CN111326254A - Method, apparatus, device and medium for index intervention - Google Patents

Method, apparatus, device and medium for index intervention Download PDF

Info

Publication number
CN111326254A
CN111326254A CN201811544043.0A CN201811544043A CN111326254A CN 111326254 A CN111326254 A CN 111326254A CN 201811544043 A CN201811544043 A CN 201811544043A CN 111326254 A CN111326254 A CN 111326254A
Authority
CN
China
Prior art keywords
index
controllable
indexes
value
response
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN201811544043.0A
Other languages
Chinese (zh)
Inventor
李京
涂威威
李南南
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
4Paradigm Beijing Technology Co Ltd
Original Assignee
4Paradigm Beijing Technology Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 4Paradigm Beijing Technology Co Ltd filed Critical 4Paradigm Beijing Technology Co Ltd
Priority to CN201811544043.0A priority Critical patent/CN111326254A/en
Publication of CN111326254A publication Critical patent/CN111326254A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/30—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for calculating health indices; for individual health risk assessment
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/50—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for simulation or modelling of medical disorders

Landscapes

  • Health & Medical Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Medical Informatics (AREA)
  • Public Health (AREA)
  • Biomedical Technology (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • Pathology (AREA)
  • Epidemiology (AREA)
  • General Health & Medical Sciences (AREA)
  • Primary Health Care (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention provides a method, a device, equipment and a medium for index intervention. Obtaining an arithmetic mechanism for providing an output result for an input sample, wherein the input sample comprises a plurality of indicators; determining a mapping relation between a controllable index and a response index in a plurality of indexes, wherein the controllable index is an index set of which the value can be adjusted, and the response index is an index set of which the value is difficult to adjust but can change according to the change of the controllable index; based on the mapping relation, the controllable indexes of the input samples are adjusted, so that the output result obtained by the operation of the operation mechanism aiming at the adjusted input samples meets the preset requirement; and providing the adjustment result of the controllable index for index intervention. Therefore, index intervention is carried out according to the adjustment result of the controllable index, expected operation output can be facilitated, and the method is easy to implement.

Description

Method, apparatus, device and medium for index intervention
Technical Field
The present invention relates generally to the field of data science and technology, and more particularly, to a method for performing index intervention, an apparatus for performing index intervention, a computing device, and a non-transitory machine-readable storage medium.
Background
Providing an output result for an input sample by constructing an operation mechanism (e.g., a prediction model (e.g., a machine learning model), a rule set of one or more rules, a function operation, etc.) is a common technical means in the fields of industrial production, finance, medical treatment, daily life, etc. For example, the production value or the loss of the machine can be predicted by constructing a prediction model so as to guide industrial production; the profit of the user can be predicted by constructing a profit prediction model so as to provide a basis for the user to make a decision; the physical condition of the user can also be managed by constructing a model for predicting the probability of illness, a model for predicting the weight, and the like.
After the calculation mechanism is established, for a given sample, a corresponding output result can be obtained. However, for the input object, the possible result is usually obtained only by the operation mechanism, but it is not known how to make the adjustment in a targeted manner to obtain the desired result.
Disclosure of Invention
An exemplary embodiment of the present invention is to provide a method and an apparatus for performing index intervention to solve the above technical problem.
According to a first aspect of the invention, a method for performing index intervention is proposed, comprising: obtaining an arithmetic mechanism for providing an output result for an input sample, wherein the input sample comprises a plurality of metrics; determining a mapping relation between a controllable index and a response index in the plurality of indexes, wherein the controllable index is an index set of which the value can be adjusted, and the response index is an index set of which the value is difficult to adjust but can change according to the change of the controllable index; based on the mapping relation, the controllable indexes of the input samples are adjusted, so that the output result obtained by the operation of the operation mechanism aiming at the adjusted input samples meets the preset requirement; and providing the adjustment result of the controllable index for index intervention.
Optionally, the operation mechanism comprises at least one of: machine learning model, rule set, function operation.
Optionally, the operation mechanism is a prediction model for providing a prediction result for the prediction problem, wherein the input sample is a prediction sample of the prediction model, and the characteristic of the input sample comprises a plurality of indexes related to the prediction problem.
Optionally, the step of determining a mapping relationship between a controllable index and a response index among the plurality of indexes comprises: according to a plurality of indexes included in the characteristics of the samples in the sample set, constructing a model for predicting the value of the response index based on the value of the controllable index so as to determine the mapping relation between the controllable index and the response index; or according to a plurality of indexes included in the characteristics of the samples in the sample set, constructing a model for predicting the value of the response index based on the value of the static index and the value of the controllable index so as to determine the mapping relation between the controllable index and the response index, wherein the static index is an index set of which the value is unchangeable.
Optionally, the step of determining a mapping relationship between a controllable index and a response index among the plurality of indexes comprises: according to the prior knowledge, the mapping relation between the controllable indexes and the response indexes in the multiple indexes is determined.
Optionally, the method further comprises: the method comprises the following steps of grouping the controllable indexes to divide the indexes with linkage relation in the controllable indexes into the same group, and adjusting the controllable indexes in the characteristics of the prediction sample, wherein the steps comprise: and in the value range of the controllable indexes, the controllable indexes are adjusted by taking the group as a unit based on the linkage relation among the indexes in the same group.
Optionally, the linkage relationship includes a correlation relationship and/or a causal relationship, and the step of grouping the controllable indicators includes: determining indexes with correlation in the controllable indexes in a calculation mode so as to divide the indexes with correlation in the controllable indexes into the same group; and/or determining the indexes with causal relationship in the controllable indexes according to external input so as to divide the indexes with causal relationship in the controllable indexes into the same group.
Optionally, the step of computationally determining the index having the correlation among the controllable indexes includes: and calculating the correlation among the indexes in the controllable indexes in an independence verification mode.
Optionally, the step of adjusting the controllable indicator among the features of the prediction samples comprises: and adjusting the controllable indexes in the characteristics of the prediction samples by taking the aim that the prediction result obtained by predicting the adjusted prediction samples by the prediction model meets the preset requirement and the adjustment cost of the controllable indexes is reduced as much as possible.
According to a second aspect of the invention, there is also provided a method for performing index intervention, comprising: acquiring a disease probability prediction model; predicting the disease probability of the user according to the characteristics of the user by using a disease probability prediction model, wherein the characteristics of the user comprise a plurality of indexes related to the predicted disease probability; under the condition that the prevalence probability of the user is predicted to be higher than a preset threshold value, based on a mapping relation between a static index and a controllable index to a response index in a plurality of indexes, adjusting the controllable index in the characteristics of the user, so that the prevalence probability predicted by a prevalence probability prediction model aiming at the adjusted characteristics of the user is lower than the preset threshold value, wherein the static index is an index set with an unchangeable value, the controllable index is an index set with an adjustable value, and the response index is an index set with a value which is difficult to adjust and can change according to the change of the controllable index; and providing the adjustment result of the controllable index so as to facilitate index intervention of a user.
Optionally, the static indicator comprises at least one of: age, gender, height, occupation; the controllable indicator comprises at least one of the following: body weight, one or more lifestyle-related characteristics, blood pressure, blood glucose; the response indicator includes at least one of: uric acid and blood fat.
Optionally, the method further comprises: and determining the mapping relation of the static indexes and the controllable indexes in the indexes to the response indexes.
Optionally, the step of determining a mapping relationship of the static index and the controllable index to the response index among the plurality of indexes comprises: according to a plurality of indexes included in the characteristics of the samples in the sample set, constructing a model for predicting the value of the response index based on the value of the static index and the value of the controllable index so as to determine a mapping relation; or determining the mapping relation between the static indexes and the controllable indexes in the multiple indexes to the response indexes according to the prior knowledge.
Optionally, the method further comprises: the method comprises the following steps of grouping the controllable indexes to divide the indexes with linkage relation in the controllable indexes into the same group, and adjusting the controllable indexes in the characteristics of the prediction sample, wherein the steps comprise: and in the value range of the controllable indexes, the controllable indexes are adjusted by taking the group as a unit based on the linkage relation among the indexes in the same group.
Optionally, the linkage relationship includes a correlation relationship and/or a causal relationship, and the step of grouping the controllable indicators includes: determining indexes with correlation in the controllable indexes in a calculation mode so as to divide the indexes with correlation in the controllable indexes into the same group; and/or determining the indexes with causal relationship in the controllable indexes according to external input so as to divide the indexes with causal relationship in the controllable indexes into the same group.
Optionally, the step of computationally determining the index having the correlation among the controllable indexes includes: and calculating the correlation among the indexes in the controllable indexes in an independence verification mode.
Optionally, the step of adjusting the controllable indicator among the features of the prediction samples comprises: and adjusting the controllable indexes in the characteristics of the prediction sample by taking the purposes that the morbidity probability obtained by predicting the adjusted characteristics of the user by the morbidity probability prediction model is lower than a preset threshold value and the adjustment cost of the controllable indexes is reduced as much as possible.
According to a third aspect of the present invention, there is also provided an apparatus for performing index intervention, comprising: an obtaining unit configured to obtain an operation mechanism for providing an output result for an input sample, wherein the input sample includes a plurality of indexes; the device comprises a determining unit, a calculating unit and a processing unit, wherein the determining unit is used for determining a mapping relation between a controllable index and a response index in the plurality of indexes, the controllable index is an index set of which the value can be adjusted, and the response index is an index set of which the value is difficult to adjust but can change according to the change of the controllable index; the adjusting unit is used for adjusting the controllable indexes of the input samples based on the mapping relation so that the output result obtained by the operation of the operation mechanism aiming at the adjusted input samples meets the preset requirement; and a providing unit for providing the adjustment result of the controllable index for index intervention.
Optionally, the operation mechanism comprises at least one of: machine learning model, rule set, function operation.
Optionally, the operation mechanism is a prediction model for providing a prediction result for the prediction problem, wherein the input sample is a prediction sample of the prediction model, and the characteristic of the input sample comprises a plurality of indexes related to the prediction problem.
Optionally, the determining unit constructs a model for predicting the value of the response index based on the value of the controllable index according to a plurality of indexes included in the characteristics of the samples in the sample set to determine a mapping relationship between the controllable index and the response index, or the determining unit constructs a model for predicting the value of the response index based on the value of the static index and the value of the controllable index according to a plurality of indexes included in the characteristics of the samples in the sample set to determine a mapping relationship between the controllable index and the response index, where the static index is an index set whose value is unchangeable.
Optionally, the determination unit determines a mapping relationship between a controllable index and a response index among the plurality of indexes according to a priori knowledge.
Optionally, the apparatus further comprises: and the grouping unit is used for grouping the controllable indexes so as to divide the indexes with the linkage relation in the controllable indexes into the same group, and the adjusting unit is used for adjusting the controllable indexes in the value range of the controllable indexes by taking the group as a unit based on the linkage relation among the indexes in the same group.
Optionally, the linkage relationship includes a correlation relationship and/or a causal relationship, the grouping unit determines the indicators having the correlation relationship among the controllable indicators in a calculation manner to divide the indicators having the correlation relationship among the controllable indicators into the same group, and/or the grouping unit determines the indicators having the causal relationship among the controllable indicators according to an external input to divide the indicators having the causal relationship among the controllable indicators into the same group.
Optionally, the grouping unit calculates the correlation between the indexes in the controllable indexes by means of independence check.
Optionally, the adjusting unit adjusts the controllable index in the characteristics of the prediction sample with the aim that a prediction result obtained by predicting the adjusted prediction sample by the prediction model meets a predetermined requirement and the adjustment cost of the controllable index is reduced as much as possible.
According to a fourth aspect of the present invention, there is also provided an apparatus for performing index intervention, comprising: an acquisition unit for acquiring a disease probability prediction model; a prediction unit for predicting an illness probability of a user according to a feature of the user using an illness probability prediction model, wherein the feature of the user includes a plurality of indexes related to the predicted illness probability; the system comprises an adjusting unit, a response index generating unit and a judging unit, wherein the adjusting unit is used for adjusting a controllable index in the characteristics of a user based on a mapping relation between a static index and a controllable index to the response index in a plurality of indexes under the condition that the prevalence probability of the user is higher than a preset threshold value, so that the prevalence probability obtained by predicting the prevalence probability prediction model aiming at the adjusted characteristics of the user is lower than the preset threshold value, the static index is an index set of which the value is unchangeable, the controllable index is an index set of which the value is adjustable, and the response index is an index set of which the value is difficult to adjust and can change according to the change of the controllable index; and the providing unit is used for providing the adjustment result of the controllable index so that the user can perform index intervention.
Optionally, the static indicator comprises at least one of: age, gender, height, occupation; the controllable indicator comprises at least one of the following: body weight, one or more lifestyle-related characteristics, blood pressure, blood glucose; the response indicator includes at least one of: uric acid and blood fat.
Optionally, the apparatus further comprises: and the determining unit is used for determining the mapping relation between the static indexes and the controllable indexes in the indexes to the response indexes.
Optionally, the determining unit constructs a model for predicting a value of the response index based on a value of the static index and a value of the controllable index according to a plurality of indexes included in the characteristics of the samples in the sample set to determine the mapping relationship; or the determining unit determines the mapping relation according to the prior knowledge.
Optionally, the apparatus further comprises: and the grouping unit is used for grouping the controllable indexes so as to divide the indexes with the linkage relation in the controllable indexes into the same group, and the adjusting unit is used for adjusting the controllable indexes in the value range of the controllable indexes by taking the group as a unit based on the linkage relation among the indexes in the same group.
Optionally, the linkage relationship includes a correlation relationship and/or a causal relationship, the grouping unit determines the indicators having the correlation relationship among the controllable indicators in a calculation manner to divide the indicators having the correlation relationship among the controllable indicators into the same group, and/or the grouping unit determines the indicators having the causal relationship among the controllable indicators according to an external input to divide the indicators having the causal relationship among the controllable indicators into the same group.
Optionally, the grouping unit calculates the correlation between the indexes in the controllable indexes by means of independence check.
Optionally, the adjusting unit adjusts the controllable index in the features of the prediction sample with a goal that the prevalence probability predicted by the prevalence probability prediction model for the adjusted features of the user is lower than a predetermined threshold and the adjustment cost of the controllable index is reduced as much as possible.
According to a fifth aspect of the present invention, there is also provided a computing device comprising: a processor; and a memory having stored thereon executable code which, when executed by the processor, causes the processor to perform a method as set forth in the first or second aspect of the invention.
According to a sixth aspect of the invention, there is also provided a non-transitory machine-readable storage medium having stored thereon executable code which, when executed by a processor of an electronic device, causes the processor to perform a method as set forth in the first or second aspect of the invention.
According to the method and the device for index intervention, a reasonable index adjustment scheme can be provided. Index intervention is carried out according to the adjustment result of the controllable index, and expected operation output can be facilitated. Compared with an index control scheme which can facilitate expected operation output by directly using an optimization method without distinguishing the characteristics of the prediction samples, the method has the advantages that the adjustment result of the controllable index is more reasonable, and the method can be realized and is easy to realize when index intervention is carried out on the basis of the adjustment result of the controllable index provided by the invention.
Additional aspects and/or advantages of the present general inventive concept will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the general inventive concept.
Drawings
The above and other objects and features of exemplary embodiments of the present invention will become more apparent from the following description taken in conjunction with the accompanying drawings which illustrate exemplary embodiments, wherein:
FIG. 1 shows a flowchart of a method for performing index intervention according to an exemplary embodiment of the present invention;
FIG. 2 shows a flow chart of a method of the invention applied to intervene on a user's physical indicators to reduce the user's risk of contracting a disease;
FIG. 3 shows a block diagram of an apparatus for performing index intervention according to an exemplary embodiment of the present invention;
FIG. 4 shows a block diagram of an apparatus for performing index intervention according to another exemplary embodiment of the present invention;
fig. 5 shows a schematic structural diagram of a computing device that can be used to implement the above method according to an exemplary embodiment of the present invention.
Detailed Description
Reference will now be made in detail to the embodiments of the present invention, examples of which are illustrated in the accompanying drawings, wherein like reference numerals refer to the like elements throughout. The embodiments are described below in order to explain the present invention by referring to the figures.
For a given sample, it is input to a particular operation mechanism to get an output result, which is typically a process that uses operations to get the desired result. Under certain constraint conditions, a proper input set is searched to optimize the objective function, which is an optimization process. How to give appropriate suggestions for the indexes contained in a given sample so that the change of the indexes can promote the expected operation output is not simply combined with the optimization method and the operation mechanism. The most important problem is that the index control scheme obtained by using the optimization method aiming at the operation mechanism cannot be realized necessarily, and is not easy to realize necessarily.
This is because the input sample may include multiple indexes related to the operation problem, and values of the multiple indexes may not be adjusted, and for an index whose value can be adjusted, when adjusting the value, there may be a certain constraint limit, for example, there are some indexes that are difficult to be directly adjusted but may change with changes of other indexes. Therefore, in the prior art, indexes of samples are not distinguished, an optimization method is directly used to obtain an index control scheme capable of promoting expected output, the obtained index control scheme cannot be guaranteed to be realized, and the obtained index control scheme cannot be guaranteed to be easy to realize.
Taking the disease risk prediction as an example, assume that the input samples of the disease risk prediction model include age, sex, height, weight, one or more lifestyle-related characteristics, blood pressure, blood glucose, uric acid, blood lipid, and other indicators. Lifestyle-related characteristics may include, but are not limited to, time of work and rest, diet, exercise, and other indicators that can be adjusted manually. If the disease risk prediction model predicts that the risk of the disease of the user is higher according to the existing index prediction of the user, a proper index adjustment suggestion needs to be given to the user, so that the disease probability of the user after index adjustment according to the index adjustment suggestion is lower. If the indexes are not distinguished, an index adjusting scheme with the lowest or lower morbidity probability is calculated by directly using an optimization method, if the obtained index adjusting scheme comprises the adjustment of the indexes such as age, sex, height, uric acid, blood fat and the like, the adjustment of the indexes such as the age, the sex and the height is obviously impossible to realize, and the adjustment of the indexes such as the uric acid, the blood fat and the like is difficult to realize for a user (because the user generally does not know how to operate to adjust the uric acid and the blood fat). Therefore, the resulting index adjustment scheme calculated directly using the optimization method is not necessarily reasonable.
In view of the above, the present invention proposes a scheme for performing index intervention. Fig. 1 shows a flowchart of a method for performing index intervention according to an exemplary embodiment of the invention. The method shown in fig. 1 can be implemented by a computer program or by a special device for performing index interventions.
In step S110, an operation mechanism for providing an output result for an input sample is obtained, wherein the input sample may include a plurality of indexes. The operation mechanism may be any operation process capable of obtaining a corresponding output result through a specific operation on an input sample. As an example, the operation mechanism may comprise at least one of: a machine learning model, a set of rules (which may include at least one rule), a function operation. Various combinations between the above items may be made, such as a combination of machine learning models and expert rules, etc. Further, the indicator may indicate any attribute or characteristic that describes the input sample itself or its association with the outside.
Specifically, the operation mechanism may be a prediction model for providing a prediction result for a prediction problem, wherein the input samples are prediction samples of the prediction model, and the characteristics include a plurality of indexes related to the prediction problem.
Here, the prediction model may be a machine learning model trained based on a machine learning algorithm, may also be a statistical model constructed based on a statistical method, and may also be a model established based on other manners, which is not limited in the present invention. In this step, a pre-trained prediction model may be received from the outside, or a prediction model may be constructed from a prediction problem. The construction process of the prediction model is not described in detail.
The prediction model can be used in various application scenarios. For example, the model may be a model for predicting the probability of illness, a model for predicting body weight, a model for predicting machine loss, a model for predicting profit, or the like. And inputting the characteristics of the prediction samples into the prediction model to obtain a prediction result provided by the prediction model aiming at the prediction problem.
The features of the prediction samples of the prediction model include a plurality of indicators related to the prediction problem. For a plurality of indexes included in the features of the prediction sample, the indexes may be some attribute information whose values have respective characteristics (e.g., whether it can be changed, a value range of the change, a cost of the change, etc.) based on factors such as business meaning and/or common sense. As an example, the plurality of indicators included in the feature of the prediction sample may be divided into a static indicator (if present), a controllable indicator, and/or a response indicator according to prior knowledge (business meaning, common sense, etc.) of the external input. The response index is an index set of which the value is difficult to adjust but can change according to the change of the controllable index.
Taking the prediction model as a disease risk prediction model as an example, the characteristics of the prediction sample may include a plurality of indicators related to disease risk prediction, such as indicators that may include, but are not limited to, the user's age, sex, height, weight, one or more lifestyle-related characteristics (such as work and rest time, diet, exercise, etc.), blood pressure, blood sugar, uric acid, blood lipid, etc. According to the priori knowledge (related medical knowledge and common sense), the indexes of age, sex and height, which cannot be changed manually, belong to static indexes, the indexes of weight, life style related characteristics, blood pressure and blood sugar, which can be adjusted, can be classified into controllable indexes, the values of uric acid and blood fat, which are difficult to adjust for users, can change along with the changes of weight, life style related characteristics, blood pressure and blood sugar, and therefore uric acid and blood fat can be classified into response indexes.
It should be noted that, when the multiple indexes related to the prediction problem included in the features of the prediction sample are classified into static indexes, controllable indexes or response indexes based on the prior knowledge (business meaning, common sense, etc.), the classification to which the indexes belong may also be determined in combination with a specific scenario. For example, the value of the index "weight" may be adjusted, and thus may be classified into a controllable index, but the "weight" may also be considered to change with the change of the index such as "diet" or "exercise", and thus may also be classified into a response index, specifically, into a controllable index or into a response index, and may be determined according to the actual situation.
In step S120, a mapping relationship between a controllable index and a response index among the plurality of indexes is determined.
When the mapping relation between the controllable indexes and the response indexes is determined, the mapping relation between the controllable indexes and the response indexes can be determined without considering the static indexes, and the mapping relation between the static indexes and the mapping relation between the controllable indexes and the response indexes can also be determined by considering the static indexes. For example, the set of indices included in the features of the predicted sample is X, and the static index is represented as XsThe controllable index is represented as xcThe response index is denoted as xd，X＝(xs,xc,xd) Then x can be constructedcTo xdCan also construct { x }c,xsTo xdTo (3) is performed. To construct { xc,xsTo xdFor example, assume Δ xcFor a change value of a controllable indicator satisfying a constraint condition, the changed response indicator is given consideration of the influence of the initial response indicator
Figure BDA0001908903340000101
Changed response indicators without considering the effect of the initial response indicator
Figure BDA0001908903340000102
f ═ E (x) are controllable index and static indexMapping to response indicators.
In the structure xcTo xdIn the mapping of (2), x can be constructedc+Δxc→xd+ΔxdCan also construct Δ xc→ΔxdIs constructed accordingly { x }c,xsTo xdIn the case of mapping of (2), a { x } can be constructedc+Δxc,xs}→xd+ΔxdCan also construct { Δ x }c，xs}→ΔxdDynamic mapping of (2). It should be noted that in either the static mapping or the dynamic mapping, the influence of the initial response index may or may not be considered according to the actual situation.
The construction mode of the mapping can include, but is not limited to, a machine learning model, a statistical model, and a human-made rule, and the specific construction mode can be set according to actual requirements. For example, in the case where the mapping relationship between the controllable index and the response index can be determined by a priori knowledge, a rule capable of characterizing the mapping relationship between the controllable index and the response index may be artificially made.
As an example, the mapping relationship between the controllable index and the response index may be determined by any one of the following two ways: in the first mode, a model for predicting the value of the response index based on the value of the controllable index may be constructed according to a plurality of indexes included in the characteristics of the sample in the sample set to determine the mapping relationship between the controllable index and the response index, or a model for predicting the value of the response index based on the value of the static index and the value of the controllable index may be constructed according to a plurality of indexes included in the characteristics of the sample in the sample set to determine the mapping relationship between the controllable index and the response index. The constructed model can be a statistical model or a machine learning model, and the sample set can be a training sample set used in training a prediction model or a sample set obtained based on other modes; and determining a mapping relation between a controllable index and a response index in the multiple indexes according to prior knowledge from a prediction problem aimed at by the prediction model, for example, determining the mapping relation between the controllable index and the response index by constructing a statistical model or making a rule according to the prior knowledge.
It should be noted that the mapping relationship between the controllable indicator and the response indicator according to the present invention may include various relationships such as one-to-one, many-to-one, one-to-many, many-to-many, and the like. When the mapping relation between the controllable indexes and the response indexes is determined by constructing a model, the traditional statistical model is generally one-to-one or many-to-one, but the mapping relation is free in a new machine learning model, such as a deep neural network, the number of input and output is arbitrary and can be automatically adjusted by the model. The present invention is not limited with respect to the specific form of the mapping relationship.
In step S130, based on the mapping relationship, the controllable index of the input sample is adjusted, so that an output result obtained by the operation mechanism performing operation on the adjusted input sample meets a predetermined requirement.
As an example, the indexes included in the input samples (e.g., prediction samples) can be divided into static indexes (if any), which are a set of indexes whose values are unchangeable, and controllable indexes, which are a set of indexes that are difficult to adjust themselves but can change according to the change of the controllable indexes. After the value of the controllable index changes, the value of the response index can be determined based on the mapping relation. Therefore, when it is necessary to provide a proper suggestion for the index included in the input sample so that the change of the index will cause the expected operation output, only the controllable index may be adjusted so that the output result obtained after inputting the adjusted input sample (composed of the new controllable index, the new response index and the original static index (if any)) into the operation mechanism satisfies the predetermined requirement.
It should be noted that there may be some indexes having linkage relationship in the controllable indexes, for example, "weight" and "blood lipid" and "blood glucose" have linkage relationship, and "blood lipid" and "blood glucose" should generally increase when the "weight" increases, and if the blood lipid and blood glucose are not changed or the blood lipid and blood glucose are decreased when the weight is increased, the actual situation is obviously violated. Thus. In order to enable the adjustment of the controllable indexes to be more accurate, the controllable indexes can be grouped so as to divide the indexes with the linkage relation in the controllable indexes into the same group, and when the controllable indexes are adjusted, the controllable indexes can be adjusted in the value range of the controllable indexes by taking the group as a unit based on the linkage relation among the indexes in the same group. The adjustment of the controllable indexes in the unit of a group means that when any one of the indexes in the same group is adjusted, other indexes in the group should be adjusted at the same time. When the value of one of the indexes is adjusted, the adjustment direction of the values of the other indexes in the group can be determined according to the actual linkage relationship, for example, when the linkage relationship between the indexes a and B in the same group is the same elevation and the same descent, and when the value of the index a is increased, the value of the index B should be increased.
The linkage relationship mainly comprises a correlation relationship and/or a causal relationship. Indexes having causal relation among the controllable indexes may be classified into the same group, and/or indexes having correlation among the controllable indexes may be classified into the same group.
The indexes having correlation among the controllable indexes may be determined by calculation to divide the indexes having correlation among the controllable indexes into the same group. As an example, the correlation relationship may be understood as a relationship in which values of two indexes have a fixed variation relationship, such as a positive correlation variation relationship (i.e., the values are the same as rising and falling), or a negative correlation variation relationship (i.e., the values are taken as rising and falling). For example, the correlation between the indexes in the controllable indexes can be calculated by means of independence check. The calculation method of the independence check may include, but is not limited to, a pearson correlation coefficient, a spearman correlation coefficient, a kender harmony coefficient, a t-test, a chi-square test, and the like. When the correlation between the indexes is calculated, the calculation can be performed based on the values of the indexes, or the calculation can be performed according to the change values of the indexes, that is, the correlation between the indexes in the controllable indexes can be calculated in a static mode and a dynamic mode, so that the static correlation and the dynamic correlation between the indexes are obtained. The detailed description of the calculation process is omitted here.
The causal relationship mainly means that logically, the change of the value of one index can cause the change of the value of another index (or other indexes). The determination of the causal relationship is closer to logical reasoning, the change of the causal relationship is generally complex, and the causal relationship is difficult to determine through a calculation mode, so that the causal relationship index in the controllable indexes can be determined according to external input, and the causal relationship indexes in the controllable indexes can be divided into the same group. For example, the indexes having causal relationship among the controllable indexes can be determined by artificial induction according to a priori knowledge.
As an example, the indexes with strong correlation (e.g., the correlation degree obtained by calculation is greater than a predetermined threshold) in the controllable indexes can be classified into the same group by way of calculation (e.g., independence check), and since the causal relationship has no particularly good calculation means, the indexes with causal relationship in the control indexes can be identified by way of external input (e.g., artificial induction), and the indexes with causal relationship can be classified into the same group to make up for the deficiency of the calculation method.
Taking an operation mechanism as an example of the prediction model, when the adjustment result of the controllable index is solved, the controllable index in the characteristics of the prediction sample can be adjusted by taking the prediction result obtained by predicting the adjusted prediction sample by the prediction model to meet the preset requirement and reducing the adjustment cost of the controllable index as much as possible as the target. The adjustment cost of the controllable index may be predetermined, for example, the adjustment cost of each index in the controllable index may be represented by a specific expression.
For example, after dividing a plurality of indexes included in the features of the prediction sample into a static index, a controllable index, and a response index, and obtaining a mapping relationship between the static index and the controllable index to the response index, based on a prediction model for providing a prediction result for a prediction problem, on the premise that a constraint related to the controllable index is satisfied, the problem of solving an adjustment result of the controllable index may be converted into the following form:
Figure BDA0001908903340000131
the method includes the steps that delta X is the characteristic change of a prediction sample, namely the change values of a plurality of indexes included in the characteristic, C is the set of all possible values of the characteristic change delta X meeting constraint conditions, and the constraint conditions include the value range of the change of controllable indexes and response indexes and the grouping constraint of the controllable indexes determined based on the linkage relation.
In consideration of the cost (Δ X) of the controllable index variation, the above equation can be finally converted into:
Figure BDA0001908903340000132
wherein, w1And w2Is a constant value used for balancing the final control effect and the control cost, w1And w2The value of (c) may be set according to a specific scenario. Min in the above formula is only for indicating the direction of optimization and is not necessarily the smallest, e.g. w1For positive timing to indicate the optimal direction is minimal, w1Negative means that the optimization direction is large. The above equation can be solved by various calculation methods. The specific solving method may be determined according to the final solving difficulty, such as linear programming, gradient descent, and simplex method, and if the function is complex, a heuristic algorithm (such as genetic algorithm, simulated annealing, ant colony algorithm, etc.) may be adopted, and details about the solving process of the equation are not repeated here.
Here, it should be noted that the above expression for solving the problem is merely an example, and for example, in the case of considering the influence of the initial response index, the above-described changed response index needs to be replaced with the one described above
Figure BDA0001908903340000133
In addition, according to different requirements met by actual needs, the above formula can also have a certain deformation, and the above formula not only can be used as a solving formula for representing and optimizing to a limit value, but also can be deformed so as to convert the above formula into a solving formula for representing and optimizing to a specific value, for example, into a solving formula for calculating the minimum value of the difference between a real value and the specific value. Details regarding specific variations of the above equation are not repeated.
In step S140, the adjustment result of the controllable index is provided for index intervention.
Therefore, only the adjustment result of the controllable index can be provided, and index intervention can be performed on relevant aspects according to the adjustment result of the controllable index, so that expected operation output can be facilitated. Compared with the index control scheme which can facilitate expected output by directly using an optimization method without distinguishing the indexes of the input samples, the adjustment result based on the controllable indexes provided by the invention is more reasonable, and the method can be realized and is easy to realize when the index intervention is carried out based on the adjustment result of the controllable indexes provided by the invention.
For example, for a model for predicting the prevalence probability, under the condition that the prevalence probability of a user is relatively high as predicted by the model according to the existing indexes of the user, the method and the device can be used for providing a suggestion for adjusting the controllable indexes in the indexes of the user, and the prevalence probability can be reduced as the user adjusts the controllable indexes according to the provided suggestion. For another example, for a model for predicting machine loss, suggestions for adjusting the controllable indexes in the existing indexes can be given by using the method, and related personnel can adjust the controllable indexes according to the given suggestions, so that the machine loss can be reduced, and the service life of the machine can be prolonged.
Fig. 2 shows a flowchart of a method applied to intervene on a user's physical indicators to reduce the risk of the user suffering from a disease in an exemplary embodiment of the present invention. The method shown in fig. 2 can be implemented by a computer program or by a special device for performing the index intervention.
Referring to fig. 2, in step S210, a prediction model is acquired. The prediction model obtained here refers to an illness probability prediction model, and the illness probability prediction model is used for predicting the illness probability of the user according to the characteristics of the user, and optionally, the illness probability prediction model may be obtained by training a machine learning algorithm. The pre-trained morbidity probability prediction model can be obtained from the outside, or the morbidity probability prediction model can be trained on line, and details are not repeated in relation to the training process of the morbidity probability prediction model.
In step S220, the prevalence probability of the user is predicted from the characteristics of the user using the prevalence probability prediction model.
The characteristics of the user include a plurality of indexes related to the predicted prevalence probability, and the plurality of indexes mainly refer to physical indexes related to the user, such as indexes that may include but are not limited to age, sex, height, weight, one or more life style related characteristics (such as work and rest time, diet, exercise), blood pressure, blood sugar, uric acid, blood fat, and the like.
In step S230, it is determined whether the predicted prevalence probability of the user is higher than a preset threshold.
Under the condition that the predicted disease probability of the user is lower than a preset threshold value, the current index of the user can be considered to be normal, and intervention is not needed. Under the condition that the predicted disease probability of the user is higher than the preset threshold value, the current index of the user is considered to be abnormal, and a proper index adjusting scheme needs to be provided for the user, so that the user can adjust the index of the user according to the given adjusting scheme, and the disease probability can be reduced.
As shown in fig. 2, in the case that the predicted prevalence probability of the user is higher than the preset threshold, steps S240 and S250 may be performed to provide the user with an adjustment result.
In step S240, based on the mapping relationship between the static index and the controllable index to the response index in the plurality of indexes, the controllable index in the features of the user is adjusted, so that the prevalence probability predicted by the prediction model for the adjusted features of the user is lower than a predetermined threshold.
For the concepts and the dividing manners of the static indicator, the controllable indicator and the response indicator, reference may be made to the above description, and details are not repeated here. As an example, the static metrics may include at least one of: age, gender, height, occupation; the controllable indicator may comprise at least one of: body weight, one or more lifestyle-related characteristics (e.g., daily life, diet, exercise), blood pressure, blood glucose; the response indicator may include at least one of: uric acid and blood fat.
The mapping relationship from the static index and the controllable index to the response index may be received from the outside, or the mapping relationship from the static index and the controllable index to the response index among the plurality of indexes may be determined.
Taking the index set included by the characteristics of the prediction samples as X, and expressing the static index as XsThe controllable index is represented as xcThe response index is denoted as xd，X＝(xs,xc,xd) For example, one can construct { x }c,xsTo xdTo (3) is performed. For example, assume Δ xcFor a change value of a controllable indicator satisfying a constraint condition, the changed response indicator is given consideration of the influence of the initial response indicator
Figure BDA0001908903340000151
Changed response indicators without considering the effect of the initial response indicator
Figure BDA0001908903340000152
And f, E (x) is the mapping relation between the controllable indexes and the static indexes to the response indexes. Wherein, in the structure { xc,xsTo xdIn the case of mapping of (2), a { x } can be constructedc+Δxc,xs}→xd+ΔxdCan also construct { Δ x }c，xs}→ΔxdDynamic mapping of (2). It should be noted that in either the static mapping or the dynamic mapping, the influence of the initial response index may or may not be considered according to the actual situation.
The construction mode of the mapping can include, but is not limited to, a machine learning model, a statistical model, and a human-made rule, and the specific construction mode can be set according to actual requirements. For example, in the case where the mapping relationship between the controllable index and the response index can be determined by a priori knowledge, a rule capable of characterizing the mapping relationship between the controllable index and the response index may be artificially made.
As an example, the mapping relationship between the controllable index and the response index may be determined by any one of the following two ways: in the first mode, a model for predicting the value of the response index based on the value of the static index and the value of the controllable index can be constructed according to a plurality of indexes included in the characteristics of the sample in the sample set, so as to determine the mapping relationship between the controllable index and the response index. The constructed model can be a statistical model or a machine learning model, and the sample set can be a training sample set used in training a prediction model or a sample set obtained based on other modes; and determining the mapping relation between the static indexes and the controllable indexes in the multiple indexes to the response indexes according to the prior knowledge. For example, the mapping relationship between the static index and the controllable index to the response index can be determined by constructing a statistical model or making a rule according to the prior knowledge (such as the related knowledge and common knowledge in the medical field).
In order to enable the adjustment of the controllable indexes to be more in line with the real situation, the controllable indexes can be grouped so as to divide the indexes with the linkage relation in the controllable indexes into the same group, and when the controllable indexes are adjusted, the controllable indexes can be adjusted in a value range of the controllable indexes by taking the group as a unit based on the linkage relation among the indexes in the same group. The implementation manner of dividing the indexes having the linkage relationship in the controllable indexes into the same group may refer to the above related description, and is not described herein again.
As an example, when the adjustment result of the controllable index is solved, the controllable index in the features of the prediction sample may be adjusted with the goal that the prevalence probability predicted by the prevalence probability prediction model for the adjusted sample of the user is lower than a predetermined threshold and the adjustment cost of the controllable index is reduced as much as possible. The adjustment cost of the controllable index may be predetermined, for example, the adjustment cost of each index in the controllable index may be represented based on a determined expression.
In step S250, the adjustment result of the controllable index is provided for the user to perform index intervention.
When the user intervenes the indexes, only the weight, one or more characteristics related to the life style, the blood pressure, the blood sugar and other controllable indexes are intervened according to the adjustment result of the provided controllable indexes, so that the purpose of reducing the sicknesses can be achieved, and the indexes of uric acid, blood fat and the like which are difficult to adjust for the user do not need to be considered.
The method for performing index intervention of the present invention can also be implemented as an apparatus for performing index intervention. Fig. 3 shows a block diagram of an apparatus for performing index intervention according to an exemplary embodiment of the present invention. The functional units of the means for performing index intervention may be implemented, inter alia, by hardware, software or by a combination of hardware and software implementing the principles of the invention. It will be appreciated by those skilled in the art that the functional units described in fig. 3 may be combined or divided into sub-units to implement the principles of the invention described above. Thus, the description herein may support any possible combination, or division, or further definition of the functional units described herein.
Functional units that the apparatus for performing index intervention may have and operations that each functional unit may perform are briefly described below, and details related thereto may be referred to the above description, and are not repeated here.
Referring to fig. 3, the apparatus 300 for performing index intervention includes an obtaining unit 310, a determining unit 320, an adjusting unit 330, and a providing unit 340.
The obtaining unit 310 is configured to obtain an operation mechanism for providing an output result for an input sample, where the input sample includes a plurality of indexes. The obtaining unit 310 may receive the operation mechanism from the outside or may establish the operation mechanism by itself. For the operation mechanism (e.g., the prediction model), the index, etc., reference may be made to the above description for step S110, which is not repeated herein.
The determining unit 320 is configured to determine a mapping relationship between a controllable indicator and a response indicator among the plurality of indicators. The controllable index is an index set of which the value can be adjusted, and the response index is an index set of which the value is difficult to adjust but can change according to the change of the controllable index.
For the determination of the mapping relationship, see the above description for step S120, which is not repeated herein. As an example of the present invention, the determining unit 320 may construct a model for predicting the value of the response index based on the value of the controllable index according to a plurality of indexes included in the characteristics of the samples in the sample set to determine the mapping relationship between the controllable index and the response index, or the determining unit 320 may also construct a model for predicting the value of the response index based on the value of the static index and the value of the controllable index according to a plurality of indexes included in the characteristics of the samples in the sample set to determine the mapping relationship between the controllable index and the response index, where the static index is an index set whose value is unchangeable. In addition, the determining unit 320 may also determine a mapping relationship between the controllable indicator and the response indicator according to the a priori knowledge.
The adjusting unit 330 is configured to adjust the controllable index of the input sample based on the mapping relationship, so that an output result obtained by the operation performed by the operation mechanism on the adjusted input sample meets a predetermined requirement.
The process of adjusting the controllable index may refer to the description of step S130, and is not described herein again. As an example of the present invention, the adjusting unit 330 may adjust the controllable index in the features of the prediction samples with the goal that a prediction result obtained by predicting the adjusted prediction samples by the prediction model satisfies a predetermined requirement and the adjustment cost of the controllable index is reduced as much as possible.
Optionally, the apparatus 300 for performing index intervention may further include a grouping unit (not shown in the figure), where the grouping unit is configured to group the controllable indexes to divide the indexes having the linkage relationship in the controllable indexes into the same group, and the adjusting unit 330 may adjust the controllable indexes in a value range of the controllable indexes, based on the linkage relationship between the indexes in the same group, by taking the group as a unit. The implementation process of grouping the controllable indexes by the grouping unit may refer to the above related description, and is not described herein again.
The providing unit 340 is configured to provide an adjustment result of the controllable indicator for performing indicator intervention.
It should be understood that the specific implementation manner of the apparatus 300 for performing index intervention according to the exemplary embodiment of the present invention may be implemented with reference to the related specific implementation manner described in conjunction with fig. 1, and will not be described in detail herein.
Fig. 4 shows a block diagram of an apparatus for index intervention according to another exemplary embodiment of the present invention. The functional units of the means for performing index intervention may be implemented, inter alia, by hardware, software or by a combination of hardware and software implementing the principles of the invention. It will be appreciated by those skilled in the art that the functional units described in fig. 4 may be combined or divided into sub-units to implement the principles of the invention described above. Thus, the description herein may support any possible combination, or division, or further definition of the functional units described herein.
Functional units that the apparatus for performing index intervention may have and operations that each functional unit may perform are briefly described below, and details related thereto may be referred to the above description, and are not repeated here.
Referring to fig. 4, an apparatus 400 for index intervention includes an obtaining unit 410, a predicting unit 420, an adjusting unit 430, and a providing unit 440.
The obtaining unit 410 is used for obtaining a prevalence probability prediction model. Specifically, see the above description for step S210, which is not repeated herein.
The prediction unit 420 is configured to predict the prevalence probability of the user according to the characteristics of the user using the prevalence probability prediction model. The characteristics of the user include a plurality of indicators related to the predicted prevalence probability, such as indicators that may include, but are not limited to, age, gender, height, weight, lifestyle, blood pressure, blood glucose, uric acid, blood lipid, and the like.
The adjusting unit 430 is configured to, when it is predicted that the prevalence probability of the user is higher than a predetermined threshold, adjust a controllable index among the features of the user based on a mapping relationship between a static index and the controllable index to a response index among the multiple indexes, so that the prevalence probability predicted by the prevalence probability prediction model for the adjusted features of the user is lower than the predetermined threshold.
For the concepts and the dividing manners of the static indicator, the controllable indicator and the response indicator, reference may be made to the above description, and details are not repeated here. As an example, the static metrics may include at least one of: age, gender, height, occupation; the controllable indicator may comprise at least one of: body weight, lifestyle, blood pressure, blood glucose; the response indicator may include at least one of: uric acid and blood fat.
The process of adjusting the controllable index may refer to the description of step S240, and is not described herein again. As an example of the present invention, the adjusting unit 430 may adjust the controllable index in the features of the prediction sample with a goal that the prevalence probability predicted by the prevalence probability prediction model for the adjusted features of the user is lower than the predetermined threshold and an adjustment cost of the controllable index is reduced as much as possible.
The providing unit 440 is configured to provide the adjustment result of the controllable index for the user to perform index intervention.
Optionally, the apparatus 400 for performing index intervention may further comprise a determination unit (not shown in the figures). The determining unit is used for determining the mapping relation between the static indexes and the controllable indexes in the indexes to the response indexes. For the construction of the mapping, reference may be made to the above description, and further description is omitted here.
It should be understood that the specific implementation manner of the apparatus 400 for performing index intervention according to the exemplary embodiment of the present invention may be implemented with reference to the related specific implementation manner described in conjunction with fig. 2, and will not be described in detail herein.
Fig. 5 shows a schematic structural diagram of a computing device that can be used to implement the above method according to an exemplary embodiment of the present invention.
Referring to fig. 5, computing device 500 includes memory 510 and processor 520.
The processor 520 may be a multi-core processor or may include a plurality of processors. In some embodiments, processor 520 may include a general-purpose host processor and one or more special coprocessors such as a Graphics Processor (GPU), a Digital Signal Processor (DSP), or the like. In some embodiments, processor 520 may be implemented using custom circuitry, such as an Application Specific Integrated Circuit (ASIC) or a Field Programmable Gate Array (FPGA).
The memory 510 may include various types of storage units, such as system memory, Read Only Memory (ROM), and permanent storage. Wherein the ROM may store static data or instructions for the processor 520 or other modules of the computer. The persistent storage device may be a read-write storage device. The persistent storage may be a non-volatile storage device that does not lose stored instructions and data even after the computer is powered off. In some embodiments, the persistent storage device employs a mass storage device (e.g., magnetic or optical disk, flash memory) as the persistent storage device. In other embodiments, the permanent storage may be a removable storage device (e.g., floppy disk, optical drive). The system memory may be a read-write memory device or a volatile read-write memory device, such as a dynamic random access memory. The system memory may store instructions and data that some or all of the processors require at runtime. Further, the memory 510 may include any combination of computer-readable storage media, including various types of semiconductor memory chips (DRAM, SRAM, SDRAM, flash memory, programmable read-only memory), magnetic and/or optical disks, may also be employed. In some embodiments, memory 510 may include a removable storage device that is readable and/or writable, such as a Compact Disc (CD), a digital versatile disc read only (e.g., DVD-ROM, dual layer DVD-ROM), a Blu-ray disc read only, an ultra-dense disc, a flash memory card (e.g., SD card, min SD card, Micro-SD card, etc.), a magnetic floppy disk, or the like. Computer-readable storage media do not contain carrier waves or transitory electronic signals transmitted by wireless or wired means.
The memory 510 has stored thereon executable code that, when executed by the processor 520, may cause the processor 520 to perform the above-mentioned methods for performing index intervention.
The method, apparatus and computing device for performing index intervention according to the present invention have been described in detail above with reference to the accompanying drawings.
Furthermore, the method according to the invention may also be implemented as a computer program or computer program product comprising computer program code instructions for carrying out the above-mentioned steps defined in the above-mentioned method of the invention.
Alternatively, the invention may also be embodied as a non-transitory machine-readable storage medium (or computer-readable storage medium, or machine-readable storage medium) having stored thereon executable code (or a computer program, or computer instruction code) which, when executed by a processor of an electronic device (or computing device, server, etc.), causes the processor to perform the steps of the above-described method according to the invention.
Those of skill would further appreciate that the various illustrative logical blocks, modules, circuits, and algorithm steps described in connection with the disclosure herein may be implemented as electronic hardware, computer software, or combinations of both.
The flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems and methods according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
Having described embodiments of the present invention, the foregoing description is intended to be exemplary, not exhaustive, and not limited to the embodiments disclosed. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen in order to best explain the principles of the embodiments, the practical application, or improvements made to the technology in the marketplace, or to enable others of ordinary skill in the art to understand the embodiments disclosed herein.

Claims (10)

1. A method for performing index intervention, comprising:
obtaining an arithmetic mechanism for providing an output result for an input sample, wherein the input sample comprises a plurality of metrics;
determining a mapping relation between a controllable index and a response index in the plurality of indexes, wherein the controllable index is an index set of which the value can be adjusted, and the response index is an index set of which the value is difficult to adjust but can change according to the change of the controllable index;
based on the mapping relation, the controllable indexes of the input samples are adjusted, so that the output result obtained by the operation of the operation mechanism aiming at the adjusted input samples meets the preset requirement; and
and providing the adjustment result of the controllable index for index intervention.
2. The method of claim 1, wherein the operational mechanism comprises at least one of: machine learning model, rule set, function operation.
3. The method of claim 1, wherein the computing mechanism is a predictive model that provides a prediction result for a predictive problem, and wherein the input samples are predictive samples of the predictive model, the characteristics of which include a plurality of metrics related to the predictive problem.
4. The method of claim 3, wherein determining a mapping between a controllable metric and a responsive metric among the plurality of metrics comprises:
according to a plurality of indexes included in the characteristics of the samples in the sample set, constructing a model for predicting the value of the response index based on the value of the controllable index so as to determine the mapping relation between the controllable index and the response index; or
And according to a plurality of indexes included in the characteristics of the samples in the sample set, constructing a model for predicting the value of the response index based on the value of a static index and the value of the controllable index so as to determine the mapping relation between the controllable index and the response index, wherein the static index is an index set of which the value is unchangeable.
5. The method of claim 3, wherein determining a mapping between a controllable metric and a responsive metric among the plurality of metrics comprises:
and determining a mapping relation between a controllable index and a response index in the plurality of indexes according to the prior knowledge.
6. A method for performing index intervention, comprising:
acquiring a disease probability prediction model;
predicting the prevalence probability of a user according to the characteristics of the user by using the prevalence probability prediction model, wherein the characteristics of the user comprise a plurality of indexes related to the predicted prevalence probability;
under the condition that the prevalence probability of the user is predicted to be higher than a preset threshold, adjusting a controllable index in the characteristics of the user based on a mapping relation between a static index and a controllable index to a response index in the multiple indexes, so that the prevalence probability predicted by the prevalence probability prediction model for the adjusted characteristics of the user is lower than the preset threshold, wherein the static index is an index set of which the value is unchangeable, the controllable index is an index set of which the value is adjustable, and the response index is an index set of which the value is difficult to adjust and can change according to the change of the controllable index; and
and providing an adjustment result of the controllable index so as to facilitate index intervention of a user.
7. An apparatus for performing index intervention, comprising:
an obtaining unit configured to obtain an operation mechanism for providing an output result for an input sample, wherein the input sample includes a plurality of indexes;
the device comprises a determining unit, a calculating unit and a processing unit, wherein the determining unit is used for determining a mapping relation between a controllable index and a response index in the plurality of indexes, the controllable index is an index set of which the value can be adjusted, and the response index is an index set of which the value is difficult to adjust but can change according to the change of the controllable index;
the adjusting unit is used for adjusting the controllable indexes of the input samples based on the mapping relation so that the output result obtained by the operation of the operation mechanism aiming at the adjusted input samples meets the preset requirement; and
and the providing unit is used for providing the adjustment result of the controllable index for index intervention.
8. An apparatus for performing index intervention, comprising:
an acquisition unit for acquiring a disease probability prediction model;
a prediction unit for predicting the prevalence probability of a user according to a feature of the user using the prevalence probability prediction model, wherein the feature of the user includes a plurality of indexes related to the predicted prevalence probability;
an adjusting unit, configured to, when it is predicted that the prevalence probability of the user is higher than a predetermined threshold, adjust a controllable index among the features of the user based on a mapping relationship between a static index and a controllable index to a response index among the multiple indexes, so that the prevalence probability predicted by the prevalence probability prediction model for the adjusted features of the user is lower than the predetermined threshold, where the static index is an index set whose value is unchangeable, the controllable index is an index set whose value is adjustable, and the response index is an index set whose value is difficult to adjust but is changeable according to a change of the controllable index; and
and the providing unit is used for providing the adjustment result of the controllable index so that the user can perform index intervention.
9. A computing device, comprising:
a processor; and
a memory having executable code stored thereon, which when executed by the processor, causes the processor to perform the method of any of claims 1 to 5.
10. A non-transitory machine-readable storage medium having stored thereon executable code, which when executed by a processor of an electronic device, causes the processor to perform the method of any of claims 1-5.
CN201811544043.0A 2018-12-17 2018-12-17 Method, apparatus, device and medium for index intervention Pending CN111326254A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201811544043.0A CN111326254A (en) 2018-12-17 2018-12-17 Method, apparatus, device and medium for index intervention

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201811544043.0A CN111326254A (en) 2018-12-17 2018-12-17 Method, apparatus, device and medium for index intervention

Publications (1)

Publication Number Publication Date
CN111326254A true CN111326254A (en) 2020-06-23

Family

ID=71170683

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201811544043.0A Pending CN111326254A (en) 2018-12-17 2018-12-17 Method, apparatus, device and medium for index intervention

Country Status (1)

Country Link
CN (1) CN111326254A (en)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104346372A (en) * 2013-07-31 2015-02-11 国际商业机器公司 Method and device for evaluating prediction models
US20160012202A1 (en) * 2014-07-14 2016-01-14 International Business Machines Corporation Predicting the risks of multiple healthcare-related outcomes via joint comorbidity discovery
CN106663224A (en) * 2014-06-30 2017-05-10 亚马逊科技公司 Interactive interfaces for machine learning model evaluations
CN107491638A (en) * 2017-07-28 2017-12-19 深圳和而泰智能控制股份有限公司 A kind of ICU user's prognosis method and terminal device based on deep learning model
CN108573309A (en) * 2018-01-04 2018-09-25 中国科学院计算技术研究所 A kind of self-adapting regulation method and system of machine learning algorithm

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104346372A (en) * 2013-07-31 2015-02-11 国际商业机器公司 Method and device for evaluating prediction models
CN106663224A (en) * 2014-06-30 2017-05-10 亚马逊科技公司 Interactive interfaces for machine learning model evaluations
US20160012202A1 (en) * 2014-07-14 2016-01-14 International Business Machines Corporation Predicting the risks of multiple healthcare-related outcomes via joint comorbidity discovery
CN107491638A (en) * 2017-07-28 2017-12-19 深圳和而泰智能控制股份有限公司 A kind of ICU user's prognosis method and terminal device based on deep learning model
CN108573309A (en) * 2018-01-04 2018-09-25 中国科学院计算技术研究所 A kind of self-adapting regulation method and system of machine learning algorithm

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
肖辉;郝元涛;徐晓;朱晓宇;: "基于随机森林算法和Logistic回归模型的糖尿病风险因素研究", 中国数字医学, no. 01 *

Similar Documents

Publication Publication Date Title
US20180260925A1 (en) Identifying group and individual-level risk factors via risk-driven patient stratification
Johnson et al. A new severity of illness scale using a subset of acute physiology and chronic health evaluation data elements shows comparable predictive accuracy
Tong et al. Comparison of predictive modeling approaches for 30-day all-cause non-elective readmission risk
US9355067B1 (en) Distribution of parameter calculation for iterative optimization methods
KR20190021471A (en) Method, Apparatus and Program for Predicting Prognosis of Gastric Cancer Using Artificial Neural Network
US20120271612A1 (en) Predictive modeling
US20180107574A1 (en) Detecting anomalous sensors
CN112633601A (en) Method, device, equipment and computer medium for predicting disease event occurrence probability
CN110969200B (en) Image target detection model training method and device based on consistency negative sample
KR20160043777A (en) Method and apparatus for disease occurrence prediction
EP3797423A1 (en) System and method for integrating genotypic information and phenotypic measurements for precision health assessments
US20220207304A1 (en) Robustness setting device, robustness setting method, storage medium storing robustness setting program, robustness evaluation device, robustness evaluation method, storage medium storing robustness evaluation program, computation device, and storage medium storing program
US11475302B2 (en) Multilayer perceptron based network to identify baseline illness risk
CN110598837A (en) Artificial neural network adjusting method and device
WO2021151273A1 (en) Disease prediction method and apparatus, electronic device, and storage medium
CN111326254A (en) Method, apparatus, device and medium for index intervention
US11651289B2 (en) System to identify and explore relevant predictive analytics tasks of clinical value and calibrate predictive model outputs to a prescribed minimum level of predictive accuracy
US20230186092A1 (en) Learning device, learning method, computer program product, and learning system
CN116757783A (en) Product recommendation method and device
CN114494263B (en) Medical image lesion detection method, system and equipment integrating clinical information
KR102303272B1 (en) System for predictting a direction of prostate cancer risk
Cox Jr What is an exposure-response curve?
CN110322055B (en) Method and system for improving grading stability of data risk model
CN113408692A (en) Network structure searching method, device, equipment and storage medium
CN109752569A (en) The system and method for condition are estimated from sensing data using random forest classification

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination