori tests to determine the clustering classification. Examples of other hikikomori tests include structured interview-based pathological hikikomori assessments and the self-administered HQ-25 (25-item Hikikomori Questionnaire).

In another embodiment, a method for using the biomarkers of the present invention as indicators for stratifying subjects is provided, comprising the steps of: acquiring the content of the hikikomori identification biomarkers of the present invention in blood samples from a plurality of hikikomori individuals and non-hikikomori individuals; generating training data by correlating the content with the results of a hikikomori clustering classification; and performing machine learning using the training data to generate a hikikomori stratification support model that outputs a hikikomori clustering classification when the content is input. Furthermore, a method for supporting the stratification of hikikomori in a subject is provided, comprising the steps of acquiring the content of the biomarkers of the present invention in a blood sample from the subject; and predicting the hikikomori clustering classification of the subject by inputting the content of the biomarkers of the subject into the hikikomori stratification support model generated by the method for generating a hikikomori stratification support model. There are no particular limitations on the content of the stratification, but preferred examples include stratification into non-hikikomori, hikikomori with depression tendencies, and other hikikomori.

The method of this embodiment makes it possible to easily and accurately evaluate the severity of a subject's hikikomori state.

According to the identification method of this embodiment, there is no limit to the age of the subject; for example, the subject may be a young subject, specifically a subject under the age of 30, or an elderly subject, specifically a subject over the age of 65. There is also no limit to the gender of the subject; the subject may be a male or a female subject. Furthermore, the subject may or may not be receiving treatment, such as medication, for a mental illness such as social withdrawal, although it is preferable that the subject is not receiving treatment.

<Hikikomori stratification support model>
FIG. 10 is a block diagram showing the schematic configuration of a hikikomori stratification support model generation system 1" of this embodiment. The hikikomori stratification support model generation system 1" is a system for implementing the hikikomori stratification support model generation method of this embodiment, and is equipped with a content measuring instrument 2, a teacher data generation device 3", and a machine learning device 4". In FIG. 10, components having the same functions as those in FIGS. 2 and 6 are given the same reference numerals, and detailed explanations will be omitted.

The content measuring device 2 measures the content of biomarkers for identifying hikikomori in blood samples from multiple hikikomori individuals and non-hikikomori individuals. The measurement method preferably used is the method described above in the "<Method for measuring the content of biomarkers in blood samples>" section. In this embodiment, the biomarkers for identifying hikikomori include at least two species selected from the group consisting of ornithine, acylcarnitine, bilirubin, arginine, arginase, uric acid, and cholesterol esters, preferably at least three species selected from the group consisting of ornithine, acylcarnitine, bilirubin, arginine, arginase, uric acid, and cholesterol esters, and more preferably at least one species selected from the group consisting of uric acid and cholesterol esters. The biomarker for identifying social withdrawal more preferably includes at least two cholesterol esters and uric acid, and particularly preferably includes at least three cholesterol esters and uric acid, and is particularly preferably selected from the group consisting of uric acid, heptadecanoic acid cholesterol ester (CE_17:0), octadecadienoic acid cholesterol ester (CE_18:2), octadecatrienoic acid cholesterol ester (CE_18:3), eicosenoic acid cholesterol ester (CE_20:1), eicosadienoic acid cholesterol ester (CE_20:2), 3-hydroxyoctadecenoylcarnitine (AC_C18:1OH), 3-hydroxyoctadecadienoylcarnitine (AC_C18:2O ... Most preferably, the biomarkers include all of the following: phospholipids (PC_32:0), eicosatetraenoyllysophosphatidylcholine (LPC_20:4), triglyceride (TG_46:2), triglyceride (TG_46:3), triglyceride (TG_48:3), triglyceride (TG_50:2), triglyceride (TG_50:3), triglyceride (TG_52:2), triglyceride (TG_52:3), phosphatidylcholine ester (PC_32:0), phosphatidylcholine ester (PC_32:1), phosphatidylcholine ester (PC_38:3), phosphatidylcholine ester (PC_40:4), phosphatidylcholine ester (PC_40:5), phosphatidylcholine ester (PC_40:6), proline, alanine, and valine. Combining the measurement results of multiple biomarkers can improve the accuracy of identifying whether or not a subject is socially withdrawn. The measured content data is sent to the training data generation device 3" along with the ID of the person being measured, etc.

The hardware configuration of the teacher data generation device 3" is the same as that of the teacher data generation device 3 shown in Figure 2 and the teacher data generation device 3' shown in Figure 6. The auxiliary storage device 30 stores a database D5 of the clustering classification determination results of hikikomori for each measurement subject measured by the content measuring device 2, as well as various programs for operating the teacher data generation device 3". The clustering classification determination result database D5 records the clustering classification determination results of hikikomori for each measurement subject determined using the principal component clustering method.

The teacher data generation device 3" has two functional blocks: an acquisition unit 31 and a teacher data generation unit 32". In this embodiment, these units are realized in software by the processor of the teacher data generation device 3" reading a predetermined program into the main memory and executing it.

The acquisition unit 31 acquires the content of hikikomori identification biomarkers in blood samples of multiple hikikomori individuals and non-hikikomori individuals from the content measuring device 2.

The teacher data generation unit 32" is a functional block that generates teacher data D6 by associating the content with the judgment result of the hikikomori clustering classification. Specifically, when the teacher data generation unit 32" receives the content from the acquisition unit 31, it reads the judgment result of the hikikomori clustering classification for the person whose content is being measured from the clustering classification judgment result database D5 by referencing the ID, etc., and generates teacher data D6 by associating the content with the read judgment result. Furthermore, the teacher data generation unit 32" stores the generated teacher data D6 in the auxiliary storage device 30.

Note that if the clustering classification judgment result database D5 is not stored in the auxiliary storage device 30 or if the judgment results for the subject are not included in the clustering classification judgment result database D5, the teacher data generation unit 32' may separately obtain the judgment results from an external source and associate them with the content.

The hardware configuration of the machine learning device 4" is the same as that of the machine learning device 4 shown in Figure 2 and the machine learning device 4' shown in Figure 6. The auxiliary storage device 40 stores the teacher data D6 transferred from the teacher data generation device 3" and various programs for operating the machine learning device 4".

The machine learning device 4" has a learning unit 41" as a functional block. The learning unit 41" performs machine learning based on the training data D6, and generates a hikikomori stratification support model M3 that outputs a clustering classification of hikikomori when a content amount is input. The learning method used by the learning unit 41" is not particularly limited, but in this embodiment, machine learning using random forest or partial least squares regression is used. The generated hikikomori stratification support model M3 is stored in the auxiliary storage device 40 and is then used to support hikikomori stratification.

Figure 11 is a flowchart showing the processing steps of the method for generating a hikikomori stratification support model in this embodiment.

In step S21, the acquisition unit 31 of the teacher data generation device 3" acquires the content of hikikomori identification biomarkers in blood samples of multiple hikikomori individuals and non-hikikomori individuals from the content measuring device 2. In step S22, the teacher data generation unit 32" associates the content with the hikikomori clustering assessment results to generate teacher data D6.

After that, steps S21 and S22 are repeated while changing the subject, and when a predetermined amount of training data D6 necessary and sufficient for machine learning has been accumulated (YES in step S23), the training data D6 is transferred to the machine learning device 4".

Next, in step S24, the learning unit 41" performs machine learning using the training data D6. As a result, when the content is input, the learning unit 41" generates a hikikomori stratification support model M3 that outputs a clustering classification of hikikomori (step S25).

<Prediction of stratification of hikikomori>
FIG. 12 is a block diagram showing a schematic configuration of a support device 5'' of this embodiment. The support device 5'' is a device for implementing the hikikomori stratification support method of this embodiment.

The hardware configuration of the support device 5" is the same as that of the support device 5 shown in Figure 4 and the support device 5' shown in Figure 8. The auxiliary storage device 50 stores the hikikomori stratification support model M3 generated by the method for generating a hikikomori stratification support model described above, as well as various programs for operating the support device 5".

The support device 5" includes an acquisition unit 51 and a prediction unit 52". In this embodiment, these units are implemented in software by the processor of the support device 5" loading a predetermined program into the main memory and executing it.

The acquisition unit 51 acquires the content of the hikikomori identification biomarker in the subject's blood sample from the content measuring device 2. The type of content acquired by the acquisition unit 51 is the same as the type of content acquired by the acquisition unit 31 shown in FIG. 10 when generating the hikikomori stratification support model M3.

The prediction unit 52" is a functional block that predicts the hikikomori clustering classification of a subject by inputting the content in the subject's blood sample into the hikikomori stratification support model M3. Specifically, when the prediction unit 52" receives the content from the acquisition unit 51, it inputs the content into the hikikomori stratification support model M3 and receives the predicted hikikomori clustering classification result output from the hikikomori stratification support model M3 in response. Furthermore, the prediction unit 52" outputs the prediction result to the display device 6.

Figure 13 is a flowchart showing the processing steps of the hikikomori stratification support method of this embodiment.

In step S26, the acquisition unit 51 of the support device 5" acquires the content of the hikikomori identification biomarker in the subject's blood sample from the content measuring device 2. In step S27, the prediction unit 52" inputs the content in the subject's blood sample into the hikikomori stratification support model M3, thereby predicting the subject's hikikomori clustering classification. In step S28, the prediction unit 52" outputs the prediction result to the display device 6, etc.

As described above, the hikikomori stratification support model M3 can be used to predict the subject's hikikomori clustering classification.

When determining the clustering classification of a subject's hikikomori state, the prediction results of the hikikomori severity prediction support model may be combined with other hikikomori tests to determine the clustering classification of the hikikomori state. Other hikikomori tests include pathological hikikomori assessments based on structured interviews and the self-administered HQ-25 (25-item Hikikomori Questionnaire).

3. Uses of the biomarker for identifying hikikomori In this embodiment, the biomarker for identifying hikikomori can be used for the following purposes in addition to identifying hikikomori, predicting severity, and stratifying them.

<Assessing the effectiveness of interventions to resolve hikikomori>
In general, the effects of intervention vary from person to person, so it is very useful to examine the effects of intervention for each individual. The effects of intervention can be easily examined by using the biomarker for identifying hikikomori of the present invention.

For example, blood is taken from a hikikomori before and after some kind of intervention is performed on the hikikomori, the amount of biomarkers used to identify hikikomori in the blood sample is measured, and the amount of biomarkers used to identify hikikomori before and after the intervention is compared. If the amount of biomarkers used to identify hikikomori after the intervention approaches the range for determining that the person is not in a hikikomori state, then the intervention can be determined to be effective.

<Preventive Intervention>
Furthermore, the biomarker for identifying hikikomori of the present invention can also be used for the prevention of or early intervention in a hikikomori state.

For example, even if a person is not a hikikomori, and does not meet the definition of hikikomori, which is "a state in which the person does not leave the house for four or more days a week, has not engaged in social participation activities for six or more consecutive months, and as a result experiences functional impairment or distress," if the content of the hikikomori identification biomarker of the present invention is within a range that determines the person as being in a "hikikomori state," it can be determined that the person is highly likely to become a hikikomori state in the future, and hikikomori state can be prevented. It also becomes possible to begin intervention at an early stage when the person has become a hikikomori state.

The present invention will be further explained below using examples, but the present invention is not limited to these examples.

All of the following examples were carried out at room temperature and pressure.

The biomarkers used in the analysis in this example are shown in Table 1 below.

In the table above, for example, AC_C2:0 indicates an acylcarnitine with two fatty acid carbon atoms and zero double bonds. For example, LPC_16:0 indicates a lysophosphatidylcholine with 16 fatty acid carbon atoms and zero double bonds. For example, PE_34:1 indicates a phosphatidylethanolamine with 34 total fatty acid carbon atoms and one double bond. For example, CE_17:0 indicates a cholesterol ester with 17 fatty acid carbon atoms and zero double bonds. For example, TG_46:2 indicates a triglyceride with 46 total fatty acid carbon atoms and two double bonds. For example, PC_28:1 indicates a phosphatidylcholine with 28 total fatty acid carbon atoms and one double bond.

[Example 1] Hikikomori Identification Support Model [1] Subject Selection and Plasma Sample Collection Data from subjects (hikikomori and non-hikikomori) in this example were collected at Kyushu University, with all subjects providing written informed consent. The subject selection method is described below. Data on the subjects is shown in Table 2.

(1) Hikikomori Individuals who fulfilled the following three criteria were defined as hikikomori, and blood samples were collected from 42 hikikomori individuals (mean age 30.2 years, SD = 8.2) who were not receiving drug treatment.
(a) Not leaving home for more than four days a week
(b) Not engaging in social participation activities for six months or more
(c) They experience functional impairment or distress as a result.

(2) Non-hikikomori individuals Those who did not meet the above three criteria and were not in a state of hikikomori were defined as non-hikikomori individuals, and blood samples were collected from 41 non-hikikomori individuals.

(3) Collection of Blood Samples Blood samples were collected by peripheral blood sampling via venipuncture.

[2] Metabolomic Analysis (1) Blood Sample Pretreatment To extract metabolites from blood samples, 250 μL of ice-cold methanol was added to 10 μL of plasma, stirred in a vortex mixer, and then centrifuged (14,000 × g, 4 °C, 15 min). The supernatant was then collected in a 1.5 mL Eppendorf microtube, diluted with an equal volume of water, and used for the analysis of the following compounds.

(2) LC/MS analysis of water-soluble compounds, including amino acids. [2] 25 μL of the supernatant obtained in (1) was diluted 3-fold with 0.1% formic acid solution and then analyzed by LC/MS. LC/MS analysis was performed using an LCMS-8060 (Shimadzu Corporation). To analyze a variety of water-soluble compounds, an HS-F5-3 column (150 × 2.1 mm, 3 μm particle size, Sigma-Aldrich) was used with a mobile phase consisting of solvent A (0.1% formic acid in water) and solvent B (0.1% formic acid in acetonitrile). The column oven temperature was set at 40°C. The mobile phase gradient was as follows:
・Flow rate 0.25ml/min. 0-2 min, 0%B; 2-5min, 0-25%B; 5-11min, 25-35%B; 11-15min, 35-95%B; 15-25min, 95%B; 25.1-30min, 0%B.
Mass spectrometry was performed in multiple reaction monitoring mode using electrospray ionization (ESI) in negative/positive ion mode under the following ESI conditions:
Dry gas flow rate 10 L/min, nebulizer gas flow rate 3 L/min, heat gas flow rate 10 L/min, interface temperature 300°C, DL temperature 250°C, heat block temperature 400°C, CID gas pressure 270 kPa.

(3) LC/MS Analysis of Acylcarnitines [2] 270 μL of ice-cold methanol was added to 30 μL of the supernatant obtained in (1), and the mixture was vortex-mixed and centrifuged (14,000 × g, 4°C, 15 minutes). The supernatant was then collected. LC/MS analysis was performed on the collected supernatant. LC/MS analysis was performed using an LCMS-8060 (Shimadzu Corporation). The column used was a Luna HILIC column 200A (150 mm × 2.0 mm, 3 μm, Phenomenex) with a mobile phase consisting of solvent A (10 mM ammonium formate) and solvent B (acetonitrile/10 mM ammonium formate (9/1, v/v)). The column oven temperature was set at 40°C. The mobile phase gradient was as follows:
・Flow rate 0.3mL/min. 0-2.5 min, 100%B; 2.5-4 min, 100-50%B; 4-7.5 min, 50-5%B; 7.5-10 min, 5%B; 10,1-12.5 min, 100%B.
Acylcarnitines were detected by precursor ion scanning at m/z 85.5 in ESI positive mode, with the collision energy (CE) adjusted according to the fatty acid chain length.
Short-chain acylcarnitine (C0-8): 20 eV
・Medium-chain acylcarnitine (C9-12): 35 eV
- Long-chain acylcarnitine (C12-18): 45 eV

(4) LC/MS Analysis of Phospholipids [2] 5 μL of the supernatant obtained in (1) was diluted 200-fold with a 20% solution of 0.1% formic acid in acetonitrile, followed by LC/MS analysis. LC/MS analysis was performed using an LCMS-8060 (Shimadzu Corporation). The column used was a Kinetex C8 column (150 × 2.1 mm, 2.6 μm particle size, Phenomenex Corporation), with a mobile phase consisting of solvent A (20 mM ammonium formate) and solvent B (acetonitrile/isopropanol (1/1, v/v)). The column oven temperature was set at 45°C. The mobile phase gradient was as follows:
・Flow rate 0.3mL/min. 0-1min, 20%B; 1-2 min, 40%B; 2-25 min, 92.5%B; 25.1-35 min, 100%B; 35.1-38 min,20%B.
Mass spectrometry was performed in electrospray ionization (ESI) positive ion mode in multiple reaction monitoring mode.
[3] Metabolite data processing
LC-MS data analysis was performed using the Lab Solutions LC-MS software program (Shimadzu Corporation). A total of 127 components (10 biochemical test items and 117 blood metabolites) were analyzed, and the data were compared between hikikomori and non-hikikomori subjects. The results are shown in Table 3 and Figure 14. Statistical analysis was performed using GraphPad Prism 8, and scatter plots were created using the Enhanced Volcano R package (ver. 1.6.0). The most significant difference between hikikomori and non-hikikomori subjects was the direct/indirect bilirubin ratio (R-Bil), which was lower in hikikomori subjects. Long-chain acylcarnitines (AC_16:0, AC_16:1, AC_18:1, AC_18:2) were higher in hikikomori subjects than in non-hikikomori subjects. The results for arginine were lower in hikikomori than in non-hikikomori, while the results for ornithine were higher in hikikomori than in non-hikikomori.

[4] Creation of a Hikikomori Identification Support Model Capable of Predicting the Hikikomori Status of Subjects Using these 127-component data, we created a random forest model for discriminating between hikikomori and non-hikikomori individuals. The random forest was created using the "Random Forest" function of the randomForest R package (ver. 4.6-14) and the "train" function of the caret R package (ver. 6.0-88) with the "mrty" parameter. All data was randomly divided into six equal parts, and five-sixths were used as training data to create the model, with one-sixth used as model validation data (Figure 15A). This random sampling procedure was repeated five times to create a model (ntree = 500; mtree = 5). The resulting classification model had a kappa coefficient of 0.7 or higher, demonstrating practical accuracy (Figure 15B). The area under the receiver operating characteristic (ROC) (AUC) was 0.854 (CI = 0.648-1.000) (Figure 15C). The performance of the prediction results was evaluated using the "auc" function of the pROC R package (ver. 1.17.0.1) or the "confusionMatrix" function of the caret R package.

Furthermore, the components that contributed most to the model are shown in Figure 15D. Ornithine, arginine, and long-chain acylcarnitines were among the top five components.

[Example 2] Blood arginase in male social withdrawal Because arginine decreases and ornithine increases in social withdrawal, and ornithine is produced from arginine through arginase, arginase can be considered as an indicator of social withdrawal, but when social withdrawal and non-social withdrawal are compared, blood arginase does not show significant difference.However, when analyzed by dividing into four groups according to gender and whether social withdrawal exists, it is found that blood arginase is high in male social withdrawal (Fig. 16C).Therefore, in men, arginine and ornithine show a negative correlation (Fig. 16D).

[Example 3] Prediction of the severity of hikikomori [1] Determination of the severity of hikikomori The severity of the subjects' hikikomori was assessed using the HQ-25 (25-item Hikikomori Questionnaire).
[2] Creation of a severity prediction support model capable of predicting the severity of a subject's hikikomori state. Using the data from the 127 components obtained in steps [1] to [3] of Example 1, a partial least squares (PLS) model was created to predict the severity of a subject's hikikomori state using the HQ-25 (Figure 17A). The partial least squares (PLS) model was created using the "plsr" or "mvr" function in the pls R package (ver. 2.7-3). The root mean square error (RMSE) of the model was 18.5, and the mean absolute error (MAR) was 15.5. Furthermore, the average correlation coefficient was 0.8. The root mean square error (RMSE) and mean absolute error (MAR) were calculated by applying the predicted results and the actual HQ-25 results to the following formula:

Figure 17B shows the components that contribute most to the model. Long-chain acylcarnitine, arginine, etc. are the top components. Long-chain acylcarnitine (AC_C16:0) showed a positive correlation with HQ-25, while arginine showed a negative correlation with HQ-25. These results also suggest that it is possible to predict the severity of hikikomori from information on blood components.

[Example 4] Stratification of Subjects Using the psychiatric test data of all subjects (data on the 23 items in Table 1), principal component clustering analysis was performed (Fig. 18A, B). As a result, the subjects were classified into three clusters.
・Cluster 1: A group consisting almost entirely of non-hikikomori.
・Cluster 2: A group consisting of half hikikomori and half non-hikikomori.
・Cluster 3: A group consisting almost entirely of hikikomori.

In Cluster 3, the depression indicators HAMD-17, PHQ-9, and BDI-II were higher than in Cluster 2. It is thought that people with a high tendency toward depression are classified in Cluster 3.

Based on these results, we searched for blood components that characterize the above clusters, and found that there were differences between each cluster in uric acid (UA) levels and cholesterol esters (CE_20:1, CE_18:3, CE_20:2) (Figures 19 and 20). The MetaboAnalyst 5.0 platform (https://www.metaboanalyst.ca/home.xhtml) was used to stratify hikikomori individuals, cluster them based on principal component analysis, and create heat maps.

1 Recognition support model generation system 1' Severity prediction support model generation system 1" Stratification support model generation system 2 Content measuring device 3 Teacher data generation device 3' Teacher data generation device 3" Teacher data generation device 30 Auxiliary storage device 31 Acquisition unit 32 Teacher data generation unit 32' Teacher data generation unit 32" Teacher data generation unit 4 Machine learning device 4' Machine learning device 4" Machine learning device 40 Auxiliary storage device 41 Learning unit 41' Learning unit 41" Learning unit 5 Support device 5' Support device 5" Support device 50 Auxiliary storage device 51 Acquisition unit 52 Prediction unit 52' Prediction unit 52" Prediction unit 6 Display device D1 Recognition result database D2 Teacher data D3 Severity judgment result database D4 Teacher data D5 Clustering classification judgment result database D6 Teacher data M1 Recognition support model M2 Severity prediction support model M3 Stratification support model

Claims (15)

A biomarker for identifying social withdrawal, comprising at least two members selected from the group consisting of long-chain acylcarnitine and ornithine, bilirubin, arginine, arginase, uric acid, and cholesterol ester. A biomarker for identifying social withdrawal, comprising long-chain acylcarnitine and at least three members selected from the group consisting of ornithine, bilirubin, arginine, arginase, uric acid and cholesterol ester. A method for using blood components containing long-chain acylcarnitine and at least one selected from the group consisting of ornithine, bilirubin, arginine, and arginase as an indicator for identifying whether or not a subject is in a state of social withdrawal. A method for determining the severity of a subject's social withdrawal state by using a blood component containing long-chain acylcarnitine and at least one selected from the group consisting of ornithine, bilirubin, arginine, and arginase as an indicator. A method for using a blood component containing at least one selected from the group consisting of uric acid and cholesterol ester as an index for stratifying a subject into a non-hikikomori group, a hikikomori group with depression tendencies, and other hikikomori groups, comprising:
A method for indicating that if the content of at least one selected from the group consisting of uric acid and cholesterol esters in a blood sample is above a standard value, the person is likely to be a hikikomori person with a tendency to depression. Obtaining the content of the biomarker for identifying hikikomori according to claim 1 or 2 in blood samples of a plurality of hikikomori individuals and non-hikikomori individuals;
generating training data by associating the content with a discrimination result of whether or not the subject is a hikikomori;
a step of performing machine learning using the training data to generate a hikikomori identification support model that outputs whether or not the person is a hikikomori when the amount of content is input;
A method for generating a hikikomori identification support model having the above. Obtaining the content of the biomarker for identifying hikikomori according to claim 1 or 2 in blood samples of a plurality of hikikomori individuals and non-hikikomori individuals;
A step of generating training data by correlating the content with the assessment result of the severity of social withdrawal;
A step of performing machine learning using the training data to generate a hikikomori severity prediction support model that outputs the severity of hikikomori when the content is input;
A method for generating a model for supporting prediction of the severity of social withdrawal. The method of claim 6 or 7, wherein the biomarkers for identifying hikikomori include at least two selected from the group consisting of ornithine, acylcarnitine, bilirubin, arginine, and arginase. Obtaining the content of the biomarker for identifying hikikomori according to claim 1 or 2 in blood samples of a plurality of hikikomori individuals and non-hikikomori individuals;
a step of generating training data by associating the content with a determination result of a clustering classification of hikikomori;
and generating a hikikomori stratification support model that outputs a clustering classification of hikikomori when the content is input by performing machine learning using the training data. The method of claim 9, wherein the biomarker for identifying hikikomori includes at least one selected from the group consisting of uric acid and cholesterol esters. The generation method described in claim 6 or 8, wherein the machine learning is machine learning using random forests. The generation method described in claim 7 or 8, wherein the machine learning is machine learning using partial least squares regression. Obtaining the content of the hikikomori identifying biomarker according to claim 1 or 2 in a blood sample of a subject;
a step of predicting whether the subject is a hikikomori by inputting the content of the subject into a hikikomori identification support model generated by the generation method according to any one of claims 6, 8, and 11;
A hikikomori identification support method comprising: Obtaining the content of the hikikomori identifying biomarker according to claim 1 or 2 in a blood sample of a subject;
A step of predicting the severity of social withdrawal of the subject by inputting the content of the subject into a social withdrawal severity prediction support model generated by the generation method according to any one of claims 7, 8, and 12;
A method for predicting the severity of social withdrawal. Obtaining the content of the hikikomori identifying biomarker according to claim 1 or 2 in a blood sample of a subject;
and predicting the clustering classification of hikikomori of the subject by inputting the content of the subject into a hikikomori stratification support model generated by the generation method of claim 9 or 10.