JP6876300B2 - Predictor - Google Patents

Description

The present invention relates generally to medical diagnostic techniques, and more specifically to predictors and methods for collecting information and predicting the onset of preterm birth in order to non-invasively predict the onset of preterm birth or detect abnormalities.

As a pregnancy-related disease, preterm birth is a typical perinatal medical disease in which the cause is unknown in about one-third, and it is a multifactorial disease that develops through complex interactions. The relationship between periodontitis in pregnant women and the birth of premature and low birth weight infants has been reported, but the detailed mechanism remains unclear.

According to Non-Patent Document 1, 124 pregnant women were measured for clinical attachment level (CAL) and Probing depth (PD) during pregnancy or within 3 days after delivery, and were PLBW (Preterm Low Birth Weight). When evaluated in the control group that did not, the CAL mean value of the case group (31 persons) was significantly larger than that of the control group. In pregnant women with CALs of 3 mm or more accounting for 60% or more, the risk of developing PLBW was 5.9 times overall and 6.7 times at the first delivery.

In addition, according to Non-Patent Document 2, a comparison of the plaque flora of the second trimester of 18 patients who had a preterm birth due to imminent preterm birth and 40 patients who did not have a term birth was one of the periodontal disease-related bacteria. The proportion of certain Tannerella forsythenesis was significantly higher. This Tannerella forsythenesis is one of three red complexes called bacteria related to periodontal disease.

Offenbacher S, et al, Periodontal Infection as a Possible Risk Factor for Preterm Low Birth Weight, J Peridontol, 67, 1103-1113, 1996 Hasegawa K, et al, Associations between systemic status, oeruidibtak statysm seryn cytokine levels and delivery outcomes in pregnant women with a diagnosis of threatened premature labor (TPL), J Peridontol, 74, 1764-1770, 2003

Starting with the report by Offenbacher et al., The association between periodontal disease and preterm birth / underweight has been reported all over the world, but there are racial differences in these relationships, and studies targeting pregnant Japanese women have been conducted. Few. According to the report by Hasegawa et al., Since only one point was measured in chronological order, the effect of changes over time on preterm birth is unknown. It also reports only relevance, but there is no assessment of how useful it is in predicting preterm birth.

In view of the above-mentioned problems, an object of the present invention is to provide a prediction device and a method for realizing prediction of the onset of preterm birth from the oral bacterial flora.

To solve the above problems, one aspect of the present invention includes a data acquisition function of acquiring flora data relating flora in the oral cavity of the pregnant women pregnant with preterm group full-term production group, based on the flora data The preterm birth prediction model generation function that generates a preterm birth prediction model, the preterm birth prediction function that executes the preterm birth prediction prediction of the pregnant woman to be predicted from the acquired bacterial flora data based on the preterm birth prediction model, and the result of the preterm birth onset prediction. It has a result notification function for notifying, and the data acquisition function obtains the bacterial flora data during a data acquisition period from any time before or after pregnancy to any time before or after childbirth of the pregnant woman to be predicted. The preterm birth prediction function, which is acquired at a plurality of time points, relates to a prediction device that executes the preterm birth onset prediction based on the bacterial flora data acquired at the plurality of time points.

According to the present invention, it is possible to provide a prediction device and a method for realizing prediction of the onset of preterm birth from the oral bacterial flora.

FIG. 1 is a schematic view showing a configuration of a prediction system according to an embodiment of the present invention. FIG. 2 is a block diagram showing a functional configuration of a prediction device according to an embodiment of the present invention. FIG. 3 is a flowchart showing a prediction process according to an embodiment of the present invention. FIG. 4 is a flowchart showing a feature amount extraction / model creation process according to an embodiment of the present invention. FIG. 5 is a flowchart showing a feature amount extraction / model creation process according to another embodiment of the present invention. FIG. 6 is a flowchart showing a feature amount extraction / model creation process according to another embodiment of the present invention.

Hereinafter, embodiments of the present invention will be described with reference to the drawings.

In the examples described later, a prediction device for predicting the onset of preterm birth is disclosed. According to the examples described below, when the bacterial flora data on the oral flora of pregnant women collected in multiple pregnancy phases is acquired, the predictor extracts the bacterial species that correlates with the onset of preterm birth and the extracted bacteria. Based on the preterm birth prediction model generated using seeds, the onset of preterm birth in pregnant women is predicted from the acquired bacterial flora data. This makes it possible to generate a preterm birth prediction model suitable for a specific race, as opposed to the preterm birth prediction from bacterial flora data, which is considered to change over time as well as being different between races.

First, a prediction system according to an embodiment of the present invention will be described with reference to FIG. FIG. 1 is a schematic view showing a configuration of a prediction system according to an embodiment of the present invention.

As shown in FIG. 1, the prediction system 10 includes a bacterial flora measuring device 50 and a predicting device 100.

The bacterial flora measuring device 50 measures the bacterial flora in the oral cavity from the collected plaque sample, and provides the generated bacterial flora data to the prediction device 100. Specifically, an inspection engineer collects a plaque sample of a pregnant woman, acquires bacterial flora data indicating the bacterial species present in the plaque sample collected using the bacterial flora measuring device 50, and obtains the acquired bacterial flora data. Provided to the prediction device 100.

More than 200 bacterial species present in the human oral cavity, including, for example, Porphyromonas gingivalis, Treponema denticola, Tabberella forsythensis, Bergeyella, Kingella denitrificans, Scardovia wiggsiae, Streptococus mutans, Selenomonas, Capnocytophaga, Treponema vincentii, etc. Known, especially Porphyromonas gingivalis, Treponema denticola, and Tabberella forsythensis are classified as a red complex related to periodontal disease, and their correlation with pregnancy-related diseases has been pointed out. The bacterial flora measuring device 50 may be any known one used for measuring the bacterial species in the oral cavity.

The prediction device 100 predicts the onset of preterm birth from the provided bacterial flora data. Specifically, the prediction device 100 creates a preterm birth prediction model from known data including preterm birth group bacterial flora data and non-preterm birth group or term birth group bacterial flora data, as described in detail below. Then, by inputting the bacterial flora data of the pregnant woman to be predicted into the created preterm birth prediction model, the presence or absence of the onset of preterm birth of the pregnant woman is predicted. The prediction of the onset of preterm birth according to the present invention is not limited to the presence or absence of the onset of preterm birth, and may include prediction of the date of birth, calculation of the risk of preterm birth, and the like, as will be described later.

Here, the prediction device 100 may be typically realized by an information processing device having a communication function such as a computer, a server, a smartphone, or a tablet. For example, a processor mounted on an information processing device executes various functions and processes described later by processing and executing data and programs stored in the memory device. However, the prediction device 100 is not limited to any specific hardware configuration, and may be realized by an appropriate hardware configuration.

Next, a prediction device according to an embodiment of the present invention will be described with reference to FIG. FIG. 2 is a block diagram showing a functional configuration of a prediction device according to an embodiment of the present invention.

As shown in FIG. 2, the prediction device 100 has a data acquisition function 110, a data storage function 120, a premature birth prediction model generation function 130, a premature birth prediction function 140, and a result notification function 150.

The data acquisition function 110 acquires bacterial flora data from the bacterial flora measuring device 50. Specifically, the data acquisition function 110 acquires bacterial flora data related to a pregnant woman at a plurality of time points during a data acquisition period from any time before and after pregnancy of a pregnant woman to any time before and after childbirth. For example, the data acquisition function 110 may acquire the bacterial flora data in the oral cavity of a pregnant woman collected in different pregnancy phases (for example, 12 to 23 weeks and 24 to 34 weeks) from the bacterial flora measuring device 50.

The data storage function 120 stores the acquired bacterial flora data in the oral cavity of a pregnant woman. Specifically, the data storage function 120 stores the acquired bacterial flora data as independent variables for each of the bacterial species (for example, 200 bacterial species) collected in each pregnancy phase. For example, when 200 bacterial species are collected in each of the two pregnancy phases, 400 (200 bacterial species x 2) can be created so that a model that takes into account changes in the composition of dental plaque bacteria at multiple time points over time can be created. The variable of pregnancy phase) may be set.

The preterm birth prediction model generation function 130 generates a preterm birth prediction model that predicts the presence or absence of the onset of preterm birth based on the acquired bacterial flora data in the oral cavity of the pregnant woman. Specifically, the preterm birth prediction model generation function 130 is a bacterial flora in the oral cavity acquired from a plurality of pregnant women including a pregnant woman who gave birth by term birth (term birth group) and a pregnant woman who gave birth by preterm birth (preterm birth group). Preterm birth prediction models may be generated and updated based on the data. For example, as described in detail below, the preterm birth prediction model generation function 130 identifies the bacterial species that distinguishes the preterm birth group and the term birth group from the acquired bacterial flora data, and reduces the variables. 10 to 20 bacterial species x pregnancy phases may be extracted as bacterial species that divide the group well. Porphyromonas gingivalis, Treponema denticola, Tabberella forsythensis, Bergeyella, Kingella denitrificans, Scardovia wiggsiae, Streptococus mutans, Selenomonas, Capnocytophaga, Treponema vinsis, etc. It may be extracted.

Subsequently, the preterm birth prediction model generation function 130 creates a preterm birth prediction model using the extracted bacterial species as a feature amount. As a specific method for creating a prediction model, for example, the preterm birth prediction model generation function 130 may use Linear SVM to select a bacterial species to be used as a feature by forward stepwise selection. For example, for the first 20 features (bacterial species for each pregnancy phase), the preterm birth prediction model generation function 130 creates preterm birth prediction models M (1) to M (20) using only one feature amount. , Identify the best performing M (A) (using feature A). Next, the preterm birth prediction model generation function 130 selects one feature amount from the remaining 19 feature amounts excluding the feature amount and combines it with A, which is a two-variable prediction model M (A, 1) to M. Create (A, 19) and select the optimal model including M (A). The premature birth prediction model generation function 130 recursively executes the above-mentioned series of processes, and finally identifies the prediction model M (A, B, C, ...).

In the above-mentioned examples, the preterm birth prediction model showing the onset of preterm birth has been described, but the preterm birth prediction model according to the present invention is not limited to this, and the birth date prediction and the preterm birth risk model for predicting the birth date are not limited to this. Alternatively, it may be a preterm birth risk model that predicts the onset probability.

The preterm birth prediction function 140 executes the prediction of the onset of preterm birth of the pregnant woman from the acquired bacterial flora data of the pregnant woman based on the preterm birth prediction model. Specifically, the preterm birth prediction function 140 inputs the pregnant woman's bacterial flora data acquired at a plurality of time points into the preterm birth prediction model, and based on the output result of the preterm birth prediction model based on the bacterial flora data acquired at the plurality of time points. Preterm birth prediction, birth date prediction, or preterm birth risk calculation for pregnant women may be performed. That is, the preterm birth prediction function 140 predicts whether the pregnant woman has a possibility of preterm birth (preterm birth onset prediction), predicts the birth date of the pregnant woman (birth date prediction), or predicts the preterm birth onset of the pregnant woman. The risk may be calculated (preterm birth risk calculation).

The result notification function 150 notifies pregnant women, medical personnel, and the like of the result of the premature birth prediction executed by the premature birth prediction function 140.

Next, the prediction process according to the embodiment of the present invention will be described with reference to FIGS. 3 to 6. The prediction process is executed by the prediction device 100 described above. FIG. 3 is a flowchart showing a prediction process according to an embodiment of the present invention.

As shown in FIG. 3, in step S100, the prediction device 100 acquires the bacterial flora data from the bacterial flora measuring device 50 and stores the acquired bacterial flora data. Specifically, the bacterial flora measuring device 50 measures a plaque sample collected from a pregnant woman, and provides the predictive device 100 with bacterial flora data indicating the amount of bacterial species contained in the plaque sample. The prediction device 100 stores the bacterial flora data acquired from the bacterial flora measuring device 50 for each pregnant woman. According to this example, bacterial flora data is collected multiple times for each pregnant woman during the data acquisition period from any time before or after pregnancy to any time before and after childbirth, for example, for each pregnancy phase.

In step s101, the prediction device 100 detects and removes abnormal values from the acquired bacterial flora data and supplements the missing values. For example, when the amount of bacterial species outside the threshold range is measured from the reference value, the prediction device 100 considers the measured value as a measurement error instead of an actual value, removes the measured value, and removes the measured value, and other bacterial species. Random noise is added to the measured value of another toothpaste sample with similar amount values and used, or multiple other toothpaste samples with similar amount values are extracted and measured. The measured value may be supplemented by any method such as calculating one estimated value from the distribution of values so as to follow the distribution. For example, when the bacterial species spA-2 of sample S is out of the threshold range or is a missing value, the variance V of the bacterial species amount of spA-2 is first calculated in the sample group other than sample S. Next, a sample in which the amount of each bacterial species in _{the bacterial species group B SUB} , which is obtained by removing spA-2 from the bacterial species group B _ALL = {spA-1, spB-1,…, spA-2, spB-2}, is close to that of the sample S. There is a method in which one sample is extracted and the value obtained by adding one random point according to the normal distribution N (0, V) to the amount of bacterial species of spA-2 is used as the estimated amount of spA-2 of sample S. At this time, in order to extract a sample close to the sample S, the Eugrid distance may be calculated by dividing each bacterial species amount by the standard deviation of each bacterial species amount and using a standardized value, or MSE (mean square error). ) Or R ² (coefficient of determination) or the like may be used. As another complementary method, for example, only 10 samples in which the amount of each bacterial species in the bacterial species group B _SUB is close to that of the sample S are extracted, the average value E2 and the dispersion V2 are obtained, and the random distribution follows the normal distribution N (E2, V2). There is a method of using one point as the estimated amount of spA-2 of the sample S. Moreover, the distribution is not necessarily a normal distribution, and may be estimated by any distribution.

In step s102, the prediction device 100 uses data of pregnant women who have already given birth and whose delivery status has been determined, and extracts variables that often separate the preterm birth group and the term birth group as feature quantities. This includes limiting variables using Student's t-test, Welch's t-test, multivariate logistic regression, chi-square test, Lasso regression, Ridge regression, Elastic Net, etc. There are methods for limiting variables based on the contribution rate of principal component analysis, but they are not limited to these.

In steps s103, s105 and s107, the prediction device 100 generates a preterm birth prediction model using the extracted features. Specifically, in step s103, the prediction device 100 generates a premature birth onset prediction model as a classification model for predicting whether or not premature birth will occur. Further, in step s105, the prediction device 100 generates a birth date prediction model as a regression model for predicting the birth date of each pregnant woman. In addition, the prediction device 100 generates a preterm birth risk calculation model for calculating the risk of preterm birth in s107. Regression models include, but are not limited to, linear regression models, logistic regression models, regression models using SVMs, and regression models using multi-layer neural networks. In addition, classification models include, but are not limited to, linear classification, logistic regression, Bayesian network, SVM, k-nearest neighbor method, random forest, and multi-layer neural network classification model. The model for calculating the risk of preterm birth includes, but is not limited to, odds ratio, attribution risk, and relative risk. Further, steps s103, s105, and s107 may be processed integrally without being separated from s102.

In steps s104, s106 and s108, the prediction device 100 inputs the bacterial flora data of the pregnant woman to be predicted into the generated preterm birth onset prediction model, birth date prediction model and preterm birth onset risk calculation model, respectively, and executes the preterm birth prediction. .. Specifically, in step s104, the prediction device 100 predicts the presence or absence of future preterm birth from the bacterial flora data of the pregnant woman to be predicted by using the preterm birth onset prediction model. Further, in step s106, the prediction device 100 predicts the birth date from the bacterial flora data of the pregnant woman to be predicted by using the birth date prediction model. Further, in step s108, the prediction device 100 calculates the risk of preterm birth from the bacterial flora data of the pregnant woman to be predicted by using the preterm birth risk calculation model. The prediction of the presence or absence of preterm birth is the final prediction of the presence or absence of preterm birth, the prediction of not developing preterm birth in the next n1 weeks (n1 is an arbitrary value), the prediction of developing preterm birth in the next n2 weeks (n2 is an arbitrary value), Etc., but not limited to them.

In step s109, the prediction device 100 notifies the pregnant woman or the medical institution of the preterm birth onset prediction result, the birth date prediction result and / or the preterm birth onset risk calculation result, and ends the prediction process.

FIG. 4 is a flowchart showing a feature amount extraction / model creation process according to an embodiment of the present invention. The feature amount extraction / model creation process can be executed in steps s102 and s103 of FIG. 3 described above, and steps s102-1 to 5 described later correspond to the feature amount extraction process of step s102, and step s103. -1 to 3 correspond to the generation process of the preterm birth onset prediction model in step s103. In the feature amount extraction / model creation process, the preterm birth prediction model generation function 130 divides the acquired data into a preterm birth group and a term birth group, and divides (that is, distinguishes, discriminates) two groups of each variable well. , Separation, etc.) Variables, for example, a predetermined number of variables that divide the two groups well are extracted as features, a preterm birth prediction model is created using the extracted features, and preterm birth is based on the best-performing features. Identify the predictive model. Next, the preterm birth prediction model generation function 130 creates a two-variable prediction model that combines the remaining feature amount excluding the feature amount and the specified feature amount, and includes these models to obtain the optimum model. select. The premature birth prediction model generation function 130 recursively executes the above-mentioned series of processes, and finally identifies the premature birth prediction model. The above is an example of a method for extracting features used in a model for predicting the presence or absence of preterm birth and a method for selecting a model for predicting the presence or absence of preterm birth, and the model for predicting the presence or absence of preterm birth is not limited thereto. For example, there is also a method of extracting features by a multi-layer neural network or Lasso regression to generate a model for predicting the presence or absence of preterm birth.

As shown in FIG. 4, in step s102-1, the predictor 100 processes the acquired bacterial flora data. Specifically, the prediction device 100 may standardize the measured amount of the bacterial species to a relative amount, or may delete the bacterial species having a small relative amount, for example.

In step s102-2, the predictor 100 extracts bacterial flora data of a pregnant woman who has already given birth for model generation.

In step s102-3, the prediction device 100 stores each bacterial species collected in each pregnancy phase as an independent variable for the bacterial flora data of the plaque sample collected in each pregnancy phase. For example, when 200 bacterial species are collected in each of the two pregnancy phases, 400 variables (200 bacterial species x 2 pregnancy phases) are set. For example, 400 data are represented by a variable sequence of "spA-1, spB-1, ..., spA-2, spB-2, ..." in association with the ID (userid) of the pregnant woman. You may.

In step S102-4, the predictor 100 divides the bacterial flora data into the preterm birth group and the term birth group, and divides the bacteria well between the two groups whose p-value is less than a (for example, a = 0.01) in the test. Extract the seeds.

In step S102-5, the prediction device 100 sorts the extracted bacterial species in ascending order of p-value, and uses these as feature quantities.

In step S103-1, the prediction device 100 divides the bacterial flora data into the preterm production group D1 and the term production group D2, and evaluates the preterm production group D1 as learning data D1a and evaluation at each ratio e (for example, e = 0.9). The term data D1b is divided, and the term production group D2 is divided into learning data D2a and evaluation data D2b.

In step S103-2, the prediction device 100 may oversample D1a in the training data D1a, D2a and / or undersample D2a in order to improve the prediction performance.

In step S103-3, the prediction device 100 selects the feature amount by forward stepwise selection using the training data D1a and D2a, and creates the preterm birth prediction model M. For example, for the first 20 features (bacterial species for each pregnancy phase), the prediction device 100 creates preterm birth prediction models M (1) to M (20) using only one feature. Then, the prediction device 100 identifies M (A) (using the feature amount A) having the best performance by using the evaluation data D1b and D2b. As the performance score, for example, AUC (Area under the curve) or f-score may be used. Next, the prediction device 100 uses the training data D1a and D2a to select one feature amount from the remaining 19 feature amounts excluding the feature amount and combine it with A as a two-variable premature delivery prediction model. Create M (A, 1) to M (A, 19), and use the evaluation data D1b and D2b to select the optimum model including M (A). The prediction device 100 recursively executes the series of processes described above, and finally identifies the premature birth prediction model M (A, B, C, ...).

The preterm birth prediction model M may be generated based on SVM, linear logistic regression, neural network, random forest, or the like. Further, the performance score is not limited to AUC or f-score, and a harmonic mean of sensitivity and specificity may be used. Further, the method of selecting the optimum feature amount is not limited to forward step selection, and backward stepwise selection or bidirectional stepwise selection may be used. In addition, after using all bacterial species and reducing the dimensions with PCA, MDS, NMDS, etc., apply forward / backward / bidirectional stepwise selection to N features of PC1, PC2, ..., PCN. May be good.

FIG. 5 is a flowchart showing a feature amount extraction / model creation process according to another embodiment of the present invention. The feature amount extraction / model creation process can be executed in steps s102 and s105 of FIG. 3 described above, and steps s102-1 to 5 described later correspond to the feature amount extraction process of step s102, and step s105. -1 to 3 correspond to the generation process of the birth date prediction model in step s105. Since steps S102-1 to 5 are the same as steps S102-1 to 5 in FIG. 4, the description thereof will be omitted.

In step S105-1, the prediction device 100 divides the bacterial flora data into the preterm production group D1 and the term production group D2, and evaluates the preterm production group D1 as learning data D1a and evaluation at each ratio e (for example, e = 0.9). The term data D1b is divided, and the term production group D2 is divided into learning data D2a and evaluation data D2b.

In step S105-2, the prediction device 100 may oversample D1a in the training data D1a, D2a and / or undersample D2a in order to improve the prediction performance.

In step S105-3, the prediction device 100 selects the feature amount by forward stepwise selection using the learning data D1a and D2a, and creates the birth date prediction model M. For example, first, for 20 feature amounts (bacterial species for each pregnancy phase), the prediction device 100 creates birth date prediction models M (1) to M (20) using only one feature amount. Then, the prediction device 100 identifies M (A) (using the feature amount A) having the best performance by using the evaluation data D1b and D2b. For example, MSE (mean square error) or R ² (coefficient of determination) may be used as the performance score. Next, the prediction device 100 uses the training data D1a and D2a to select one feature amount from the remaining 19 feature amounts excluding the feature amount and combine it with A to predict the birth date of two variables. Models M (A, 1) to M (A, 19) are created, and the optimum model including M (A) is selected using the evaluation data D1b and D2b. The prediction device 100 recursively executes the series of processes described above, and finally identifies the birth date prediction model M (A, B, C, ...).

The birth date prediction model M may be generated based on SVM, linear logistic regression, neural network, or the like. Also, the performance score does not have to be limited to MSE (mean square error) or R ^{2 (coefficient of determination).} Further, the method of selecting the optimum feature amount is not limited to forward step selection, and backward stepwise selection or bidirectional stepwise selection may be used. In addition, after using all bacterial species and reducing the dimensions with PCA, MDS, NMDS, etc., apply forward / backward / bidirectional stepwise selection to N features of PC1, PC2, ..., PCN. May be good.

FIG. 6 is a flowchart showing a feature amount extraction / model creation process according to another embodiment of the present invention. The feature amount extraction / model creation process can be executed in steps s102 and s107 of FIG. 3 described above, and steps s102-1 to 5 described later correspond to the feature amount extraction process of step s102, and step s107. -1 to 2 correspond to the process of generating the preterm birth risk model in step s107. Since steps S102-1 to 5 are the same as steps S102-1 to 5 in FIG. 4, the description thereof will be omitted.

_{In step S107-1, the predictor 100 determines the amount threshold range S spA-1} = [for each bacterial species (spA-1, spB-1, ..., spA-2, spB-2, ...). Set A _spA-1 , B _spa-1 ] and calculate the risk ratio (RR _{spA-1) to preterm birth.} Here, the risk ratio is RR = a (c + d) / c (a + b), where a is the number of plaque samples in the preterm birth group (the amount of bacterial species) within the threshold range. b is the number of plaque samples in the term birth group and within the threshold range, c is the number of plaque samples in the preterm birth group and outside the threshold range, and d is the term birth group and threshold. The number of plaque samples that are out of range.

In step S107-2, when the amount of bacterial species of the pregnant woman to be predicted is (x _spA-1 , x _spB-1 , ...), the prediction device 100
RR = Π _sp RR _sp ^{f (x} _sp ⁾
f (x _sp ) = 1 if x _sp is in S _sp
f (x _sp ) = -1 if x _sp is not in S _sp
May generate a preterm birth risk prediction model.

According to the results of the above example, Porphyromonas gingivalis, Treponema denticola, Tabberella forsythensis, Bergeyella, Kingella denitrificans, Scardovia wiggsiae, Streptococus mutans, Selenomonas, Capnocytophaga, Treponema The performance of the preterm birth onset prediction model using these as feature quantities was 0.714 for sensitivity and 0.751 for specificity.

Although the examples of the present invention have been described in detail above, the present invention is not limited to the above-mentioned specific embodiments, and various modifications are made within the scope of the gist of the present invention described in the claims.・ Can be changed.

10 Prediction system 50 Bacterial flora measurement device 100 Prediction device 110 Data acquisition function 120 Data storage function 130 Preterm birth prediction model generation function 140 Preterm birth prediction function 150 Result notification function

Claims (4)

A data acquisition function of acquiring flora data relating flora in the oral cavity of the pregnant full-term production group and preterm group pregnant women,
A preterm birth prediction model generation function that generates a preterm birth prediction model based on the bacterial flora data,
Based on the preterm birth prediction model, the preterm birth prediction function that executes the prediction of the onset of preterm birth of the pregnant woman to be predicted from the acquired bacterial flora data, and
A result notification function that notifies the result of the prediction of the onset of preterm birth,
Have,
The data acquisition function acquires the bacterial flora data at a plurality of time points during the data acquisition period from any time before and after pregnancy to any time before and after childbirth of the pregnant woman to be predicted.
The preterm birth prediction function is a prediction device that executes the preterm birth onset prediction based on the bacterial flora data acquired at the plurality of time points. The preterm birth prediction model generation function sets each bacterial species of the bacterial flora data acquired at the plurality of time points as independent variables, and extracts a predetermined number of high-ranking variables that often separate the preterm birth group and the term birth group. And
The prediction device according to claim 1, wherein the premature birth prediction model generation function generates the premature birth prediction model from the extracted variables according to a forward stepwise selection method. The Preterm Prediction model is, Porphyromonas gingivalis, Treponema denticola, and Tabberella forsythensis, Bergeyella, Kingella denitrificans, Scardovia wiggsiae, Streptococus mutans, Selenomonas, Capnocytophaga, is generated so as to extract one or more of Treponema vincentii, claim 1 or 2. The prediction device according to 2. The Preterm Prediction model generation function generates a premature onset whether prediction model及beauty preterm risk model,
The Preterm Prediction, based respectively on the generated preterm onset whether prediction model及beauty preterm risk model, executes the onset whether pre Haka及beauty premature risk calculating preterm claims 1 to 3 set forth in any one Predictor.

Images