WO2023139718A1 - 特徴量選定装置、特徴量選定方法、身体状態推定システム、および記録媒体 - Google Patents
特徴量選定装置、特徴量選定方法、身体状態推定システム、および記録媒体 Download PDFInfo
- Publication number
- WO2023139718A1 WO2023139718A1 PCT/JP2022/001953 JP2022001953W WO2023139718A1 WO 2023139718 A1 WO2023139718 A1 WO 2023139718A1 JP 2022001953 W JP2022001953 W JP 2022001953W WO 2023139718 A1 WO2023139718 A1 WO 2023139718A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- unit
- feature
- data
- values
- data sets
- Prior art date
Links
- 238000010187 selection method Methods 0.000 title claims 2
- 238000000034 method Methods 0.000 claims abstract description 93
- 238000004458 analytical method Methods 0.000 claims abstract description 45
- 238000010276 construction Methods 0.000 claims abstract description 34
- 230000002776 aggregation Effects 0.000 claims abstract description 6
- 238000004220 aggregation Methods 0.000 claims abstract description 6
- 239000011159 matrix material Substances 0.000 claims description 90
- 238000012545 processing Methods 0.000 claims description 50
- 230000008569 process Effects 0.000 claims description 34
- 238000011156 evaluation Methods 0.000 claims description 16
- 230000004931 aggregating effect Effects 0.000 claims description 5
- 230000005021 gait Effects 0.000 description 95
- 230000001133 acceleration Effects 0.000 description 76
- 238000010586 diagram Methods 0.000 description 46
- 238000010606 normalization Methods 0.000 description 30
- 238000004891 communication Methods 0.000 description 22
- 239000000284 extract Substances 0.000 description 22
- 238000005259 measurement Methods 0.000 description 14
- 230000006870 function Effects 0.000 description 12
- 230000010365 information processing Effects 0.000 description 12
- 239000013598 vector Substances 0.000 description 12
- 238000000605 extraction Methods 0.000 description 8
- 210000002414 leg Anatomy 0.000 description 8
- 210000003205 muscle Anatomy 0.000 description 7
- 238000004422 calculation algorithm Methods 0.000 description 6
- 238000010988 intraclass correlation coefficient Methods 0.000 description 6
- 238000004364 calculation method Methods 0.000 description 4
- 238000012937 correction Methods 0.000 description 4
- 239000000203 mixture Substances 0.000 description 4
- 238000013459 approach Methods 0.000 description 3
- 239000000470 constituent Substances 0.000 description 3
- 230000037230 mobility Effects 0.000 description 3
- 210000002303 tibia Anatomy 0.000 description 3
- 238000012549 training Methods 0.000 description 3
- 230000001364 causal effect Effects 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 230000000052 comparative effect Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000012417 linear regression Methods 0.000 description 2
- 238000000691 measurement method Methods 0.000 description 2
- 238000007637 random forest analysis Methods 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 230000003068 static effect Effects 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 230000005856 abnormality Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 239000003814 drug Substances 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000012854 evaluation process Methods 0.000 description 1
- 230000007717 exclusion Effects 0.000 description 1
- 230000005484 gravity Effects 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 210000003127 knee Anatomy 0.000 description 1
- 210000003141 lower extremity Anatomy 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 230000001404 mediated effect Effects 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 230000017105 transposition Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
Definitions
- the present disclosure relates to a feature quantity selection device or the like that selects feature quantities used for estimation.
- IoT Internet of Things
- IoT devices Internet of Things
- machine learning By applying machine learning to the information collected by IoT devices, that information can be used for applications such as estimating physical conditions.
- IoT devices are often placed in places where power supply is difficult, advanced power saving is required.
- the proportion of power consumed for communication is large. For example, if it is possible to reduce the amount of information of the feature amount used for estimating the physical state, the amount of information transmitted from the IoT device can be reduced, and the power consumption of the IoT device can be reduced.
- Patent Document 1 discloses a technique for reducing data that has a weak causal relationship with device abnormality prediction from sensor data collected in factories and the like.
- Patent Literature 1 discloses a technique of reducing data with sparse causal relationships by utilizing a plurality of sparse estimation techniques.
- Patent Document 1 mentions LASSO (Least Absolute Shrinkage and Selection Operator) regression (also called Lasso regression) as an example of the sparse estimation method.
- LASSO Least Absolute Shrinkage and Selection Operator
- a model that applies a plurality of sparse estimation methods and outputs a predetermined output learns the degree of application of each of the sparse estimation methods.
- an appropriate sparsity estimation technique can be selected according to the problem, and the degree of sparsity can be adjusted for the selected sparsity estimation technique.
- An object of the present disclosure is to provide a feature value selection device and the like that can select feature values that are highly robust against outliers and outliers.
- a feature amount selection device includes an acquisition unit that acquires multiple data sets, a building unit that builds multiple re-extracted data sets by changing the distribution of data included in the data sets, an analysis unit that analyzes the multiple re-extracted data sets using the Lasso regression method, and aggregating the values of elements included in the multiple re-extracting data sets according to the analysis results of the multiple re-extracting data sets, and setting logical values to the elements included in the multiple re-extracting data sets according to the aggregation results of the element values.
- a statistical unit a selection unit that selects a combination of feature amounts according to a logical value set for the element according to a preset specific rule, and an output unit that outputs selection information about the combination of the selected feature amounts.
- a plurality of data sets are acquired, the distribution of data included in the data sets is changed to construct a plurality of re-extracted data sets, the plurality of re-extracted data sets are analyzed using the Lasso regression technique, the values of elements included in the plurality of re-extracted data sets are aggregated according to the analysis results of the plurality of re-extracted data sets, the logical values are set for the elements included in the multiple re-extracted data sets according to the aggregation results of the element values, and a predetermined specific rule is set.
- a combination of feature amounts is selected according to the value of the logical value set to the element, and selection information regarding the combination of the selected feature amounts is output.
- a program includes a process of acquiring multiple data sets, a process of changing the distribution of data contained in the data sets to construct multiple re-extracted data sets, a process of analyzing the multiple re-extracted data sets using the Lasso regression technique, a process of aggregating the values of elements included in the multiple re-extracted data sets according to the analysis results of the multiple re-extracted data sets, and a process of setting logical values to the elements contained in the multiple re-extracted data sets according to the results of the aggregation of the values of the elements.
- a computer is caused to execute a process of selecting a combination of feature amounts corresponding to the value of a logical value set to an element and a process of outputting selection information about the combination of the selected feature amounts.
- a feature quantity selection device or the like that can select a feature quantity that is highly robust against outliers and outliers.
- FIG. 1 is a block diagram showing an example of the configuration of a feature selection device according to a first embodiment
- FIG. FIG. 4 is a conceptual diagram for explaining a first matrix generated by the feature quantity selection device according to the first embodiment
- FIG. 4 is a conceptual diagram for explaining a plurality of patterns of first matrices generated by the feature quantity selection device according to the first embodiment
- FIG. 4 is a conceptual diagram for explaining aggregated values of cells of a first matrix of a plurality of patterns generated by the feature quantity selection device according to the first embodiment
- FIG. 4 is a conceptual diagram for explaining a second matrix generated by the feature quantity selection device according to the first embodiment
- This is an estimation example using an estimation model generated using feature values selected by a general Lasso regression method.
- 7 is a graph for explaining the effects of outliers and outliers that may be included in sensor data measured for multiple subjects.
- 4 is a flowchart for explaining an example of the operation of the feature selection device according to the first embodiment; 4 is a flowchart for explaining an example of the operation of the feature selection device according to the first embodiment; It is a block diagram showing an example of a configuration of a feature amount selection device according to a second embodiment.
- FIG. 12 is a block diagram showing an example of the configuration of a feature selection device according to a fourth embodiment;
- FIG. FIG. 21 is a block diagram showing an example of the configuration of a learning system according to a fifth embodiment;
- FIG. FIG. 12 is a block diagram showing an example of a configuration of a learning device included in a learning system according to a fifth embodiment;
- FIG. FIG. 11 is a conceptual diagram for explaining an example of learning by a learning device included in a learning system according to a fifth embodiment;
- FIG. 21 is a block diagram showing an example of the configuration of a physical condition estimation system according to a sixth embodiment;
- FIG. FIG. 11 is a block diagram showing an example of the configuration of a gait measuring device included in a physical state estimation system according to a sixth embodiment;
- FIG. 12 is a conceptual diagram for explaining an example of arrangement of gait measuring devices included in the physical state estimation system according to the sixth embodiment
- FIG. 11 is a conceptual diagram for explaining a coordinate system set in a gait measuring device included in a physical state estimation system according to a sixth embodiment
- FIG. 12 is a conceptual diagram for explaining a human body plane used in the explanation of the gait measuring device included in the physical state estimation system according to the sixth embodiment
- FIG. 11 is a conceptual diagram for explaining a walking cycle used in explaining a gait measuring device included in a physical state estimation system according to a sixth embodiment
- FIG. 14 is a graph for explaining an example of time-series data of sensor data measured by a gait measuring device included in the physical state estimation system according to the sixth embodiment
- FIG. 20 is a diagram for explaining an example of normalization of walking waveform data extracted from time-series data of sensor data measured by a gait measuring device included in the physical state estimation system according to the sixth embodiment
- FIG. 11 is a conceptual diagram for explaining an example of a walking phase cluster from which feature amounts are extracted by a feature amount data generation unit of a gait measuring device included in a physical state estimation system according to a sixth embodiment
- FIG. 12 is a block diagram showing an example of the configuration of an estimation device included in a physical state estimation system according to a sixth embodiment
- FIG. FIG. 11 is a block diagram showing an example of estimation of a physical condition score by an estimating device included in a physical condition estimating system according to a sixth embodiment
- FIG. 16 is a flowchart for explaining an example of the operation of a gait measuring device included in the physical state estimation system according to the sixth embodiment;
- FIG. 16 is a flowchart for explaining an example of the operation of an estimation device included in the physical state estimation system according to the sixth embodiment;
- FIG. 12 is a conceptual diagram for explaining an application example of the physical condition estimation system according to the sixth embodiment; It is a block diagram showing an example of hardware constitutions which perform processing concerning each embodiment.
- the feature amount selection device of the present embodiment uses LASSO (Least Absolute Shrinkage and Selection Operator) regression (hereinafter referred to as Lasso regression) to select feature amounts to be used for estimating a physical state or the like. Lasso regression is also called L1 regularization.
- LASSO Least Absolute Shrinkage and Selection Operator
- the feature quantity used for estimating the physical state is extracted based on sensor data relating to the movement of the user's legs as they walk.
- sensor data related to foot movement is measured by a measurement device installed on the footwear.
- the measuring device includes an acceleration sensor and an angular velocity sensor.
- sensor data is not limited to sensor data relating to leg movements, and may include features relating to gait.
- the sensor data may be sensor data including features related to gait that are measured using motion capture, smart apparel, or the like.
- the following method can be applied not only to the selection of feature amounts related to gaits, but also to the use of selecting feature amounts from arbitrary sensor data.
- FIG. 1 is a block diagram showing an example of the configuration of a feature selection device 10 according to this embodiment.
- the feature quantity selection device 10 includes an acquisition unit 11 , a construction unit 12 , an analysis unit 13 , a statistics unit 15 , a selection unit 17 and an output unit 19 .
- the acquisition unit 11 acquires a data set used for estimating physical conditions, measured for multiple subjects.
- a data set is data obtained by combining an explanatory variable and an objective variable corresponding to the explanatory variable.
- a data set is data in which measured values or feature values relating to a subject are associated with the physical state of the subject.
- the explanatory variables used for estimating the physical state are feature quantities extracted from sensor data relating to leg movements and gaits.
- the constructing unit 12 constructs a new dataset (also called a re-extracted dataset) by changing the distribution of datasets regarding multiple subjects. For example, the constructing unit 12 constructs the re-extracted data set using a Leave-One-Subject-Out (also called LOSO) method. When using the LOSO approach, one is removed from the multiple datasets and the remaining datasets are used to construct the reconstructed dataset. When using the LOSO approach, reconstructed datasets are generated for the number of subjects. For example, if there are 50 subjects, 50 re-extracted datasets can be constructed using the LOSO approach.
- a new dataset also called a re-extracted dataset
- the construction unit 12 may construct the re-extracted dataset using the bootstrap method.
- the bootstrap method estimates the properties of a population based on values randomly sampled from a sample population by sampling with replacement.
- the bootstrap method involves iteratively generating new data sets using randomly sampled values from a sample population and calculating statistical values. For example, after 1000 iterations of generating new datasets, a re-extracted dataset of 1000 data can be constructed.
- the analysis unit 13 executes Lasso regression on the re-extracted data set constructed by the construction unit 12 .
- the analysis unit 13 uses a loss function represented by Equation 1 below.
- N is the number of observations.
- i is the observation number.
- x i is a vector (data) of length p at observation i.
- y i is the response data (correct value) of the observed value i.
- ⁇ is a non-negative regularization parameter (Lagrange multiplier) corresponding to one value.
- ⁇ 0 is a scalar.
- ⁇ is a vector of length p.
- j is a feature quantity number. If there are p feature quantities, the feature quantity number j is one of 1 to p.
- ⁇ j correspond to the coefficients of the polynomial function (also called model parameters) used as the estimation model.
- T indicates a transposition process.
- the first term on the right side of Equation 1 is the term for the sum-of-squares error.
- the second term on the right side of Equation 1 is the regularization term.
- the regularization term is a function defined to return a larger value as the model parameter ⁇ j increases.
- the regularization term corresponds to a penalty on the magnitude of the model parameters ⁇ j .
- the regularization parameter ⁇ is a meta parameter set during model learning.
- the regularization parameter ⁇ adjusts the strength of regularization (penalty).
- the penalty of the regularization term becomes strong, and overfitting is suppressed more strongly.
- the value of the regularization parameter ⁇ is too large, keeping the model parameters small is prioritized and the expressive power of the model decreases. As a result, too large a value of the regularization parameter ⁇ leaves a large bias.
- Equation 2 shows the minimum value when ⁇ 0 and the coefficient vector ⁇ are variables. Equation 2 determines the magnitude of the second term (regularization term) related to the penalty according to the magnitude of the absolute value of the model parameter ⁇ j .
- Equation 3 is a limiting condition for each element of the coefficient vector. Equation 2 above corresponds to determining the model parameter ⁇ j when the limiting condition of Equation 3 is provided for each element (model parameter ⁇ j ) of the coefficient vector ⁇ when the multiple regression coefficient vector ⁇ is determined by the least squares method.
- the regularization parameter ⁇ has one corresponding coefficient vector ⁇ . As the regularization parameter ⁇ increases, the coefficient vector ⁇ has fewer non-zero elements. That is, as the regularization parameter ⁇ increases, the number of zero elements in the coefficient vector ⁇ increases, resulting in an increase in unnecessary features. On the other hand, when the regularization parameter ⁇ becomes smaller, the number of non-zero elements of the coefficient vector ⁇ increases and the required feature amount increases. If an appropriate regularization parameter ⁇ is set, unnecessary zero elements can be reduced while leaving nonzero elements necessary for estimation.
- the analysis unit 13 executes Lasso regression on the re-extracted data set constructed by the construction unit 12.
- the analysis unit 13 changes the regularization parameter ⁇ and executes Lasso regression on the re-extracted data set for each subject.
- the analysis unit 13 generates a matrix (also referred to as a first matrix) composed of columns for the number of regularization parameters ⁇ and rows for the number of features. For example, if there are P regularization parameters ⁇ , each regularization parameter ⁇ is given a number from 1 to P (also called a ⁇ number) (P is a natural number).
- the first matrix has rows of the number of feature values used for estimating the physical condition or the like. When the number of feature quantities is p, each feature quantity is given a number from 1 to p (also called a feature quantity number) (p is a natural number).
- FIG. 2 is a conceptual diagram showing an example of the first matrix.
- hatched cells indicate non-zero elements.
- unhatched blank cells indicate zero elements.
- FIG. 3 is an example of the first matrix generated for 50 subjects using the reconstructed data set constructed by LOSO.
- a first matrix for each subject is generated.
- the first matrix of 50 patterns is generated according to the number of subjects (50).
- hatched cells indicate non-zero elements.
- blank cells that are not hatched indicate zero elements.
- the statistics unit 15 assigns a logical value (0, 1) to each cell of the first matrix of the generated multiple patterns.
- the process of assigning logical values (0, 1) to each cell of the first matrix of multiple patterns generated by Lasso regression is also called first statistical process.
- the statistical unit 15 sets non-zero elements to TRUE (1) and zero elements to FALSE (0) for the plurality of first matrices.
- the statistic unit 15 aggregates logical values (0, 1) for each cell with respect to all first matrices.
- the statistic unit 15 aggregates the logical values (0, 1) for each cell by adding the logical values (1) of non-zero elements for all the first matrices.
- FIG. 4 is an example in which the aggregate value of the logical values for the first matrix of 50 subjects is entered in each cell of the matrix (also called the second matrix) corresponding to all the first matrices.
- Each cell of the second matrix is filled with the number of non-zero elements (number of TRUEs) of the first matrix for multiple patterns.
- the statistics unit 15 assigns a logical value (0 or 1) to each cell of the second matrix according to the total value of each cell included in all of the first matrix. If the aggregate value for each cell in the second matrix is equal to or greater than a predetermined threshold, the statistic unit 15 sets that cell to TRUE (1). On the other hand, if the aggregated value is below the predetermined threshold, the statistic unit 15 sets the cell to FALSE (0).
- the process of totaling the logical values (0, 1) of each cell for all the first matrixes and assigning the logical value (0 or 1) according to the totaled value to each cell of the second matrix is also called second statistical processing.
- FIG. 5 is a conceptual diagram showing an example of assigning logical values (0 or 1) to the total values in FIG.
- cells with a total value of 49 or more in FIG. 4 are set to TRUE (1).
- cells set to TRUE (1) by the second statistical processing are hatched.
- cells set to FALSE (0) by the second statistical process are blank.
- the result of the second statistical processing as shown in FIG. 5 may be displayed on a screen that the user can confirm. In that case, the user can select a desired combination of feature amounts by selecting the ⁇ number according to the result of the second statistical processing displayed on the screen.
- the statistics unit 15 may assign a logical value corresponding to the average value of the aggregated values to each cell of the second matrix. For example, the statistic unit 15 sets TRUE(1) to cells in which the average value of the total values is equal to or greater than a predetermined threshold for each cell in the second matrix. On the other hand, the statistic unit 15 sets FALSE (0) to cells in which the average value of the total values is less than the predetermined threshold. Such processing is also included in the second statistical processing.
- the selection unit 17 selects the ⁇ number according to preset specific rules.
- a specific rule is a rule for determining the ⁇ number to be selected.
- the specific rule is to select a ⁇ number whose number of cells set to TRUE (1) corresponds to a preset reference value.
- the reference value may be set according to restrictions on the amount of calculation and the amount of communication.
- the reference value is set to a value that does not exceed the load that can be assigned to the amount of computation and the amount of communication.
- the reference value is set to a value that does not exceed a ratio (for example, 50 to 80%) of the load that can be assigned to the amount of computation or communication.
- the selection unit 17 selects a combination of feature amounts in which the selected ⁇ number cell is set to TRUE (1) based on the specific rule. For example, the selection unit 17 may select a combination of feature amounts according to a reference value set by the user.
- the output unit 19 outputs information (also referred to as selection information) regarding the feature quantity selected by the selection unit 17 .
- the selection information is information about a combination of feature quantities used for estimating a physical condition or the like.
- the selection information includes information indicating from which walking phase the feature amount is to be extracted in the time-series data of acceleration and angular velocity for one step cycle.
- a gait phase indicates a gait cycle (percent) when the step cycle is normalized to 0 to 100 percent.
- a feature amount over a plurality of consecutive walking phases may be extracted.
- a cluster of a plurality of continuous walking phases from which feature values are extracted is also called a walking phase cluster.
- the selection information output from the output unit 19 is used as a condition for extracting feature amounts from sensor data measured by a measuring device or the like.
- the selection information may be stored in a storage unit (not shown) by the selection unit 17 .
- the feature amount extracted according to the selection information is used for learning an estimation model for estimating the physical condition and the like.
- the feature amount to be extracted is extracted from sensor data measured by a measuring device or the like worn by the user whose physical condition is to be estimated.
- FIGS. 6 and 7 are conceptual diagrams for explaining differences in estimated values by an estimation model generated by learning using feature values selected using a general Lasso regression (comparative example) method and feature values selected using the method of this embodiment.
- FIGS. 6 and 7 show an example of estimating TUG (Time Up and Go) test results as a subject's mobility ability. The performance of the TUG test is the time it takes to stand up from a chair, walk to a landmark 3 meters away, turn around, and sit back in the chair (also called TUG duration).
- TUG Time Up and Go
- Fig. 6 is an example of estimation using an estimation model generated using nine feature values selected by a general Lasso regression (comparative example) method.
- the correlation intraclass correlation coefficient ICC Intraclass Correlation Coefficients
- MAE Mean Absolute Error
- FIG. 7 is an estimation example using an estimation model generated using nine feature values selected by the method of this embodiment.
- the intra-class correlation coefficient ICC was 0.682 for the true value (measured value) and the estimated value of the TUG required time.
- the mean absolute error MAE between the true value (measured value) and the estimated value of the TUG required time was 0.63.
- both ICC and MAE were larger when the method of the present embodiment was used. That is, by using the method of this embodiment, the robustness against outliers and outliers is improved.
- FIG. 8 is a graph for explaining the influence of outliers and outliers that may be included in sensor data measured for multiple subjects.
- the data within the range enclosed by the dashed circle corresponds to outliers and outliers.
- L1 is a regression line obtained by performing linear regression on a plurality of sensor data, including outliers and outliers.
- L2 is a regression line obtained by arbitrarily excluding outliers and outliers and linearly regressing a plurality of sensor data.
- the regression line L1 is affected by outliers and outliers and does not fit most of the sensor data.
- the regression line L2 is not affected by outliers and outliers and fits most sensor data.
- Lasso regression is performed after changing the distribution of the dataset using a method such as LOSO or the bootstrap method.
- a method such as LOSO or the bootstrap method.
- the above-described first statistical processing and second statistical processing are executed.
- an average solution in which the effects of outliers and outliers are reduced is obtained.
- FIGS. 9 and 10 are flowcharts for explaining an example of the operation of the feature selection device 10.
- FIG. In the explanation using the flow charts of FIGS. 9 and 10, the feature quantity selection device 10 will be explained as an operator.
- the feature quantity selection device 10 acquires N data sets (step S111).
- the data set number corresponds to the explanatory variable (feature quantity) number (feature quantity number) included in the data set.
- the feature quantity selection device 10 sets the feature quantity number n to 1 (step S112).
- n is the number of the data set (feature amount).
- the feature quantity selection device 10 excludes the data of the n-th subject (step S113).
- the feature quantity selection device 10 executes Lasso regression on the N-1 data sets from which the data of the n-th subject is excluded (step S114).
- the feature quantity selection device 10 executes the first statistical processing (step S115).
- the feature quantity selection device 10 assigns a logical value to each cell of the first matrix (matrix B n ) generated by Lasso regression.
- the feature selection device 10 sets the non-zero elements of the matrix B n to TRUE (1) and sets the zero elements of the matrix B n to FALSE (0).
- the feature selection device 10 may set TRUE (1) to cells whose matrix B n element values are equal to or greater than the threshold T 0 , and FALSE (0) to cells whose matrix B n element values are less than the threshold T 0 .
- the feature quantity selection device 10 increments (+1) the feature quantity number n (step S116).
- step S117 if the feature quantity number n is smaller than the number N of data sets (Yes in step S117), the process returns to step S113. On the other hand, if the feature quantity number n is greater than or equal to the number N of data sets (No in step S117), the process proceeds to step S121 in FIG.
- the feature quantity selection device 10 executes the second statistical processing (step S121).
- the feature quantity selection device 10 aggregates logical values (0 or 1) for each cell with respect to all the first matrices.
- the feature quantity selection device 10 sets a logical value (0 or 1) to each cell of the aggregated second matrix according to the relationship between the aggregate value of the logical values for each cell and a predetermined threshold value. For example, the feature quantity selection device 10 sets TRUE (1) to cells whose total value is greater than or equal to a predetermined threshold. On the other hand, the feature quantity selection device 10 sets FALSE (0) to cells whose total value is less than the predetermined threshold.
- the feature quantity selection device 10 selects a ⁇ number based on a specific rule according to the result of the second statistical processing (step S122).
- the feature amount selection device 10 selects a combination of feature amounts corresponding to the selected ⁇ number (step S123).
- the feature amount selection device 10 outputs information (selection information) on the selected feature amount (step S124).
- the selection information output from the feature quantity selection device 10 is used as conditions for extracting feature quantities from sensor data measured by a measuring device or the like.
- the feature quantity selection device of this embodiment includes an acquisition unit, a construction unit, an analysis unit, a statistics unit, a selection unit, and an output unit.
- the acquisition unit acquires multiple data sets.
- the constructing unit constructs a plurality of re-extracted datasets by changing the distribution of data contained in the datasets.
- the analysis unit analyzes multiple re-extracted data sets using the Lasso regression technique.
- the statistical unit aggregates the values of the elements included in the plurality of re-extracted data sets according to the analysis results of the plurality of re-extracted data sets.
- the statistical unit sets logical values to the elements included in the plurality of re-extracted data sets according to the result of counting the values of the elements.
- the selection unit selects a combination of feature quantities according to the logical values set for the elements, according to a preset specific rule.
- the output unit outputs selection information about the selected combination of feature amounts.
- the analysis unit executes Lasso regression for each of a plurality of preset regularization parameters for a plurality of re-extracted datasets.
- the analysis unit generates a first matrix of multiple patterns configured by columns corresponding to regularization parameters used in Lasso regression and rows corresponding to feature amounts.
- the statistical unit performs a first statistical process of setting a first logical value of non-zero element cells to 1 and setting a first logical value of zero element cells to 0 for the first matrix of the plurality of patterns.
- the statistic unit aggregates the first logical values for each cell that constitutes the first matrix of the plurality of patterns.
- the statistical unit executes a second statistical process for generating a second matrix in which 1 is set as a second logical value to cells whose aggregated first logical values satisfy a predetermined condition, and 0 is set as a second logical value to cells whose aggregated first logical values do not satisfy the predetermined condition.
- the selection unit selects columns of the second matrix according to a preset specific rule, and selects combinations of feature amounts corresponding to the selected columns.
- the construction unit constructs multiple re-extracted datasets using the Leave-One-Subject-Out technique. According to this aspect, by artificially changing the data distribution using the Leave-One-Subject-Out method, the data distribution can be brought closer to the original distribution of the population.
- the construction unit constructs multiple re-extracted datasets using the bootstrap method.
- the data distribution can be approximated to the distribution of the population estimated from the sample group by artificially changing the data distribution using the bootstrap method.
- the statistical unit calculates the total value of the first logical values for each cell that constitutes the first matrix of multiple patterns.
- the statistic unit generates a second matrix in which the second logical values of cells whose total first logical values are equal to or greater than a predetermined threshold are set to 1, and the cells whose total first logical values are less than the predetermined threshold are set to 0.
- a combination of feature amounts can be selected based on the logical values of the second matrix set according to the total value of the first logical values.
- the statistical unit calculates the average value of the first logical values for each cell that constitutes the first matrix of multiple patterns.
- the statistic unit generates a second matrix in which the second logical value of cells whose average first logical value is equal to or greater than a predetermined threshold is set to 1, and 0 is set to cells whose average first logical value is less than the predetermined threshold.
- a combination of feature amounts can be selected based on the logical values of the second matrix set according to the average value of the first logical values.
- the feature quantity selection device of the present embodiment constructs an estimation model using the feature quantity selected by the method of the first embodiment.
- the feature amount selection device of the present embodiment selects feature amounts according to the estimation result of the constructed estimation model.
- FIG. 11 is a block diagram showing an example of the configuration of the feature selection device 20 according to this embodiment.
- the feature selection device 20 includes an acquisition unit 21 , a construction unit 22 , an analysis unit 23 , a statistics unit 25 , an estimation model construction unit 26 , a selection unit 27 and an output unit 29 .
- the acquisition unit 21 has the same configuration as the acquisition unit 11 of the first embodiment. Acquisition unit 21 acquires a data set used for estimating physical conditions, measured for a plurality of subjects.
- the construction unit 22 has the same configuration as the construction unit 12 of the first embodiment.
- the constructing unit 22 constructs a new dataset (also called a re-extracted dataset) by changing the distribution of datasets for multiple subjects.
- the constructing unit 22 constructs the re-extracted data set using a Leave-One-Subject-Out (also called LOSO) technique.
- the construction unit 12 may construct the re-extracted data set using the bootstrap method.
- the analysis unit 23 has the same configuration as the analysis unit 13 of the first embodiment.
- the analysis unit 23 executes Lasso regression on the re-extracted data set constructed by the construction unit 22 .
- the analysis unit 23 generates a matrix (also referred to as a first matrix) composed of columns for the number of regularization parameters ⁇ and rows for the number of features. As a result, a first matrix is generated that has as many columns as the number of varied regularization parameters ⁇ .
- the statistic unit 25 has the same configuration as the statistic unit 15 of the first embodiment.
- the statistical unit 25 performs a first statistical process of assigning a logical value (0, 1) to each cell of the first matrix of multiple patterns generated for each subject. In the first statistical processing, the statistical unit 25 sets non-zero elements to TRUE (1) and zero elements to FALSE (0) for the plurality of first matrices.
- the statistical unit 25 aggregates the logical values (0, 1) for each cell with respect to all the first matrices, and performs the second statistical processing for adding the logical value (1) of non-zero elements for each cell.
- the statistics unit 25 assigns a logical value (0 or 1) to each cell of the second matrix according to the aggregated value of each cell of the first matrix. For each cell in the second matrix, if the total value is equal to or greater than a predetermined threshold, the statistic unit 25 sets the cell to TRUE (1). On the other hand, when the total value is below the predetermined threshold, the statistic unit 25 sets the cell to FALSE (0).
- the estimation model construction unit 26 constructs an estimation model by learning using the feature amount selected by the selection unit 27.
- the estimation model construction unit 26 evaluates the constructed estimation model. For example, the estimation model construction unit 26 calculates evaluation indices such as mean squared error, mean absolute error, mean relative error, coefficient of determination, and correlation coefficient.
- the estimation model construction unit 26 outputs the calculated evaluation index to the selection unit 27 .
- the estimation model evaluation results may be displayed on a screen that the user can check. In that case, the user can select the most likely combination of feature amounts according to the evaluation results displayed on the screen.
- the selection unit 27 selects a combination of feature quantities with the highest evaluation index calculated by the estimation model construction unit 26 .
- the selection unit 27 may select the most likely feature amount according to an instruction input by the user.
- the output unit 29 has the same configuration as the output unit 19 of the first embodiment.
- the output unit 29 outputs information (also referred to as selection information) regarding the feature amount selected by the selection unit 27 .
- the selection information output from the output unit 29 is used as a condition for extracting feature amounts from sensor data measured by a measuring device or the like.
- the selection information may be stored in a storage unit (not shown) by the selection unit 27 .
- the feature amount extracted according to the selection information is used for learning an estimation model for estimating the physical condition and the like.
- the feature amount to be extracted is extracted from sensor data measured by a measuring device or the like worn by the user whose physical condition is to be estimated.
- FIGS. 12 and 13 are flowcharts for explaining an example of the operation of the feature selection device 20.
- FIG. In the description using the flowcharts of FIGS. 12 and 13, the feature quantity selection device 20 will be described as an operating entity.
- the feature quantity selection device 20 acquires N data sets (step S211).
- the data set number corresponds to the explanatory variable (feature quantity) number (feature quantity number) included in the data set.
- the feature quantity selection device 20 sets the feature quantity number n to 1 (step S212).
- n is the number of the data set (feature amount).
- the feature quantity selection device 20 excludes the data of the n-th subject (step S213).
- the feature quantity selection device 20 executes Lasso regression on the N-1 data sets from which the data of the n-th subject is excluded (step S214).
- the feature quantity selection device 20 executes the first statistical processing (step S215).
- the feature quantity selection device 20 assigns a logical value to each cell of the first matrix (matrix B n ) generated by Lasso regression.
- the feature selection device 20 sets the non-zero elements of the matrix B n to TRUE (1) and sets the zero elements of the matrix B n to FALSE (0).
- the feature selection device 20 may set TRUE (1) to cells whose matrix B n element values are equal to or greater than the threshold T 0 , and FALSE (0) to cells whose matrix B n element values are less than the threshold T 0 .
- the feature quantity selection device 20 increments (+1) the feature quantity number n (step S216).
- step S217 if the feature quantity number n is smaller than the number N of data sets (Yes in step S217), the process returns to step S213. On the other hand, if the feature quantity number n is greater than or equal to the number N of data sets (No in step S217), the process proceeds to step S221 in FIG.
- the feature quantity selection device 20 executes the second statistical processing (step S221).
- the feature quantity selection device 20 aggregates logical values (0, 1) for each cell with respect to all the first matrices.
- the feature quantity selection device 20 assigns the sum of the aggregated logical values for each cell of the first matrix to each cell of the second matrix.
- the feature quantity selection device 20 sets a logical value (0 or 1) to each cell of the aggregated second matrix according to the value of each cell of the second matrix. For example, the feature selection device 20 sets TRUE (1) to cells whose total value is greater than or equal to a predetermined threshold. On the other hand, the feature quantity selection device 20 sets the cells whose total value is below the predetermined threshold to FALSE (0).
- step S222 the feature quantity selection device 20 executes model evaluation processing.
- the model evaluation processing in step S222 will be described later (FIG. 14).
- the feature quantity selection device 20 searches for the ⁇ number corresponding to the evaluation index obtained by the model evaluation process (step S223).
- the feature amount selection device 20 selects a combination of feature amounts corresponding to the retrieved ⁇ numbers (step S224).
- the feature amount selection device 20 outputs information (selection information) on the selected feature amount (step S225).
- the selection information output from the feature quantity selection device 20 is used as conditions for extracting feature quantities from sensor data measured by a measuring device or the like.
- FIG. 14 is a flowchart for explaining model evaluation processing.
- the feature quantity selection device 20 will be explained as an operator.
- the feature quantity selection device 20 sets the ⁇ number m to 1 (step S231). m is the number of the regularization parameter ⁇ .
- the feature amount selection device 20 selects a combination of feature amounts corresponding to the ⁇ number m (step S232).
- the feature quantity selection device 20 constructs an estimation model using the selected feature quantity (step S233).
- the feature quantity selection device 20 evaluates the constructed estimation model (step S234).
- the feature quantity selection device 20 outputs the evaluation index of the estimation model (step S235).
- the feature quantity selection device 20 increments (+1) the ⁇ number m (step S236).
- step S237 if the ⁇ number m is smaller than the number P of regularization parameters ⁇ (Yes in step S237), the process returns to step S232. On the other hand, if the ⁇ number m is greater than or equal to the number P of regularization parameters ⁇ (Yes in step S237), the process proceeds to step S223 in FIG.
- the feature quantity selection device of the present embodiment includes an acquisition unit, construction unit, analysis unit, statistics unit, selection unit, estimation model construction unit, and output unit.
- the acquisition unit acquires multiple data sets.
- the constructing unit constructs a plurality of re-extracted datasets by changing the distribution of data contained in the datasets.
- the analysis unit analyzes multiple re-extracted data sets using the Lasso regression technique.
- the statistical unit aggregates the values of the elements included in the plurality of re-extracted data sets according to the analysis results of the plurality of re-extracted data sets.
- the statistical unit sets logical values to the elements included in the plurality of re-extracted data sets according to the result of counting the values of the elements.
- the selection unit selects a combination of feature quantities according to the logical values set for the elements, according to a preset specific rule.
- the estimation model construction unit constructs an estimation model by learning using the selected feature amount, and evaluates the constructed estimation model.
- the selection unit selects a combination of feature amounts according to the evaluation result of the estimation model.
- the output unit outputs selection information about the selected combination of feature amounts.
- the feature quantity is selected according to the evaluation result of the estimation model constructed using the feature quantity selected by the selection unit. Therefore, according to the present embodiment, it is possible to select a highly reliable feature amount by using the evaluation result of the estimation model.
- the feature quantity selection apparatus of the present embodiment differs from the first embodiment in that the first statistical processing is omitted and the second statistical processing is executed on the average value of each cell regarding a plurality of first matrices.
- FIG. 15 is a block diagram showing an example of the configuration of the feature selection device 30 according to this embodiment.
- the feature amount selection device 30 includes an acquisition unit 31 , a construction unit 32 , an analysis unit 33 , a statistics unit 35 , a 3-selection unit 37 and an output unit 39 .
- the acquisition unit 31 has the same configuration as the acquisition unit 11 of the first embodiment.
- the acquisition unit 31 acquires a data set used for estimating physical conditions, which is measured for a plurality of subjects.
- the construction unit 32 has the same configuration as the construction unit 12 of the first embodiment.
- the constructing unit 32 constructs a new dataset (also called a re-extracted dataset) by changing the distribution of datasets for multiple subjects.
- the constructing unit 32 constructs the re-extracted data set using a Leave-One-Subject-Out (also called LOSO) technique.
- the construction unit 32 may construct the re-extracted data set using the bootstrap method.
- the analysis unit 33 has the same configuration as the analysis unit 13 of the first embodiment.
- the analysis unit 33 executes Lasso regression on the re-extracted data set constructed by the construction unit 32 .
- the analysis unit 33 generates a matrix (also referred to as a first matrix) composed of columns for the number of regularization parameters ⁇ and rows for the number of features. As a result, a first matrix is generated that has as many columns as the number of varied regularization parameters ⁇ .
- the statistical unit 35 calculates the average value of each cell for the first matrix of multiple patterns generated for each subject.
- the statistics unit 35 generates a second matrix in which an average value of each cell is assigned to each cell of the first matrix of multiple patterns. For example, the statistic unit 35 sets TRUE(1) to cells in which the average value of the total values is equal to or greater than a predetermined threshold for each cell in the second matrix. On the other hand, the statistics unit 35 sets FALSE (0) to cells in which the average value of the aggregated values is less than the predetermined threshold. This processing is included in the second statistical processing.
- the selection unit 37 has the same configuration as the selection unit 17 of the first embodiment.
- the selection unit 17 selects a ⁇ number based on a preset specific rule.
- the selection unit 17 selects a combination of feature amounts in which the selected ⁇ number cell is set to TRUE (1) based on the specific rule.
- the output unit 39 has the same configuration as the output unit 19 of the first embodiment.
- the output unit 39 outputs information (also referred to as selection information) regarding the feature amount selected by the selection unit 27 .
- the selection information output from the output unit 39 is used as conditions for extracting feature amounts from sensor data measured by a measuring device or the like.
- the selection information may be stored in a storage unit (not shown) by the selection unit 37 .
- the feature amount extracted according to the selection information is used for learning an estimation model for estimating the physical condition and the like.
- the feature amount to be extracted is extracted from sensor data measured by a measuring device or the like worn by the user whose physical condition is to be estimated.
- FIGS. 16 and 17 are flowcharts for explaining an example of the operation of the feature selection device 30.
- FIG. In the description using the flowcharts of FIGS. 16 and 17, the feature quantity selection device 30 will be described as an operating body.
- the feature quantity selection device 30 acquires N data sets (step S311).
- the data set number corresponds to the explanatory variable (feature quantity) number (feature quantity number) included in the data set.
- the feature quantity selection device 30 sets the feature quantity number n to 1 (step S312).
- n is the number of the data set (feature quantity).
- the feature quantity selection device 30 excludes the data of the n-th subject (step S313).
- the feature quantity selection device 30 executes Lasso regression on the N-1 data sets from which the data of the n-th subject is excluded (step S314).
- the feature quantity selection device 30 increments (+1) the feature quantity number n (step S315).
- step S316 if the feature quantity number n is smaller than the number N of data sets (Yes in step S316), the process returns to step S313. On the other hand, if the feature quantity number n is greater than or equal to the number N of data sets (No in step S316), the process proceeds to step S321 in FIG.
- the feature quantity selection device 30 executes the second statistical processing (step S321).
- the feature quantity selection device 30 As the first stage of the second statistical processing, the feature quantity selection device 30 generates a second matrix in which the average value of each cell is assigned to each cell of the first matrix of multiple patterns generated for each subject.
- the feature quantity selection device 30 sets logical values (0, 1) for each cell with respect to the generated second matrix as the second statistical processing. For example, the feature quantity selection device 30 sets TRUE (1) to cells in which the average value of the total values is equal to or greater than a predetermined threshold value for each cell of the second matrix.
- the feature quantity selection device 30 sets FALSE (0) to cells in which the average value of the total values is less than the predetermined threshold.
- the feature quantity selection device 30 selects a ⁇ number based on a specific rule according to the result of the second statistical processing (step S322).
- the feature amount selection device 30 selects a combination of feature amounts corresponding to the selected ⁇ number (step S323).
- the feature amount selection device 30 outputs information (selection information) on the selected feature amount (step S324).
- the selection information output from the feature quantity selection device 30 is used as conditions for extracting feature quantities from sensor data measured by a measuring device or the like.
- the feature quantity selection device of this embodiment includes an acquisition unit, a construction unit, an analysis unit, a statistics unit, a selection unit, and an output unit.
- the acquisition unit acquires multiple data sets.
- the constructing unit constructs a plurality of re-extracted datasets by changing the distribution of data contained in the datasets.
- the analysis unit analyzes multiple re-extracted data sets using the Lasso regression technique.
- the statistical unit aggregates the values of the elements included in the plurality of re-extracted data sets according to the analysis results of the plurality of re-extracted data sets.
- the statistical unit sets logical values to the elements included in the plurality of re-extracted data sets according to the result of counting the values of the elements.
- the selection unit selects a combination of feature quantities according to the logical values set for the elements, according to a preset specific rule.
- the output unit outputs selection information about the selected combination of feature amounts.
- the analysis unit executes Lasso regression for each of a plurality of preset regularization parameters for each of the plurality of re-extracted data sets.
- the analysis unit generates a first matrix of multiple patterns configured by columns corresponding to regularization parameters used in Lasso regression and rows corresponding to feature amounts.
- the statistical unit aggregates element values for each cell that constitutes the first matrix of the plurality of patterns.
- the statistic unit performs a second statistical process of generating a second matrix in which 1 is set as a second logical value to cells in which the average element value is equal to or greater than a predetermined threshold, and 0 is set as a second logical value to cells in which the average element value is less than the predetermined threshold.
- the selection unit selects columns of the second matrix according to a preset specific rule, and selects combinations of feature amounts corresponding to the selected columns.
- the feature quantity selection device of this embodiment has a simplified configuration of the feature quantity selection devices of the first to third embodiments.
- FIG. 18 is a block diagram showing an example of the configuration of the feature selection device 40 according to this embodiment.
- the feature quantity selection device 40 includes an acquisition unit 41 , a construction unit 42 , an analysis unit 43 , a statistics unit 45 , a selection unit 47 and an output unit 49 .
- the acquisition unit 41 acquires multiple data sets.
- the constructing unit 42 constructs a plurality of re-extracted datasets by changing the distribution of data included in the datasets.
- the analysis unit 43 analyzes a plurality of re-extracted data sets using the Lasso regression technique.
- the statistics unit 45 aggregates the values of the elements included in the plurality of re-extracted data sets according to the analysis results of the plurality of re-extracted data sets.
- the statistics unit 45 sets logical values to the elements included in the plurality of re-extracted data sets according to the results of counting the values of the elements.
- the selection unit 47 selects a combination of feature amounts according to the values of the logical values set to the elements according to a preset specific rule.
- the output unit 49 outputs selection information regarding the selected combination of feature amounts.
- the learning system of this embodiment executes learning using the feature values selected by the feature value selection devices of the first to fourth embodiments.
- FIG. 19 is a block diagram showing an example of the configuration of the learning system 5 according to this embodiment.
- the learning system 5 includes a gait measuring device 50 and a learning device 55 .
- the gait measuring device 50 and the learning device 55 may be wired or wirelessly connected.
- the gait measuring device 50 and the learning device 55 may be configured as a single device.
- the learning system 5 may be configured by only the learning device 55 by excluding the gait measuring device 50 from the configuration of the learning system 5 .
- the learning device 55 may be configured to perform learning using feature amount data that is not connected to the gait measuring device 50 and that is generated in advance by the gait measuring device 50 and stored in a database.
- the gait measuring device 50 is installed on at least one of the left and right feet.
- the gait measuring device 50 has the same configuration as the gait measuring device 50 of the first embodiment.
- Gait measuring device 50 includes an acceleration sensor and an angular velocity sensor.
- the gait measuring device 50 converts the measured physical quantity into digital data (also called sensor data).
- the gait measuring device 50 generates normalized gait waveform data for one step cycle from time-series data of sensor data.
- the gait measuring device 50 generates feature amount data used for estimating the physical condition.
- the gait measuring device 50 transmits the generated feature amount data to the learning device 55 .
- the gait measuring device 50 may be configured to transmit feature amount data to a database (not shown) accessed by the learning device 55 .
- the feature amount data accumulated in the database is used for learning by the learning device 55 .
- the learning device 55 receives feature amount data from the gait measuring device 50 .
- the feature quantity data received by the learning device 55 includes the feature quantities selected by the feature quantity selection devices of the first to fourth embodiments.
- the learning device 55 receives the feature amount data from the database.
- the learning device 55 performs learning using the received feature amount data. For example, the learning device 55 learns teacher data using feature amount data extracted from walking waveform data of a plurality of subjects as explanatory variables and values relating to physical conditions according to the feature amount data as objective variables.
- the learning algorithm executed by the learning device 55 is not particularly limited.
- the learning device 55 generates an estimated model trained using teacher data regarding a plurality of subjects.
- the learning device 55 stores the generated estimation model.
- the estimation model learned by the learning device 55 may be stored in a storage device external to the learning device 55 .
- FIG. 20 is a block diagram showing an example of the detailed configuration of the learning device 55. As shown in FIG. The learning device 55 has a receiving section 551 , a learning section 553 and a storage section 555 .
- the receiving unit 551 receives feature amount data from the gait measuring device 50 .
- the receiving unit 551 outputs the received feature amount data to the learning unit 553 .
- the receiving unit 551 may receive the feature amount data from the gait measurement device 50 via a cable such as a cable, or may receive the feature amount data from the gait measurement device 50 via wireless communication.
- the receiving unit 551 is configured to receive feature amount data from the gait measuring device 50 via a wireless communication function (not shown) conforming to standards such as Bluetooth (registered trademark) and WiFi (registered trademark).
- the communication function of the receiving unit 551 may conform to standards other than Bluetooth (registered trademark) and WiFi (registered trademark).
- the learning unit 553 acquires feature amount data from the receiving unit 551 .
- the learning unit 553 performs learning using the acquired feature amount data.
- the learning unit 553 learns, as teaching data, a data set in which feature amount data extracted regarding the subject's gait is used as an explanatory variable and the subject's physical state is used as an objective variable.
- the learning unit 553 learns physical conditions such as grip strength, whole-body muscle strength, lower-limb muscle strength, mobility, dynamic balance, static balance, and the like of the subject.
- the learning unit 553 generates an estimation model for estimating the physical state according to input of feature amount data learned about a plurality of users.
- the learning unit 553 uses explanatory variables including attribute data such as gender, age, height, and weight to generate an estimation model that performs estimation according to attributes.
- the learning unit 553 causes the storage unit 555 to store the estimated models learned for the plurality of subjects.
- the learning unit 553 performs learning using a linear regression algorithm.
- the learning unit 553 performs learning using a Support Vector Machine (SVM) algorithm.
- the learning unit 553 performs learning using a Gaussian Process Regression (GPR) algorithm.
- the learning unit 553 performs learning using a random forest (RF) algorithm.
- the learning unit 553 may perform unsupervised learning for classifying subjects who generated the feature amount data according to the feature amount data.
- a learning algorithm executed by the learning unit 553 is not particularly limited.
- the learning unit 553 may perform learning using the walking waveform data (sensor data) for one step cycle as explanatory variables. For example, the learning unit 553 performs supervised learning using walking waveform data of acceleration in three-axis directions, angular velocity about three axes, and angle (posture angle) about three axes as explanatory variables, and correct values of the body state to be estimated as objective variables.
- FIG. 21 is a conceptual diagram for explaining learning for generating an estimation model.
- FIG. 21 is a conceptual diagram showing an example of learning by the learning unit 553 using a data set of feature values F1 to Fn, which are explanatory variables, and scores relating to physical conditions, which are objective variables, as teacher data.
- the learning unit 553 learns data about a plurality of subjects, and generates an estimation model that outputs an output (estimated value) regarding the subject's physical condition according to the input of the feature amount extracted from the sensor data.
- the storage unit 555 stores estimated models learned for a plurality of subjects.
- the storage unit 555 stores an estimation model for estimating the physical condition learned for a plurality of subjects.
- the estimation model stored in the storage unit 555 is used for body condition estimation by a body condition estimation system according to a sixth embodiment, which will be described later.
- the learning system of this embodiment includes a gait measuring device and a learning device.
- a gait measuring device acquires time-series data of sensor data relating to leg movements.
- the gait measuring device extracts walking waveform data for one step cycle from time-series data of sensor data, and normalizes the extracted walking waveform data.
- the gait measuring device extracts a feature amount related to the body state of an estimation target from the normalized walking waveform data.
- the gait measuring device extracts feature quantities selected by the feature quantity selecting devices of the first to fourth embodiments.
- the gait measuring device generates feature amount data including the extracted feature amount.
- the gait measuring device outputs the generated feature amount data to the learning device.
- the learning device has a receiving unit, a learning unit, and a storage unit.
- the receiving unit acquires feature amount data generated by the gait measuring device.
- the learning unit performs learning using the feature amount data.
- the learning unit generates an estimation model that outputs a physical state according to an input of feature values extracted from time-series data of sensor data measured as the user walks.
- the estimation model generated by the learning unit is stored in the storage unit.
- the learning system of this embodiment uses the feature amount data measured by the gait measuring device to generate an estimation model.
- the learning system of this embodiment executes learning using the feature values selected by the feature value selection devices of the first to fourth embodiments. Therefore, according to this aspect, it is possible to generate an estimation model that enables appropriate estimation of the physical condition in daily life using highly robust feature values.
- the physical state estimation system of this embodiment measures sensor data related to the movement of the user's feet as they walk.
- the physical condition estimation system of the present embodiment uses the measured sensor data to estimate the physical condition of the user.
- the physical condition estimation system of the present embodiment estimates muscle strength indices such as grip strength and knee extension strength, dynamic balance, leg muscle strength, mobility, static balance, and the like, as physical conditions.
- the sensor data may be sensor data including features related to gait that are measured using motion capture, smart apparel, or the like.
- FIG. 22 is a block diagram showing an example of the configuration of the physical state estimation system 6 according to this embodiment.
- the physical state estimation system 6 includes a gait measurement device 60 and an estimation device 63 .
- the gait measuring device 60 and the estimating device 63 are configured as separate hardware will be described.
- the gait measuring device 60 is installed on footwear or the like of a subject (user) whose body condition is to be estimated.
- the function of the estimation device 63 is installed in a mobile terminal carried by a subject (user).
- the configurations of the gait measuring device 60 and the estimating device 63 will be individually described below.
- FIG. 23 is a block diagram showing an example of the configuration of the gait measuring device 60.
- the gait measuring device 60 has a sensor 61 and a feature quantity data generator 62 .
- the sensor 61 and the feature amount data generation unit 62 are integrated will be given.
- the sensor 61 and feature amount data generator 62 may be provided as separate devices.
- the sensor 61 has an acceleration sensor 611 and an angular velocity sensor 612.
- FIG. 23 shows an example in which the sensor 61 includes an acceleration sensor 611 and an angular velocity sensor 612 .
- Sensors 61 may include sensors other than acceleration sensor 611 and angular velocity sensor 612 . Description of sensors other than the acceleration sensor 611 and the angular velocity sensor 612 that may be included in the sensor 61 is omitted.
- the acceleration sensor 611 is a sensor that measures acceleration in three axial directions (also called spatial acceleration).
- the acceleration sensor 611 measures acceleration (also referred to as spatial acceleration) as a physical quantity related to foot movement.
- the acceleration sensor 611 outputs the measured acceleration to the feature quantity data generator 62 .
- the acceleration sensor 611 can be a sensor of a piezoelectric type, a piezoresistive type, a capacitive type, or the like. As long as the sensor used as the acceleration sensor 611 can measure acceleration, the measurement method is not limited.
- the angular velocity sensor 612 is a sensor that measures angular velocities around three axes (also called spatial angular velocities).
- the angular velocity sensor 612 measures angular velocity (also called spatial angular velocity) as a physical quantity related to foot movement.
- the angular velocity sensor 612 outputs the measured angular velocity to the feature amount data generator 62 .
- the angular velocity sensor 612 can be a vibration type sensor or a capacitance type sensor. As long as the sensor used as the angular velocity sensor 612 can measure the angular velocity, the measurement method is not limited.
- the sensor 61 is realized, for example, by an inertial measurement device that measures acceleration and angular velocity.
- An example of an inertial measurement device is an IMU (Inertial Measurement Unit).
- the IMU includes an acceleration sensor 611 that measures acceleration along three axes and an angular velocity sensor 612 that measures angular velocity around three axes.
- the sensor 61 may be implemented by an inertial measurement device such as VG (Vertical Gyro) or AHRS (Attitude Heading).
- the sensor 61 may be realized by GPS/INS (Global Positioning System/Inertial Navigation System).
- the sensor 61 may be implemented by a device other than an inertial measurement device as long as it can measure physical quantities related to foot movement.
- FIG. 24 is a conceptual diagram showing an example in which the gait measuring device 60 is arranged in the shoe 600 of the right foot.
- the gait measuring device 60 is installed at a position corresponding to the back side of the arch.
- the gait measuring device 60 is placed on an insole that is inserted into the shoe 600 .
- the gait measuring device 60 may be arranged on the bottom surface of the shoe 600 .
- the gait measuring device 60 may be embedded in the body of the shoe 600 .
- the gait measuring device 60 may be detachable from the shoe 600 or may not be detachable from the shoe 600 .
- the gait measuring device 60 may be installed at a position other than the back side of the arch as long as it can measure sensor data relating to the movement of the foot. Also, the gait measuring device 60 may be installed on a sock worn by the user or an accessory such as an anklet worn by the user. Also, the gait measuring device 60 may be attached directly to the foot or embedded in the foot. FIG. 24 shows an example in which the gait measuring device 60 is installed on the shoe 600 of the right foot. The gait measuring device 60 may be installed on the shoes 600 of both feet.
- a local coordinate system including the x-axis in the horizontal direction, the y-axis in the front-back direction, and the z-axis in the vertical direction is set with the gait measuring device 60 (sensor 61) as a reference.
- the x-axis is positive to the left
- the y-axis is positive to the rear
- the z-axis is positive to the top.
- the directions of the axes set in the sensors 61 may be the same for the left and right feet, or may be different for the left and right feet.
- the vertical directions (Z-axis direction) of the sensors 61 placed in the left and right shoes 600 are the same.
- the three axes of the local coordinate system set in the sensor data derived from the left leg and the three axes of the local coordinate system set in the sensor data derived from the right leg are the same on the left and right.
- FIG. 25 is a conceptual diagram for explaining the local coordinate system (x-axis, y-axis, z-axis) set in the gait measuring device 60 (sensor 61) installed on the back side of the foot arch and the world coordinate system (X-axis, Y-axis, Z-axis) set with respect to the ground.
- the world coordinate system X-axis, Y-axis, Z-axis
- the user's lateral direction is set to the X-axis direction (leftward is positive)
- the user's back direction is set to the Y-axis direction (rearward is positive)
- the direction of gravity is set to the Z-axis direction (vertically upward is positive) when the user is standing upright facing the direction of travel. Note that the example in FIG.
- 25 conceptually shows the relationship between the local coordinate system (x-axis, y-axis, z-axis) and the world coordinate system (X-axis, Y-axis, Z-axis), and does not accurately show the relationship between the local coordinate system and the world coordinate system, which fluctuate according to the user's walking.
- FIG. 26 is a conceptual diagram for explaining a plane set for the human body (also called a human body plane).
- a sagittal plane that divides the body left and right a coronal plane that divides the body front and back, and a horizontal plane that divides the body horizontally are defined.
- the world coordinate system and the local coordinate system coincide with each other in a state in which the center line of the foot is directed in the direction of travel.
- rotation in the sagittal plane with the x-axis as the rotation axis is defined as roll
- rotation in the coronal plane with the y-axis as the rotation axis is defined as pitch
- rotation in the horizontal plane with the z-axis as the rotation axis is defined as yaw
- the rotation angle in the sagittal plane with the x-axis as the rotation axis is defined as the roll angle
- the rotation angle in the coronal plane with the y-axis as the rotation axis is defined as the pitch angle
- the rotation angle in the horizontal plane with the z-axis as the rotation axis is defined as the yaw angle.
- the feature amount data generation unit 62 (also called a feature amount data generation device) has an acquisition unit 621, a normalization unit 622, an extraction unit 623, a generation unit 625, and a feature amount data output unit 627.
- the feature amount data generator 62 is implemented by a microcomputer or microcontroller that performs overall control and data processing of the gait measuring device 60 .
- the feature amount data generator 62 has a CPU (Central Processing Unit), RAM (Random Access Memory), ROM (Read Only Memory), flash memory, and the like.
- the feature amount data generator 62 controls the acceleration sensor 611 and the angular velocity sensor 612 to measure angular velocity and acceleration.
- the feature amount data generator 62 may be mounted on the side of a mobile terminal (not shown) carried by the subject (user).
- the acquisition unit 621 acquires acceleration in three axial directions from the acceleration sensor 611 . Also, the acquisition unit 621 acquires angular velocities around the three axes from the angular velocity sensor 612 . For example, the acquisition unit 621 performs AD conversion (Analog-to-Digital Conversion) on physical quantities (analog data) such as the acquired angular velocity and acceleration. The physical quantities (analog data) measured by acceleration sensor 611 and angular velocity sensor 612 may be converted into digital data by acceleration sensor 611 and angular velocity sensor 612, respectively. The acquisition unit 621 outputs converted digital data (also referred to as sensor data) to the normalization unit 622 . The acquisition unit 621 may be configured to store sensor data in a storage unit (not shown).
- the sensor data includes at least acceleration data converted into digital data and angular velocity data converted into digital data.
- the acceleration data includes acceleration vectors in three axial directions.
- the angular velocity data includes angular velocity vectors around three axes. Acceleration data and angular velocity data are associated with acquisition times of those data. Further, the acquisition unit 621 may apply corrections such as mounting error correction, temperature correction, and linearity correction to the acceleration data and the angular velocity data.
- the normalization unit 622 acquires sensor data from the acquisition unit 621.
- the normalization unit 622 extracts time-series data (also referred to as walking waveform data) for one step cycle from the time-series data of the acceleration in the three-axis direction and the angular velocity around the three axes included in the sensor data.
- the normalization unit 622 normalizes (also referred to as first normalization) the time of the extracted walking waveform data for one step cycle to a walking cycle of 0 to 100% (percentage). Timings such as 1% and 10% included in the 0-100% walking cycle are also called walking phases.
- the normalization unit 622 normalizes (also referred to as second normalization) the walking waveform data for the first normalized step cycle so that the stance phase is 60% and the swing phase is 40%.
- the stance phase is the period during which at least part of the sole of the foot is in contact with the ground.
- the swing phase is the period during which the sole of the foot is off the ground.
- FIG. 27 is a conceptual diagram for explaining the step cycle based on the right foot.
- the step cycle based on the left foot is also the same as the right foot.
- the horizontal axis of FIG. 27 represents one gait cycle of the right foot starting when the heel of the right foot lands on the ground and ending when the heel of the right foot lands on the ground.
- the horizontal axis of FIG. 27 is first normalized with the stride period as 100%.
- the horizontal axis in FIG. 27 is second normalized so that the stance phase is 60% and the swing phase is 40%.
- One walking cycle of one leg is roughly divided into a stance phase in which at least part of the sole of the foot is in contact with the ground, and a swing phase in which the sole of the foot is separated from the ground.
- the stance phase is further subdivided into a load response period T1, a middle stance period T2, a final stance period T3, and an early swing period T4.
- the swing phase is further subdivided into early swing phase T5, middle swing phase T6, and final swing phase T7.
- FIG. 27 is an example, and does not limit the periods constituting the one-step cycle, the names of these periods, and the like.
- E1 represents an event (heel contact) in which the heel of the right foot touches the ground (HC: Heel Contact).
- E2 represents an event in which the toe of the left foot leaves the ground while the sole of the right foot is in contact with the ground (OTO: Opposite Toe Off).
- E3 represents an event (heel rise) in which the heel of the right foot is lifted while the sole of the right foot is in contact with the ground (HR: Heel Rise).
- E4 is an event in which the heel of the left foot touches the ground (opposite heel strike) (OHS: Opposite Heel Strike).
- E5 represents an event (toe off) in which the toe of the right foot leaves the ground while the sole of the left foot is in contact with the ground (TO: Toe Off).
- E6 represents an event (Foot Adjacent) in which the left foot and the right foot cross each other while the sole of the left foot is in contact with the ground (FA: Foot Adjacent).
- E7 represents an event (tibia vertical) in which the tibia of the right foot becomes almost vertical to the ground while the sole of the left foot is in contact with the ground (TV: Tibia Vertical).
- E8 represents an event (heel contact) in which the heel of the right foot touches the ground (HC: Heel Contact).
- E8 corresponds to the end point of the walking cycle starting from E1 and the starting point of the next walking cycle. Note that FIG. 27 is an example, and does not limit the events that occur during walking and the names of those events.
- FIG. 28 is a diagram for explaining an example of detecting heel contact HC and toe off TO from time-series data (solid line) of traveling direction acceleration (Y-direction acceleration).
- the timing of heel contact HC is the timing of the minimum peak immediately after the maximum peak appearing in the time-series data of traveling direction acceleration (Y-direction acceleration).
- the maximum peak that marks the timing of heel contact HC corresponds to the maximum peak of the walking waveform data for one step cycle.
- the interval between successive heel strikes HC is the stride period.
- the timing of the toe-off TO is the timing of the rise of the maximum peak that appears after the period of the stance phase in which no change appears in the time-series data of the acceleration in the traveling direction (the Y-direction acceleration).
- time-series data (dashed line) of the roll angle (angular velocity around the X-axis).
- the midpoint timing between the timing when the roll angle is minimum and the timing when the roll angle is maximum corresponds to the middle stage of stance.
- parameters also called gait parameters
- walking speed stride length
- circumcision internal rotation/external rotation
- plantarflexion/dorsiflexion etc.
- FIG. 29 is a diagram for explaining an example of walking waveform data normalized by the normalization unit 622.
- the normalization unit 622 detects heel contact HC and toe off TO from the time-series data of traveling direction acceleration (Y-direction acceleration).
- the normalization unit 622 extracts the interval between consecutive heel strikes HC as walking waveform data for one step cycle.
- the normalization unit 622 converts the horizontal axis (time axis) of the walking waveform data for one step cycle into a walking cycle of 0 to 100% by the first normalization.
- the walking waveform data after the first normalization is indicated by a dashed line.
- the timing of the toe take-off TO deviates from 60%.
- the normalization unit 622 normalizes the section from the heel contact HC at 0% in the walking phase to the toe-off TO following the heel contact HC to 0-60%. Further, the normalization unit 622 normalizes the section from the toe-off TO to the heel-contact HC in which the walking phase subsequent to the toe-off TO is 100% to 60% to 100%.
- the gait waveform data for one step cycle is normalized into a section of 0 to 60% of the gait cycle (stance phase) and a section of 60 to 100% of the gait cycle (swing phase).
- the walking waveform data after the second normalization is indicated by a solid line. In the second normalized walking waveform data (solid line), the timing of the toe take-off TO coincides with 60%.
- Figures 28 and 29 show an example of extracting/normalizing the walking waveform data for one step cycle based on the traveling direction acceleration (Y-direction acceleration).
- the normalization unit 622 extracts/normalizes walking waveform data for one step cycle in accordance with the walking cycle of the traveling direction acceleration (Y direction acceleration).
- the normalization unit 622 may also generate time-series data of angles about three axes by integrating time-series data of angular velocities about three axes.
- the normalization unit 622 also extracts/normalizes the walking waveform data for the one-step cycle in accordance with the walking cycle of the acceleration in the direction of travel (acceleration in the Y direction) with respect to the angles around the three axes.
- the normalization unit 622 may extract/normalize the walking waveform data for one step cycle based on the acceleration/angular velocity other than the traveling direction acceleration (Y-direction acceleration) (not shown). For example, the normalization unit 622 may detect heel contact HC and toe off TO from time series data of vertical direction acceleration (Z direction acceleration).
- the timing of the heel contact HC is the timing of a sharp minimum peak appearing in the time-series data of vertical acceleration (Z-direction acceleration). At the timing of the sharp minimum peak, the value of the vertical acceleration (Z-direction acceleration) becomes almost zero.
- the minimum peak that marks the timing of heel contact HC corresponds to the minimum peak of walking waveform data for one step cycle.
- the interval between successive heel strikes HC is the stride period.
- the timing of the toe-off TO is the timing of the inflection point in the middle of the time-series data of the vertical acceleration (Z-direction acceleration) gradually increasing after passing through a section with small fluctuations after the maximum peak immediately after the heel contact HC.
- the normalization unit 622 may extract/normalize the walking waveform data for one step cycle based on both the traveling direction acceleration (Y direction acceleration) and the vertical direction acceleration (Z direction acceleration). Further, the normalization unit 622 may extract/normalize the walking waveform data for one step cycle based on acceleration, angular velocity, angle, etc. other than the traveling direction acceleration (Y direction acceleration) and vertical direction acceleration (Z direction acceleration).
- the extraction unit 623 acquires the walking waveform data for the one-step cycle normalized by the normalization unit 622 .
- the extraction unit 623 extracts a feature amount used for estimating the physical condition from the walking waveform data for one step cycle.
- the extraction unit 623 extracts a feature amount for each walking phase cluster from walking phase clusters obtained by integrating temporally continuous walking phases based on preset conditions.
- a walking phase cluster includes at least one walking phase.
- a gait phase cluster also includes a single gait phase. The walking waveform data and the walking phase from which the feature amount used for estimating the physical condition is extracted will be described later.
- FIG. 30 is a conceptual diagram for explaining extraction of a feature amount for estimating a physical condition from walking waveform data for one step cycle.
- the extraction unit 623 extracts temporally consecutive walking phases I to I+m as a walking phase cluster C (I and m are natural numbers).
- the walking phase cluster C includes m walking phases (components). That is, the number of walking phases (constituent elements) constituting the walking phase cluster C (also referred to as the number of constituent elements) is m.
- FIG. 30 shows an example in which the walking phase is an integer value, the walking phase may be subdivided to decimal places.
- the number of constituent elements of the walking phase cluster C is a number corresponding to the number of data points in the section of the walking phase cluster.
- the extraction unit 623 extracts feature amounts from each of the walking phases I to I+m.
- the extraction unit 623 extracts the feature quantity from the single walking phase J (J is a natural number).
- the generation unit 625 applies the feature amount constitutive formula to the feature amount (first feature amount) extracted from each of the walking phases that make up the walking phase cluster, and generates the feature amount (second feature amount) of the walking phase cluster.
- the feature quantity constitutive formula is a calculation formula set in advance to generate the feature quantity of the walking phase cluster.
- the feature quantity configuration formula is a calculation formula regarding the four arithmetic operations.
- the second feature amount calculated using the feature amount construction formula is the integral average value, arithmetic average value, inclination, variation, etc. of the first feature amount in each walking phase included in the walking phase cluster.
- the generation unit 625 applies a calculation formula for calculating the slope and variation of the first feature amount extracted from each of the walking phases forming the walking phase cluster as the feature amount configuration formula. For example, if the walking phase cluster is composed of a single walking phase, the inclination and the variation cannot be calculated, so a feature value constitutive formula that calculates an integral average value or an arithmetic average value may be used.
- the feature amount data output unit 627 outputs feature amount data for each walking phase cluster generated by the generation unit 625 .
- the feature amount data output unit 627 outputs the feature amount data of the generated walking phase cluster to the estimation device 63 that uses the feature amount data.
- FIG. 31 is a block diagram showing an example of the configuration of the estimation device 63. As shown in FIG. The estimation device 63 has a data acquisition section 631 , a storage section 632 , an estimation section 633 and an output section 635 .
- the data acquisition unit 631 acquires feature amount data from the gait measurement device 60 .
- the data acquisition unit 631 outputs the received feature amount data to the estimation unit 633 .
- the data acquisition unit 631 may receive the feature amount data from the gait measurement device 60 via a cable such as a cable, or may receive the feature amount data from the gait measurement device 60 via wireless communication.
- the data acquisition unit 631 is configured to receive feature amount data from the gait measurement device 60 via a wireless communication function (not shown) conforming to standards such as Bluetooth (registered trademark) and WiFi (registered trademark).
- the communication function of the data acquisition unit 631 may conform to standards other than Bluetooth (registered trademark) and WiFi (registered trademark).
- the storage unit 632 stores an estimation model for estimating the physical condition using the feature amount data extracted from the walking waveform data.
- the storage unit 632 stores feature amount data relating to physical conditions of a plurality of subjects and an estimation model that has learned the relationship between the physical conditions.
- the storage unit 632 stores an estimation model for estimating the physical condition learned for a plurality of subjects.
- the storage unit 632 may store estimation models according to attributes.
- the estimation model may be stored in the storage unit 632 at times such as when the product is shipped from the factory or during calibration before the user uses the physical condition estimation system 6 .
- an estimation model stored in a storage device such as an external server may be used.
- the estimated model may be used via an interface (not shown) connected to the storage device.
- the estimation unit 633 acquires feature amount data from the data acquisition unit 631 .
- the estimation unit 633 estimates the physical state using the acquired feature amount data.
- the estimation unit 633 inputs the feature amount data to the estimation model stored in the storage unit 632 .
- the estimation unit 633 outputs an estimation result according to the physical condition output from the estimation model.
- the estimating unit 633 uses the estimated model via an interface (not shown) connected to the storage device.
- the output unit 635 outputs the estimation result of the physical condition by the estimation unit 633 .
- the output unit 635 displays the estimation result of the physical condition on the screen of the subject's (user's) mobile terminal.
- the output unit 635 outputs the estimation result to an external system or the like that uses the estimation result. No particular limitation is imposed on the use of the physical condition output from the estimating device 63 .
- the estimation device 63 is connected to an external system built on a cloud or server via a mobile terminal (not shown) carried by the subject (user).
- a mobile terminal (not shown) is a portable communication device.
- the mobile terminal is a mobile communication device having a communication function such as a smart phone, a smart watch, or a mobile phone.
- the estimating device 63 is connected to the mobile terminal via a wire such as a cable.
- the estimating device 63 is connected to the mobile terminal via wireless communication.
- the estimating device 63 is connected to the mobile terminal via a wireless communication function (not shown) conforming to standards such as Bluetooth (registered trademark) and WiFi (registered trademark).
- the communication function of the estimation device 63 may conform to standards other than Bluetooth (registered trademark) and WiFi (registered trademark).
- the body state estimation result may be used by an application installed on the mobile terminal. In that case, the mobile terminal executes processing using the estimation result by application software or the like installed in the mobile terminal.
- FIG. 32 is a conceptual diagram showing an example in which the learning system of the fifth embodiment inputs the feature quantities F1 to Fn extracted from the sensor data measured as the user walks into the estimation model 651 pre-built for estimating the physical condition, and outputs a score regarding the physical condition.
- the estimation model 651 outputs a physical condition score according to the input of the feature quantities F1 to Fn.
- the estimation model 651 is generated by learning using teacher data with the physical condition as the objective variable and the feature quantities F1 to Fn used for estimating the physical condition as explanatory variables.
- the estimation result of the estimation model 651 is not limited as long as the estimation result related to the physical condition is output according to the input of the feature amount data for estimating the physical condition.
- the estimation model 651 may be a model for estimating the physical condition using attributes such as gender, age, height, and weight as explanatory variables in addition to the feature quantities F1 to Fn used for estimating the physical condition.
- the storage unit 632 stores an estimation model for estimating the physical condition using the multiple regression prediction method.
- the storage unit 632 stores parameters for estimating the physical condition score S using Equation 1 below.
- S f1 x F1 + f2 x F2 + + fn x Fn + f0
- F1, F2, . f1, f2, . . . , fn are coefficients by which F1, F2, . f0 is a constant term.
- the storage unit 632 stores coefficients such as f1, f2, . . . , fn.
- the operation of the physical condition estimation system 6 will be described with reference to the drawings.
- the gait measuring device 60 and the estimating device 63 included in the physical state estimating system 6 will be individually described.
- the operation of the feature amount data generation unit 62 included in the gait measuring device 60 will be described.
- FIG. 33 is a flow chart for explaining the operation of the feature amount data generator 62 included in the gait measuring device 60. As shown in FIG. In the description according to the flowchart of FIG. 33, the feature amount data generation unit 62 will be described as an operating entity.
- the feature amount data generation unit 62 acquires time-series data of sensor data related to gait (step S601).
- the feature amount data generation unit 62 extracts walking waveform data for one step cycle from the time-series data of the sensor data (step S602).
- the feature amount data generator 62 detects heel contact and toe off from the time-series data of the sensor data.
- the feature amount data generator 62 extracts the time-series data of the section between successive heel strikes as walking waveform data for one step cycle.
- the feature amount data generation unit 62 normalizes the extracted walking waveform data for one step cycle (step S603).
- the feature amount data generator 62 normalizes the walking waveform data for one step cycle to a walking cycle of 0 to 100% (first normalization). Further, the feature amount data generator 62 normalizes the ratio of the stance phase and the swing phase of the walking waveform data for the first normalized step cycle to 60:40 (second normalization).
- the feature amount data generation unit 62 extracts feature amounts from the walking phases used for estimating the physical condition with respect to the normalized walking waveform (step S604). For example, the feature amount data generation unit 62 extracts feature amounts to be input to the estimation model constructed for each gender.
- the feature quantity data generation unit 62 uses the extracted feature quantity to generate a feature quantity for each walking phase cluster (step S605).
- the feature amount data generation unit 62 integrates the feature amounts for each walking phase cluster to generate feature amount data for the one step cycle (step S606).
- the feature amount data generation unit 62 outputs the generated feature amount data to the estimation device 63 (step S607).
- FIG. 33 is a flowchart for explaining the operation of the estimating device 63.
- the estimation device 63 will be described as an operating entity.
- the estimating device 63 acquires feature amount data generated using sensor data relating to gait (step S631).
- the estimating device 63 inputs the acquired feature amount data to the estimation model for estimating the physical state (step S632).
- the estimation device 63 estimates the user's physical condition according to the output (estimated value) from the estimation model (step S633).
- the estimation device 63 outputs information on the estimated physical condition (step S634).
- the physical condition is output to a terminal device (not shown) carried by the user.
- the physical state is output to a system that performs processing using the physical state.
- the function of the estimating device 63 installed in the mobile terminal carried by the user estimates the physical condition using feature amount data measured by the gait measuring device 60 placed on the shoe.
- FIG. 35 is a conceptual diagram showing an example of displaying the estimation result by the estimation device 63 on the screen of the mobile terminal 660 carried by the user walking wearing the shoes 600 on which the gait measurement device 60 is arranged.
- FIG. 35 shows an example of displaying on the screen of the portable terminal 660 information corresponding to the result of estimating the physical condition using feature amount data corresponding to sensor data measured while the user is walking.
- FIG. 35 is an example of information displayed on the screen of the mobile terminal 660 according to the estimated value of the physical condition.
- a score digitized according to a preset criterion is displayed on the display unit of the mobile terminal 660 as the estimated result of the physical condition.
- the information about the estimation result of the physical condition is displayed on the display unit of the portable terminal 660 according to the estimated value of the physical condition, such as "the total muscle strength of the whole body is declining.”
- the display unit of the mobile terminal 660 displays recommendation information according to the estimated total muscle strength, such as “Training A is recommended. Please see the video below.” After confirming the information displayed on the display unit of the mobile terminal 660, the user can practice training that leads to improvement of the total muscle strength of the whole body by exercising with reference to the training A video in accordance with the displayed recommendation information.
- the physical condition estimation system of this embodiment includes a gait measuring device and a physical condition estimation device.
- a gait measuring device includes a sensor and a feature amount data generator.
- the sensor has an acceleration sensor and an angular velocity sensor.
- the sensor measures spatial acceleration using an acceleration sensor.
- the sensor measures the spatial angular velocity using an angular velocity sensor.
- the sensor uses the measured spatial acceleration and spatial angular velocity to generate sensor data regarding foot movement.
- the sensor outputs the generated sensor data to the feature data generator.
- the feature amount data generation unit acquires time-series data of sensor data related to foot movement.
- the feature amount data generation unit extracts walking waveform data for one step cycle from the time-series data of the sensor data.
- the feature amount data generator normalizes the extracted walking waveform data.
- the feature amount data generation unit extracts, from the normalized walking waveform data, a feature amount related to the body condition of the estimation target from a walking phase cluster composed of at least one temporally continuous walking phase.
- the feature quantity data generator extracts the feature quantities selected by the feature quantity selection devices of the first to fourth embodiments.
- the feature amount data generation unit generates feature amount data including the extracted feature amount.
- the feature amount data generation unit outputs the generated feature amount data.
- a physical condition estimation device includes a data acquisition unit, a storage unit, an estimation unit, and an output unit.
- the data acquisition unit acquires feature amount data including feature amounts used for estimating the physical state of the user, which are extracted from the features of the user's gait.
- the storage unit stores an estimation model that outputs a physical state according to input of feature amount data.
- the estimation unit inputs the acquired feature amount data to the estimation model to estimate the physical state of the user.
- the output unit outputs information about the estimated physical condition.
- the physical condition estimation system of this embodiment estimates the user's physical condition using the feature amount extracted from the user's gait characteristics. Therefore, according to the physical condition estimation system of the present embodiment, it is possible to appropriately estimate the physical condition in daily life using feature values with high robustness.
- the information processing device 90 includes a processor 91, a main storage device 92, an auxiliary storage device 93, an input/output interface 95, and a communication interface 96.
- the interface is abbreviated as I/F (Interface).
- Processor 91 , main storage device 92 , auxiliary storage device 93 , input/output interface 95 , and communication interface 96 are connected to each other via bus 98 so as to enable data communication.
- the processor 91 , the main storage device 92 , the auxiliary storage device 93 and the input/output interface 95 are connected to a network such as the Internet or an intranet via a communication interface 96 .
- the processor 91 loads the program stored in the auxiliary storage device 93 or the like into the main storage device 92 .
- the processor 91 executes programs developed in the main memory device 92 .
- a configuration using a software program installed in the information processing device 90 may be used.
- the processor 91 executes processing according to each embodiment.
- the main storage device 92 has an area in which programs are expanded.
- a program stored in the auxiliary storage device 93 or the like is developed in the main storage device 92 by the processor 91 .
- the main memory device 92 is realized by a volatile memory such as a DRAM (Dynamic Random Access Memory). Further, as the main storage device 92, a non-volatile memory such as MRAM (Magnetoresistive Random Access Memory) may be configured/added.
- the auxiliary storage device 93 stores various data such as programs.
- the auxiliary storage device 93 is implemented by a local disk such as a hard disk or flash memory. It should be noted that it is possible to store various data in the main storage device 92 and omit the auxiliary storage device 93 .
- the input/output interface 95 is an interface for connecting the information processing device 90 and peripheral devices based on standards and specifications.
- a communication interface 96 is an interface for connecting to an external system or device through a network such as the Internet or an intranet based on standards and specifications.
- the input/output interface 95 and the communication interface 96 may be shared as an interface for connecting with external devices.
- Input devices such as a keyboard, mouse, and touch panel may be connected to the information processing device 90 as necessary. These input devices are used to enter information and settings.
- a touch panel is used as an input device, the display screen of the display device may also serve as an interface of the input device. Data communication between the processor 91 and the input device may be mediated by the input/output interface 95 .
- the information processing device 90 may be equipped with a display device for displaying information.
- the information processing device 90 is preferably provided with a display control device (not shown) for controlling the display of the display device.
- the display device may be connected to the information processing device 90 via the input/output interface 95 .
- the information processing device 90 may be equipped with a drive device. Between the processor 91 and a recording medium (program recording medium), the drive device mediates reading of data and programs from the recording medium, writing of processing results of the information processing device 90 to the recording medium, and the like.
- the drive device may be connected to the information processing device 90 via the input/output interface 95 .
- the above is an example of the hardware configuration for enabling the processing according to each embodiment of the present invention.
- the hardware configuration of FIG. 36 is an example of a hardware configuration for executing processing according to each embodiment, and does not limit the scope of the present invention.
- the scope of the present invention also includes a program that causes a computer to execute the processing according to each embodiment.
- the scope of the present invention also includes a program recording medium on which the program according to each embodiment is recorded.
- the recording medium can be implemented as an optical recording medium such as a CD (Compact Disc) or a DVD (Digital Versatile Disc).
- the recording medium may be implemented by a semiconductor recording medium such as a USB (Universal Serial Bus) memory or an SD (Secure Digital) card.
- the recording medium may be realized by a magnetic recording medium such as a flexible disk, or other recording medium.
- each embodiment may be combined arbitrarily. Also, the components of each embodiment may be realized by software or by circuits.
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Mathematical Physics (AREA)
- Pure & Applied Mathematics (AREA)
- Computational Mathematics (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Operations Research (AREA)
- Algebra (AREA)
- Evolutionary Biology (AREA)
- Databases & Information Systems (AREA)
- Bioinformatics & Computational Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- Computing Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Artificial Intelligence (AREA)
- Image Analysis (AREA)
Abstract
Description
まず、第1の実施形態に係る特徴量選定装置について、図面を参照しながら説明する。本実施形態の特徴量選定装置は、LASSO(Least Absolute Shrinkage and Selection Operator)回帰(以下、ラッソ回帰と呼ぶ)の手法を用いて、身体状態などの推定に用いられる特徴量を選定する。ラッソ回帰は、L1正則化とも呼ばれる。
図1は、本実施形態に係る特徴量選定装置10の構成の一例を示すブロック図である。特徴量選定装置10は、取得部11、構築部12、解析部13、統計部15、選定部17、および出力部19を備える。
次に、本実施形態の特徴量選定装置10の動作について、図面を参照しながら説明する。図9~図10は、特徴量選定装置10の動作の一例について説明するためのフローチャートである。図9~図10のフローチャートを用いた説明においては、特徴量選定装置10を動作主体として説明する。
次に、第2の実施形態に係る特徴量選定装置について、図面を参照しながら説明する。本実施形態の特徴量選定装置は、第1の実施形態の手法で選定された特徴量を用いて、推定モデルを構築する。本実施形態の特徴量選定装置は、構築された推定モデルの推定結果に応じて、特徴量を選定する。
図11は、本実施形態に係る特徴量選定装置20の構成の一例を示すブロック図である。特徴量選定装置20は、取得部21、構築部22、解析部23、統計部25、推定モデル構築部26、選定部27、および出力部29を備える。
次に、本実施形態の特徴量選定装置20の動作について、図面を参照しながら説明する。図12~図13は、特徴量選定装置20の動作の一例について説明するためのフローチャートである。図12~図13のフローチャートを用いた説明においては、特徴量選定装置20を動作主体として説明する。
次に、図13のステップS222のモデル評価処理について、図面を参照しながら説明する。図14は、モデル評価処理について説明するためのフローチャートである。図14のフローチャートを用いた説明においては、特徴量選定装置20を動作主体として説明する。
次に、第3の実施形態に係る特徴量選定装置について図面を参照しながら説明する。本実施形態の特徴量選定装置は、第1統計処理を省略し、複数の第1行列に関する各セルの平均値に関して第2統計処理を実行する点において、第1の実施形態とは異なる。
図15は、本実施形態に係る特徴量選定装置30の構成の一例を示すブロック図である。特徴量選定装置30は、取得部31、構築部32、解析部33、統計部35、3選定部37、および出力部39を備える。
次に、本実施形態の特徴量選定装置30の動作について、図面を参照しながら説明する。図16~図17は、特徴量選定装置30の動作の一例について説明するためのフローチャートである。図16~図17のフローチャートを用いた説明においては、特徴量選定装置30を動作主体として説明する。
次に、第4の実施形態に係る特徴量選定装置について図面を参照しながら説明する。本実施形態の特徴量選定装置は、第1~第3の実施形態の特徴量選定装置を簡略化した構成である。
次に、第5の実施形態に係る学習システムについて図面を参照しながら説明する。本実施形態の学習システムは、第1~第4の実施形態の特徴量選定装置によって選定された特徴量を用いた学習を実行する。
次に、学習装置55の詳細について図面を参照しながら説明する。図20は、学習装置55の詳細構成の一例を示すブロック図である。学習装置55は、受信部551、学習部553、および記憶部555を有する。
次に、第6の実施形態に係る身体状態推定システムについて図面を参照しながら説明する。本実施形態の身体状態推定システムは、ユーザの歩行に応じた足の動きに関するセンサデータを計測する。本実施形態の身体状態推定システムは、計測されたセンサデータを用いて、そのユーザの身体状態を推定する。例えば、本実施形態の身体状態推定システムは、身体状態として、握力や膝伸展力などの筋力指標や、動的バランス、下肢筋力、移動能力、静的バランスなどを推定する。なお、センサデータは、モーションキャプチャーやスマートアパレル等を用いて計測された、歩容に関する特徴を含むセンサデータであってもよい。
図22は、本実施形態に係る身体状態推定システム6の構成の一例を示すブロック図である。身体状態推定システム6は、歩容計測装置60と推定装置63を備える。本実施形態においては、歩容計測装置60と推定装置63が別々のハードウェアに構成される例について説明する。例えば、歩容計測装置60は、身体状態の推定対象である被験者(ユーザ)の履物等に設置される。例えば、推定装置63の機能は、被験者(ユーザ)の携帯する携帯端末にインストールされる。以下においては、歩容計測装置60および推定装置63の構成について、個別に説明する。
図23は、歩容計測装置60の構成の一例を示すブロック図である。歩容計測装置60は、センサ61と特徴量データ生成部62を有する。本実施形態においては、センサ61と特徴量データ生成部62が一体化された例を挙げる。センサ61と特徴量データ生成部62は、別々の装置として提供されてもよい。
図31は、推定装置63の構成の一例を示すブロック図である。推定装置63は、データ取得部631、記憶部632、推定部633、および出力部635を有する。
S=f1×F1+f2×F2+・・・+fn×Fn+f0・・・(1)
上記の式1において、F1、F2、・・・、Fnは、身体状態の推定に用いられる歩行フェーズクラスターごとの特徴量である。f1、f2、・・・、fnは、F1、F2、・・・、Fnに掛け合わされる係数である。f0は、定数項である。例えば、記憶部632には、f1、f2、・・・、fnなどの係数を記憶させておく。
次に、身体状態推定システム6の動作について図面を参照しながら説明する。ここでは、身体状態推定システム6に含まれる歩容計測装置60および推定装置63について、個別に説明する。歩容計測装置60に関しては、歩容計測装置60に含まれる特徴量データ生成部62の動作について説明する。
図33は、歩容計測装置60に含まれる特徴量データ生成部62の動作について説明するためのフローチャートである。図33のフローチャートに沿った説明においては、特徴量データ生成部62を動作主体として説明する。
図33は、推定装置63の動作について説明するためのフローチャートである。図33のフローチャートに沿った説明においては、推定装置63を動作主体として説明する。
次に、本実施形態に係る適用例について図面を参照しながら説明する。以下の適用例において、靴に配置された歩容計測装置60によって計測された特徴量データを用いて、ユーザが携帯する携帯端末にインストールされた推定装置63の機能が、身体状態を推定する例を示す。
ここで、本開示の各実施形態に係る処理を実行するハードウェア構成について、図36の情報処理装置90を一例として挙げて説明する。なお、図36の情報処理装置90は、各実施形態の処理を実行するための構成例であって、本開示の範囲を限定するものではない。
6 身体状態推定システム
10、20、30、40 特徴量選定装置
11、21、31、41 取得部
12、22、32、42 構築部
13、23、33、43 解析部
15、25、35、45 統計部
17、27、37、47 選定部
19、29、39、49 出力部
26 推定モデル構築部
50、60 歩容計測装置
55 学習装置
61 センサ
62 特徴量データ生成部
63 推定装置
551 受信部
553 学習部
555 記憶部
631 データ取得部
632 記憶部
633 推定部
635 出力部
Claims (10)
- 複数のデータセットを取得する取得手段と、
前記データセットに含まれるデータの分布を変更して複数の再抽出データセットを構築する構築手段と、
ラッソ回帰の手法を用いて、複数の前記再抽出データセットを解析する解析手段と、
複数の前記再抽出データセットの解析結果に応じて複数の前記再抽出データセットに含まれる要素の値を集計し、前記要素の値の集計結果に応じて複数の前記再抽出データセットに含まれる前記要素に論理値を設定する統計手段と、
予め設定された特定ルールに従って、前記要素に設定された前記論理値の値に応じた組み合わせの特徴量を選定する選定手段と、
選定された前記特徴量の組み合わせに関する選定情報を出力する出力手段と、を備える特徴量選定装置。 - 前記構築手段は、
Leave-One-Subject-Outの手法を用いて、複数の前記再抽出データセットを構築する請求項1に記載の特徴量選定装置。 - 前記構築手段は、
ブートストラップ法を用いて、複数の前記再抽出データセットを構築する請求項1に記載の特徴量選定装置。 - 前記解析手段は、
複数の前記再抽出データセットに関して、予め設定された複数の正則化パラメータごとに前記ラッソ回帰を実行し、
前記ラッソ回帰で用いられた前記正則化パラメータに対応する列と、前記特徴量に対応する行とによって構成される複数パターンの第1行列を生成し、
前記統計手段は、
複数パターンの前記第1行列に関して、非ゼロ要素のセルの第1論理値を1に設定して、ゼロ要素のセルの第1論理値を0に設定する第1統計処理を実行し、
複数パターンの前記第1行列を構成するセルごとに前記第1論理値を集計して、前記第1論理値の集計値が所定条件を満たすセルに第2論理値として1が設定され、前記第1論理値の集計値が前記所定条件を満たさないセルに前記第2論理値として0が設定された第2行列を生成する第2統計処理を実行し、
前記選定手段は、
予め設定された前記特定ルールに従って前記第2行列の列を選択し、選択された列に対応する前記特徴量の組み合わせを選定する請求項1乃至3のいずれか一項に記載の特徴量選定装置。 - 前記統計手段は、
前記第2統計処理において、
複数パターンの前記第1行列を構成するセルごとに前記第1論理値の合計値を計算し、
前記第1論理値の合計値が所定閾値以上のセルの前記第2論理値が1に設定され、前記第1論理値の合計値が所定閾値未満のセルが0に設定された前記第2行列を生成する請求項4に記載の特徴量選定装置。 - 前記統計手段は、
前記第2統計処理において、
複数パターンの前記第1行列を構成するセルごとに前記第1論理値の平均値を計算し、
前記第1論理値の平均値が所定閾値以上のセルの前記第2論理値が1に設定され、前記第1論理値の平均値が所定閾値未満のセルが0に設定された前記第2行列を生成する請求項4に記載の特徴量選定装置。 - 前記解析手段は、
複数の前記再抽出データセットの各々に関して、予め設定された複数の正則化パラメータごとに前記ラッソ回帰を実行し、
前記ラッソ回帰で用いられた前記正則化パラメータに対応する列と、前記特徴量に対応する行とによって構成される複数パターンの第1行列を生成し、
前記統計手段は、
複数パターンの前記第1行列を構成するセルごとに前記要素の値を集計して、前記要素の値の平均値が所定閾値以上のセルに第2論理値として1が設定され、前記要素の値の平均値が前記所定閾値未満のセルに前記第2論理値として0が設定された第2行列を生成する第2統計処理を実行し、
前記選定手段は、
予め設定された前記特定ルールに従って前記第2行列の列を選択し、選択された列に対応する前記特徴量の組み合わせを選定する請求項1乃至3のいずれか一項に記載の特徴量選定装置。 - 選定された前記特徴量を用いた学習により推定モデルを構築し、構築された前記推定モデルを評価する推定モデル構築手段を備え、
前記選定手段は、
前記推定モデルの評価結果に応じて、前記特徴量の組み合わせを選定する請求項1乃至7のいずれか一項に記載の特徴量選定装置。 - コンピュータが、
複数のデータセットを取得し、
前記データセットに含まれるデータの分布を変更して複数の再抽出データセットを構築し、
ラッソ回帰の手法を用いて、複数の前記再抽出データセットを解析し、
複数の前記再抽出データセットの解析結果に応じて複数の前記再抽出データセットに含まれる要素の値を集計し、
前記要素の値の集計結果に応じて複数の前記再抽出データセットに含まれる前記要素に論理値を設定し、
予め設定された特定ルールに従って、前記要素に設定された前記論理値の値に応じた組み合わせの特徴量を選定し、
選定された前記特徴量の組み合わせに関する選定情報を出力する特徴量選定方法。 - 複数のデータセットを取得する処理と、
前記データセットに含まれるデータの分布を変更して複数の再抽出データセットを構築する処理と、
ラッソ回帰の手法を用いて、複数の前記再抽出データセットを解析する処理と、
複数の前記再抽出データセットの解析結果に応じて複数の前記再抽出データセットに含まれる要素の値を集計する処理と、
前記要素の値の集計結果に応じて複数の前記再抽出データセットに含まれる前記要素に論理値を設定する処理と、
予め設定された特定ルールに従って、前記要素に設定された前記論理値の値に応じた組み合わせの特徴量を選定する処理と、
選定された前記特徴量の組み合わせに関する選定情報を出力する処理と、をコンピュータに実行させるプログラムを記録させた非一過性の記録媒体。
Priority Applications (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2023574969A JPWO2023139718A5 (ja) | 2022-01-20 | 特徴量選定装置、特徴量選定方法、およびプログラム | |
PCT/JP2022/001953 WO2023139718A1 (ja) | 2022-01-20 | 2022-01-20 | 特徴量選定装置、特徴量選定方法、身体状態推定システム、および記録媒体 |
US18/717,019 US20250045354A1 (en) | 2022-01-20 | 2022-01-20 | Feature quantity selection device, feature quantity selection method, and recording medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
PCT/JP2022/001953 WO2023139718A1 (ja) | 2022-01-20 | 2022-01-20 | 特徴量選定装置、特徴量選定方法、身体状態推定システム、および記録媒体 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2023139718A1 true WO2023139718A1 (ja) | 2023-07-27 |
Family
ID=87348296
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2022/001953 WO2023139718A1 (ja) | 2022-01-20 | 2022-01-20 | 特徴量選定装置、特徴量選定方法、身体状態推定システム、および記録媒体 |
Country Status (2)
Country | Link |
---|---|
US (1) | US20250045354A1 (ja) |
WO (1) | WO2023139718A1 (ja) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2025032627A1 (ja) * | 2023-08-04 | 2025-02-13 | 日本電気株式会社 | 情報提供装置、情報提供システム、情報提供方法、および記録媒体 |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2018151883A (ja) * | 2017-03-13 | 2018-09-27 | 株式会社東芝 | 解析装置、解析方法、及び、プログラム |
WO2021229648A1 (ja) * | 2020-05-11 | 2021-11-18 | 日本電気株式会社 | 数式モデル生成システム、数式モデル生成方法および数式モデル生成プログラム |
-
2022
- 2022-01-20 US US18/717,019 patent/US20250045354A1/en active Pending
- 2022-01-20 WO PCT/JP2022/001953 patent/WO2023139718A1/ja active Application Filing
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2018151883A (ja) * | 2017-03-13 | 2018-09-27 | 株式会社東芝 | 解析装置、解析方法、及び、プログラム |
WO2021229648A1 (ja) * | 2020-05-11 | 2021-11-18 | 日本電気株式会社 | 数式モデル生成システム、数式モデル生成方法および数式モデル生成プログラム |
Non-Patent Citations (1)
Title |
---|
ENDO, AKIHIRO; NAGATA, KENJI; KIDO, SHOJI; SHOUNO, HAYARU: "Feature Selection for Diffuse Lung Disease Images by Bolasso", IPSJ JOURNAL, vol. 12, no. 3, 1 December 2019 (2019-12-01), pages 68 - 77, XP009547814, ISSN: 1882-7780 * |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2025032627A1 (ja) * | 2023-08-04 | 2025-02-13 | 日本電気株式会社 | 情報提供装置、情報提供システム、情報提供方法、および記録媒体 |
Also Published As
Publication number | Publication date |
---|---|
US20250045354A1 (en) | 2025-02-06 |
JPWO2023139718A1 (ja) | 2023-07-27 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
JPWO2018003910A1 (ja) | 歩行状態判定装置、歩行状態判定システム、歩行状態判定方法及びプログラム | |
US20240350034A1 (en) | Estimation device, information presentation system, estimation method, and recording medium | |
WO2023139718A1 (ja) | 特徴量選定装置、特徴量選定方法、身体状態推定システム、および記録媒体 | |
WO2023286106A1 (ja) | 歩容評価装置、歩容評価方法、歩容計測システム、および記録媒体 | |
WO2020240752A1 (ja) | 情報処理装置、体重推定装置、体重推定システム、情報処理方法及び記憶媒体 | |
WO2023127015A1 (ja) | 筋力評価装置、筋力評価システム、筋力評価方法、および記録媒体 | |
WO2022038663A1 (ja) | 検出装置、検出システム、検出方法、およびプログラム記録媒体 | |
WO2022201338A1 (ja) | 特徴量生成装置、歩容計測システム、特徴量生成方法、および記録媒体 | |
US20240138757A1 (en) | Pelvic inclination estimation device, estimation system, pelvic inclination estimation method, and recording medium | |
US20240172966A1 (en) | Harmonic index estimation device, estimation system, harmonic index estimation method, and recording medium | |
JP7704216B2 (ja) | 特徴量データ生成装置、歩容計測装置、身体状態推定システム、特徴量データ生成方法、およびプログラム | |
US20250031995A1 (en) | Dynamic balance estimation device, dynamic balance estimation system, dynamic balance estimation method, and recording medium | |
US20250046464A1 (en) | Static balance estimation device, static balance estimation system, static balance estimation method, and recording medium | |
US20240138710A1 (en) | Waist swinging estimation device, estimation system, waist swinging estimation method, and recording medium | |
US20250032000A1 (en) | Fall probability estimation device, fall probability estimation system, fall probability estimation method, and recording medium | |
WO2023127007A1 (ja) | 筋力指標推定装置、筋力指標推定システム、筋力指標推定方法、および記録媒体 | |
WO2022208838A1 (ja) | 生体情報処理装置、情報処理システム、生体情報処理方法、および記録媒体 | |
WO2023105740A1 (ja) | 特徴量データ生成装置、歩容計測装置、身体状態推定システム、特徴量データ生成方法、および記録媒体 | |
US20240237922A1 (en) | Estimation device, estimation system, estimation method, and recording medium | |
Gibson et al. | Foot orientation and trajectory variability in locomotion: effects of real-world terrain | |
Chen et al. | Smart Gait Detection and Analysis | |
WO2023127010A1 (ja) | 移動能力推定装置、移動能力推定システム、移動能力推定方法、および記録媒体 | |
JP7626221B2 (ja) | 補間装置、歩容計測システム、補間方法、およびプログラム | |
WO2023157161A1 (ja) | 検出装置、検出システム、歩容計測システム、検出方法、および記録媒体 | |
WO2023127009A1 (ja) | 下肢筋力推定装置、下肢筋力推定システム、下肢筋力推定方法、および記録媒体 |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 22921876 Country of ref document: EP Kind code of ref document: A1 |
|
WWE | Wipo information: entry into national phase |
Ref document number: 18717019 Country of ref document: US |
|
ENP | Entry into the national phase |
Ref document number: 2023574969 Country of ref document: JP Kind code of ref document: A |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 22921876 Country of ref document: EP Kind code of ref document: A1 |