JP7630184B2 - System and method for predicting exercise effect - Google Patents

Description

Application of Article 30, Paragraph 2 of the Patent Act Poster presentation at the 77th Annual Meeting of the Japan Society of Physical Fitness and Sports Medicine, which was held entirely online, on September 22, 2022 [Publications, etc.] Published in the "Program and Abstracts of the 77th Annual Meeting of the Japan Society of Physical Fitness and Sports Medicine" published by the 77th Annual Meeting Secretariat in August 2022 [Publications, etc.] Published on the 77th Annual Meeting of the Japan Society of Physical Fitness and Sports Medicine's website, "https://right-stuff.biz/jspfsm77/," on September 16, 2022 [Publications, etc.] Published in "The Journal of Physical Fitness and Sports Medicine (JPFSM) Vol. 11, No. 6," published by the Editorial Board of the Japan Society of Physical Fitness and Sports Medicine on November 25, 2022 [Publications, etc.] Posted on November 25, 2022 on the website of the Japan Science and Technology Agency (J-STAGE) at https://www.jstage.jst.go.jp/browse/jpfsm/11/6/_contents/-char/en

The present invention relates to a system and method for predicting exercise effects.

Patent Document 1 proposes an exercise effect prediction system that accurately predicts the degree to which measured values such as body weight, muscle strength, and blood sugar level will improve if a predetermined exercise is performed continuously for a certain period of time. In this exercise effect prediction system, initial measurement values measured before the exercise is prescribed or immediately before and average weekly fast walking time are input into an initial condition input screen. The input initial measurement values are determined by an initial attribute group determination means to which of the initial attribute groups classified based on statistical data the initial measurement values belong. When the initial attribute group is determined, an effect prediction means substitutes the average weekly fast walking time into an effect estimation formula prepared in advance for each initial attribute group to calculate the amount of change (effect) in the initial measurement values after about five months. The effect estimation formula used to calculate the effect is obtained by analyzing statistical data for each initial attribute group.

JP 2008-073339 A

It is known that exercise is effective in improving physical fitness and preventing lifestyle-related diseases. However, it is not clear to what extent exercise duration or intensity is effective in improving physical fitness or preventing lifestyle-related diseases. For example, even if increasing exercise intensity helps improve physical fitness, there are known dangers and disadvantages to increasing exercise intensity too much, and on the other hand, it is possible that no effect will be obtained if the exercise intensity is not increased. Furthermore, even if the exercise intensity is increased, there is a possibility that no effect will be obtained if the exercise time is insufficient, so it is important to provide an appropriate exercise prescription that is suited to the user's current situation and future (goals).

One aspect of the present invention is a system for predicting the effect of continuing training in which the time during which a given intensity is reached reaches a given time on average per week. By analyzing the exercise effect of a user who continued this training, the inventors of the present application found that, with regard to the effect of the predicted items related to improving physical fitness or improving lifestyle-related diseases, there are items that depend on the exercise intensity (exercise intensity-dependent group) and items that do not depend on the exercise intensity (exercise intensity-independent group). In this specification, the definition of "independent of exercise intensity" or "independent of exercise intensity" means that the exercise intensity has reached a "predetermined intensity" of the exercise specified as the target training, and is not dependent on or independent of exercise intensity equal to or greater than this "predetermined intensity."

Based on this new viewpoint, the system of the present invention has an input interface for acquiring prediction target items related to improving physical fitness or lifestyle-related diseases, a first prediction unit for acquiring, via the input interface, at least the initial characteristics of the user for the prediction target item, the weekly average time planned during continued training, and the exercise intensity planned during continued training if the prediction target item is included in the exercise intensity-dependent group, and outputting a predicted effect of the prediction target item after continuing training for a predetermined period of time using an exercise intensity-dependent estimation formula that includes at least the initial characteristics, the weekly average time, and the exercise intensity as elements, and a second prediction unit for acquiring, via the input interface, at least the initial characteristics of the user for the prediction target item and the weekly average time planned during continued training if the prediction target item is included in the exercise intensity-independent group, and outputting a predicted effect of the prediction target item after continuing training for a predetermined period of time using an exercise intensity-independent estimation formula that includes at least the initial characteristics and the weekly average time as elements.

Furthermore, the inventors have found that the effect of the predicted item in the exercise intensity-dependent group is related to the gender of the user. Therefore, the first prediction unit may acquire the gender of the user via the input interface, and output the predicted effect of the predicted item after continuing training for a predetermined period of time using an exercise intensity-dependent estimation equation that includes at least gender, initial characteristics, average weekly time, and exercise intensity as elements.

Examples of prediction target items included in the exercise intensity-dependent group are maximum oxygen intake, body mass index, and minimum blood pressure, and examples of prediction target items included in the exercise intensity-independent group are blood glucose level, HDL cholesterol, and arteriosclerosis index. An example of training is interval walking training with a specified intensity set to 70% of maximum oxygen intake and a specified average weekly time set to 60 minutes.

The system may further include a first interface that provides the user with information including that the predicted effect depends on exercise intensity when the prediction target item is included in the exercise intensity-dependent group. The system may also include a second interface that provides the user with information including that the predicted effect does not depend on exercise intensity when the prediction target item is included in the exercise intensity-independent group. By clearly indicating to the user whether the effect of the prediction target item depends on exercise intensity, for example, by explicitly adding or omitting an input item for exercise intensity, or by clearly indicating that the effect depends or does not depend on exercise intensity, it is possible to encourage the user to actively increase exercise intensity or not to increase exercise intensity unnecessarily.

Another aspect of the present invention is a method for predicting the effect of continuing training in which the time to reach a predetermined intensity reaches a predetermined time on a weekly average, using an effect prediction system. The effect prediction system has an input interface for acquiring prediction target items related to improving physical fitness or improving lifestyle-related diseases, and a library that stores, for prediction target items included in an exercise intensity-dependent group, an exercise intensity-dependent estimation formula that includes at least the initial characteristics of the user for the prediction target item, the weekly average time planned during the continuation of the training, and the exercise intensity planned during the continuation of the training as elements, and stores, for prediction target items included in an exercise intensity-independent group, an exercise intensity-independent estimation formula that includes at least the initial characteristics and the weekly average time as elements.

The method includes, if the prediction target item is included in the exercise intensity-dependent group, acquiring at least the initial characteristics, weekly average time, and exercise intensity via an input interface, and outputting a predicted effect of the prediction target item in the exercise intensity-dependent group after continuing training for a predetermined period of time using an exercise intensity-dependent estimation equation that includes at least these as elements; and, if the prediction target item is included in the exercise intensity-independent group, acquiring at least the initial characteristics and weekly average time via the input interface, and outputting a predicted effect of the prediction target item in the exercise intensity-independent group after continuing training for a predetermined period of time using an exercise intensity-independent estimation equation that includes at least these as elements.

The library may store an exercise intensity-dependent estimation formula for a prediction target item included in the exercise intensity-dependent group, the formula including at least the user's gender, the user's initial characteristics of the prediction target item, the average weekly time planned during the continuation of training, and the exercise intensity planned during the continuation of training as elements. Outputting the predicted effect of the prediction target item in the exercise intensity-dependent group may include outputting the predicted effect of the prediction target item after continuing training for a predetermined period of time using the exercise intensity-dependent estimation formula including at least the gender, initial characteristics, average weekly time, and exercise intensity as elements.

This method may include providing the user with information including that the predicted effect depends on exercise intensity when the prediction target item is included in the exercise intensity-dependent group, and may include providing the user with information including that the predicted effect does not depend on exercise intensity when the prediction target item is included in the exercise intensity-independent group.

Another aspect of the present invention is a method for predicting the effect of continuing training involving exercise in which the time at which a predetermined intensity is reached reaches a predetermined time on average per week, the method comprising the steps of:
- Obtaining prediction items related to improving physical fitness or improving lifestyle-related diseases.
- If the item to be predicted is included in the exercise intensity-dependent group, at least the user's initial characteristics for the item to be predicted, the average weekly time planned during the continuation of training, and the exercise intensity planned during the continuation of training are obtained via the input interface, and a predicted effect of the item to be predicted after continuing training for a specified period is output using an exercise intensity-dependent estimation equation that includes at least the initial characteristics, the average weekly time, and the exercise intensity as elements.
- If the item to be predicted is included in the exercise intensity-independent group, at least the user's initial characteristics for the item to be predicted and the weekly average time planned during the continuation of training are obtained via the input interface, and the predicted effect of the item to be predicted after continuing training for a specified period is output using an exercise intensity-independent estimation equation that includes at least the initial characteristics and the weekly average time as elements.

Another aspect of the present invention is a program for predicting the effect of continuing training in which the time to reach a predetermined intensity reaches a predetermined time on average per week, by a computer, the program having instructions for the computer to execute the following: acquiring a prediction target item related to improving physical fitness or improving lifestyle-related diseases; if the prediction target item is included in the exercise intensity-dependent group, acquiring at least the initial characteristics of the user of the prediction target item, the weekly average time planned during the continuation of the training, and the exercise intensity planned during the continuation of the training via the input interface, and outputting the predicted effect of the prediction target item after continuing the training for a predetermined period of time using an exercise intensity-dependent estimation formula that includes at least the initial characteristics, the weekly average time, and the exercise intensity as elements; and if the prediction target item is included in the exercise intensity-independent group, acquiring at least the initial characteristics of the user of the prediction target item and the weekly average time planned during the continuation of the training via the input interface, and outputting the predicted effect of the prediction target item after continuing the training for a predetermined period of time using an exercise intensity-independent estimation formula that includes at least the initial characteristics and the weekly average time as elements.

FIG. 1 is a block diagram showing an overview of an exercise effect prediction system. A diagram showing an overview of interval walking training. A graph showing the effects of interval walking training on knee extension strength, knee flexion strength, and endurance index (peak oxygen uptake). A graph showing the effects of interval walking training on systolic and diastolic blood pressure. 1 is a flowchart showing an analysis procedure. A graph showing characteristic values of subjects included in the determinant identification group and the accuracy validation group. FIG. 13 is a diagram showing an estimated equation obtained by analysis. Analysis results including total walking time, change in peak oxygen intake ΔVO2peak, change in body mass index ΔBMI, change in diastolic blood pressure ΔDBP, change in blood glucose level ΔFBG, change in blood cholesterol ΔHDL, and change in arteriosclerosis index Δ(LDL/HDL) as independent variables. Analysis results including weekly brisk walking time as independent variables for change in maximum oxygen intake ΔVO2peak, change in body mass index ΔBMI, change in diastolic blood pressure ΔDBP, change in blood glucose level ΔFBG, change in blood cholesterol ΔHDL, and change in arteriosclerosis index Δ(LDL/HDL). Analysis results including weekly brisk walking time and brisk walking exercise intensity as independent variables for change in maximum oxygen intake ΔVO2peak, change in body mass index ΔBMI, change in diastolic blood pressure ΔDBP, change in blood glucose level ΔFBG, change in blood cholesterol ΔHDL, and change in arteriosclerosis index Δ(LDL/HDL). Analysis results including change in body mass index ΔBMI, change in diastolic blood pressure ΔDBP, change in blood glucose level ΔFBG, change in blood cholesterol ΔHDL, change in arteriosclerosis index Δ(LDL/HDL), and change in estimated maximum oxygen intake ΔVO2peak as independent variables. 13 shows the results of evaluation of the actual and estimated values of the determinant identification group (a) and accuracy verification group (b) of the estimation formula that includes the exercise intensity of the maximum oxygen intake change ΔVO2peak as an independent variable. 13 shows the results of evaluation of the actual and estimated values of the determinant identification group (a) and accuracy verification group (b) of an estimation formula that does not include the exercise intensity of the maximum oxygen uptake change ΔVO2peak as an independent variable. Evaluation results of the actual measured values and estimated values of the group (a) for identifying the determinants of the estimation formula for body mass index change ΔBMI and the group (b) for verifying accuracy. Evaluation results of the actual measured values and estimated values of the determinant factor identification group (a) and the accuracy verification group (b) of the estimation formula for diastolic blood pressure change ΔDBP. Evaluation results of the actual measured values and estimated values of the determinant factor identification group (a) and the accuracy verification group (b) of the estimation formula for blood glucose change ΔFBG. Evaluation results of the actual measured values and estimated values of the determinant identification group (a) and the accuracy verification group (b) of the estimated formula for blood cholesterol change ΔHDL. 13 shows the evaluation results of the actual measured values and estimated values of the determinant identification group (a) and the accuracy verification group (b) of the estimation formula for the arteriosclerosis index change Δ(LDL/HDL). 1 is a flowchart showing an overview of the processing of the effect prediction system.

Figure 1 shows an example of a system for predicting the effect of continuing a specified training. In this system 1, a user 6 uses a terminal 5 such as a PC or smartphone to access a server (exercise effect prediction information providing system, effect prediction system) 10 of a provider that provides a service of providing exercise effect prediction information via the Internet 9, and inputs the specified user's information (user information) 7 to obtain exercise effect prediction information 8. The exercise effect prediction provided by the effect prediction system 10 is a predicted effect when training is continued in which the time at which a specified intensity is reached reaches a specified time on average per week. One example of training is interval walking training (hereinafter, interval walking) in which the specified intensity is set to 70% of the maximum oxygen intake and the specified time on average per week is set to 60 minutes or more.

The effect prediction system (information provision system) 10 includes computer resources such as a CPU and memory. The effect prediction system 10 includes an input/output interface 15 that acquires user information 7 and provides predicted effect information 8. The user information 7 includes prediction target items related to improving physical fitness or improving lifestyle-related diseases. The information provision system 10 includes a library 20 that stores estimation formulas for predicting exercise effects for each prediction target item. The library 20 stores, for prediction target items included in the exercise intensity-dependent group 21, an exercise intensity-dependent estimation formula that includes at least the initial characteristics of the user for the prediction target item, the average weekly time planned during the continuation of training, and the exercise intensity planned during the continuation of training as elements, and, for prediction target items included in the exercise intensity-independent group 22, an exercise intensity-independent estimation formula that includes at least the initial characteristics and the average weekly time as elements.

Examples of prediction target items included in the exercise intensity dependent group 21 are maximum oxygen intake, body mass index, and minimum blood pressure, and the library 20 includes estimation equations that estimate the maximum oxygen intake change ΔVO2peak as the predicted effect of maximum oxygen intake, the body mass index change ΔBMI as the predicted effect of body mass index, and the minimum blood pressure change ΔDBP as the predicted effect of minimum blood pressure.

Examples of prediction target items included in the exercise intensity independent group 22 are blood glucose level, HDL cholesterol, and arteriosclerosis index, and the library 20 includes estimation formulas that estimate blood glucose level change ΔFBG as the predicted effect of blood glucose level, HDL cholesterol change ΔHDL as the predicted effect of HDL cholesterol, and arteriosclerosis index change Δ(LDL/HDL) as the predicted effect of arteriosclerosis index. In this specification, exercise intensity independence means that exercise at a specified intensity specified to obtain the effect of training is performed, and the effect of training does not depend on exercise intensity exceeding the specified intensity, and does not allow exercise intensity below the specified intensity.

The effect prediction system 10 includes a first prediction unit 11 that predicts the predicted effect of a target item in the exercise intensity-dependent group 21 using an estimation formula, and a second prediction unit 12 that predicts the predicted effect of a target item in the exercise intensity-independent group 22 using an estimation formula. If the prediction target item is included in the exercise intensity-dependent group 21, the first prediction unit 11 acquires user information 7 including the initial characteristics of the user of the prediction target item, the average weekly time planned during the continuation of training, and the exercise intensity planned during the continuation of training via the input/output interface 15, and outputs the predicted effect of the prediction target item after continuing training for a predetermined period, in this example, five months, using the estimation formula of the exercise intensity-dependent group 21 that includes these as elements. If the prediction target item is included in the exercise intensity-independent group 22, the second prediction unit 12 acquires the initial characteristics of the user of the prediction target item and the average weekly time planned during the continuation of training via the input interface 15, and outputs the predicted effect of the prediction target item after continuing training for a predetermined period using the estimation formula of the exercise intensity-independent group 22 that includes these as elements.

The estimation formula for the exercise intensity-dependent group 21 may include the user's gender S as an element, in which case the first prediction unit 11 acquires user information 7 including the user's gender S via the input interface 15, and uses the estimation formula for the exercise intensity-dependent group 21 including the user information 7 to output the predicted effect of the predicted item after continuing training for a specified period of time.

The input/output interface 15 includes a first interface 16 that provides the user with information including that the predicted effect depends on exercise intensity when the prediction target item is included in the exercise intensity-dependent group, and a second interface 17 that provides the user with information including that the predicted effect does not depend on exercise intensity (exercise intensity equal to or greater than a predetermined exercise intensity (minimum exercise intensity) required for training) when the prediction target item is included in the exercise intensity-independent group. These interfaces 16 and 17 clearly indicate to the user 6 whether the predicted effect of the prediction target item depends on exercise intensity (exercise intensity equal to or greater than a predetermined exercise intensity (minimum exercise intensity) required for training). For example, these interfaces 16 and 17 explicitly add or omit the input item for exercise intensity, or clearly indicate that the information depends or does not depend on exercise intensity, thereby encouraging the user 6 to actively increase the exercise intensity in order to obtain the predicted effect, or not to increase the exercise intensity unnecessarily. Only the first interface 16 or the second interface 17 may be provided.

Figure 2 shows an overview of interval walking training (IWT). Interval walking training is training in which the predetermined intensity (exercise intensity) is set to 70% of the maximum oxygen intake VO2peak, and the predetermined weekly average time is set to 60 minutes or more. As shown in Figure 2, the training is performed by repeating walking at a predetermined intensity or less and fast walking at a predetermined intensity or more at appropriate intervals, for example, 5-minute intervals, for 15 minutes a day, 4 days a week or more, so that the average (weekly average time) of fast walking in one week is 60 minutes or more. Therefore, the weekly average time (planned weekly average time) Wa planned by the user 6 included in the user information 7 of this effect prediction system 10 is the weekly average time of fast walking in interval walking training, and the exercise intensity (planned exercise intensity) Wi planned during the training is the ratio to the maximum oxygen intake VO2peak of fast walking in interval walking training. For this reason, the weekly average time Wa included in the user information 7 is assumed to be 60 minutes or more, and the exercise intensity Wi is assumed to be 70% or more.

It is known that the value of the maximum oxygen intake VO2peak, which is an index of training, can be estimated by several methods. Of course, it may be an exhaled gas analysis method using equipment for measuring physical fitness such as a bicycle ergometer or a treadmill. One method for obtaining the maximum oxygen intake VO2peak is the three-stage step-up walking method, in which the load is gradually increased for three minutes each for rest, slow walking, medium walking, and fastest walking, and the energy consumption (oxygen intake) for the last three minutes of the fastest walking is obtained from the output of a three-axis accelerometer and an altimeter. This method significantly simplifies the measurement of the maximum oxygen intake, and allows physical fitness to be measured with sufficient accuracy. The oxygen intake per unit time during walking eVO2W can be estimated by the following formula. The estimation formula (7) for calculating the estimated value eVO2W of oxygen intake during walking is disclosed in “Yamazaki T, Gen-no H, Kamijo Y, Okazaki K, Masuki S, and Nose H. A new device to estimate VO2 during incline walking by accelerometry and barometry. Med Sci Sports Exerc, 41: 2213-2219, 2009” by the inventors of the present application.
eVO2W=w1・VM+w2・Hu+w3・Hd...(1)
VM is the vector norm of the three-axis acceleration sensor, i.e., the accumulated acceleration per unit time (G/min, G is the gravitational acceleration) obtained by accumulating the acceleration in each direction, Hu is the amount of rise per unit time (rising speed, m/min), and Hd is the amount of fall per unit time (falling speed, m/min). When a barometer is used, the rising speed and falling speed H (amount of rise and fall per unit time) are used. w1, w2, and w3 are coefficients.

The value (estimated value) of oxygen intake VO2 and the walking time during interval walking training can be automatically recorded by a measuring device 2 worn by the user 6 or a smartphone app, etc. Therefore, the user 6 can perform interval walking training so as to satisfy the weekly average time Wa and exercise intensity Wi that he or she has set (planned).

Figures 3 and 4 show an example of the changes observed by the inventors in improving physical fitness and improving lifestyle-related diseases when interval walking training is continued for five months. White indicates the tendency for men, diagonal lines indicate the tendency for women, and black indicates the overall tendency. Interval walking training is effective in improving physical fitness and improving lifestyle-related diseases, such as knee extension strength, knee flexion strength, endurance index, systolic blood pressure, and diastolic blood pressure, and it is clear that interval walking training is more effective than exercising 10,000 steps a day, which has traditionally been said to be effective.

According to the inventors' analysis, five months of interval walking training (IWT) for middle-aged and elderly people improves the maximum oxygen intake (VO2peak) by an average of 10% and improves the symptoms of lifestyle-related diseases by 20%. Furthermore, in this invention, it was clarified to what extent the initial characteristics of each subject and the intensity and amount of exercise during interval walking training affect the effect. In other words, the factors that determine the effect of interval walking training were identified and their reliability was verified.

Figure 5 shows the flow of the analysis. Among the participants who joined the interval walking training project (this project) between 2005 and 2016, 485 middle-aged and elderly people (up to 64 years old, determinant factor identification group) who joined this project in 2005 and performed interval walking training for five months were included in the determinant factor identification group. A multiple regression analysis was performed with the subjects' initial characteristics and the intensity and amount of exercise during the intervention as candidates for independent variables, and each effect parameter as a dependent variable to identify the independent variable with the highest correlation between the estimated value and the actual measured value (Analysis 1). Furthermore, the reliability was verified for another 609 middle-aged and elderly people (up to 65 years old, accuracy verification group) who joined this project between 2006 and 2016 and performed interval walking training for five months (Analysis 2).

Figure 6 shows the average values of pre-training (pre), post-training (post) and change (Δ) of each characteristic value of the subjects in the determinant identification group and accuracy verification group that were the subject of the analysis. The changes in the maximum oxygen intake VO2peak, body mass index BMI, diastolic blood pressure DBP, blood glucose level FBG, HDL cholesterol HDL and arteriosclerosis index (LDL/HDL) over a 5-month period in the determinant identification group were analyzed (Analysis 1) to obtain a regression equation, and the initial values were involved as all effect determining factors. Furthermore, the intensity and amount of fast walking were identified as determining factors for the maximum oxygen intake VO2peak and body mass index BMI, the amount of fast walking for blood glucose level FBG, HDL and arteriosclerosis index (LDL/HDL), and the change in maximum oxygen intake ΔVO2peak due to interval fast walking training for the diastolic blood pressure DBP were identified as determining factors. Furthermore, in the accuracy verification group, 95% prediction intervals were calculated from the estimated values and actual measured values using each regression equation obtained in the determinant factor identification group (Analysis 2), and these were almost consistent with those in the determinant factor identification group.

Therefore, it was found that it is possible to estimate (predict) the changes (Δ, exercise effect) in the maximum oxygen intake VO2peak, body mass index BMI, diastolic blood pressure DBP, blood glucose level FBG, HDL cholesterol HDL, and arteriosclerosis index (LDL/HDL) after five months of continuing interval walking training. In other words, when the exercise effects are the maximum oxygen intake change ΔVO2peak, body mass index change ΔBMI, diastolic blood pressure change ΔDBP, blood glucose level change ΔFBG, HDL cholesterol change ΔHDL, and arteriosclerosis index change Δ(LDL/HDL), it was found that the exercise effect can be predicted (estimated) by acquiring specific user information 7 suitable for predicting the effects of these prediction target items.

Figure 7 shows the estimation formula (prediction formula) of the exercise effect (item to be predicted) that changes due to training. The estimation formula for the maximum oxygen intake change ΔVO2peak and the body mass index change ΔBMI includes the gender S (male = 1, female = 0), the initial value (initial characteristic), the weekly fast walking time (weekly average time) Wa, and the exercise intensity of fast walking Wi as determining factors (parameters). The estimation formula for the minimum blood pressure change ΔDBP includes the initial value (initial characteristic) and the maximum oxygen intake change ΔVO2peak estimated above as determining factors (parameters). It was found that the estimation formula for the blood glucose change ΔFBG, the HDL cholesterol change ΔHDL, and the arteriosclerosis index change Δ(LDL/HDL) includes the initial value (initial characteristic) and the weekly fast walking time (weekly average time) Wa as determining factors (parameters). Therefore, it was found that the maximum oxygen intake change ΔVO2peak, the body mass index change ΔBMI, and the diastolic blood pressure change ΔDBP are included in the exercise intensity-dependent group 21, while the blood glucose change ΔFBG, the HDL cholesterol change ΔHDL, and the arteriosclerosis index change Δ(LDL/HDL) are included in the exercise intensity-independent group 22, as exercise effects to be predicted.

8 to 11 show the process of identifying the determinants in the estimation equation (multiple regression equation) of the exercise effect of each target item. From the initial values and training records of the subjects included in the determinant identification group, factors that determine the change (Δ, exercise effect) of each measured parameter (each prediction target item) before and after training were identified. An example of the multiple regression equation is shown in the following equation (2).
yj=a(j,1)・x1+a(j,2)・x2+...
+a(j,i)・xi+cj (2)
Here, the candidates for the independent variables xi are age, sex, initial values of each measurement parameter, training performance over the past five months (total weekly walking time, weekly fast walking time, fast walking exercise intensity), and maximum change in oxygen content (ΔVO2peak). The dependent variables yj are the changes in each measurement parameter before and after training (ΔVO2peak, body mass index (ΔBMI), systolic blood pressure (ΔSBP), diastolic blood pressure (ΔDBP), mean blood pressure (ΔMBP), blood glucose level (ΔFBG), blood cholesterol (ΔHDL, ΔLDL), blood triglyceride (ΔTG), and arteriosclerosis index (Δ(LDL/HDL)).

Next, for each dependent variable, candidates for the independent variable were substituted into the regression equation, and the correlation coefficient (r) between the estimated value and the actual measured value was determined to be significant and to have the highest value a(j,i). Furthermore, the following linear regression equation (equation (3)) between the estimated value and the actual measured value was determined, and the 95% prediction interval for the reliability of the estimation equation was calculated.
yi'=mj・yj+nj (3)
In addition, when a Bland-Altman analysis was performed between the estimated values and the measured values, a bias occurred in the error, so the estimated values were corrected by the Brace method, which takes into account the variations on both the X-axis and Y-axis.

Figure 8 shows the analysis results when the independent variables include the total walking time, with the maximum oxygen intake change ΔVO2peak, the body mass index change ΔBMI, the diastolic blood pressure change ΔDBP, the blood glucose change ΔFBG, the blood cholesterol change ΔHDL, and the arteriosclerosis index change Δ(LDL/HDL) selected as dependent variables (amount of change) after five months of training. Figure 9 shows the analysis results when the weekly brisk walking time is included as an independent variable. Figure 10 shows the analysis results when the weekly brisk walking time and brisk walking exercise intensity are included as independent variables. Figure 11 shows the analysis results when the estimated maximum oxygen intake change ΔVO2peak is included as an independent variable.

The maximum oxygen intake change ΔVO2peak is correlated with gender, initial value, total weekly walking time, weekly fast walking time, and fast walking exercise intensity, but the estimation formula with gender, initial value, weekly fast walking time, and fast walking exercise intensity as independent variables was found to be the most accurate. The body mass index change ΔBMI is correlated with gender, initial value, total weekly walking time, weekly fast walking time, fast walking exercise intensity, and estimated maximum oxygen intake change, but the estimation formula with gender, initial value, weekly fast walking time, and fast walking exercise intensity as independent variables was found to be the most accurate. The minimum blood pressure change ΔDBP is correlated with initial value, weekly fast walking time, and estimated maximum oxygen intake change, but the estimation formula with initial value and estimated maximum oxygen intake change as independent variables was found to be the most accurate. Note that, because exercise intensity is necessary to estimate the maximum oxygen intake change ΔVO2peak, the minimum blood pressure change ΔDBP is also included in the exercise intensity dependent group 21.

For blood glucose change ΔFBG, blood cholesterol change ΔHDL, and arteriosclerosis index change Δ(LDL/HDL), correlations were found with the initial values and weekly fast walking time, but no correlations were found with other parameters. Therefore, these changes (exercise effects, predicted items) are included in exercise intensity independent group 22.

Figures 8 to 11 also show the 95% confidence intervals of the coefficients for each independent variable, and the following can be used as the estimation equation for each predicted target item. Here, S is gender (male = 1, female = 0), Wa is the average weekly time (weekly brisk walking time, min/week), and Wi is exercise intensity (brisk walking exercise intensity, %VO2peak). The initial value of the target item is the value of the target item at the start of training for the user to be estimated, for example, the maximum oxygen intake at the start of training. Note that the units of each coefficient are omitted, but they match the independent and dependent variables.

Maximum oxygen intake change ΔVO2peak (ml/kg/min)
ΔVO2peak=a11·S+a12·VO2peak initial value+a13·Wa
+a14・Wi+c1 (11)
-1.873<a11<-0.495
-0.301<a12<-0.149
0.006<a13<0.014
0.091<a14<0.152
-6.469<c1<0.542
The median value of each coefficient is as shown in Figure 7. The same applies to the following coefficients.

Change in body mass index ΔBMI (kg/m ² )
ΔBMI = a21·S + a22·BMI initial value + a23·Wa + a24·Wi
+ c2 (12)
0.051<a21<0.306
-0.079<a22<-0.041
-0.003<a23<-0.001
-0.011<a24<-0.001
1.014<c2<2.227

Diastolic blood pressure change ΔDBP (mmHg)
ΔDBP=a32·DBP initial value+a35·ΔVO2peak+c3 (13)
-0.564<a32<-0.427
-0.657<a35<-0.072
31.587<c3<42.612

Blood glucose change ΔFBG (mg/dl)
ΔFBG=a42·FBG initial value+a43·Wa+c4 (14)
-0.256<a42<-0.150
-0.029<a43<-0.005
13.483<c4<25.261

Blood cholesterol change ΔHDL (mg/dl)
ΔHDL = a52 · HDL initial value + a53 · Wa + c5 (15)
-0.147<a52<-0.058
0.002<a53<0.017
3.829<c5<9.989

Arteriosclerosis index change Δ (LDL/HDL)
Δ(LDL/HDL)=a62·(LDL/HDL) initial value+a63·Wa
+ c6 (16)
-0.170<a62<-0.079
-0.001<a63<0.000
0.124<c6<0.343

Next, in analysis 2, the independent variables of the accuracy verification group were substituted into the estimation equation for the predictive effect of each prediction target item determined above using the determinant factor identification group to obtain estimates, which were then plotted against the actual measured values of the accuracy verification group to calculate the 95% prediction interval of the estimation equation. Finally, the obtained 95% prediction intervals were compared between the determinant factor identification group and the accuracy verification group.

Figures 12 and 13 show the evaluation results of the actual measured values and estimated values for the group for identifying the determinants of the maximum oxygen uptake change ΔVO2peak and the group for verifying accuracy. Figure 12 shows the evaluation results of the estimation formula that includes the exercise intensity described above as an independent variable, and Figure 13 shows, for reference, the evaluation results of the estimation formula that does not include exercise intensity as an independent variable (including only gender, initial value, and weekly fast walking time). It can be seen that the correlation coefficient r of the estimation formula that includes exercise intensity as an independent variable shown in Figure 12 is higher than the correlation coefficient of the estimation formula shown in Figure 13, and the exercise effect can be estimated with higher accuracy.

Figure 14 shows the evaluation results of the actual and estimated values of the body mass index change ΔBMI in the determinant identification group and the accuracy verification group. Figure 15 shows the evaluation results of the actual and estimated values of the diastolic blood pressure change ΔDBP in the determinant identification group and the accuracy verification group. Figure 16 shows the evaluation results of the actual and estimated values of the blood glucose change ΔFBG in the determinant identification group and the accuracy verification group. Figure 17 shows the evaluation results of the actual and estimated values of the blood cholesterol change ΔHDL in the determinant identification group and the accuracy verification group. Figure 18 shows the evaluation results of the actual and estimated values of the arteriosclerosis index change Δ(LDL/HDL) in the determinant identification group and the accuracy verification group. In all prediction target items, the determinant identification group and the accuracy verification group showed high correlation, and further, in the accuracy verification group, when the 95% prediction interval was calculated from the estimated values and actual values using each regression equation (estimation equation) calculated in the determinant identification group, it almost matched those in the determinant identification group. Therefore, it is possible to use each of the estimation equations described above to predict the exercise effect (predicted effect) of each prediction target item with a high degree of accuracy.

Figures 12 to 18 show the linear regression equation (equation (3)) between the estimated values and actual values for each forecast target item. The accuracy of the estimated values can be improved by correcting the coefficients and constants in the estimation equations for each forecast target item shown in equations (11) to (16) using the Brace method as follows. Note that Δ' indicates the corrected value, and the number in parentheses indicates the median.

Maximum oxygen intake change ΔVO2peak (ml/kg/min) (Estimation formula (11))
Δ′VO2peak=1.75ΔVO2peak−2.10
-3.278<a11(-2.07)<-0.866
-0.527<a12(-0.394)<-0.261
0.011<a13(0.018)<0.025
0.159<a14(0.214)<0.266
-13.421<c1(-7.28)<-1.152

Change in body mass index ΔBMI (kg/m ² ) (Estimation formula (12))
Δ′BMI=2.55ΔBMI+0.65
0.130<a21(0.456)<0.780
-0.201<a22(-0.153)<-0.105
-0.008<a23(-0.005)<-0.003
-0.028<a24(-0.015)<-0.003
3.236<c2(4.78)<6.329

Diastolic blood pressure change ΔDBP (mmHg) (Estimation formula (13))
Δ′DBP=1.79ΔDBP+2.51
-1.010<a32(-0.886)<-0.764
-1.176<a35(-0.653)<-0.129
59.051<c3(68.9)<78.785

Blood glucose change ΔFBG (mg / dl) (Estimation formula (14))
Δ′FBG=2.76ΔFBG+7.18
-0.707<a42(-0.560)<-0.414
-0.080<a43(-0.047)<-0.014
44.393<c4(60.7)<76.900

Blood cholesterol change ΔHDL (mg/dl) (Estimated formula (15))
Δ´HDL=4.22ΔHDL-4.08
-0.620<a52(-0.430)<-0.245
0.008<a53(0.040)<0.072
12.078<c5(25.1)<38.074

Arteriosclerosis index change Δ(LDL/HDL) (estimation formula (16))
Δ′(LDL/HDL)=3.73Δ(LDL/HDL)+0.22
-0.634<a62(-0.466)<-0.295
-0.004<a63(-0.002)<0.000
0.683<c6(1.09)<1.499

Figure 19 shows a flowchart outlining the process of determining predicted effects for predicted effect items in the effect prediction system 10. This process may be provided as a program (program product) including computer-executable instructions via a computer-readable medium such as a recording medium. The library 20 of the effect prediction system 10 includes, as the exercise intensity-dependent group 21, the above estimation formulas (11) to (13) determined for each of the maximum oxygen intake change ΔVO2peak of the predicted effect of the maximum oxygen intake VO2peak, the body mass index change ΔBMI (kg/m2) of the predicted effect of the body mass index BMI, and the diastolic blood pressure change ΔDBP (mmHg) of the predicted effect of the diastolic blood pressure DBP. The library 20 also includes, as an exercise intensity-independent group 22, the above estimation formulas (14) to (16) calculated for the predicted effect of blood glucose FBG change ΔFBG (mg/dl), the predicted effect of HDL cholesterol change ΔHDL (mg/dl), and the predicted effect of arteriosclerosis index change Δ(LDL/HDL).

First, in step 51, the effect prediction system 10 acquires the item for predicting the exercise effect (prediction target item) via the input/output interface 15. In step 52, if the prediction target item is an exercise intensity dependent group 21, i.e., at least one of maximum oxygen intake, body mass index, and diastolic blood pressure, the input/output interface 15 or the first prediction unit 11 acquires user information 7 including gender S, the user's initial value (initial characteristics) of the prediction target item, the average weekly time Wa planned during the training, and the exercise intensity Wi planned during the training, via the input/output interface 15 in step 53. At this time, in step 53a, information including the fact that the predicted effect (exercise effect) of the prediction target item being queried depends on the exercise intensity may be provided to the user 6 via the first interface 16. It is possible to show the user 6 that the exercise effect is more likely to be obtained by actively increasing the exercise intensity.

On the other hand, in step 52, if the prediction target item is at least one of the exercise intensity independent group 22, i.e., blood glucose level, HDL cholesterol, and arteriosclerosis index, the input/output interface 15 or the second prediction unit 12 acquires user information 7 including the user's initial value (initial characteristic) of the prediction target item and the weekly average time Wa planned during the training via the input/output interface 15 in step 54. At this time, in step 54a, information including that the predicted effect (exercise effect) of the prediction target item being queried does not depend on the exercise intensity may be provided to the user 6 via the second interface 17. In other words, if the exercise intensity is a predetermined intensity (70% or more of maximum physical strength (VO2peak)) required for training, information including that it does not depend on the exercise intensity may be provided. It is possible to show the user 6 that the exercise effect can be obtained by securing the weekly average time even if the exercise intensity is not increased to a predetermined intensity or more (70% or more of maximum physical strength).

In step 55, the first prediction unit 11 and the second prediction unit 12 estimate the predicted effect of the queried prediction target item based on the user information 7 using the estimation equation for the exercise intensity-dependent group 21 or the exercise intensity-independent group 22 pre-stored in the library 20, and in step 56 provide the predicted effect (expected exercise effect) to the terminal 5 of the user 6 via the Internet 9 or the like.

The LSD score (Lifestyle-related disease index, LSD risk factor, Lifestyle-related disease) is known as one of the indexes for lifestyle-related diseases. The LSD score (Lifestyle-related disease index) is a so-called metabolic index, and one point is added if one of the following four diagnostic criteria is met. Therefore, if all four items are met, the maximum score is four points.
1) Systolic blood pressure ≥ 130mmHg or diastolic blood pressure ≥ 85mmHg
2) Fasting blood sugar ≧100mg/dl,
3) BMI≧25kg/ ^m2 ,
4) Triglycerides ≧150 mg/dl or HDL cholesterol ≦40 mg/dl.

For example, when user 6 tries to improve his/her LSD score, if 1) or 3) applies, it is desirable to improve the body mass index and/or diastolic blood pressure. When considering improving the applicable items through interval walking training, it is possible to predict the exercise effect using effect prediction system 10. In this case, body mass index and diastolic blood pressure are exercise intensity dependent groups 21, and by providing user information 7 including the planned exercise intensity Wi in addition to the initial value and the planned weekly average time Wa to effect prediction system 10, the predicted training effect, in this example, body mass index change ΔBMI and diastolic blood pressure change ΔDBP, can be estimated and provided to user 6, and a training program including a predicted training effect and the exercise intensity for obtaining the training effect desired by the user can be generated.

On the other hand, when the LSD score corresponds to 4), it is desirable to improve HDL cholesterol. HDL cholesterol is in the exercise intensity independent group 22, and by providing user information 7 including the initial value and the planned weekly average time Wa to the effect prediction system 10, the predicted training effect, in this example, the HDL cholesterol change ΔHDL, can be estimated and provided to the user 6, and a training program including a predicted training effect and the exercise intensity for obtaining the training effect desired by the user can be generated. In this case, it can be conveyed to the user 6 that the exercise effect remains almost unchanged even if the exercise intensity Wi is increased, and that increasing the weekly average training time Wa is important as training.

10. Effect prediction system

Claims (15)

A system for predicting an effect of continuing training in which an average time during which a predetermined intensity is reached reaches a predetermined time per week, comprising:
an input interface for acquiring prediction target items related to improving physical fitness or improving lifestyle-related diseases;
a first prediction unit that, if the prediction target item is included in an exercise intensity dependent group, acquires at least an initial characteristic of the user for the prediction target item, a weekly average time planned during the continuation of the training, and an exercise intensity planned during the continuation of the training via the input interface, and outputs a predicted effect of the prediction target item after the training is continued for a predetermined period using an exercise intensity dependent estimation equation that includes at least the initial characteristic, the weekly average time, and the exercise intensity as elements;
and a second prediction unit that, if the item to be predicted is included in an exercise intensity-independent group, acquires via the input interface at least the user's initial characteristics for the item to be predicted and the weekly average time planned during the continuation of the training, and outputs a predicted effect of the item to be predicted after the training has been continued for a predetermined period using an exercise intensity-independent estimation equation that includes at least the initial characteristics and the weekly average time as elements. In claim 1,
The system includes a first prediction unit that acquires the gender of the user via the input interface, and outputs a predicted effect of the item to be predicted after the training is continued for a predetermined period of time using an exercise intensity-dependent estimation equation that includes at least the gender, the initial characteristics, the average weekly time, and the exercise intensity as elements. In claim 1,
A system comprising a first interface that, if the item to be predicted is included in the exercise intensity dependent group, provides the user with information including that the predicted effect depends on exercise intensity. In claim 1,
The system includes a second interface that, if the item to be predicted is included in the exercise intensity-independent group, provides the user with information including that the predicted effect is independent of exercise intensity. In claim 1,
The training includes interval walking training with the predetermined intensity set to 70% of maximum oxygen intake and the predetermined duration set to 60 minutes on a weekly average. In claim 5,
The prediction target items included in the exercise intensity dependent group include maximum oxygen intake, body mass index, and minimum blood pressure,
The prediction target items included in the exercise intensity-independent group include blood glucose level, HDL cholesterol, and an arteriosclerosis index. A method for predicting an effect of continuing training in which an exercise duration during which a predetermined intensity is reached reaches a predetermined time on average per week, using an effect prediction system, comprising:
The effect prediction system includes an input interface for acquiring prediction target items related to physical fitness improvement or lifestyle-related disease improvement;
a library that stores an exercise intensity-dependent estimation equation for the prediction target item included in an exercise intensity-dependent group, the estimation equation including at least an initial characteristic of a user of the prediction target item, a weekly average time planned during the continuation of the training, and an exercise intensity planned during the continuation of the training as elements, and a library that stores an exercise intensity-independent estimation equation for the prediction target item included in an exercise intensity-independent group, the estimation equation including at least an initial characteristic and the weekly average time as elements,
If the prediction target item is included in the exercise intensity-dependent group, at least the initial characteristics, the weekly average time, and the exercise intensity are acquired via the input interface, and a predicted effect of the prediction target item of the exercise intensity-dependent group after the training is continued for a predetermined period is output using the exercise intensity-dependent estimation equation including at least the initial characteristics, the weekly average time, and the exercise intensity as elements.
If the item to be predicted is included in the exercise intensity-independent group, at least the initial characteristics and the weekly average time are acquired via the input interface, and a predicted effect of the item to be predicted in the exercise intensity-independent group after continuing the training for a predetermined period is output using the exercise intensity-independent estimation equation that includes at least these as elements. In claim 7,
the library stores, for the prediction target item included in the exercise intensity dependent group, an exercise intensity dependent estimation equation including at least the user's gender, an initial characteristic of the user for the prediction target item, an average weekly time planned during the continuation of the training, and an exercise intensity planned during the continuation of the training as elements;
The method, wherein outputting the predicted effect of the predicted item of the exercise intensity-dependent group includes outputting the predicted effect of the predicted item after continuing the training for a predetermined period of time using an exercise intensity-dependent estimation equation that includes at least the gender, the initial characteristics, the average weekly time, and the exercise intensity as elements. In claim 7,
The method includes, if the item to be predicted is included in the exercise intensity dependent group, providing the user with information including that the predicted effect depends on exercise intensity. In claim 7,
A method comprising, if the item to be predicted is included in the exercise intensity-independent group, providing the user with information including that the predicted effect is independent of exercise intensity. In claim 7,
The method of claim 1, wherein the training comprises interval walking training, the predetermined intensity being set to 70% of maximum oxygen intake, and the predetermined duration being set to 60 minutes on a weekly average. In claim 11,
The prediction target items included in the exercise intensity dependent group include maximum oxygen intake, body mass index, and minimum blood pressure,
A method according to claim 1, wherein the prediction target items included in the exercise intensity-independent group include blood glucose level, HDL cholesterol, and an atherosclerosis index. In claim 12,
the library includes the following estimation formulas for estimating a peak oxygen intake change ΔVO2peak (ml/kg/min) as the predicted effect of the peak oxygen intake, a body mass index change ΔBMI (kg/ ^m2 ) as the predicted effect of the body mass index, a diastolic blood pressure change ΔDBP (mmHg) as the predicted effect of the diastolic blood pressure, a blood glucose change ΔFBG (mg/dl) as the predicted effect of the blood glucose level, a blood cholesterol change ΔHDL (mg/dl) as the predicted effect of the HDL cholesterol, and an atherosclerosis index change Δ(LDL/HDL) as the predicted effect of the atherosclerosis index.
ΔVO2peak=a11·S+a12·VO2peak initial value+a13·Wa
+a14・Wi+c1
ΔBMI = a21·S + a22·BMI initial value + a23·Wa + a24·Wi + c2
ΔDBP = a32 · DBP initial value + a35 · ΔVO2peak + c3
ΔFBG = a42 · FBG initial value + a43 · Wa + c4
ΔHDL = a52 · HDL initial value + a53 · Wa + c5
Δ(LDL/HDL)=a62·(LDL/HDL) initial value+a63·Wa+c6
where S is the gender (male = 1, female = 0), Wa is the weekly average time (min/week), Wi is the exercise intensity (%VO2peak), and a11 to a63 and c1 to c6 are coefficients in the following ranges.
-3.278<a11<-0.866
-0.527<a12<-0.261
0.011<a13<0.025
0.159<a14<0.266
-13.421<c1<-1.152
0.130<a21<0.780
-0.201<a22<-0.105
-0.008<a23<-0.003
-0.028<a24<-0.003
3.236<c2<6.329
-1.010<a32<-0.764
-1.176<a35<-0.129
59.051<c3<78.785
-0.707<a42<-0.414
-0.080<a43<-0.014
44.393<c4<76.900
-0.620<a52<-0.245
0.008<a53<0.072
12.078<c5<38.074
-0.634<a62<-0.295
-0.004<a63<0.000
0.683<c6<1.499 A method for predicting , by a computer , the effect of continuing training in which an exercise time at a predetermined intensity reaches a predetermined time on average per week, comprising:
The computer acquires prediction target items related to improving physical strength or improving lifestyle-related diseases via an input interface ;
If the prediction target item is included in an exercise intensity dependent group, at least an initial characteristic of the user for the prediction target item, an average weekly time planned during the continuation of the training, and an exercise intensity planned during the continuation of the training are acquired via the input interface, and a predicted effect of the prediction target item after the training is continued for a predetermined period is output using an exercise intensity dependent estimation equation including at least the initial characteristic, the average weekly time, and the exercise intensity as elements;
If the item to be predicted is included in an exercise intensity-independent group, the method comprises: acquiring, via the input interface, at least the user's initial characteristics for the item to be predicted and the weekly average time planned during the continuation of the training; and outputting a predicted effect of the item to be predicted after continuing the training for a predetermined period of time using an exercise intensity-independent estimation equation that includes at least the initial characteristics and the weekly average time as elements. A program for predicting, by a computer, an effect of continuing training in which an exercise time during which a predetermined intensity is reached reaches a predetermined time on average per week, comprising:
The computer acquires prediction target items related to improving physical strength or improving lifestyle-related diseases via an input interface ;
If the prediction target item is included in an exercise intensity dependent group, at least an initial characteristic of the user for the prediction target item, an average weekly time planned during the continuation of the training, and an exercise intensity planned during the continuation of the training are acquired via the input interface, and a predicted effect of the prediction target item after the training is continued for a predetermined period is output using an exercise intensity dependent estimation equation including at least the initial characteristic, the average weekly time, and the exercise intensity as elements;
If the item to be predicted is included in an exercise intensity-independent group, then via the input interface, at least the user's initial characteristics for the item to be predicted and the weekly average time planned during the continuation of the training, and outputting a predicted effect of the item to be predicted after continuing the training for a predetermined period of time using an exercise intensity-independent estimation equation that includes at least the initial characteristics and the weekly average time as elements.

Images