Laboratory data fault-tolerant control method considering total error
By generating a dataset containing all errors, the laboratory data fault-tolerant control method is optimized, solving the problem of incomplete error simulation in existing technologies and realizing the reliability and accuracy of the laboratory system.
Patent Information
- Application Number
- CN202411133525.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-19
- Publication Date
- 2026-03-03
AI Technical Summary
Existing fault-tolerant processing methods cannot fully simulate the various types of errors in laboratory data, resulting in inaccurate predictions or untimely responses from laboratory systems in practical applications.
By establishing a fault-tolerant mechanism, a dataset containing proportional, constant, and exponential errors is generated to optimize and validate the parameters of real-time statistical algorithms, ensuring the effectiveness and reliability of the algorithms.
It achieves full-error simulation of the laboratory system, enabling timely detection of data vulnerabilities and ensuring the reliability and accuracy of the algorithm.
Smart Images
Figure CN121597451A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of medical data statistics, and specifically relates to a laboratory data fault-tolerant control method that considers total error. Background Technology
[0002] In real-time statistical algorithms based on laboratory analytical data, ensuring the robustness of the laboratory system is crucial. These algorithms need to be able to handle various types of errors, including proportional error, constant error, exponential error, and extreme errors, to ensure their effectiveness and reliability in practical applications. Existing fault-tolerant methods often cannot fully simulate these errors, leading to problems such as inaccurate predictions or untimely responses in real-world applications. Therefore, there is an urgent need to develop a fault-tolerant control method that can simulate various types of errors and verify the robustness of laboratory systems. Summary of the Invention
[0003] To address the aforementioned technical problems, this invention proposes a laboratory data fault-tolerant control method that considers all errors. This control method can generate a dataset containing different types of errors by uploading raw laboratory analysis data. This error dataset can be used to optimize and verify the parameters of real-time statistical algorithms, ensuring the effectiveness and reliability of the algorithms in practical applications.
[0004] According to one aspect of the present invention, a fault-tolerant control method for laboratory data considering total error is proposed, comprising the following steps:
[0005] S1: Establish a fault tolerance mechanism, which is used to define fault tolerance objects and fault types;
[0006] S2: Input the fault tolerance mechanism and its related raw data into the first prediction model to establish a dataset, and divide the dataset into a training dataset and a test dataset according to the proportion.
[0007] S3: Calculate the total error of each data sample in the training dataset, where the total error includes the proportional error, constant error, and exponential error;
[0008] S4: Simulated error injection: Input the proportional error, the constant error, and the exponential error into the dataset of the first prediction model to obtain the second prediction model;
[0009] S5: When the error between the prediction values of the first prediction model and the second prediction model is higher than the threshold, return to step S1 and define a new fault-tolerant object or a new fault type.
[0010] According to one embodiment of the present invention, the fault-tolerant object in step S1 is the medical procedure, the number of patients, or the number of test tubes.
[0011] According to one embodiment of the present invention, the fault type in step S1 is data lag, data interruption, or data loss.
[0012] According to one embodiment of the present invention, the threshold is 3% or 5%.
[0013] According to one embodiment of the present invention, the first prediction model in step S3 is the following formula (1):
[0014]
[0015] Among them, y n,t It is the fault tolerance object of the nth department and the tth hour, y n,t-p It is the fault type of the nth department and the tpth hour, θ n,t It is the time series composition structure of n departments on day t, where A is the coefficient corresponding to the time series composition structure; B1 and B p x is the coefficient corresponding to the fault type. m,t Let m be the environmental index of the m-th department at the t-th hour, and C be the coefficient corresponding to the environmental index, where m and n are both natural numbers greater than 2.
[0016] According to one embodiment of the present invention, the second prediction model in step S4 is the following formula (2):
[0017]
[0018] Where, ε n This indicates the total error.
[0019] According to one embodiment of the present invention, the ratio of the training dataset to the test dataset in step S2 is 80%:20% or 90%:10%.
[0020] According to one embodiment of the present invention, the constant error includes a maximum error and a minimum error.
[0021] In summary, compared with the prior art, the above-described technical solutions conceived by this invention mainly possess the following technical advantages:
[0022] 1. The fault tolerance mechanism of this invention takes into account all error modes that exist in the experimental data, which can not only ensure the reliability of the algorithm, but also promptly detect data vulnerabilities based on the magnitude of the error between the final models.
[0023] 2. The prediction model of this invention is simple and practical, and the error calculation method is quick. Attached Figure Description
[0024] Figure 1 This is a schematic diagram of the method flow of the present invention. Detailed Implementation
[0025] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.
[0026] This invention proposes a fault-tolerant control method for laboratory data that considers total error, comprising the following steps:
[0027] S1: Establish a fault tolerance mechanism, which is used to define fault tolerance objects and fault types;
[0028] S2: Input the fault tolerance mechanism and its related raw data into the first prediction model to establish a dataset, and divide the dataset into a training dataset and a test dataset according to the proportion.
[0029] S3: Calculate the total error of each data sample in the training dataset, where the total error includes the proportional error, constant error, and exponential error;
[0030] S4: Simulated error injection: Input the proportional error, the constant error, and the exponential error into the dataset of the first prediction model to obtain the second prediction model;
[0031] S5: When the error between the prediction values of the first prediction model and the second prediction model is higher than the threshold, return to step S1 and define a new fault-tolerant object or a new fault type.
[0032] In some embodiments, the fault-tolerant object in step S1 is the medical examination item, the number of patients, or the number of test tubes, wherein the medical examination item is a common examination item in the art, such as height, weight, vision, ENT, internal medicine, etc.
[0033] In some embodiments, the test tube may be a blood routine test tube or a urine routine test tube, etc.
[0034] In some embodiments, the fault type in step S1 is data lag, data interruption, or data loss.
[0035] In some embodiments, the threshold is 3% or 5%.
[0036] In some embodiments, the first prediction model in step S2 is the following equation (1):
[0037]
[0038] Among them, y n,tIt is the fault tolerance object of the nth department and the tth hour, y n,t-p It is the fault type of the nth department and the tpth hour, θ n,t It is the time series composition structure of n departments on day t, where A is the coefficient corresponding to the time series composition structure; B1 and B p x is the coefficient corresponding to the fault type. m,t Let m be the environmental index of the m-th department at the t-th hour, and C be the coefficient corresponding to the environmental index, where m and n are both natural numbers greater than 2.
[0039] In some embodiments, the second prediction model in step S4 is the following equation (2):
[0040]
[0041] Where, ε n This indicates the total error.
[0042] In some embodiments, the ratio of the training dataset to the test dataset in step S2 is 80%:20% or 90%:10%.
[0043] In some embodiments, the proportional error, constant error, and exponential error of the training dataset can be calculated using mathematical methods commonly used in the art, which will not be elaborated here.
[0044] In some embodiments, constant error includes maximum error and minimum error.
[0045] The following are specific examples:
[0046] Example 1
[0047] A fault-tolerant control method for laboratory data that considers total error is as follows:
[0048] S1: Define the fault-tolerant object as a blood routine test tube, and define the fault type as hysteresis;
[0049] S2: Input the number of blood routine test tubes and the lagged data of the Department of Internal Medicine of Peking Union Medical College Hospital on a certain date into the first prediction model of the following formula (1) to establish a dataset, and divide the dataset into a training dataset and a test dataset in a ratio of 80%:20%.
[0050]
[0051] Among them, y n,t It represents the number of blood test tubes in the nth department and the tth hour, y n,t-p It is the fault type of the nth department and the tpth hour, θ n,tIt is the time series composition structure of n departments on day t, where A is the coefficient corresponding to the time series composition structure; B1 and B p x is the coefficient corresponding to the fault type. m,t Let m be the environmental index of the m-th department at the t-th hour, and C be the coefficient corresponding to the environmental index, where m and n are both natural numbers greater than 2.
[0052] S3: Calculate the total error of each data sample in the training dataset, where the total error includes the proportional error, constant error, and exponential error;
[0053] S4: Input the proportional error, the constant error, and the exponential error into the dataset of the first prediction model to obtain the second prediction model, which is the following equation (2), where ε n Indicates total error:
[0054]
[0055] S5: The error between the predictions of the first and second prediction models is calculated to be 2%.
[0056] Example 2
[0057] Unlike Example 1, in another day's routine inspection, the error between the prediction values of the first prediction model and the second prediction model is 8%. At this time, return to step S1 to redefine the fault type as missing, and the error returns to 3%.
[0058] Example 3
[0059] Unlike Example 1, during routine testing on a certain day, it was found that the prediction error between the first prediction model and the second prediction model was 13%. At this time, the system returned to step S1 and redefined the fault type as missing, and the error returned to only 10%. Step S1 was then continued to redefine the fault type as stopped, and the error returned to 7%. At this time, the system returned to step S1 again and redefined the fault tolerance object as the natural number of people who took blood, keeping the fault type unchanged, and the error returned to 1%. The subsequent system error analysis report suggested that duplicate blood draws were suspected.
[0060] Example 4
[0061] Unlike Example 3, the fault tolerance object is defined as blood pressure. In a routine test on a certain day, it was found that the error between the prediction values of the first prediction model and the second prediction model was 52%. At this time, the system returned to step S1 and redefined the fault tolerance object as weight. The error returned to 2%. The subsequent system error analysis report suggested that there was a suspected error in the operator's medical examination item input.
[0062] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.
Claims
1. A laboratory data fault-tolerant control method considering total error, characterized in that, Includes the following steps: S1: Establish a fault tolerance mechanism, which is used to define fault tolerance objects and fault types; S2: Input the fault tolerance mechanism and its related raw data into the first prediction model to establish a dataset, and divide the dataset into a training dataset and a test dataset according to the proportion. S3: Calculate the total error of each data sample in the training dataset, where the total error includes the proportional error, constant error, and exponential error; S4: Simulated error injection: Input the proportional error, the constant error, and the exponential error into the dataset of the first prediction model to obtain the second prediction model; S5: When the error between the prediction values of the first prediction model and the second prediction model is higher than the threshold, return to step S1 and define a new fault-tolerant object or a new fault type.
2. The laboratory data fault-tolerant control method considering total error as described in claim 1, characterized in that, The fault-tolerant object mentioned in step S1 is the medical service item, the number of patients, or the number of test tubes.
3. The laboratory data fault-tolerant control method considering total error as described in claim 1, characterized in that, The fault types mentioned in step S1 are data lag, data interruption, or data loss.
4. The laboratory data fault-tolerant control method considering total error as described in claim 1, characterized in that, The threshold is 3% or 5%.
5. The laboratory data fault-tolerant control method considering total error as described in claim 1, characterized in that, In step S2, the first prediction model is the following equation (1): Among them, y n,t It is the fault tolerance object of the nth department and the tth hour, y n,t-p It is the fault type of the nth department and the tpth hour, θ n,t It is the time series composition structure of n departments on day t, where A is the coefficient corresponding to the time series composition structure; B1 and B p x is the coefficient corresponding to the fault type. m,t Let m be the environmental index of the m-th department at the t-th hour, and C be the coefficient corresponding to the environmental index, where m and n are both natural numbers greater than 2.
6. The laboratory data fault-tolerant control method considering total error as described in claim 1, characterized in that, In step S4, the second prediction model is the following equation (2): Where, ε n This indicates the total error.
7. The laboratory data fault-tolerant control method considering total error as described in claim 1, characterized in that, In step S2, the ratio of the training dataset to the test dataset is 80%:20% or 90%:10%.
8. The laboratory data fault-tolerant control method considering total error as described in claim 1, characterized in that, The constant error includes the maximum error and the minimum error.