JP3787149B1 - Method and apparatus for evaluating true effect of intervention - Google Patents

Abstract

[Purpose] To provide a method for evaluating the true effects of interventions that can eliminate or reduce the use of control populations such as patient populations receiving placebos or subject populations not receiving health promotion activities. .
[Configuration] Before the intervention, a first step of actually measuring the items related to the health of each individual in the intervention target group and obtaining an average value of the whole group of the actual measurement values of each individual in the intervention target group; The second step of obtaining the standard deviation of the entire group of the measured values of each individual of the intervention target group, and the standard deviation of the group of the actual measured values of each individual of the intervention target group obtained in the second step are used. Then, a third step of estimating a “regression to the average” that is a value corresponding to the return to the average of the intervention target group, and after or after the step 2 or step 3, The fourth step of measuring the items related to the health of each individual of the intervention target group to obtain the average value of the entire group of the actual measurement values of each individual of the intervention target group; At the same time, the average value of the whole group of the measured values of each individual of the intervention target group obtained in the first step and the actual measurement of each individual of the intervention target group obtained in the fourth step after the intervention. A fifth step of obtaining a difference between the average value of the whole group of values, an average value of the whole group of the measured values of each individual of the intervention target group before the intervention obtained in the fifth step, and the intervention target group after the intervention The difference between the measured value of each individual and the average value of the entire group is compared with the “regression to the mean” of the intervention target group estimated in the third step, and the true value of the intervention in the intervention target group is compared. A method for evaluating the true effect of the intervention, comprising a sixth step of evaluating the effect of the intervention.
[Selection] Figure 1

Description

  The present invention relates to a method and apparatus for evaluating the effects of various interventions such as clinical medicines and health promotion activities in the workplace or community.

  Conventionally, when a new drug is introduced, a group of patients who receive administration (intervention) of a new drug in a medical institution, and a pseudo drug that is indistinguishable from a new drug but does not contain a useful substance as a pseudo drug. Divided into two groups, the patient group administered without notification (control group), and continued administration for a certain period to each group, then matters related to illness before and after administration for each group By comparing these measured values, clinical trials are conducted to determine whether the effects of drug intake are truly meaningful as drug effects.

  In addition, when deciding whether to introduce new health promotion activities (intervention) such as exercise, dietary habits improvement and smoking cessation movement in the region and industrial sites (workplace), health promotion activities (intervention) Divided into a group of subjects to be performed and a group of subjects (control group) who are not subject to health promotion activities (intervention implementation), and after each intervention, hypertension before and after intervention for each group It is said that it is necessary to evaluate the effects of health promotion activities (interventions) by comparing measurements of health-related matters such as hyperlipidemia and hyperlipidemia.

Conventionally, a system has been proposed in which clinical data of each patient is accumulated in a database for each patient, and the data is analyzed and evaluated and fed back to medical measures for each patient (see, for example, Patent Document 1).
JP 2002-95650 A

  However, in order to evaluate the effect of a new drug, it is not humane to administer a prescribed amount of a placebo without informing the control group (no intervention) of the placebo in the trial of the new drug. There was a big problem from the viewpoint.

  In addition, in order to evaluate the effects of health promotion activities at local and industrial sites, groups of subjects (control groups) who do not perform health promotion activities (interventions), that is, conventional and no lifestyle habits at local or industrial sites. It is practically very difficult to secure a group of subjects who will not change, to have the group conduct only the test, and to perform the same test again several months later, because very few subjects will respond to such a test. It was difficult.

  Moreover, it was not possible to evaluate the true effects of the above-mentioned new drugs and health promotion activities only by utilizing the clinical data of each patient by utilizing the computer system as shown in Patent Document 1. .

  The present invention has been made paying attention to such problems of the prior art, and does it eliminate the need to use a control group such as a patient group to which a placebo is administered or a subject group to which health promotion activities are not performed? It is an object to provide a method and apparatus for assessing the true effect of performing an intervention that can be reduced or reduced.

  The apparatus for evaluating the true effect of implementation of the intervention according to the present invention for solving the above-mentioned problems is “administration of a predetermined amount of a drug within a predetermined period, regular movement of a predetermined amount within a predetermined period” , At least one of the change of the type and amount of nutrients to be taken within a predetermined period and the regular reduction of smoking amount during a predetermined period "concerning the health of each individual of the target population before implementing" Based on the first input means for inputting the result of actual measurement of the items, and the actual measurement result input by the first input means, the average value of the entire group of the actual measurement values of each individual of the implementation target group before the implementation First storage means for storing the average value of the entire group of actually measured values of each individual of the implementation target group before the execution determined by the first calculation means, 1st input A second computing means for obtaining a standard deviation of the entire group of actually measured values of each individual of the implementation target group before the implementation based on the actual measurement results input by the stage; and the second computation means obtained by the second computing means Second storage means for storing the standard deviation of the entire group of actual measurement values of each individual of the implementation target group before the implementation, and the actual measurement values of each individual of the implementation target group before the implementation read from the second storage means An estimation means for estimating “regression to the mean”, which is a value corresponding to the regression to the average of the target population before the implementation, using the standard deviation of the entire population of A second input means for inputting a result of actual measurement of items related to health of each individual of the implementation target group; and each of the implementation target group after the implementation based on the measurement result input by the second input means Whole group of individual measured values And a third storage means for storing an average value of the entire group of actually measured values of each individual of the implementation target group after the execution obtained by the third calculation means. And the average value of the entire group of the measured values of each individual of the implementation target group before the implementation from the first storage unit, and the measured value of each individual of the implementation target group after the implementation from the third storage unit 4th calculating means for reading out the average value of the whole group and calculating the difference between them, and the average of the whole group of the measured values of each individual of the implementation target group before the execution outputted from the fourth calculating means The difference between the value and the average value of the entire group of actual measurement values of each individual of the implementation target group after the implementation, and the "regression to the average" of the implementation target group before the implementation output from the estimation means Comparison means for comparison and comparison by the comparison means An apparatus for evaluating the true effect of performing intervention on a group to be implemented, realized by a computer system, comprising output means for outputting a result.

  In addition, the method for evaluating the true effect of the implementation of the intervention according to the present invention for solving the above-described problems is as follows. The first input means performs “administration of a predetermined amount of medicine within a predetermined period, Subject to implementation of “at least one of regular exercise in a certain amount, change in the type and amount of nutrients consumed in a prescribed period, and reduction in regular smoking in a prescribed period” A first step of inputting a result of actual measurement of items related to health of each individual of the group, and a first computing means based on the actual measurement result input by the first input means, An average value of the entire group of actually measured values of each individual is obtained, and this is stored in the first storage means, and at the same time as or before or after the second step, the second calculation means provides the first input. By means Based on the actually measured result, the third step of obtaining the standard deviation of the whole group of the actual measurement values of each individual of the implementation target group before the implementation, and storing this in the second storage means, the estimation means, It is a value corresponding to the return to the average of the implementation target group before the implementation using the standard deviation of the entire group of the measured values of each individual of the implementation target group before the implementation read out from the second storage means. The fourth step of estimating the "regression to the average" and the execution target after the execution by the second input means simultaneously with or before or after at least one of the step 2, step 3, and step 4 A fifth step of inputting an actual measurement result of each item of health of each individual of the group, and at least one of the step 2, step 3 and step 4 simultaneously or in succession, Is a group of measured values of each individual of the implementation target group after the implementation based on the actual measurement results of the respective health items of the individual of the implementation target group after the implementation input by the second input means A sixth step of obtaining an overall average value and storing the average value in the third storage means, and a group of actually measured values of each individual of the implementation target group before the execution read by the fourth calculation means from the first storage means A seventh step of obtaining a difference between the overall average value and the average value of the entire group of the measured values of each individual of the implementation target group after the execution read out from the third storage unit; and the comparison unit, the fourth calculation The difference between the average value of the entire group of the actual measurement values of each individual of the implementation target group before the implementation output from the means and the average value of the entire group of the actual measurement values of each individual of the implementation target group after the implementation, The actual output from the estimation means before the execution. Use of a computer system comprising: an eighth step of comparing “regression to the mean” of the application target group, and a ninth step of outputting the comparison result output from the comparison means by an output means such as a display This is a method for evaluating the true effect of the intervention in the target population.

  FIG. 1 is a conceptual block diagram showing an embodiment of a true effect evaluation apparatus used for the true effect evaluation method according to the present invention. In FIG. 1, 1 is “administration of a predetermined amount of medicine periodically within a predetermined period, exercise of a predetermined amount regularly within a predetermined period, change of the type and amount of nutrients taken within a predetermined period, and It is the 1st input part for inputting the result of having measured the matter about each individual of each individual of the implementation object group before carrying out "at least one of the periodic reduction of smoking amount in a predetermined period". Reference numeral 2 denotes a first calculation unit for obtaining an average value of the entire group of actual measurement values of each individual of the implementation target group before the implementation based on the actual measurement result input by the first input unit 1. Reference numeral 3 denotes a first storage unit for storing the average value obtained by the first calculation unit 2. Reference numeral 4 denotes a second calculation unit for obtaining a standard deviation of the entire group of actual measurement values of each individual of the implementation target group before the implementation based on the actual measurement result input by the first input unit 1. Reference numeral 5 denotes a second storage unit for storing a standard deviation of the entire group of actually measured values of each individual of the group to be implemented before the execution obtained by the second calculation unit 4. 6 is a value corresponding to the return to the average of the target group using the standard deviation of the entire group of the measured values of each individual of the target group before the execution read from the second storage unit 5. This is an estimation unit for estimating a certain “return to average”. Reference numeral 7 denotes a second input unit for inputting a result of actual measurement of items related to health of each individual in the implementation target group after the implementation. Reference numeral 8 denotes a third calculation unit for obtaining an average value of the entire group of actual measurement values of each individual of the implementation target group after the execution based on the actual measurement result input by the second input unit 7. Reference numeral 9 denotes a third storage unit for storing an average value of the entire group of actually measured values of each individual of the implementation target group after the execution obtained by the third calculation unit 8. 10 reads out the average value of the entire group of actual measurement values of each individual of the implementation target group before the implementation from the first storage unit 3 and each individual of the implementation target group after the implementation from the third storage unit 9 This is a fourth calculation unit for reading out the average value of the entire group of actually measured values and obtaining the difference between them. 11 is an average value of the entire group of actually measured values of each individual of the implementation target group before the execution output from the fourth arithmetic unit 10 and the entire group of actual measurement values of each individual of the target group to be implemented after the execution. It is a comparison unit for comparing the difference from the average value and the “regression to the average” of the implementation target group before the execution output from the estimation unit 6. Reference numeral 12 denotes an output unit such as a display for outputting a comparison result by the comparison unit 11.

In addition, in the apparatus or method for evaluating the true effect of the intervention of the present invention, the improvement of “measurement items that are meaningful in decreasing or decreasing the value” such as the blood pressure and neutral fat of the individual is performed. In the case where it is an object, it is desirable that the estimation means estimate the “regression to the average” of the implementation target group before the implementation by the following procedure.
(1) A hypothesis is made that “all measurement subjects have participated because the measurement values at the time of participation are higher or higher than the true values”, and deviations from the true values of the measured values in virtual individuals (u _i ) is defined in advance as follows.
u _i = (Xi−μi) / σ
In this equation, X is a measured value of a virtual individual, μ is a true value of the virtual individual, and σ is a standard deviation of the entire group of actually measured values of each individual in the implementation target group before the implementation.
(2) Since u _i has a normal distribution, one normal random number with an average = 0 and a standard deviation = 1 is generated as a deviation u from a true value of a measured value possessed by a virtual individual.
(3) By substituting one normal random number in (2) above into u of the following estimation formula that “the probability that an individual whose measurement value is higher or more than the true value is higher than the true value is high” is: A probability rate that a virtual individual having the deviation u participates in the virtual group R satisfying the hypothesis (1) is obtained.
Probability to participate rate = 1 / (1 + exp (−a × u))
In this expression, exp is an exponential function, and a is a constant.
(4) One random number from 0 to 1 (r) is generated as a cut-off value for determining whether or not a virtual individual having the deviation u is allowed to participate in the virtual group R.
(5) The probability rate of participation determined in (3) above is compared with r as the cutoff value generated in (4) above, and only when rate> r, a virtual individual having the deviation u Are adopted as virtual individuals constituting the virtual group R.
(6) The above operations (2) to (5) are repeated a plurality of times using a computer. As a result, a group that satisfies the above hypothesis (1), that is, a virtual group R in which virtual individuals whose measured values are higher or more than the true value participate, is created in a simulated manner.
(7) The average u _{ave of the} entire virtual group R of deviation u from the true value μ of the measured value X of each virtual individual included in the virtual group R simulated in (6) above Ask for.
(8) Based on the above equation (1), Δ which is “regression to average” of the virtual group R is obtained by the following equation.
Δ = average X _{ave of the} entire virtual population R of the measured value X of each virtual individual X _{ave −average} value of the entire virtual group R of the true value μ of each virtual individual μ _ave = true of the measured value of each virtual individual The average u _{ave of the} entire virtual group R of deviations u from the value of x × the standard deviation σ of the entire group of the actual measurement values of each individual of the target group before the execution
(9) The above operations (2) to (8) are carried out a plurality of times using a computer, an average value of each Δ obtained in each implementation is obtained, and the average value of this Δ is determined as the subject of implementation. Estimate the “regression to the mean” of the population.

Further, in the apparatus or method for evaluating the true effect of the intervention of the present invention, the purpose of the improvement is to improve “measurement items that are meaningful in decreasing or decreasing the value” such as blood pressure and neutral fat of an individual. In this case, it is desirable that the estimation means estimate the “regression to the average” of the implementation target group before the implementation by the following procedure.
(1) A hypothesis is made that “all measurement subjects have participated because the measurement values at the time of participation are higher or higher than the true values”, and deviations from the true values of the measured values in virtual individuals (u _i ) is defined in advance as follows.
u _i = (Xi−μi) / σ
In this equation, X is a measured value of a virtual individual, μ is a true value of the virtual individual, and σ is a standard deviation of the entire group of actually measured values of each individual in the implementation target group before the implementation.
(2) Since the calculation means A has a normal distribution, u _i takes one normal random number with an average = 0 and a standard deviation = 1 as a deviation u from a true value of a measured value possessed by a virtual individual. And this is stored in the storage means A.
(3) The calculation means B reads one normal random number generated in the above (2) from the storage means A, and determines this one normal random number as “an individual whose measured value is higher or more than the true value is a group. By substituting in u of the following estimation formula that “the probability of participating in is high”, a probability rate that a virtual individual having the deviation u participates in a virtual group R that satisfies the hypothesis of (1) is obtained, This is stored in the storage means B.
Probability to participate rate = 1 / (1 + exp (−a × u))
In this expression, exp is an exponential function, and a is a constant.
(4) The calculation means C uses one random number (set as r) of 0 to 1 as a cut-off value for determining whether or not a virtual individual having the deviation u is allowed to participate in the virtual group R. Generated and stored in the storage means C.
(5) The cut that the computing means D reads out the probability rate of participation determined in (3) above from the storage means B, and that is generated in (4) above and stored in the storage means C. Compared with r as an off value, only when rate> r, a virtual individual having the deviation u is adopted as a virtual individual constituting the virtual group R and stored in the storage means D.
(6) The control means controls the arithmetic means A, B, C and D, and repeats the operations (2) to (5) a plurality of times, thereby satisfying the hypothesis (1). Such a group, that is, a virtual group R in which virtual individuals whose measured values are higher or more than the true value participate is created in a simulated manner, and this is stored in the storage means E.
(7) The virtual group R of the deviation u from the true value μ of the measured value X of each virtual individual included in the virtual group R simulated by the calculation means E in (6) above. The overall average u _ave is obtained and stored in the storage means F.
(8) The average of the entire virtual group R of deviation u from the true value of the measured value of each virtual individual read out from the storage means F by the calculation means F based on the formula (1) Based on u _ave and the standard deviation σ read from the second storage means (refer to claim 1), Δ which is the “regression to the mean” of the virtual group R is obtained by the following equation, Is stored in the storage means G.
Δ = average X _{ave of the} entire virtual population R of the measured value X of each virtual individual X _{ave −average} value of the entire virtual group R of the true value μ of each virtual individual μ _ave = true of the measured value of each virtual individual The average u _{ave of the} entire virtual group R of deviations u from the value of x × the standard deviation σ of the entire group of the actual measurement values of each individual of the target group before the execution
(9) The control means controls the arithmetic means A, B, C, D, E, and F, and performs the operations (2) to (8) a plurality of times. The average value of each Δ is obtained, and the average value of Δ is estimated as “regression to the average” of the implementation target group, and this is stored in the storage means H.

Further, in the apparatus or method for evaluating the true effect of the intervention according to the present invention, the intervention is performed in such a way that the value increases or increases, such as the maximum oxygen intake that is an index representing the individual's HDL cholesterol and physical fitness. In the case where the purpose is to improve the “measurement item possessed”, the estimation means estimates the “regression to the mean” of the implementation target group before the implementation by the following procedure. Is desirable.
(1) The hypothesis that “all measurement subjects participated because the measurement value at the time of participation was lower or less than the true value” and a deviation from the true value of the measurement value in the virtual individual (u _i ) is defined in advance as follows.
u _i = (Xi−μi) / σ
In this equation, X is a measured value of a virtual individual, μ is a true value of the virtual individual, and σ is a standard deviation of the entire group of actually measured values of each individual in the implementation target group before the implementation.
(2) Since u _i has a normal distribution, one normal random number with an average = 0 and a standard deviation = 1 is generated as a deviation u from a true value of a measured value possessed by a virtual individual.
(3) By substituting one normal random number in the above (2) into u of the following estimation formula that “the probability that an individual whose measurement value is lower or less than a true value participates in a group is high” is: A probability rate that a virtual individual having the deviation u participates in the virtual group R satisfying the hypothesis (1) is obtained.
Probability to participate rate = 1 / (1 + exp (a × u))
In this expression, exp is an exponential function, and a is a constant.
(4) One random number from 0 to 1 (r) is generated as a cut-off value for determining whether or not a virtual individual having the deviation u is allowed to participate in the virtual group R.
(5) The probability rate of participation determined in (3) above is compared with r as the cutoff value generated in (4) above, and only when rate> r, a virtual individual having the deviation u Are adopted as virtual individuals constituting the virtual group R.
(6) The above operations (2) to (5) are repeated a plurality of times using a computer. As a result, a group that satisfies the above hypothesis (1), that is, a virtual group R in which virtual individuals whose measured values are lower or less than the true value participate, is created in a simulated manner.
(7) The average u _{ave of the} entire virtual group R of deviation u from the true value μ of the measured value X of each virtual individual included in the virtual group R simulated in (6) above Ask for.
(8) Based on the above equation (1), Δ which is “regression to average” of the virtual group R is obtained by the following equation.
Δ = average X _{ave of the} entire virtual population R of the measured value X of each virtual individual X _{ave −average} value of the entire virtual group R of the true value μ of each virtual individual μ _ave = true of the measured value of each virtual individual The average u _{ave of the} entire virtual group R of deviations u from the value of x × the standard deviation σ of the entire group of the actual measurement values of each individual of the target group before the execution
(9) The above operations (2) to (8) are carried out a plurality of times using a computer, an average value of each Δ obtained in each implementation is obtained, and the average value of this Δ is determined as the subject of implementation. Estimate the “regression to the mean” of the population.

Further, in the apparatus or method for evaluating the true effect of the intervention according to the present invention, the intervention is “significant to increase or increase the value,” such as the maximum oxygen intake that is an index representing the individual's HDL cholesterol and physical fitness. If the purpose is to improve the `` measurement item having '', the estimation means estimates the `` regression to the mean '' of the implementation target group before the implementation by the following procedure, It is desirable.
(1) The hypothesis that “all measurement subjects participated because the measurement value at the time of participation was lower or less than the true value” and a deviation from the true value of the measurement value in the virtual individual (u _i ) is defined in advance as follows.
u _i = (Xi−μi) / σ
In this equation, X is a measured value of a virtual individual, μ is a true value of the virtual individual, and σ is a standard deviation of the entire group of actually measured values of each individual in the implementation target group before the implementation.
(2) Since the calculation means A has a normal distribution, u _i takes one normal random number with an average = 0 and a standard deviation = 1 as a deviation u from a true value of a measured value possessed by a virtual individual. And this is stored in the storage means A.
(3) The calculation means B reads one normal random number generated in the above (2) from the storage means A, and reads this one normal random number as “an individual whose measured value is lower or less than the true value becomes a group. By substituting in u of the following estimation formula that “the probability of participation is high”, a probability rate that a virtual individual having the deviation u participates in a virtual group R satisfying the hypothesis of (1) above is obtained. Is stored in the storage means B.
Probability to participate rate = 1 / (1 + exp (a × u))
In this expression, exp is an exponential function, and a is a constant.
(4) The calculation means C uses one random number (set as r) of 0 to 1 as a cut-off value for determining whether or not a virtual individual having the deviation u is allowed to participate in the virtual group R. Generated and stored in the storage means C.
(5) The cut-off value that the computing means D reads out the probability rate of participation obtained in (3) above from the storage means B, and that is generated in (4) and stored in the storage means C. The virtual individual having the deviation u is adopted as a virtual individual constituting the virtual group R and stored in the storage means D only when rate> r.
(6) The control means controls the arithmetic means A, B, C, and D, and repeats the operations from (2) to (5) a plurality of times, whereby the hypothesis of (1) is obtained. A satisfying group, that is, a virtual group R in which virtual individuals whose measured values are lower or less than the true value participate is created in a simulated manner, and this is stored in the storage means E.
(7) The virtual means of the deviation u from the true value μ of the measured value X of each virtual individual included in the virtual group R created and stored by the arithmetic means E in the above (6) The average u _ave of the entire group R is obtained and stored in the storage means F.
(8) The average of the entire virtual group R of deviation u from the true value of the measured value of each virtual individual read out from the storage means F by the calculation means F based on the formula (1) Based on u _ave and the standard deviation σ read from the second storage means (refer to claim 1), Δ which is the “regression to the mean” of the virtual group R is obtained by the following equation, Is stored in the storage means G.
Δ = average X _{ave of the} entire virtual population R of the measured value X of each virtual individual X _{ave −average} value of the entire virtual group R of the true value μ of each virtual individual μ _ave = true of the measured value of each virtual individual The average u _{ave of the} entire virtual group R of deviations u from the value of x × the standard deviation σ of the entire group of the actual measurement values of each individual of the target group before the execution
(9) The control means controls the arithmetic means A, B, C, D, E, and F, and performs the operations (2) to (8) a plurality of times. The average value of each Δ is obtained, and the average value of Δ is estimated as “regression to the average” of the implementation target group, and this is stored in the storage means H.

  According to the present invention, it becomes possible to eliminate or reduce the use of a control group such as a patient group to which a placebo is administered or a subject group to which health promotion activities are not performed. Therefore, it is possible to eliminate or reduce the need for setting a control group as in the past, so that it is possible to eliminate ethical problems associated with administering a placebo to a patient, Since inspections are unnecessary, a great saving effect can be obtained in terms of budget and labor.

  In the present invention, as described above, a normal random number is generated in the above (2) and a random number r of 0 to 1 is generated as a cutoff value in the above (4). Therefore, in the present invention, not only the probability rate of participation is determined at random, but also a cutoff value for determining whether or not a virtual individual having the deviation u is included in the virtual group R is determined at random. Therefore, the virtual group R that satisfies the above-mentioned hypothesis that “each measurement subject participated because the measured value at the time of participation is higher or more than the true value” can be created more randomly. Therefore, according to the present invention, it is possible to further improve the accuracy of the simulation (see (8) above) for obtaining the “regression to average” of the virtual group R.

  As described above, conventionally, the true effect of new drugs and health promotion activities was evaluated using a control group (control group) in which no intervention was performed. The reason why the control group is required to evaluate the true effect of the new drug or health promotion activity is to eliminate the influence of “return to the average”.

  “Regression to the mean” means that when a test is performed on a sample population and the mean value of the test is far from the mean value of the population, the second test is the standard population. Means that the average value of approaches the average value of the population.

  In other words, the return to the average means a phenomenon in which an object having an extreme score with respect to a certain variable takes a value close to the average rather than an extreme in the retest. The reason for the return to the average is that the observed score is composed of two components: a true score and an error score (observed score = true). Score + error score). The above error components are generated by many causes such as differences in measurement equipment, physiological fluctuations of the subject, motivation, fatigue, and errors in recording. This error component varies stochastically (randomly). That is, it is assumed that sometimes it is greater than the true value and sometimes less than the true value. In the regression to the average, the retest is performed in the average direction because the probability of having such a distant value becomes smaller as the error distribution becomes farther from the average value. In addition, the description in this paragraph was published in November 21, 2000 by Medical Science International Co., Ltd. (1-228-36 Hongo, Bunkyo-ku, Tokyo, Japan). Cited from pages 72-73 of "Print".

  Such a return to the average is one of the natural phenomena commonly seen in living organisms. In particular, the phenomenon of regression to the average is likely to occur for items that tend to fluctuate during measurement.

  For example, assume that the result of measuring systolic blood pressure for a certain group is as shown in FIG. Assume that the average value of the group indicated by the oblique lines in FIG. 2A is, for example, 148 mmHg. In this case, when the group indicated by the oblique line is then reexamined without any intervention, the average value of the result may be, for example, 144 mmHg. In this case, a blood pressure decrease of 4 mmHg (= 148 mmHg-144 mmHg) is observed without intervention such as health education for the group indicated by the oblique lines (see FIG. 2B). Therefore, in this case, in order to evaluate the true effect of health education, a group (control group) that does not perform health education (intervention) for determining the effect of return to the average is required.

  That is, before giving health education, measure systolic blood pressure for both the group receiving the health education and the control group, then, after giving the health education, both the group receiving the health education and the control group Take the same measurement. Then, as shown in FIG. 3, blood pressure reduction due to “return to average” is also measured for the control group. In FIG. 3, for the group that received the health education, the blood pressure reduction effect still exists even if the blood pressure reduction due to the return to the average measured for the control group is removed. It can be evaluated as the true effect of education. 2 and 3 above are Shinohara Shuppan Shinsha (MD Building, 2-4-9 Yushima, Bunkyo-ku, Tokyo, Japan) written by Takashi Muto and Kei Fukuwatari “Health Education and Health Promotion”. Is quoted from "Evaluation of the first edition, third edition".

  Thus, it is said that a control group is needed to remove the effect of regression on the mean and examine the net (true) intervention effect. However, in order to evaluate the true effects of health promotion activities for a certain group, another group becomes a control group, and that group is left as it is without changing lifestyles. It is very difficult in practice to perform only re-examination because few people cooperate with such inspection and re-examination.

  Therefore, the present inventor has invented a method for statistically inferring the effect of regression on the average to evaluate the true effect of interventions such as health promotion activities without using a control group. The inventor used a Monte Carlo simulation to statistically infer the effect of regression on the mean.

  Here, the Monte Carlo method is a method of trying to obtain a solution of a problem to be obtained or an approximation of the law by applying a sufficiently large number of random extraction operations or random experiment results. In other words, the Monte Carlo method is an empirical study of statistics using random numbers. The Monte Carlo method creates a model of the problem to be solved and uses it to execute many cases that can actually occur, and translates the problem captured in analytical mathematical form into a problem of stochastic process Then, it is a method of trying to obtain the numerical solution approximately by a statistical experiment.

A method for estimating the circumference ratio π by the Monte Carlo method will be described based on FIG. 4 and the following equation.

A program for estimating the above-mentioned π by the Monte Carlo method is as follows. In this program, the mathematical software “Mathcad” (Mathsoft Engineering & Education, Inc. 101 Main Street Cambridge, MA 02142-1521, Great Breat Tel: 617-444-18000 Fax: 617-444: 17: Fax is used. In this program, random x and y between 0 and 1 are generated, and the above determination of whether x ² + y ² is ≦ 1 or> 1 is repeated 10,000 times to obtain a quarter circle. The above-mentioned probability r is obtained by counting the number contained therein.

  Next, an example of a method for evaluating the true effect by the intervention according to the embodiment of the present invention will be described. In the following, if the intervention is aimed at improving “measurements that are meaningful in decreasing or decreasing the value” such as an individual's blood pressure or neutral fat, that is, evaluating the effect of the intervention on the target population A method for obtaining a regression to the average using the Monte Carlo method in the case where the directionality is expected to “decrease or decrease” will be described.

  First, assuming that the direction of evaluation of the effect of intervention is expected to “decrease or decrease”, “All measurement subjects have higher or more measured values at the time of participation than the true values. I made a hypothesis that I participated. In other words, “intra-individual variation in measured values is considered to have a normal distribution centered on the true value of each measurement subject, and all measurement subjects have measured values at the time of participation from the true value. I made a hypothesis that I participated because of the large gap. By doing so, when the direction of evaluation by intervention is expected to “decrease or decrease”, it is considered that the degree of decrease in the measured value increases when measured again. Therefore, this hypothesis leads to overestimation of the return to the average. If there is still a significant effect after subtracting the regression to the overestimated average in this way, it is a meaningful intervention. Conversely, as a drawback of making such a hypothesis, there is a possibility that a slight intervention effect cannot be evaluated. However, the present inventor thinks that such a slight intervention effect may be ignored as having a low impact.

  Next, a procedure for creating a formula for satisfying the above hypothesis and performing a simulation for estimating the regression to the average will be described.

(1) First, as described above, “intra-individual variation in measured values takes a normal distribution centered on the true value of each measurement subject, and all of the measurement subjects are measured when participating than the true value. I hypothesized that I participated because of the high or high value.

  In addition, when the intervention is aimed at improving “measurement items that are meaningful in increasing or increasing the value” such as good cholesterol (HDL cholesterol) and maximum oxygen intake which is an index representing physical fitness, In other words, when the direction of evaluation of the effect of intervention is expected to “rise or increase”, “intra-individual variation in measured values has a normal distribution centered on the true value of each measurement subject, A hypothesis may be made that all measurement subjects participated because the measured value at the time of participation was lower or lower than the true value.

Simultaneously or in parallel with the above hypothetical work, the deviation (u _i ) from the true value of the measured value in the virtual individual is defined as follows.

u _i = (Xi−μi) / σ

  In this equation, X is a measured value of a virtual individual, μ is a true value of the virtual individual, and σ is a standard deviation of the entire group of actually measured values of each individual of the intervention target group before the intervention. The intervention target group is a “group performing intervention” that is not a control group. The intervention target group is a “group that is actually measured” that is not a virtual group.

(2) Since u _i has a normal distribution, one normal random number with an average = 0 and a standard deviation = 1 is generated as a deviation u from a true value of a measured value possessed by a virtual individual.

(3) By substituting one normal random number in (2) above into u of the following estimation formula that “the probability that an individual whose measurement value is higher or more than the true value is higher than the true value is high” is: A probability rate that a virtual individual having the deviation u participates in the virtual group R satisfying the hypothesis (1) is obtained.

Probability to participate rate = 1 / (1 + exp (−a × u))

  In this expression, exp is an exponential function, and a is a constant.

  As shown in FIG. 5B, the above estimation formula is a formula in which the probability of participation increases as the deviation from the true value increases. In the above estimation formula, the S-shaped curve in FIG. 5B changes depending on the value of the constant a. As a contrivance for determining this value, the above constant a is determined so that it is close to the results of previous studies conducted with the control group in the studies that have been conducted so far. . By doing in this way, it leads to improving the reliability of simulation.

  In addition, when the intervention is aimed at improving “measurement items that are meaningful in increasing or increasing the value” such as good cholesterol (HDL cholesterol) and maximum oxygen intake which is an index representing physical fitness, In other words, when the direction of evaluation of the effect of intervention is expected to “rise or increase”, an estimation formula that “there is a high probability that individuals with lower or fewer measured values will join the group” That is, an estimation formula in which the above (−a × u) is replaced with (a × u) may be used.

(4) One random number from 0 to 1 (r) is generated as a cut-off value for determining whether or not a virtual individual having the deviation u is allowed to participate in the virtual group R.

(5) The probability rate of participation determined in (3) above is compared with r as the cutoff value generated in (4) above, and only when rate> r, a virtual individual having the deviation u Are adopted as virtual individuals constituting the virtual group R. That is, if rate> r, the virtual individual having the deviation u is adopted and incorporated into the virtual group R. If rate = <r, the virtual individual having the deviation u is not adopted and is not included in the virtual group R.

(6) The above operations (2) to (5) are repeated a plurality of times using a computer. As a result, a group that satisfies the above hypothesis (1), that is, a virtual group R in which virtual individuals whose measured values are higher or more than the true value participate, is created in a simulated manner. In this (6), the number of times that the operations from (2) to (5) are repeated using a computer is determined by the number of virtual individuals incorporated in the virtual group R being the actual number of intervention target groups. It is desirable that the number be close to the number of people. For example, it is desirable to repeat the operation as many times as the number of the actual intervention target group.

(7) The average u _{ave of the} entire virtual group R of deviation u from the true value μ of the measured value X of each virtual individual included in the virtual group R simulated in (6) above Ask for.

(8) Based on the above equation (1), Δ which is “regression to average” of the virtual group R is obtained by the following equation.

Δ = average X _{ave of the} entire virtual population R of the measured value X of each virtual individual X _{ave −average} value of the entire virtual group R of the true value μ of each virtual individual μ _ave = true of the measured value of each virtual individual The average u _{ave of the} entire virtual group R of deviations u from the value of × × standard deviation σ of the entire group of actual measurements of each individual of the intervention target group before the intervention

(9) The operations from (2) to (8) are performed a plurality of times using a computer. The present inventor is currently conducting a trial using computer software called Mathcad. The number of times of performing this work (computer simulation) is not particularly limited, but it is desirable that the number of times is as large as possible, for example, from several tens of thousands to several millions. Then, an average value of each Δ obtained in each implementation is obtained.

  The average value of each Δ obtained in the above (7) is estimated as “regression to the average” of the intervention target group. As described above, the amount of decrease (return to the average) without intervention was obtained by computer simulation. The difference between the average of improvement values by actual intervention (for example, the blood pressure decrease amount actually measured by intervention) and Δ (regression to the average) obtained by the simulation can be evaluated as the true effect of the intervention.

  As described above, in the present embodiment, by generating a normal random number in the above (2) and generating a random number r of 0 to 1 as a cutoff value in the above (4), the participation probability rate is randomly determined. In addition to determining, a cut-off value for determining whether or not a virtual individual having the deviation u is to be included in the virtual group R is also determined at random. Therefore, in the present embodiment, the virtual group R that satisfies the above-mentioned hypothesis that “each measurement subject participated because the measurement value at the time of participation is higher or more than the true value” can be created more randomly. Become. Therefore, according to the present embodiment, it is possible to further improve the accuracy of the simulation (see (8) above) for obtaining the “regression to average” of the virtual group R.

  Next, an example will be described in which the method of the embodiment of the present invention described above is applied to the evaluation of the results of specific health promotion activities performed by the inventor on a predetermined group of individuals. The present inventor measured the effect on hypertension improvement by exercise and nutrition guidance for 12 weeks for 94 subjects. Of the 94 people, 32 were hypertensive. The contents of exercise guidance in health promotion activities are exercises that are about 50% of the maximum oxygen intake, such as bicycle ergometer, walking, mild aerobics, etc. once an hour, twice a week, 12 weeks It was to do. In addition, the content of nutritional guidance in health promotion activities is changed by changing the type and amount of nutrients that should be consumed within the above period (by the subject) while conducting individual guidance by the nutrition manager once in the first half and in the middle. It was to do.

  Among the 94 persons, blood pressure was measured for 32 hypertensive persons before and after the start of the instruction. As a result, the average of each measured value was 151.2 ± 13.2 mmHg before guidance, and 142.6 ± 13.5 mmHg after guidance, and blood pressure was significantly reduced by guidance (Paired t test; p <0. 01). The effect of lowering blood pressure by the guidance was 8.6 mmHg.

  However, the above effect of 8.6 mmHg is considered to be “effect by health promotion activity + return to average”. Therefore, the work of estimating this "return to average" was performed as follows.

  First, multiple virtual hypertensive individuals (for example, 32 people) participated in the health promotion activity program because their measured values at the time of participation were higher than their true values (blood pressure). I made a hypothesis. Under this hypothesis, if the measurement is performed again without any intervention, the blood pressure measurement value will decrease (see FIG. 6).

  Next, the deviation from the true value possessed by each virtual individual in the virtual group consisting of a plurality of virtual hypertensive individuals was defined as follows (see FIG. 5A).

u _i = (Xi−μi) / σ

  In this equation, X is a measured value of a virtual individual, μ is a true value of the virtual individual, σ is an entire group of actually measured values of each individual of the intervention target group (32 actual hypertensive persons) before the intervention. Standard deviation.

Next, since u _i has a normal distribution, one normal random number having an average = 0 and a standard deviation = 1 is generated as a deviation u from a true value of a measured value possessed by a virtual individual ( (See FIG. 8 (a)).

  Then, this normal random number is substituted into u in the following estimation expression that “the individual whose measurement value is higher than the true value is high in the probability of participation”, and the virtual individual having the deviation u is The probability rate of participating in the virtual group was obtained (see FIG. 7).

Probability to participate rate = 1 / (1 + exp (−a × u))

In this expression, exp is an exponential function, and a is a constant.

  Next, a random number from 0 to 1 (denoted r) was generated as a cut-off value for determining whether or not a virtual individual having the deviation u is to be included in the virtual group R. Then, the probability rate of participation determined by the estimation formula is compared with r as the cutoff value, and a virtual individual having the deviation u is incorporated into the virtual group R only when rate> r. It was.

  The operations from the step of generating one normal random number to the step of incorporating a virtual individual having the deviation u into the virtual group R only when the participation rate rate is greater than r A virtual group R in which a virtual individual whose measured value is higher than the true value participates was simulated by repeating the number of times close to twice the number of persons in the group (see FIG. 8B).

Next, the average u _ave of the entire virtual group R of deviation u from the true value μ of the measured value X of each virtual individual was obtained. Then, on the basis of the above formula u _i = (Xi−μi) / σ, Δ which is “regression to the average” of the virtual group R was obtained by the following formula.

Δ = average X _{ave of the} entire virtual population R of the measured value X of each virtual individual X _{ave −average} value of the entire virtual group R of the true value μ of each virtual individual μ _ave = true of the measured value of each virtual individual The average u _{ave of the} entire virtual group R of deviations u from the value of × × standard deviation σ of the entire group of actual measurements of each individual of the intervention target group before the intervention

  Next, the simulation for obtaining Δ which is the “regression to the average” of the above-mentioned virtual group R was repeatedly performed many times. Next, an average value of each Δ obtained by each simulation was obtained, and this average value of Δ was estimated as “regression to average” of the group of 32 actual hypertensives (intervention target group). . From the above, the average of the whole group of the blood pressure decrease amount of the actual measurement value of each individual of the group of 32 actual hypertensives (intervention target group) and the average value of each Δ ("regression to the average") The difference was considered a substantial intervention effect.

  As a result of actually performing a simulation by executing the Monte Carlo method for the above example, the return to the average was 2.9 ± 2.4 mmHg. The result of the Mann-Whitney U test (a non-parametric version of the so-called “unpaired t-test”, one of the statistical methods for testing the difference between the two groups) was p = 0.0336. Therefore, as a result of testing the difference between the actual measured value before and after the intervention and the average of Δ obtained by simulation, the null hypothesis that they are equal to each other is rejected, and the difference is significant. Proved. From the above, it was possible to evaluate that the result of the intervention in this example had a significant decrease in the actual measurement value. Moreover, it was able to be evaluated that the true effect of the intervention in a present Example is 5.7 mmHg.

  In the above description, exercise of a predetermined type and amount within the period of about 12 weeks actually conducted by the present inventor and execution of predetermined nutritional guidance within a period of about 12 weeks (taken by meals) The example in which the present Example 1 was used to evaluate the true effect on the improvement of hypertension due to the change in the type and amount of nutrients to be performed has been described. Real effects on health improvement by, for example, “implementing activities to reduce the amount of smoking (including reducing smoking to zero) in the workplace and home within a given period” without being limited to implementation It can also be used to evaluate the true effect on the improvement of illness / symptoms due to the evaluation of "Implementation of new drugs to patients within a predetermined period".

The conceptual block diagram for demonstrating the evaluation apparatus used for the evaluation method of the true effect by this invention. The figure for demonstrating the evaluation method of the true effect by intervention by embodiment of this invention. The figure for demonstrating this embodiment. The figure for demonstrating this embodiment. The figure for demonstrating this embodiment. The figure for demonstrating Example 1 of this invention for evaluating the true effect by health promotion activity. The figure for demonstrating Example 1 of this invention. The figure for demonstrating Example 1 of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 1st input part 2 1st calculating part 3 1st memory | storage part 4 2nd calculating part 5 2nd memory | storage part 6 Estimating part 7 2nd input part 8 3rd calculating part 9 3rd memory | storage part 10 4th calculating part 11 Comparison Part 12 Output part

Claims (4)

The real effect of the intervention when the intervention is implemented in the target population for the purpose of improving the “measurement items that are meaningful to decrease or decrease the value” such as individual blood pressure and neutral fat. A device including a computer system for evaluating
“Regular administration of a predetermined amount of drug within a predetermined period, regular predetermined amount of exercise during a predetermined period, changes in the type and amount of nutrients taken within a predetermined period, and regular smoking during a predetermined period A first input means for inputting a result of actual measurement of the health-related items of each individual of the implementation target group before performing "at least one of the reductions in quantity";
Based on the actual measurement result input by the first input means, a first computing means for obtaining an average value of the entire group of actual measurement values of each individual of the implementation target group before the implementation;
First storage means for storing an average value of the entire group of actual measurement values of each individual of the implementation target group before the execution obtained by the first calculation means;
Second computing means for obtaining a standard deviation σ of the entire group of actually measured values of each individual of the implementation target group before the implementation based on the actual measurement result input by the first input means;
Second storage means for storing a standard deviation of the entire group of actually measured values of each individual of the implementation target group before the execution obtained by the second calculation means;
It is a value corresponding to the return to the average of the implementation target group before the implementation using the standard deviation of the entire group of the actual measurement values of each individual of the implementation target group before the implementation read from the second storage means. An estimation means for estimating "return to average";
A second input means for inputting a result of actual measurement of items related to health of each individual of the implementation target group after the implementation;
Third computing means for obtaining an average value of the entire group of actually measured values of each individual of the implementation target group after the implementation based on the actual measurement result input by the second input means;
Third storage means for storing an average value of the entire group of measured values of each individual of the implementation target group after the execution obtained by the third calculation means;
The average value of the entire group of actual measurement values of each individual of the implementation target group before the implementation is read from the first storage unit, and the group of actual measurement values of each individual of the implementation target group after the implementation from the third storage unit A fourth calculating means for reading out the average value of the whole and calculating the difference between the two,
The average value of the whole group of the actual measurement values of each individual of the implementation target group before the execution outputted from the fourth calculation means and the average value of the whole group of the actual measurement values of each individual of the implementation target group after the execution Comparison means for comparing the difference with the “regression to the average” of the implementation target group before the execution output from the estimation means;
Output means for outputting a comparison result from the comparison means;
The estimation means estimates the “regression to the mean” of the implementation target group before the implementation according to the following procedures (1) to (8). , A device for evaluating the true effects of implementing interventions in the target population.
(1) operation means A is from taking u _i is the normal distribution in the following expression defining deviation (u _i) from the true value of the measured values in the individual virtual, mean = 0, standard deviation = 1 One normal random number is generated as a deviation u from the true value of the measured value of the virtual individual, and this is stored in the storage means A.
u _i = (Xi−μi) / σ (where X is the measured value of the virtual individual, μ is the true value of the virtual individual, and σ is the measured value of each individual in the implementation target group before the execution) Standard deviation of the entire population.)
(2) The calculation means B reads one normal random number generated in the above (1) from the storage means A, and determines this one normal random number as “an individual whose measurement value is higher or more than the true value is a group. By substituting into u of the following estimation formula that “the probability of participating in is high”, a virtual individual having the deviation u has “the measured value at the time of participation of all the measurement subjects is higher than the true value or The probability rate of participating in the virtual group R that satisfies the hypothesis that “participated because there are many” is obtained, and this is stored in the storage means B.
Probability of participation rate = 1 / (1 + exp (−a × u)) (where exp is an exponential function and a is a constant)
(3) The calculation means C uses one random number (set as r) of 0 to 1 as a cut-off value for determining whether or not a virtual individual having the deviation u is allowed to participate in the virtual group R. Generated and stored in the storage means C.
(4) The cut that the calculation means D reads out the probability of participation rate obtained in (2) above from the storage means B, and that is generated in (3) and stored in the storage means C. Compared with r as an off value, only when rate> r, a virtual individual having the deviation u is adopted as a virtual individual constituting the virtual group R, and this is stored in the storage means D.
(5) The control means controls the calculation means A, B, C, and D, and repeats the operations (1) to (4) a plurality of times. A group that satisfies the hypothesis that `` participated because there are more or more measured values at the time of participation than the true value '', that is, a virtual group R in which virtual individuals whose measured values are higher or more than the true value participate Create a simulation and store it in the storage means E.
(6) The virtual group R of the deviation u from the true value μ of the measured value X of each virtual individual included in the virtual group R simulated by the calculation means E in the above (5) The overall average u <sub>ave</sub> is obtained and stored in the storage means F.
(7) The average of the entire virtual group R of deviation u from the true value of the measured value of each virtual individual read out from the storage means F by the calculation means F based on the formula (1) Based on u <sub>ave</sub> and the standard deviation σ read out from the second storage means, Δ which is the “regression to average” of the virtual group R is obtained by the following equation and stored in the storage means G. Let
Δ = average X <sub>ave of the</sub> entire virtual population R of the measured value X of each virtual individual X <sub>ave −average</sub> value of the entire virtual group R of the true value μ of each virtual individual μ <sub>ave</sub> = true of the measured value of each virtual individual The average u <sub>ave of the</sub> entire virtual group R of deviations u from the value of x × the standard deviation σ of the entire group of the actual measurement values of each individual of the target group before the execution
(8) The control means controls the arithmetic means A, B, C, D, E, and F, and performs the operations (1) to (7) a plurality of times, and is obtained in each implementation. The average value of each Δ is obtained, and the average value of Δ is estimated as “regression to the average” of the implementation target group, and this is stored in the storage means H. When performing interventions in the target population for the purpose of improving “measurement items that are meaningful to increase or increase the value” such as the maximum oxygen intake that is an indicator of individual HDL cholesterol and physical fitness A device, including a computer system, for assessing the true effect of performing the intervention,
“Regular administration of a predetermined amount of drug within a predetermined period, regular predetermined amount of exercise during a predetermined period, changes in the type and amount of nutrients taken within a predetermined period, and regular smoking during a predetermined period A first input means for inputting a result of actual measurement of the health-related items of each individual of the implementation target group before performing "at least one of the reductions in quantity";
Based on the actual measurement result input by the first input means, a first computing means for obtaining an average value of the entire group of actual measurement values of each individual of the implementation target group before the implementation;
First storage means for storing an average value of the entire group of actual measurement values of each individual of the implementation target group before the execution obtained by the first calculation means;
Second computing means for obtaining a standard deviation σ of the entire group of actually measured values of each individual of the implementation target group before the implementation based on the actual measurement result input by the first input means;
Second storage means for storing a standard deviation of the entire group of actually measured values of each individual of the implementation target group before the execution obtained by the second calculation means;
It is a value corresponding to the return to the average of the implementation target group before the implementation using the standard deviation of the entire group of the actual measurement values of each individual of the implementation target group before the implementation read from the second storage means. An estimation means for estimating "return to average";
A second input means for inputting a result of actual measurement of items related to health of each individual of the implementation target group after the implementation;
Third computing means for obtaining an average value of the entire group of actually measured values of each individual of the implementation target group after the implementation based on the actual measurement result input by the second input means;
Third storage means for storing an average value of the entire group of measured values of each individual of the implementation target group after the execution obtained by the third calculation means;
The average value of the entire group of actual measurement values of each individual of the implementation target group before the implementation is read from the first storage unit, and the group of actual measurement values of each individual of the implementation target group after the implementation from the third storage unit A fourth calculating means for reading out the average value of the whole and calculating the difference between the two,
The average value of the whole group of the actual measurement values of each individual of the implementation target group before the execution outputted from the fourth calculation means and the average value of the whole group of the actual measurement values of each individual of the implementation target group after the execution Comparison means for comparing the difference with the “regression to the average” of the implementation target group before the execution output from the estimation means;
Output means for outputting a comparison result from the comparison means;
The estimation means estimates the “regression to the mean” of the implementation target group before the implementation according to the following procedures (1) to (8). A device for evaluating the true effects of interventions.
(1) operation means A is from taking u <sub>i</sub> is the normal distribution in the following expression defining deviation (u <sub>i)</sub> from the true value of the measured values in the individual virtual, mean = 0, standard deviation = 1 One normal random number is generated as a deviation u from the true value of the measured value of the virtual individual, and this is stored in the storage means A.
u <sub>i</sub> = (Xi−μi) / σ (where X is the measured value of the virtual individual, μ is the true value of the virtual individual, and σ is the measured value of each individual in the implementation target group before the execution) Standard deviation of the entire population.)
(2) The calculation means B reads one normal random number generated in the above (1) from the storage means A, and reads this one normal random number as “an individual whose measured value is lower or less than the true value becomes a group. By substituting into u of the following estimation formula that “the probability of participation is high”, a virtual individual having the deviation u has “the measurement value at the time of participation is lower or less than the true value of all measurement subjects” The probability rate of participating in the virtual group R that satisfies the hypothesis that "participated for" is obtained, and this is stored in the storage means B.
Probability of participation rate = 1 / (1 + exp (a × u)) (where exp is an exponential function and a is a constant)
(3) The calculation means C uses one random number (set as r) of 0 to 1 as a cut-off value for determining whether or not a virtual individual having the deviation u is allowed to participate in the virtual group R. Generated and stored in the storage means C.
(4) The cut-off value obtained by the calculation means D reading the probability of participation rate obtained in the above (2) from the storage means B and generating it in the above (3) and storing it in the storage means C. And only when rate> r, a virtual individual having the deviation u is adopted as a virtual individual constituting the virtual group R, and this is stored in the storage means D.
(5) The control means controls the arithmetic means A, B, C, and D, and repeats the operations from (1) to (4) a plurality of times. Is a group that satisfies the hypothesis that “participation was due to lower or lower measured value at the time of participation than the true value”, that is, a virtual group R in which virtual individuals whose measured values are lower or lower than the true value Is simulated and stored in the storage means E.
(6) The virtual means of the deviation u from the true value μ of the measured value X of each virtual individual included in the virtual group R created and stored in the simulation (5) by the computing means E An average u <sub>ave</sub> of the entire group R is obtained and stored in the storage means F.
(7) The average of the entire virtual group R of deviation u from the true value of the measured value of each virtual individual read out from the storage means F by the calculation means F based on the formula (2) Based on u <sub>ave</sub> and the standard deviation σ read out from the second storage means, Δ which is the “regression to average” of the virtual group R is obtained by the following equation and stored in the storage means G. Let
Δ = average X <sub>ave of the</sub> entire virtual population R of the measured value X of each virtual individual X <sub>ave −average</sub> value of the entire virtual group R of the true value μ of each virtual individual μ <sub>ave</sub> = true of the measured value of each virtual individual The average u <sub>ave of the</sub> entire virtual group R of deviations u from the value of x × the standard deviation σ of the entire group of the actual measurement values of each individual of the target group before the execution
(8) The control means controls the arithmetic means A, B, C, D, E, and F, and performs the operations (1) to (7) a plurality of times, and is obtained in each implementation. The average value of each Δ is obtained, and the average value of Δ is estimated as “regression to the average” of the implementation target group, and this is stored in the storage means H. The real effect of the intervention when the intervention is implemented in the target population for the purpose of improving the “measurement items that are meaningful to decrease or decrease the value” such as individual blood pressure and neutral fat. A method using a computer system for evaluating
By means of the first input means, “administration of a predetermined amount of medicine periodically within a predetermined period, regular predetermined amount of exercise within a predetermined period, change of the type and amount of nutrients taken within a predetermined period, and predetermined A first step of inputting results of actual measurements of the health of each individual of the target population before performing “at least one of the regular reductions in smoking during the period”;
Based on the actual measurement result input by the first input means, the first calculation means obtains an average value of the entire group of actual measurement values of each individual of the implementation target group before the execution, and stores this in the first storage means. A second step of storing;
At the same time as or before or after the second step, the second calculation means, based on the actual measurement result input by the first input means, of the entire group of actual measurement values of each individual of the implementation target group before the implementation Obtaining a standard deviation σ and storing it in the second storage means;
The estimation means uses the standard deviation of the entire group of the measured values of each individual of the implementation target group before the execution read from the second storage means, and corresponds to the regression to the average of the implementation target group before the execution A fourth step of estimating the value of “regression to average”,
At the same time as or before or after at least one of Step 2, Step 3, and Step 4, the result of actual measurement of the items related to the health of each individual in the implementation target group after the implementation by the second input means. A fifth step of input,
Each individual of the implementation target group after the implementation input by the second input means by the third calculation means at the same time as or at least one of the steps 2, 3 and 4 A sixth step of obtaining an average value of the entire group of actual measurement values of each individual of the implementation target group after the implementation based on the actual measurement result of each item relating to health, and storing the average value in a third storage unit;
The fourth computing means reads the average value of the whole group of the measured values of each individual of the implementation target group before the execution read from the first storage means and the execution target group after the execution read from the third storage means. A seventh step for determining a difference between the measured value of each individual and the average value of the entire group;
The comparison means outputs the average value of the whole group of the actual measurement values of each individual of the implementation target group before the execution output from the fourth calculation means and the entire group of the actual measurement values of each individual of the implementation target group after the execution An eighth step of comparing the difference between the average value and the “regression to the average” of the execution target group before the execution output from the estimation means;
A ninth step of outputting the comparison result output from the comparison means by output means such as a display;
The estimation means in the fourth step is for estimating the “regression to the mean” of the implementation target group before the implementation according to the following procedure. To evaluate the true effect of the intervention.
(1) The calculation means A defines the deviation (u <sub>i</sub> ) from the true value of the measured value in the virtual individual. Since ui in the following equation has a normal distribution, the normal = 0 having the average = 0 and the standard deviation = 1 One random number is generated as a deviation u from the true value of the measured value possessed by the virtual individual, and this is stored in the storage means A.
u <sub>i</sub> = (Xi−μi) / σ (where X is the measured value of the virtual individual, μ is the true value of the virtual individual, and σ is the measured value of each individual in the implementation target group before the execution) Standard deviation of the entire population.)
(2) The calculation means B reads one normal random number generated in the above (1) from the storage means A, and determines this one normal random number as “an individual whose measurement value is higher or more than the true value is a group. By substituting into u of the following estimation formula that “the probability of participating in is high”, a virtual individual having the deviation u has “the measured value at the time of participation of all the measurement subjects is higher than the true value or The probability rate of participating in the virtual group R that satisfies the hypothesis that “participated because there are many” is obtained, and this is stored in the storage means B.
Probability of participation rate = 1 / (1 + exp (−a × u)) (where exp is an exponential function and a is a constant)
(3) The calculation means C uses one random number (set as r) of 0 to 1 as a cut-off value for determining whether or not a virtual individual having the deviation u is allowed to participate in the virtual group R. Generated and stored in the storage means C.
(4) The cut that the calculation means D reads out the probability of participation rate obtained in (2) above from the storage means B, and that is generated in (3) and stored in the storage means C. Compared with r as an off value, only when rate> r, a virtual individual having the deviation u is adopted as a virtual individual constituting the virtual group R, and this is stored in the storage means D.
(5) The control means controls the arithmetic means A, B, C, and D, and repeats the operations (1) to (4) a plurality of times. A group that satisfies the hypothesis that “participated because the measured value at the time of participation is higher or higher than the value of the user”, that is, a virtual group R in which virtual individuals whose measured values are higher or higher than the true value participate. Are created and stored in the storage means E.
(6) The virtual group R of the deviation u from the true value μ of the measured value X of each virtual individual included in the virtual group R simulated by the calculation means E in the above (5) The overall average u <sub>ave</sub> is obtained and stored in the storage means F.
(7) The average of the entire virtual group R of deviation u from the true value of the measured value of each virtual individual read out from the storage means F by the calculation means F based on the formula (1) Based on u <sub>ave</sub> and the standard deviation σ read out from the second storage means, Δ which is the “regression to average” of the virtual group R is obtained by the following equation and stored in the storage means G. Let
Δ = average X <sub>ave of the</sub> entire virtual population R of the measured value X of each virtual individual X <sub>ave −average</sub> value of the entire virtual group R of the true value μ of each virtual individual μ <sub>ave</sub> = true of the measured value of each virtual individual The average u <sub>ave of the</sub> entire virtual group R of deviations u from the value of x × the standard deviation σ of the entire group of the actual measurement values of each individual of the target group before the execution
(8) The control means controls the arithmetic means A, B, C, D, E, and F, and performs the operations (1) to (7) a plurality of times, and is obtained in each implementation. The average value of each Δ is obtained, and the average value of Δ is estimated as “regression to the average” of the implementation target group, and this is stored in the storage means H. When performing interventions in the target population for the purpose of improving “measurement items that are meaningful to increase or increase the value” such as the maximum oxygen intake that is an indicator of individual HDL cholesterol and physical fitness , Using a computer system to evaluate the true effect of implementing the intervention,
By means of the first input means, “administration of a predetermined amount of medicine periodically within a predetermined period, regular predetermined amount of exercise within a predetermined period, change of the type and amount of nutrients taken within a predetermined period, and predetermined A first step of inputting results of actual measurements of the health of each individual of the target population before performing “at least one of the regular reductions in smoking during the period”;
Based on the actual measurement result input by the first input means, the first calculation means obtains an average value of the entire group of actual measurement values of each individual of the implementation target group before the execution, and stores this in the first storage means. A second step of storing;
At the same time as or before or after the second step, the second calculation means, based on the actual measurement result input by the first input means, of the entire group of actual measurement values of each individual of the implementation target group before the implementation Obtaining a standard deviation σ and storing it in the second storage means;
The estimation means uses the standard deviation of the entire group of the measured values of each individual of the implementation target group before the execution read from the second storage means, and corresponds to the regression to the average of the implementation target group before the execution A fourth step of estimating the value of “regression to average”,
At the same time as or before or after at least one of Step 2, Step 3, and Step 4, the result of actual measurement of the items related to the health of each individual in the implementation target group after the implementation by the second input means. A fifth step of input,
Each individual of the implementation target group after the implementation input by the second input means by the third calculation means at the same time as or at least one of the steps 2, 3 and 4 A sixth step of obtaining an average value of the entire group of actual measurement values of each individual of the implementation target group after the implementation based on the actual measurement result of each item relating to health, and storing the average value in a third storage unit;
The fourth computing means reads the average value of the whole group of the measured values of each individual of the implementation target group before the execution read from the first storage means and the execution target group after the execution read from the third storage means. A seventh step for determining a difference between the measured value of each individual and the average value of the entire group;
The comparison means outputs the average value of the whole group of the actual measurement values of each individual of the implementation target group before the execution output from the fourth calculation means and the entire group of the actual measurement values of each individual of the implementation target group after the execution An eighth step of comparing the difference between the average value and the “regression to the average” of the execution target group before the execution output from the estimation means;
A ninth step of outputting the comparison result output from the comparison means by output means such as a display;
And the estimation means in the fourth step estimates the “regression to the mean” of the implementation target group before the implementation by the following procedures (1) to (8) To evaluate the true effect of the intervention.
(1) The calculation means A defines a deviation (u <sub>i</sub> ) from a true value of a measured value in a virtual individual. Since ui in the following equation has a normal distribution, the normal means with mean = 0 and standard deviation = 1 One random number is generated as a deviation u from the true value of the measured value possessed by the virtual individual, and this is stored in the storage means A.
u <sub>i</sub> = (Xi−μi) / σ (where X is the measured value of the virtual individual, μ is the true value of the virtual individual, and σ is the measured value of each individual in the implementation target group before the execution) Standard deviation of the entire population.)
(2) The calculation means B reads one normal random number generated in the above (1) from the storage means A, and reads this one normal random number as “an individual whose measured value is lower or less than the true value becomes a group. By substituting into u of the following estimation formula that “the probability of participation is high”, a virtual individual having the deviation u has “the measurement value at the time of participation is lower or less than the true value of all measurement subjects” The probability rate of participating in the virtual group R that satisfies the hypothesis that "participated for" is obtained, and this is stored in the storage means B.
Probability of participation rate = 1 / (1 + exp (a × u)) (where exp is an exponential function and a is a constant)
(3) The calculation means C uses one random number (set as r) of 0 to 1 as a cut-off value for determining whether or not a virtual individual having the deviation u is allowed to participate in the virtual group R. Generated and stored in the storage means C.
(4) The cut-off value obtained by the calculation means D reading the probability of participation rate obtained in the above (2) from the storage means B and generating it in the above (3) and storing it in the storage means C. And only when rate> r, a virtual individual having the deviation u is adopted as a virtual individual constituting the virtual group R, and this is stored in the storage means D.
(5) The control means controls the arithmetic means A, B, C, and D, and repeats the operations (1) to (4) a plurality of times. A group that satisfies the hypothesis that `` participated because the measured value at participation is lower or lower than the true value '', that is, a virtual group R in which virtual individuals whose measured values are lower or less than the true value participate Create a simulation and store it in the storage means E.
(6) The virtual means of the deviation u from the true value μ of the measured value X of each virtual individual included in the virtual group R created and stored in the simulation (5) by the computing means E An average u _ave of the entire group R is obtained and stored in the storage means F.
(7) The average of the entire virtual group R of deviation u from the true value of the measured value of each virtual individual read out from the storage means F by the calculation means F based on the formula (1) Based on u _ave and the standard deviation σ read out from the second storage means, Δ which is the “regression to average” of the virtual group R is obtained by the following equation and stored in the storage means G. Let
Δ = average X _{ave of the} entire virtual population R of the measured value X of each virtual individual X _{ave −average} value of the entire virtual group R of the true value μ of each virtual individual μ _ave = true of the measured value of each virtual individual The average u _{ave of the} entire virtual group R of deviations u from the value of x × the standard deviation σ of the entire group of the actual measurement values of each individual of the target group before the execution
(8) The control means controls the arithmetic means A, B, C, D, E, and F, and performs the operations (1) to (7) a plurality of times, and is obtained in each implementation. The average value of each Δ is obtained, and the average value of Δ is estimated as “regression to the average” of the implementation target group, and this is stored in the storage means H.

Images