Medical potential related indicator risk monitoring system, method, terminal and storage medium

Abstract

The present invention belongs to the technical field of medical indicator data identification, and discloses a medical potential related indicator risk monitoring system, method, terminal and storage medium. The monitoring method includes: establishing a medical potential related indicator risk case database; establishing a preventive screening artificial intelligence algorithm model based on medical potential related indicator risk data according to the database, and verifying the established preventive screening artificial intelligence algorithm model according to the actual medical potential related indicator risk data obtained; inputting the medical potential related indicator risk data into the report output system program, and predicting according to the established preventive screening artificial intelligence algorithm model, the report output system outputs the report of the medical potential related indicator risk preventive screening, and visualizes it on the client interface. The present invention can help CINV prevention screening subjects and doctors who receive HEC to conduct CINV risk management and control, reduce the incidence of CINV, and reduce the social burden.

Description

Technical Field

[0001] The present invention belongs to the technical field of medical indicator data identification, and in particular relates to a medical potential related indicator risk monitoring system, method, terminal and storage medium. Background Art

[0002] Chemotherapy-induced nausea and vomiting (CINV) is the most common adverse reaction to chemotherapy, which may lead to dehydration, electrolyte imbalance, nutritional deficiency and emotional depression, significantly affecting the patient's quality of life and daily function, reducing the patient's tolerance and compliance to chemotherapy, and affecting the effect of chemotherapy. Although the development of modern antiemetic therapy has reduced the incidence of CINV patients and improved the quality of life of cancer patients, some patients still suffer from CINV, especially those receiving highly emetogenic chemotherapy (HEC). Studies have found that even with the use of guideline-recommended antiemetics, HEC may still cause CINV in 30% of patients. At present, research on CINV mainly focuses on the acute phase (within 24 hours of chemotherapy) and the delayed phase (2-5 days of chemotherapy), but in clinical work, it is found that some patients still experience nausea or vomiting in the persistent phase (6-14 days of chemotherapy). The persistent phase occurs more often after the patient is discharged from the hospital, and it is more difficult for doctors to provide help compared with acute and delayed phase CINV. There have been studies on CINV risk assessment tools in the past, but the predictive ability is general and there is a lack of research on the persistent phase.

[0003] In recent years, with the advent of the big data era and the improvement of computer application capabilities, machine learning algorithms have developed rapidly. Machine learning algorithms can learn patterns and decision rules from data and infer important connections between data sets that are difficult to relate. Compared with traditional statistical methods, machine learning algorithms have higher accuracy.

[0004] Through the above analysis, the problems and defects of the prior art are as follows:

[0005] (1) The data on preventive screening of potential CINV-related indicators by existing technologies are incomplete and have low accuracy.

[0006] (2) Existing technologies for CINV prevention and screening take a long time and are costly, and cannot provide timely data support for CINV risk management. Summary of the invention

[0007] In order to overcome the problems existing in the related art, the disclosed embodiments of the present invention provide a medical potential related indicator risk monitoring system, method, terminal and storage medium, and specifically relate to a CINV prevention screening method for HEC based on artificial intelligence.

[0008] The technical solution is as follows: A method for monitoring the risk of potential medical related indicators is applied to a client, and the method for monitoring the risk of potential medical related indicators includes the following steps:

[0009] S1: Establish a medical potential related indicator risk case database to store the collected medical potential related indicator risk data;

[0010] S2: According to the medical potential relevant indicator risk case database established in step S1, a preventive screening artificial intelligence algorithm model based on the medical potential relevant indicator risk data is established, and the established preventive screening artificial intelligence algorithm model is verified according to the actual medical potential relevant indicator risk data obtained;

[0011] S3: Input the medical potential related indicator risk data into the report output system program, and make predictions based on the established preventive screening artificial intelligence algorithm model. The report output system outputs a report on the medical potential related indicator risk preventive screening and displays it visually on the client interface.

[0012] In one embodiment, the method for establishing a medical potential relevant indicator risk case database in step S1 includes the following steps:

[0013] S11: Collect potentially relevant medical indicators and record different levels; the potentially relevant indicators include demographic data, liver and kidney function markers, treatment methods, and daily monitoring indicators;

[0014] S12: putting the collected data into an Excel table to form a medical potential related indicator risk case database;

[0015] S13: preprocessing the data of the medical potential related indicator risk case database formed in step S12;

[0016] S14: The data of the medical potential relevant indicator risk case database is divided, 70% of the data is used as a training data set, and 30% of the data is used as a verification data set.

[0017] In one embodiment, the preprocessing in step S13 includes the following steps:

[0018] Data screening: removing training samples and features whose missing data ratio exceeds the threshold;

[0019] Use the SMOTE algorithm to balance the data. For each sample in the minority class, use the Euclidean distance as the standard to calculate its distance to all samples in the minority class sample set to obtain its nearest neighbors. Set a sampling ratio according to the sample imbalance ratio to determine the sampling rate. For each minority class sample, randomly select several samples from its nearest neighbors to construct a new sample.

[0020] Feature selection: select feature subsets; the feature subsets are demographic data, liver and kidney function markers, treatment methods, and daily monitoring indicators.

[0021] In one embodiment, the method for establishing a preventive screening artificial intelligence algorithm model in step S2 includes the following steps:

[0022] S21: Model building: Based on the training data set divided by the data of the medical potential related indicator risk case database, the artificial intelligence gcForest algorithm is used to build a preventive screening artificial intelligence algorithm model;

[0023] S22: Model optimization: Optimizing the preventive screening artificial intelligence algorithm model obtained in step S21: Model training uses automatic tuning, using grid search technology to automatically adjust the network structure;

[0024] S23: Model training: Train the preventive screening AI algorithm model into a risk prediction model;

[0025] S24: Model evaluation: The risk prediction model obtained in step S23 is evaluated using a validation data set. Evaluation indicators include accuracy, precision, recall, F1 value, and AUC.

[0026] In one embodiment, in step S3, the input report output system includes: a terminal device of the fully automatic biochemical analyzer and a mobile terminal program software; the mobile terminal program software is built into the terminal device of the fully automatic biochemical analyzer.

[0027] In one embodiment, the result output of the input report output system includes: the terminal device based on the fully automatic biochemical analyzer performs liver and kidney function biomarker detection and analysis to obtain quantitative blood biochemical results, and at the same time combines the demographic data, treatment methods, and daily monitoring indicators of the screening object, and finally presents an intelligent prevention screening system report based on the prevention screening artificial intelligence algorithm model;

[0028] The preventive screening system report includes: demographic data, liver and kidney function markers, treatment methods, details of daily monitoring indicators, risk factors, result interpretation and treatment advice information.

[0029] In one embodiment, the preventive screening artificial intelligence algorithm model in step S3 includes:

[0030] Input to multi-grained scanning, scan with 5-dim, 10-dim, and 15-dim windows respectively, and obtain 16 5-dimensional vectors, 11 10-dimensional vectors, and 6 15-dimensional vectors, which are input to completely random forest and random forest, and the output results are aggregated into 66-dimensional vectors; the 66-dimensional vector is input to the cascade forest, and the 66-dimensional vector is aggregated with the output results of the first and second layers of the cascade forest, the results of the third layer are averaged, and then the maximum value is taken to obtain the prediction result;

[0031] The cascade forest aggregation process includes: the 66-dimensional features are first input into 5 forests to obtain 5 2-dimensional feature vectors, which are connected with the original 66-dimensional feature vectors to form the first-layer 76-dimensional lay features. The lay1 features are also passed through 5 forests to obtain 5 2-dimensional feature vectors, which are connected with the 66-dimensional original feature vectors to form the second-layer 76-dimensional feature vectors. In the third layer, the 76-dimensional features are input and passed through 5 forests to obtain 5 2-dimensional feature vectors. The 5 vectors are averaged and the maximum is taken.

[0032] Another object of the present invention is to provide a medical potential relevant indicator risk monitoring system for implementing the medical potential relevant indicator risk monitoring method, which is applied to a client, and the medical potential relevant indicator risk monitoring system comprises:

[0033] A medical potential related indicator risk case database, used to store the collected medical potential related indicator risk data;

[0034] A preventive screening artificial intelligence algorithm model establishment module is used to establish a preventive screening artificial intelligence algorithm model based on the established medical potential related indicator risk case database, and to verify the established preventive screening artificial intelligence algorithm model based on the collected actual medical potential related indicator risk data;

[0035] The input report output system is used to input the collected medical potential related indicator risk data into the built-in program, and make predictions based on the established preventive screening artificial intelligence algorithm model, output the preventive screening report, and display it visually.

[0036] Another object of the present invention is to provide a computer device, the computer device comprising a memory and a processor, the memory storing a computer program, and when the computer program is executed by the processor, the processor executes the medical potential related indicator risk monitoring method.

[0037] Another object of the present invention is to provide a computer-readable storage medium storing a computer program, which, when executed by a processor, enables the processor to execute the method for monitoring the risk of medical potential related indicators.

[0038] Combining all the above technical solutions, the advantages and positive effects of the present invention are as follows:

[0039] In order to solve the technical problems existing in the prior art, the present invention provides a CINV prevention and screening program software for patients receiving HEC, which uses 20 CINV potential related indicators such as demographic data, liver and kidney function markers, and daily monitoring indicators of treatment methods of CINV prevention and screening subjects receiving HEC, and outputs CINV prevention and screening reports and issues early warnings on the system program according to the established artificial intelligence algorithm model. The advantage of this method is that it can help CINV prevention and screening subjects and doctors receiving HEC to manage CINV risks, reduce CINV incidence, and reduce social burden. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the present disclosure.

[0041] Figure 1 is a flow chart of a method for monitoring the risk of medical potential related indicators provided by an embodiment of the present invention;

[0042] Figure 2 is the optimal hyperparameter provided by the embodiment of the present invention Figure 2 ;

[0043] Figure 3 is a cascade forest flow chart provided by an embodiment of the present invention;

[0044] Figure 4 Schematic diagram of a medical potential related indicator risk monitoring system provided by an embodiment of the present invention;

[0045] Figure 5 is a Deep Forest performance diagram of the present invention compared with a classic machine learning model provided by an embodiment of the present invention;

[0046] In the figure: 1. Medical potential related indicator risk case database; 2. Prevention screening artificial intelligence algorithm model establishment module; 3. Input report output system. DETAILED DESCRIPTION

[0047] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below in conjunction with the accompanying drawings. In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without violating the connotation of the present invention, so the present invention is not limited by the specific implementation disclosed below.

[0048] 1. Explanation of the embodiment:

[0049] The embodiment of the present invention provides a method for monitoring the risk of medical potential related indicators, which is applied to a client. The method for monitoring the risk of medical potential related indicators includes the following steps:

[0050] Step 1: Establish a medical potential related indicator risk case database to store the collected medical potential related indicator risk data;

[0051] Step 2: Based on the medical potential relevant indicator risk case database established in step 1, establish a preventive screening artificial intelligence algorithm model based on the medical potential relevant indicator risk data, and verify the established preventive screening artificial intelligence algorithm model based on the actual medical potential relevant indicator risk data obtained;

[0052] Step three: Input the medical potential related indicator risk data into the report output system program, and make predictions based on the established preventive screening artificial intelligence algorithm model. The report output system outputs a report on the medical potential related indicator risk preventive screening and displays it visually on the client interface.

[0053] The following is a further description of the technical solution of the present invention by taking the CINV prevention screening of HEC as an example.

[0054] Example 1

[0055] like Figure 1 As shown, the medical potential related indicator risk monitoring method provided by the embodiment of the present invention (CINV prevention screening method for HEC recipients based on artificial intelligence) includes the following steps:

[0056] S101: Patients receiving HEC were included in the database to collect potential CINV-related indicators, and the overall CINV observation period was set to 0 to 14 days;

[0057] S102: establishing an artificial intelligence algorithm model for CINV prevention and screening based on HEC according to the CINV case database of patients receiving HEC established in step S1, and verifying the artificial intelligence algorithm model for CINV prevention and screening based on HEC according to the actual incidence of CINV cases receiving HEC;

[0058] S103: Input the potential relevant indicators of CINV of patients receiving HEC into the report output system program, and output the CINV prevention screening report of patients receiving HEC based on the prevention screening artificial intelligence algorithm model.

[0059] Example 2

[0060] Based on the medical potential related indicator risk monitoring method provided by the embodiment of the present invention provided in Example 1, preferably, the specific method of establishing the CINV case database of HEC in step S101 includes the following steps:

[0061] S11: Collect potential CINV-related indicators and record the nausea and vomiting levels during the overall observation period; the potential related indicators include demographic data, liver and kidney function markers, treatment methods, and daily monitoring indicators; the grading of nausea and vomiting is assessed using the National Cancer Institute's Common Terminology Criteria for Adverse Events (NCI-CTCAE, version 5.0);

[0062] S12: putting the collected data into an Excel table to form a database;

[0063] S13: preprocessing the data of the case database of HEC formed in step S12;

[0064] S14: The data of the case database of HEC was divided, 70% of the data was used as a training data set, and 30% of the data was used as a validation data set.

[0065] Example 3

[0066] Based on the medical potential related indicator risk monitoring method provided by the embodiment of the present invention provided in Example 2, preferably, the preprocessing in step S13 includes:

[0067] Data screening: removing training samples and features whose missing data ratio exceeds the threshold;

[0068] Balance the data by using the SMOTE algorithm to analyze and simulate a small number of samples and add new artificially simulated samples to the data set so that the categories in the original data are no longer severely unbalanced;

[0069] Feature selection: select feature subsets; the feature subsets are demographic data, liver and kidney function markers, treatment methods, and daily monitoring indicators;

[0070] Preferably, the demographic data include: gender, age, drinking history, history of morning sickness; liver and kidney function markers include: creatinine clearance, total protein, globulin, albumin, aspartate aminotransferase, alanine aminotransferase, total bilirubin, direct bilirubin, alkaline phosphatase; treatment methods include: chemotherapy regimen, antiemetic regimen; daily monitoring indicators include: chemotherapy cycle, expected nausea and vomiting, sleep time the night before chemotherapy, nausea or vomiting during the previous cycle of chemotherapy, and use of over-the-counter antiemetics at home.

[0071] Example 4

[0072] Based on the medical potential related indicator risk monitoring method provided by the embodiment of the present invention provided in Example 1, preferably, the specific method of establishing the CINV prevention screening artificial intelligence algorithm model based on HEC in step S102 includes the following steps:

[0073] S21: Establishing a model: Based on the training data set in step S14, using the gsForest algorithm of artificial intelligence, an artificial intelligence algorithm model for CINV prevention screening based on HEC is established;

[0074] S22: Model optimization: Optimize the AI ​​algorithm model for CINV prevention screening based on HEC obtained in step S21: Model training uses automatic tuning, grid search technology, and automatically adjusts the network structure to improve the model's recognition accuracy for CINV risks based on HEC, and the optimal hyperparameters, such as Figure 2 shown.

[0075] S23: Model training: The HEC-based CINV prevention screening artificial intelligence algorithm model is trained as a risk prediction model;

[0076] S24: Model evaluation: The HEC-based CINV risk prediction model was evaluated using the validation dataset. The evaluation indicators included accuracy, precision, recall, F1 value, and AUC.

[0077] Example 5

[0078] Based on the medical potential related indicator risk monitoring method provided by the embodiment of the present invention provided in Example 4, preferably, the positive data for the effectiveness evaluation of the HEC-based CINV risk prediction model will be screened from the historical data of patients who developed CINV within 14 days after chemotherapy in the data set; the negative data will be selected from the historical data of patients who did not develop CINV within 14 days after chemotherapy in the data set; and the supervised learning model will be used for model training.

[0079] Example 6

[0080] Based on the medical potential related indicator risk monitoring method provided by the embodiment of the present invention provided in Example 4, preferably, in step S103, the input report output system includes: a terminal device based on a fully automatic biochemical analyzer and a mobile terminal program software.

[0081] Example 7

[0082] Based on the medical potential related indicator risk monitoring method provided by the embodiment of the present invention provided in Example 6, preferably, the result output process of the input report output system is: based on the terminal device of the fully automatic biochemical analyzer, the liver and kidney function biomarker detection and analysis is performed to obtain quantitative blood biochemical results, and at the same time, the demographic data, treatment methods, and daily monitoring indicators of the screening object are combined, and according to the HEC-based CINV prevention screening artificial intelligence algorithm model, finally presenting the intelligent CINV prevention screening system report;

[0083] The preventive screening system report includes: demographic data, liver and kidney function markers, treatment methods, details of daily monitoring indicators, risk factors, result interpretation, and treatment advice information.

[0084] Example 8

[0085] Based on the medical potential related indicator risk monitoring method provided by the embodiment of the present invention provided in Example 7, preferably, the preventive screening artificial intelligence algorithm model is as shown in Table 1 below.

[0086] Table 1 Preventive screening artificial intelligence algorithm model

[0087]

[0088] Example 9

[0089] Based on the preventive screening artificial intelligence algorithm model provided in Example 8, in step S103, preferably, according to the preventive screening artificial intelligence algorithm model, the method for predicting potential related indicators of CINV for receiving HEC includes:

[0090] A total of 20 variables are included, so one sample has 20 dimensions. One sample is first input into multi-grained sliding window scanning, and 5-dim, 10-dim, and 15-dim windows are scanned respectively to obtain 16 5-dimensional vectors, 11 10-dimensional vectors, and 6 15-dimensional vectors, which are input into completely random forests and random forests, and the output results are aggregated into 66-dimensional vectors. The 66-dimensional vector is input into the cascade forest, and the 66-dimensional vector is aggregated with the output results of the first and second layers of the cascade forest. The results of the third layer are averaged, and then the maximum value is taken to obtain the prediction result.

[0091] like Figure 3As shown in the figure, the cascade forest process is as follows: the 66-dimensional feature is first input into 5 forests to obtain 5 2-dimensional feature vectors, which are connected with the original 66-dimensional feature vector to form the 76-dimensional lay1 (first layer) feature. The lay1 feature also passes through 5 forests to obtain 5 2-dimensional feature vectors, which are connected with the 66-dimensional original feature vector to form a 76-dimensional feature vector (second layer). In the third layer, the 76-dimensional feature is input and passed through 5 forests to obtain 5 2-dimensional feature vectors. The 5 vectors are averaged and the maximum is taken.

[0092] Example 10

[0093] like Figure 4 As shown, the medical potential related indicator risk monitoring system provided by the embodiment of the present invention includes:

[0094] A medical potential related indicator risk case database 1, used to store the collected medical potential related indicator risk data;

[0095] The preventive screening artificial intelligence algorithm model establishment module 2 is used to establish the preventive screening artificial intelligence algorithm model according to the established medical potential related indicator risk case database, and verify the established preventive screening artificial intelligence algorithm model according to the collected actual medical potential related indicator risk data;

[0096] The input report output system 3 is used to input the collected medical potential related indicator risk data into the built-in program, and make predictions based on the established preventive screening artificial intelligence algorithm model, output the preventive screening report, and display it visually.

[0097] Preferably, the medical potential relevant indicator risk case database 1 includes:

[0098] The nausea and vomiting grade data collection module is used to collect potential CINV-related indicators and record the nausea and vomiting grades during the overall observation period; the potential related indicators include demographic data, liver and kidney function markers, treatment methods, and daily monitoring indicators; the grading of nausea and vomiting is evaluated using the Common Terminology Criteria for Adverse Events of the National Cancer Institute of the United States (NCI-CTCAE, version 5.0);

[0099] The database forming module is used to put the collected data into an Excel table to form a database;

[0100] A data preprocessing module, used for preprocessing the data of the formed case database of HEC;

[0101] The case database data division module is used to divide the data of the case database of HEC, 70% of the data is used as a training data set, and 30% of the data is used as a verification data set.

[0102] Preferably, the preventive screening artificial intelligence algorithm model building module 2 includes:

[0103] Establish a model module to use the artificial intelligence gsForest algorithm based on the training data set to establish an artificial intelligence algorithm model for CINV prevention screening based on HEC;

[0104] The model optimization module is used to optimize the obtained AI algorithm model for CINV prevention screening based on HEC: the model training adopts automatic tuning, and uses grid search technology to automatically adjust the network structure to improve the model's recognition accuracy for HEC-based CINV risks, and the optimal hyperparameters, such as Figure 2 shown.

[0105] Model training module, used to train the HEC-based CINV prevention screening artificial intelligence algorithm model into a risk prediction model;

[0106] The model evaluation module is used to evaluate the HEC-based CINV risk prediction model using a validation dataset. The evaluation indicators include accuracy, precision, recall, F1 value, and AUC.

[0107] Preferably, the input report output system 3 includes: a terminal device based on a fully automatic biochemical analyzer and a mobile terminal program software; the terminal device based on the fully automatic biochemical analyzer performs liver and kidney function biomarker detection and analysis to obtain quantitative blood biochemical results, and at the same time combines the demographic data, treatment methods, and daily monitoring indicators of the screening object, and based on the HEC-based CINV prevention screening artificial intelligence algorithm model, finally presents an intelligent CINV prevention screening system report; the prevention screening system report includes: demographic data, liver and kidney function markers, treatment methods, specific conditions of daily monitoring indicators, risk factors, result interpretation, and processing opinion information.

[0108] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0109] Since the information interaction, execution process, etc. between the above-mentioned devices / units are based on the same concept as the method embodiment of the present invention, their specific functions and technical effects can be found in the method embodiment part and will not be repeated here.

[0110] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated into a processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0111] 2. Application examples:

[0112] Application Example 1

[0113] An application embodiment of the present invention also provides a computer device, which includes: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the processor implements the steps in any of the above-mentioned method embodiments when executing the computer program.

[0114] Application Example 2

[0115] The application embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in the above-mentioned method embodiments can be implemented.

[0116] Application Example 3

[0117] The application embodiment of the present invention also provides an information data processing terminal, which is used to provide a user input interface to implement the steps in the above method embodiments when executed on an electronic device. The information data processing terminal is not limited to mobile phones, computers, and switches.

[0118] Application Example 4

[0119] The application embodiment of the present invention further provides a server, which is used to provide a user input interface to implement the steps in the above method embodiments when executed on an electronic device.

[0120] Application Example 5

[0121] An application embodiment of the present invention provides a computer program product. When the computer program product runs on an electronic device, the electronic device can implement the steps in the above-mentioned method embodiments when executing the computer program product.

[0122] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, which can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, the steps of the above-mentioned various method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may at least include: any entity or device that can carry the computer program code to the camera / terminal device, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium. For example, a USB flash drive, a mobile hard disk, a disk or an optical disk.

[0123] III. Evidence of the relevant effects of the embodiments:

[0124] The trained deep random forest model was applied to the test set to calculate the AUC, Accuracy, Precision, and F1 score. The deep forest has an AUC of 0.8041 and an accuracy of 0.8916, which has excellent classification ability and is better than typical machine learning models, including SVM (RBF), SVM (linear), random forest, naive Bayes, and logistic regression.

[0125] Table 1. Comparison with classic machine learning models

[0126]

[0127]

[0128] like Figure 5 As shown in the performance diagram of Deep Forest compared with the classic machine learning model, that is, the ROC curve diagram of the model, the Deep Forest has an AUC of 0.8041 and an accuracy of 0.8916, and has excellent classification ability, which is better than typical machine learning models, including SVM (RBF), SVM (linear), random forest, naive Bayes and logistic regression.

[0129] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any modifications, equivalent substitutions and improvements made by any technician familiar with the technical field within the technical scope disclosed by the present invention and within the spirit and principles of the present invention should be covered within the protection scope of the present invention.

Claims

1. A method for monitoring the risk of potential medical related indicators, characterized in that: Applied to the client, the method for monitoring the risk of potential medical related indicators includes the following steps: S1: Establish a medical potential related indicator risk case database to store the collected medical potential related indicator risk data; The medical potential related indicator risk case database is a CINV case database of patients receiving HEC, and patients receiving HEC are included in the database to collect CINV potential related indicators; S2: According to the medical potential relevant indicator risk case database established in step S1, a preventive screening artificial intelligence algorithm model based on the medical potential relevant indicator risk data is established, and the established preventive screening artificial intelligence algorithm model is verified according to the actual medical potential relevant indicator risk data obtained; S3: Input the medical potential related indicator risk data into the report output system program, and make predictions based on the established preventive screening artificial intelligence algorithm model. The report output system outputs a report on the medical potential related indicator risk preventive screening, and displays it visually on the client interface; The method for establishing a preventive screening artificial intelligence algorithm model in step S2 includes the following steps: S21: Model building: Based on the training data set divided by the data of the medical potential related indicator risk case database, the artificial intelligence gcForest algorithm is used to build a preventive screening artificial intelligence algorithm model; S22: Model optimization: Optimizing the preventive screening artificial intelligence algorithm model obtained in step S21: Model training uses automatic tuning, using grid search technology to automatically adjust the network structure; S23: Model training: Train the preventive screening AI algorithm model into a risk prediction model; S24: Model evaluation: using the validation data set to evaluate the risk prediction model obtained in step S23, the evaluation indicators include accuracy, precision, recall rate, F1 value and AUC; The preventive screening artificial intelligence algorithm model in step S3 includes: Input to multi-granularity sliding window scanning, 5-dim, 10-dim, 15-dim windows are scanned respectively, and 16 5-dimensional vectors, 11 10-dimensional vectors, and 6 15-dimensional vectors are obtained. Input to completely random forest and random forest, and the output results are aggregated into 66-dimensional vectors; input the 66-dimensional vector to the cascade forest, and aggregate the 66-dimensional vector with the output results of the first and second layers of the cascade forest, average the results of the third layer, and then take the maximum value to get the prediction result; The cascade forest aggregation process includes: the 66-dimensional features are first input into 5 forests to obtain 5 2-dimensional feature vectors, which are connected with the original 66-dimensional feature vectors to form the first-layer 76-dimensional lay features. The lay1 features are also passed through 5 forests to obtain 5 2-dimensional feature vectors, which are connected with the 66-dimensional original feature vectors to form the second-layer 76-dimensional feature vectors. In the third layer, the 76-dimensional features are input and passed through 5 forests to obtain 5 2-dimensional feature vectors. The 5 vectors are averaged and the maximum is taken.

2. The method for monitoring the risk of medical potential related indicators according to claim 1, characterized in that: The method for establishing a medical potential relevant indicator risk case database in step S1 comprises the following steps: S11: Collect potentially relevant medical indicators and record different levels; the potentially relevant indicators include demographic data, liver and kidney function markers, treatment methods, and daily monitoring indicators; S12: putting the collected data into an Excel table to form a medical potential related indicator risk case database; S13: preprocessing the data of the medical potential related indicator risk case database formed in step S12; S14: The data of the medical potential relevant indicator risk case database is divided, 70% of the data is used as a training data set, and 30% of the data is used as a verification data set.

3. The method for monitoring the risk of medical potential related indicators according to claim 2, characterized in that: The preprocessing in step S13 includes the following steps: Data screening: removing training samples and features whose missing data ratio exceeds the threshold; Use the SMOTE algorithm to balance the data. For each sample in the minority class, use the Euclidean distance as the standard to calculate its distance to all samples in the minority class sample set to obtain its nearest neighbors. Set a sampling ratio according to the sample imbalance ratio to determine the sampling rate. For each minority class sample, randomly select several samples from its nearest neighbors to construct a new sample. Feature selection, select feature subsets; the feature subsets are demographic data, liver and kidney function markers, treatment methods, and daily monitoring indicators.

4. The method for monitoring the risk of medical potential related indicators according to claim 1, characterized in that: In step S3, the input report output system includes: a terminal device of the fully automatic biochemical analyzer and a mobile terminal program software; the mobile terminal program software is built in the terminal device of the fully automatic biochemical analyzer.

5. The method for monitoring the risk of medical potential related indicators according to claim 4, characterized in that: The output of the input report output system includes: quantitative blood biochemical results obtained by detecting and analyzing liver and kidney function biomarkers based on the terminal equipment of the fully automatic biochemical analyzer, and combining the demographic data, treatment methods, and daily monitoring indicators of the screening subjects, and finally presenting the intelligent prevention screening system report based on the prevention screening artificial intelligence algorithm model; The preventive screening system report includes: demographic data, liver and kidney function markers, treatment methods, details of daily monitoring indicators, risk factors, result interpretation and treatment advice information.

6. A medical potential relevant indicator risk monitoring system implementing the medical potential relevant indicator risk monitoring method according to any one of claims 1 to 5, characterized in that: Applied to the client, the medical potential related indicator risk monitoring system includes: A medical potential related indicator risk case database (1), used to store the collected medical potential related indicator risk data; A preventive screening artificial intelligence algorithm model establishment module (2) is used to establish a preventive screening artificial intelligence algorithm model based on the established medical potential related indicator risk case database, and to verify the established preventive screening artificial intelligence algorithm model based on the collected actual medical potential related indicator risk data; The input report output system (3) is used to input the collected medical potential related indicator risk data into the built-in program, and make predictions based on the established preventive screening artificial intelligence algorithm model, output the preventive screening report, and visualize it.

7. A computer device, characterized in that: The computer device includes a memory and a processor, the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the medical potential related indicator risk monitoring method according to any one of claims 1-5.

8. A computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor executes the medical potential related indicator risk monitoring method according to any one of claims 1 to 5.

Images