CN110648764A

CN110648764A - Method and device for obtaining clinical data prediction model, readable medium and electronic equipment

Info

Publication number: CN110648764A
Application number: CN201910770436.1A
Authority: CN
Inventors: 郎超; 刘水清; 温馨; 梁玮; 李潇
Original assignee: Nanjing Yiyi Yunda Data Technology Co Ltd; Nanjing Medical Duyun Medical Technology Co Ltd
Current assignee: Nanjing Yiyi Yunda Data Technology Co Ltd; Nanjing Medical Duyun Medical Technology Co Ltd
Priority date: 2019-08-20
Filing date: 2019-08-20
Publication date: 2020-01-03

Abstract

The invention discloses a method, a device, a readable medium and electronic equipment for acquiring a clinical data prediction model, wherein the method comprises the following steps: carrying out format processing on the sample data to obtain a sample set; performing data training through the sample set to establish a plurality of prediction models; testing the prediction models by using specific types of test data to obtain test indexes of the prediction models for the specific types; determining the prediction model with the highest test index as a target prediction model corresponding to the specific type; thereby determining which predictive model is relatively more accurate in a particular scenario; and then, the target prediction model is selected to complete prediction under a specific scene, and the accuracy of a prediction result is ensured.

Description

Method and device for obtaining clinical data prediction model, readable medium and electronic equipment

Technical Field

The invention relates to the technical field of computers, in particular to a method, a device, a readable medium and electronic equipment for acquiring a clinical data prediction model.

Background

In medicine, the importance of clinical data is self evident. The majority of patients' situations may be directly or indirectly manifested by clinical data. Some clinical data have intuitive relevance to the patient's condition, so that a doctor can make a direct diagnosis with the help of medical data. In other cases, a combination of clinical data may implicitly indicate certain conditions or potential health risks to the patient. For this case, it is difficult to find by manual data analysis.

The prediction model is established based on the current advanced technologies such as artificial intelligence, machine learning, big data analysis and the like, and clinical data are analyzed, so that diseases or potential risks can be discovered earlier than before, and treatment and rehabilitation are facilitated. Therefore, the application of the prediction model in the medical field has very important medical value.

However, the type of clinical data, the nature of the disease, and the type of outcome predicted are all diverse. Meanwhile, in the fields of artificial intelligence, machine learning and the like, various basic algorithms exist, and the performance and the advantages and the disadvantages of the basic algorithms are different. It is difficult to have a predictive model that can achieve accurate predictions in a variety of situations. In practical applications, it is very difficult to determine which prediction model should be used to predict specific situations.

Disclosure of Invention

The invention provides a method, a device, a readable medium and electronic equipment for obtaining a clinical data prediction model, wherein a plurality of prediction models are obtained based on different machine learning algorithm training, and each model is evaluated through testing.

In a first aspect, the present invention provides a method of obtaining a predictive model of clinical data, comprising:

carrying out format processing on the sample data to obtain a sample set; performing data training through the sample set to establish a plurality of prediction models; testing the prediction models by using specific types of test data to obtain test indexes of the prediction models for the specific types; and determining the prediction model with the highest test index as the target prediction model corresponding to the specific type.

Preferably, the training of data through the sample set to establish a plurality of predictive models comprises:

and respectively carrying out data training through the sample set based on a plurality of machine learning algorithms to establish a prediction model corresponding to each machine learning algorithm.

Preferably, the sample data includes known feature data, and the training of data through the sample set to establish a plurality of predictive models includes:

and performing supervised learning training through the sample set to obtain a functional relation between the known characteristic data and the data characteristics, and establishing the prediction model through the functional relation.

Preferably, the training of data through the sample set to establish a plurality of predictive models further comprises:

substituting the known characteristic data into the prediction model to obtain the fitting degree of the prediction model;

and when the fitting degree is lower than a preset fitting degree standard, correcting the function relation through the supervised learning training.

Preferably, the sample data includes known feature data and unknown feature data, and the training data through the sample set to build a plurality of prediction models includes:

and performing semi-supervised learning training through the sample set to obtain a functional relation between the known characteristic data and the data characteristics, and establishing the prediction model through the functional relation.

Preferably, the performing format processing on the sample data to obtain a sample set includes:

and according to a preset format template, carrying out format conversion processing on the sample data to obtain the sample set.

Preferably, the method further comprises the following steps:

preprocessing the sample data; the preprocessing includes data supplement, data modification and/or data dimension reduction.

In a second aspect, the present invention provides an apparatus for obtaining a predictive model of clinical data, comprising:

the format processing module is used for carrying out format processing on the sample data to obtain a sample set;

the model training module is used for carrying out data training through the sample set so as to establish a plurality of prediction models;

the model testing module is used for testing the prediction models by using specific types of test data to obtain test indexes of the prediction models for the specific types;

and the model determining module is used for determining the prediction model with the highest test index as the target prediction model corresponding to the specific type.

In a third aspect, the invention provides a readable medium comprising executable instructions, which when executed by a processor of an electronic device, perform the method according to any of the first aspect.

In a fourth aspect, the present invention provides an electronic device, comprising a processor and a memory storing execution instructions, wherein when the processor executes the execution instructions stored in the memory, the processor performs the method according to any one of the first aspect.

The invention provides a method, a device, a readable medium and electronic equipment for acquiring a clinical data prediction model, which can meet the requirements under various different scenes by establishing a prediction model based on a plurality of machine learning algorithms and predict different specific contents aiming at different data inputs; testing the prediction model by using specific types of test data to obtain a test index, so as to determine which prediction model is relatively more accurate in a specific scene; and then, the target prediction model is selected to complete prediction under a specific scene, and the accuracy of a prediction result is ensured.

Further effects of the above-mentioned unconventional preferred modes will be described below in conjunction with specific embodiments.

Drawings

In order to more clearly illustrate the embodiments or the prior art solutions of the present invention, the drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments described in the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without inventive labor.

FIG. 1 is a flow chart illustrating a method for obtaining a predictive model of clinical data according to an embodiment of the invention;

FIG. 2 is a schematic flow chart of another method for obtaining a predictive model of clinical data according to an embodiment of the invention;

FIG. 3 is a schematic structural diagram of an apparatus for obtaining a clinical data prediction model according to an embodiment of the present invention;

fig. 4 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention more apparent, the technical solutions of the present invention will be described in detail and completely with reference to the following embodiments and accompanying drawings. It is to be understood that the described embodiments are merely exemplary of the invention, and not restrictive of the full scope of the invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

As is known from the foregoing, the types of clinical data, the nature of the disease, and the types of predicted outcomes are all diverse. Meanwhile, in the fields of artificial intelligence, machine learning and the like, various basic algorithms exist, and the performance and the advantages and the disadvantages of the basic algorithms are different. It is difficult to have a predictive model that can achieve accurate predictions in a variety of situations.

For example, there are cases where a binary result (yes or no) is output using the predictive model, and there are cases where a continuous or discrete value (e.g., risk of illness) is output using the predictive model. It is clear that models based on different machine learning algorithms behave differently for different clinical data inputs and different predicted content in different situations. However, in practical applications, it is very difficult to determine which prediction model should be used to predict which specific situation. In the invention, various prediction models are evaluated in a quantitative mode so as to determine which prediction model has relatively higher accuracy under a specific condition.

Referring to fig. 1, a method for obtaining a clinical data prediction model according to an embodiment of the present invention is shown. The method in this embodiment includes the following steps:

step 101, performing format processing on the sample data to obtain a sample set.

And 102, performing data training through the sample set to establish a plurality of prediction models.

Step 103, testing the prediction models by using test data of a specific type to obtain test indexes of the prediction models for the specific type.

And step 104, determining the prediction model with the highest test index as the target prediction model corresponding to the specific type.

In this embodiment, first, format processing is performed on sample data, specifically, format conversion processing is performed on the sample data according to a preset format template, so as to obtain the sample set. That is, in order to perform subsequent training based on various machine learning algorithms, the uniformity of the data in the sample set must be ensured by format processing. In addition, in order to ensure the quality of the sample set, the data can be preprocessed. The preprocessing comprises processing operations such as data supplement, data correction, data dimension reduction and the like so as to make up for defects in the originally acquired sample data. In the field of machine learning, the above process may also be referred to as feature engineering.

The sample set obtained by format processing may be a two-dimensional table. One dimension of the two-dimensional table represents the patient and the other dimension represents the specific values of the clinical data. An exemplary table of a sample set two-dimensional table is shown in the following table.

	Body weight	Blood pressure	Blood sugar	Heart rate
					Patient 1	a1	b1	c1	d1
Patient 2	a2	b2	c2	d2
					Patient 3	a3	b3	c3	d3
Patient 4	a4	b4	c4	d4

In this embodiment, the data training is performed through the sample set, which may be based on a plurality of machine learning algorithms, and the data training is performed through the sample set, so as to establish a prediction model corresponding to each machine learning algorithm. Due to the fact that the performance of prediction models built based on different machine learning algorithms is different, the prediction models are applicable to different scenes. Therefore, a plurality of prediction models are established in the embodiment, and the requirements under various different scenes can be met theoretically. I.e. different specific content can be predicted for different data inputs.

However, to determine which prediction model is more suitable for which specific scenario, the steps of testing and determining by using the prediction model in this embodiment are also required.

In this embodiment, after the prediction models are established, the prediction models are further tested by using test data of a specific type, so as to obtain test indexes of the prediction models for the specific type. The specific type is determined, i.e. the type of one of the input clinical data, the type of the output prediction result, and the specific content of the prediction. In other words, the usage scenario of a fixed prediction model is predetermined.

In the test process, it can be understood that in the above scenario, the accuracy of the test model is analyzed by predicting the test data of the known actual result and comparing whether the test result is consistent with the actual result, so as to determine the test index of the test model. This makes it possible to evaluate which prediction model is relatively more accurate in the above-described scenario. And determining the prediction model with the highest test index as the target prediction model corresponding to the specific type. The number of the target prediction models may be one or more according to the requirement, and is not limited herein.

In addition, after the target prediction model corresponding to the specific type is determined, in an actual prediction process, in a scene corresponding to the specific type, the data to be predicted of the specific type is input to the target prediction model corresponding to the specific type, and a prediction result is obtained. Thereby completing the actual prediction process. Since the target prediction model is the relatively most accurate prediction model for the specific type, the output prediction result can better meet the requirement of medical prediction.

According to the technical scheme, the beneficial effects of the embodiment are as follows: the method comprises the steps that a prediction model based on a plurality of machine learning algorithms is established, so that the requirements under various different scenes can be met, and different specific contents can be predicted according to different data inputs; testing the prediction model by using specific types of test data to obtain a test index, so as to determine which prediction model is relatively more accurate in a specific scene; and then, the target prediction model is selected to complete prediction under a specific scene, and the accuracy of a prediction result is ensured.

Fig. 1 shows only a basic embodiment of the method of the present invention, and based on this, certain optimization and expansion can be performed, and other preferred embodiments of the method can also be obtained.

FIG. 2 shows another embodiment of the method for clinical data prediction model acquisition according to the present invention. On the basis of the previous embodiment, the embodiment performs more detailed description and a certain degree of optimization on the training modeling process. The method in this embodiment includes the following steps:

step 201, performing format processing on the sample data to obtain a sample set.

In this embodiment, the sample data includes known feature data, and preferably may further include part of unknown feature data.

Known characteristic data is clinical data that can determine whether it corresponds to a particular disease characteristic. In this embodiment, the predictive model is trained to predict certain disease features from clinical data. For example, to predict whether a patient is "at risk for hypertension". Therefore, the training process is a process for finding the implicit relationship between clinical data and "hypertension". Then there is a need in the known signature data to include data that partially matches the signature of the disease (e.g., clinical data from hypertensive patients) and data that partially does not match the signature of the disease (e.g., clinical data from non-hypertensive patients). Thereby learning specific associations between clinical data and data characteristics.

While unknown characteristic data, i.e., clinical data that cannot be determined whether it corresponds to a particular disease characteristic.

Step 202, performing supervised learning training through the known characteristic data to obtain a functional relation between the known characteristic data and the data characteristic, and establishing the prediction model through the functional relation.

In some cases, the predictive model may be built by supervised learning training. The supervised learning training, namely training the known characteristic data through artificial intelligence calculation, finds out how the correlation exists between the clinical data and the known data characteristics.

Assume that a piece of known feature data is denoted as (x, y). Where x represents a numerical value of data, and x ═ x (x) can be used specifically₁,x₂…x_n) To express x₁～x_nN clinical indexes are the numerical values of the clinical indexes. y represents a data characteristic, and in the present embodiment, it can be considered that when y is 1, the data characteristic is "hypertensive patient", and when y is 0, the data characteristic is "non-hypertensive patient".

Through the supervised learning training, the functional relationship y of the known characteristic data and the data characteristic is obtained as f (x). I.e. to obtain the prediction model. For clinical data with unknown data characteristics, the data characteristics y can be obtained only by substituting the numerical value of the clinical data with x into the model.

And 203, performing semi-supervised learning training through the known characteristic data and the unknown characteristic data to obtain a functional relation between the known characteristic data and the data characteristics, and establishing the prediction model through the functional relation.

In other cases, the predictive model may also be built by semi-supervised learning training. The semi-supervised learning training is to train part of known characteristic data and part of unknown characteristic data together, and find out how the correlation exists between the clinical data and the known data characteristics.

And 204, testing the prediction models by using test data of a specific type to obtain test indexes of the prediction models for the specific type.

And step 205, determining the prediction model with the highest test index as the target prediction model corresponding to the specific type.

According to the technical solutions above, on the basis of the embodiment shown in fig. 1, the present embodiment further has the following beneficial effects: the embodiment of the invention discloses a process for establishing a prediction model by using a supervised learning training method and a semi-supervised learning training method in detail, so that the overall technical scheme of the method is more complete and the disclosure is more sufficient.

Fig. 3 shows an embodiment of the apparatus for obtaining a clinical data prediction model according to the present invention. The apparatus of this embodiment is a physical apparatus for performing the method described in fig. 1-2. The technical solution is essentially the same as that in the above embodiment, and the corresponding description in the above embodiment is also applicable to this embodiment. The device in this embodiment includes:

the format processing module 301 is configured to perform format processing on the sample data to obtain a sample set.

A model training module 302, configured to perform data training through the sample set to establish a plurality of prediction models.

The model testing module 303 is configured to test the prediction models by using specific types of test data to obtain test indexes of the prediction models for the specific types.

And a model determining module 304, configured to determine the prediction model with the highest test index as the target prediction model corresponding to the specific type.

Fig. 4 is a schematic structural diagram of an electronic device according to an embodiment of the present invention. On the hardware level, the electronic device comprises a processor and optionally an internal bus, a network interface and a memory. The Memory may include a Memory, such as a Random-Access Memory (RAM), and may further include a non-volatile Memory, such as at least 1 disk Memory. Of course, the electronic device may also include hardware required for other services.

The processor, the network interface, and the memory may be connected to each other via an internal bus, which may be an ISA (Industry standard architecture) bus, a PCI (Peripheral component interconnect) bus, an EISA (Extended Industry standard architecture) bus, or the like. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of illustration, only one double-headed arrow is shown in FIG. 4, but that does not indicate only one bus or one type of bus.

And the memory is used for storing the execution instruction. In particular, a computer program that can be executed by executing instructions. The memory may include both memory and non-volatile storage and provides execution instructions and data to the processor.

In a possible implementation manner, the processor reads corresponding execution instructions from the nonvolatile memory into the memory and then executes the corresponding execution instructions, and corresponding execution instructions can also be acquired from other equipment, so as to form a device for acquiring the clinical data prediction model on a logic level. The processor executes the execution instructions stored in the memory to implement the method for obtaining a clinical data prediction model provided in any embodiment of the present invention by executing the execution instructions.

The method performed by the apparatus for acquiring a clinical data prediction model according to the embodiment of the present invention shown in fig. 3 may be applied to or implemented by a processor. The processor may be an integrated circuit chip having signal processing capabilities. In implementation, the steps of the above method may be performed by integrated logic circuits of hardware in a processor or instructions in the form of software. The Processor may be a general-purpose Processor, including a Central Processing Unit (CPU), a Network Processor (NP), and the like; but also Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field Programmable Gate Arrays (FPGAs) or other Programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. The various methods, steps and logic blocks disclosed in the embodiments of the present invention may be implemented or performed. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.

The steps of the method disclosed in connection with the embodiments of the present invention may be directly implemented by a hardware decoding processor, or implemented by a combination of hardware and software modules in the decoding processor. The software module may be located in ram, flash memory, rom, prom, or eprom, registers, etc. storage media as is well known in the art. The storage medium is located in a memory, and a processor reads information in the memory and completes the steps of the method in combination with hardware of the processor.

Embodiments of the present invention further provide a readable storage medium, which stores an execution instruction, and when the stored execution instruction is executed by a processor of an electronic device, the electronic device can be caused to execute the method for obtaining a clinical data prediction model provided in any embodiment of the present invention, and is specifically configured to execute the method shown in fig. 1 to fig. 2.

The electronic device described in the foregoing embodiments may be a computer.

It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as a method or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects.

The embodiments of the present invention are described in a progressive manner, and the same and similar parts among the embodiments can be referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, as for the apparatus embodiment, since it is substantially similar to the method embodiment, the description is relatively simple, and for the relevant points, reference may be made to the partial description of the method embodiment.

It should also be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.

The above description is only an example of the present invention, and is not intended to limit the present invention. Various modifications and alterations to this invention will become apparent to those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the scope of the claims of the present invention.

Claims

1. A method of obtaining a predictive model of clinical data, comprising:

carrying out format processing on the sample data to obtain a sample set;

performing data training through the sample set to establish a plurality of prediction models;

testing the prediction models by using specific types of test data to obtain test indexes of the prediction models for the specific types;

and determining the prediction model with the highest test index as the target prediction model corresponding to the specific type.

2. The method of claim 1, wherein the training of data through the sample set to create a plurality of predictive models comprises:

3. The method of claim 1, wherein the sample data comprises known feature data, and wherein the training of data through the set of samples to create a plurality of predictive models comprises:

performing supervised learning training through the known feature data to obtain a functional relation between the known feature data and data features;

and establishing the prediction model through the functional relation.

4. The method of claim 1, wherein the sample data includes known feature data and unknown feature data, and wherein the training of data through the set of samples to create a plurality of predictive models comprises:

performing semi-supervised learning training through the known characteristic data and the unknown characteristic data to obtain a functional relation between the known characteristic data and the data characteristic;

and establishing the prediction model through the functional relation.

5. The method according to any one of claims 1 to 4, wherein the formatting the sample data to obtain the sample set comprises:

6. The method of claim 5, further comprising:

7. The method according to any one of claims 1 to 4, further comprising:

and inputting the data to be predicted of the specific type into a target prediction model corresponding to the specific type to obtain a prediction result.

8. An apparatus for obtaining a predictive model of clinical data, comprising:

9. A readable medium comprising executable instructions which, when executed by a processor of an electronic device, cause the electronic device to perform the method of any of claims 1 to 7.

10. An electronic device comprising a processor and a memory storing execution instructions, the processor performing the method of any of claims 1-7 when the processor executes the execution instructions stored by the memory.