CN114242259A - Advanced influenza condition prediction system, program product and establishing and using method thereof - Google Patents
Advanced influenza condition prediction system, program product and establishing and using method thereof Download PDFInfo
- Publication number
- CN114242259A CN114242259A CN202111038166.9A CN202111038166A CN114242259A CN 114242259 A CN114242259 A CN 114242259A CN 202111038166 A CN202111038166 A CN 202111038166A CN 114242259 A CN114242259 A CN 114242259A
- Authority
- CN
- China
- Prior art keywords
- medical
- prediction
- influenza
- data
- information system
- 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
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
- G16H50/80—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for detecting, monitoring or modelling epidemics or pandemics, e.g. flu
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
- G06F18/24147—Distances to closest patterns, e.g. nearest neighbour classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- 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
- G16H10/00—ICT specially adapted for the handling or processing of patient-related medical or healthcare data
- G16H10/60—ICT specially adapted for the handling or processing of patient-related medical or healthcare data for patient-specific data, e.g. for electronic patient records
-
- 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
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Public Health (AREA)
- Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Epidemiology (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Databases & Information Systems (AREA)
- Pathology (AREA)
- Biomedical Technology (AREA)
- Measuring And Recording Apparatus For Diagnosis (AREA)
Abstract
The invention relates to a system and a program product for predicting the state of an advanced influenza disease and a method for establishing and using the same. The establishment of the advanced flu illness state prediction system comprises the steps of firstly obtaining flu medical data of advanced flu patients from a medical database, and then carrying out AI learning by using characteristic variables such as vital signs, medical history, patient action states, blood inspection values and the like to obtain prediction models such as hospitalization transfer probability, concurrent pneumonia probability, concurrent septicemia or shock probability, intensive care unit transfer probability, death probability and the like; a medical information system service interface, a characteristic value acquisition service program and an illness state prediction service program are established on a servo host. The doctor uses a medical information system to enable the characteristic value acquisition service program to acquire a medical characteristic value of the elderly flu patient from the medical database, the disease condition prediction service program predicts by the prediction model according to the medical characteristic value, and the doctor can refer to the prediction result to perform subsequent treatment on the elderly flu patient.
Description
Technical Field
The invention relates to a disease state prediction system and a program product for advanced influenza and a building and using method thereof, in particular to an invention for predicting the development of the disease state of an advanced influenza patient by using AI to assist doctors to perform subsequent treatment on the advanced influenza patient.
Background
Taiwan is one of the most rapidly aging regions of the world population. In 2018, taiwan elderly account for 14% of the total population, and a sharp rise to 20% is expected in 2025. The clinical experience of emergency treatment shows that the old is complex and rapid in disease change, and the challenge is how to timely and properly treat the old, wherein influenza is a common seasonal disease of the old, severe people may need hospitalization, and septicemia, secondary bacterial infection, or complications such as respiratory tract and ischemic heart disease may be caused in the process, so that the old can die.
Due to limited medical resources in the prevailing season of influenza, prediction of prognosis and subsequent treatment of elderly influenza patients becomes a very important topic.
In recent years, the explosive development of Artificial Intelligence (AI), including machine learning and natural language, can process more variables without limiting data distribution, so that introducing AI technology and establishing disease prediction mode through electronic health record can provide better patient prognosis reference for physicians.
Chinese patent No. CN110051324A proposes a method and system for predicting the mortality of acute respiratory distress syndrome by AI.
Disclosure of Invention
Based on uncertainty of disease state development of the old-age influenza patient, the influence variables influencing the disease state development of the old-age influenza patient are found out, and the disease state development of the old-age influenza patient is predicted by adopting an AI learning mode so as to assist doctors to perform subsequent treatment on the old-age influenza patient.
Accordingly, the invention provides a method for establishing a disease prediction system for advanced influenza, comprising the following steps:
acquiring medical data of influenza: retrieving flu medical data associated with flu from a raw data of a medical database.
And (3) carrying out model training by AI learning: screening out the aged flu patients with the ages of more than 65 years to be diagnosed from the flu medical data, and excluding the aged flu patients with cardiac and pulmonary function stop before hospital arrival; and further cleaning and converting the medical influenza data to obtain a plurality of characteristic variables, entering the characteristic variables into a big data database, and performing model training by using AI according to the characteristic variables, wherein the characteristic variables comprise vital signs, medical history, patient action state and blood test values.
A step of obtaining a prediction model: and training according to the model to obtain a prediction model, wherein the prediction model comprises hospitalization transfer probability, pneumonia complication probability, septicemia or shock complication probability, intensive care unit transfer probability and death probability.
Establishing a network service: providing a medical information system service interface, a characteristic value acquisition service program and a disease condition prediction service program; the medical information system service interface is connected to a medical information system, the medical information system calls the medical information system service interface, the characteristic value acquisition service program acquires a medical characteristic value of the elderly flu patient related to the characteristic variable from the medical database, the illness state prediction service program predicts by the prediction model according to the medical characteristic value, and a prediction result is transmitted back to the medical information system service interface.
Further, the cleansing transformation of the medical data of the flu modifies the medical data of the flu that does not conform to a standard data type to conform to the standard data type. Furthermore, the data type not conforming to the standard data type includes one or a combination of incomplete data, mixed content, repeated data, error data generated by no check in input, incorrect format, different report units of null value or different check methods.
Further, among the characteristic variables: vital signs include respiratory rate and coma index; the medical history comprises hypertension, coronary artery disease and malignant tumor; the patient is in bed; blood test values include white blood cell count, rod-shaped nuclear granulocytes, hemoglobin and C-reactive protein.
Further, the AI learning algorithm uses one of a Random Forest algorithm (Random Forest), a Support Vector Machine (SVM), a K-Neighbor (KNN), a Multilayer Perceptron (MLP), a Light Gradient Boosting model (LightGBM), an eXtreme Gradient Boosting (XGBoost), and a Logistic Regression analysis (Logistic Regression). Furthermore, in the AI learning process, the medical data of the flu is divided into a training set and a test set, and seventy percent of the medical data of the flu is used for the training set and thirty percent is used for the test set, and the test set is used for verification.
Further, the prediction model also comprises the probability of returning to the emergency again within one day, the probability of returning to the emergency again within three days and the probability of returning to the emergency again within fourteen days.
Furthermore, the prediction result is displayed on the medical information system service interface in parallel by using a visual graph and numbers.
The invention further provides an old-age influenza disease state prediction system established by the establishing method of the old-age influenza disease state prediction system, which comprises the following steps:
the server host is provided with the big data database, is connected with the medical information system, is connected with the medical database together, provides a service interface of the medical information system to the medical information system and executes the characteristic value acquisition service program and the illness state prediction service program.
Further, a physiological monitoring instrument is connected with the medical database.
The invention further provides a program product, which loads an application program into a computer to build the advanced influenza disease prediction system.
The invention further provides a use method of the advanced influenza disease prediction system, which comprises the following steps:
the medical information system service interface is called by the medical information system. The characteristic value capturing service program captures the medical characteristic value of the elderly influenza patient from the medical database. The disease condition prediction service program predicts according to the medical characteristic value by using the prediction model. And transmitting the prediction result back to the medical information system service interface.
Furthermore, the disease condition prediction service program predicts by a plurality of different algorithms and returns the most same prediction result.
Furthermore, a physiological monitoring instrument is used for acquiring the medical characteristic value of the elderly flu patient at any time, so that the disease condition prediction service program can continuously predict.
The following effects can be achieved through the technical characteristics:
1. the invention finds out the influence factors influencing the disease development of the elderly influenza patients, including respiratory rate, coma index, hypertension, coronary artery diseases, malignant tumors, bedridden patients, white blood cell count, rod-shaped nuclear granulocytes, heme and C-reactive protein, and carries out AI learning by taking the influence factors as characteristic variables, so as to predict the disease development of the elderly influenza patients by utilizing AI, assist doctors, particularly emergency doctors, to carry out subsequent treatment on the elderly influenza patients and improve the cure rate of the elderly influenza patients.
2. In the AI learning process, seventy percent of the medical data is used for the training set, thirty percent of the medical data is used for the testing set, and the testing set is used for verification, so that the Accuracy (Accuracy), the Sensitivity (Sensitivity), the Specificity (Specificity) and the AUC (AUC) of the medical data are high, and the satisfaction degree of a clinician who is applied to clinic is high.
3. The invention further obtains physiological data related to medical characteristic values of the elderly influenza patients at any time through a physiological monitoring instrument, so that the disease condition prediction service program can continuously predict, correct treatment can be performed in real time, and complications are reduced.
4. The invention can adopt a plurality of different algorithms to predict and return the most same prediction result, thereby improving the accuracy of AI prediction.
Drawings
Fig. 1 is a schematic diagram illustrating an overall architecture of an advanced influenza disease prediction system according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of the establishment and prediction of a prediction system for an advanced influenza condition according to an embodiment of the present invention;
FIG. 3 is a diagram illustrating a service interface of a medical information system that can be connected to a server host according to an embodiment of the present invention;
FIG. 4 is a schematic diagram of a service interface of a medical information system of a server according to an embodiment of the present invention;
fig. 5 is a schematic diagram of continuously monitoring physiological data related to medical characteristic values of an elderly influenza patient by a physiological monitoring apparatus for continuous prediction according to an embodiment of the present invention.
Description of the symbols:
1: servo host
11 big data database
Medical information system service interface 12
13, characteristic value acquisition service program
14 disease condition prediction service program
Medical information system
21 medical record interface
Medical database
31 external data
32 in-hospital calendar data
33 structural medical data in the hospital
34 non-structural medical data in hospital
4: physiological monitoring instrument
A is a characteristic variable
A1 respiratory rate
A2 coma index
A3 hypertension
A4 coronary artery disease
A5 malignant tumor
A6 bed-ridden patient
A7 white blood cell count
A8 rod-shaped nuclear granulocytes
A9 heme
A10C-reactive protein
B, algorithm
B1 random forest Algorithm
B2 support vector machine
B3K-neighbor algorithm
B4 multilayer sensor
B5 lightweight gradient lifting model
B6 extreme gradient Lift
B7 logistic regression analysis
C prediction model
C1 chance of hospitalization
C2 probability of complicated pneumonia
C3 complication of sepsis or shock
C4 chance of transferring to intensive care unit
C5 probability of death
C6 Emergency call probability again in one day
C7 Emergency recovery probability within three days
And C8, the emergency treatment probability is returned within fourteen days.
Detailed Description
In view of the above technical features, the main functions of the system for predicting an advanced influenza disease, the program product and the method for establishing and using the same according to the present invention will be apparent from the following examples.
Referring to fig. 1, the system for predicting an advanced influenza disease state of the present embodiment includes a server 1 for AI prediction, wherein the server 1 has a big data database 11 and provides a medical information system service interface 12, a characteristic value capturing service program 13 and a disease state prediction service program 14; the server 1 is connected to a medical information system 2(HIS) of a hospital, and the server 1 and the HIS 2 are connected to a medical database 3.
Referring to fig. 1 and 2, the establishment of the advanced influenza disease prediction system first requires the server 1, and specifically includes the following steps:
procedure for acquiring medical data of influenza
Retrieving influenza medical data related to influenza from a raw data of the medical database 3. The medical database 3 may include: the medical data of the flu is acquired from data of a Qimei hospital general, a Suiyi hospital area and a Jiali hospital area in the interval from 2009 to 2018, and is acquired by emergency medical advice, nursing records, a batch price, a medical history and an inspection system established by a medical information system 2 in the hospital.
Step for model training in AI learning
According to the definition of the World Health Organization (WHO) of the United nations, the population over 65 years is the elderly population, so that the elderly influenza patients with the age of more than 65 years before arrival are screened out from the influenza medical data, and the elderly influenza patients with cardiac and pulmonary function stoppage before arrival are excluded. In emergency influenza patients, physicians will first set up influenza drugs and input relevant diagnosis of influenza in the diagnosis field, but sometimes in emergency situations, physicians will first perform relevant treatment orally, so to determine whether advanced influenza patients can be extracted from the doctor's advice diagnosis or influenza drug pricing system.
A plurality of characteristic variables a are further obtained from the medical data of influenza and stored in the big data database 11, and model training is performed with AI using the characteristic variables a according to the fact that about 6000 advanced influenza patients in the last decade are recruited by the emergency doctor of the Qimei hospital. The above characteristic variables A are selected by analyzing ten influencing factors which may influence the disease development, and can be classified into vital signs, medical history, patient action status and blood test value, wherein: vital signs include respiratory rate a1 (Tachypnea) and Coma index a2(Severe Coma, Glasgow Coma Scale); the medical history includes Hypertension A3(Hypertension), Coronary artery disease a4(Coronary artery disease) and malignancy a5 (Cancer); the patient's action state is bed A6 (Bedridden); the blood test values include the white blood cell count A7(Leukocytosis, white blood cell count), the rod-shaped nuclear cell A8(Bandemia, white blood cell count band form), the hemoglobin A9(Anemia, hemoglobin), and the C-reactive protein A10 (elongated CRP, C-reactive protein). Wherein the respiratory rate A1 is greater than 20 times per minute; the coma index A2 is less than 8; white blood cell count A7 is greater than 12000 per CC; the number of the rod-shaped nuclear granulocyte A8 is more than 10 percent; hemoglobin A9 is collected to less than 12 mg/dL; c-reactive protein A10 was collected at greater than 10 mg/dL. Each characteristic variable A is classified according to whether it meets the above-mentioned influence factors such as vital signs, medical history, patient action status and blood test value, for example, each characteristic variable A is classified as a vital sign when it falls into the category of A1 more than 20 times per minute.
When the medical data of the influenza does not conform to a standard data type, the medical data of the influenza is cleaned and converted to conform to the standard data type, and the standard data type shows that the data is complete, the arrangement format is correct, the data is not repeated, the data does not exceed the range of possible values, the unit conforms to the standard and the like. If the data type does not conform to the standard data type, for example, the data is incomplete, the content is mixed, the data is duplicated, the data is not checked in the input process to generate error data, the format is incorrect, the report units of null values or different checking methods are different, and the like. The cleansing conversion of medical data for influenza may include giving incomplete null portions of data to values to be filled in according to clinical practice, such as normal values that are not within the range of classification of the above-mentioned vital signs, medical history, patient action status, blood test values, and other influencing factors, i.e., the missing values are not classified into any of the above-mentioned influencing factors, and therefore the missing values do not influence the subsequent AI learning, such as the respiratory rate a1 represented at 12; the coma index a2 is denoted by 15; white blood cell count a7 is expressed in 7000 per CC; rod-shaped nuclear granulocytes A8 is expressed at 0%; heme A9 is expressed at 12 g/dL; c-reactive protein A10 was expressed at 2.5 mg/L.
The characteristic variables a are statistically classified by AI learning, and the algorithms B used in this embodiment include a Random Forest algorithm B1(Random Forest), a Support Vector Machine B2(Support Vector Machines, SVM), a K-Neighbor algorithm B3(K Nearest Neighbor, KNN), a Multilayer Perceptron B4 (MLP), a Light Gradient Boosting model B5(Light Gradient Boosting Machine, LightGBM), a limit Gradient Boosting B6 (exxtreme Gradient Boosting, XGBoost), and a Logistic Regression analysis B7 (logic Regression). In the AI learning process, the medical data of the influenza is divided into a training set and a test set, seventy percent of the medical data of the influenza is used for the training set, thirty percent of the medical data of the influenza is used for the test set, and the test set is used for verification. Referring to tables 1 to 8 below, according to the AI learning and verifying results, the prediction Accuracy (Accuracy), Sensitivity (Sensitivity), Specificity (Specificity), and auc (area under the curve) are all as high as 70% to 90%.
Table 1: the probability of returning to emergency again in one day
Method | Accuracy rate | Sensitivity of the probe | Degree of specificity | AUC |
Logistic regression analysis | 0.688 | 0.807 | 0.569 | 0.703 |
Random forest algorithm | 0.980 | 0.985 | 0.975 | 0.997 |
Support vector machine | 0.678 | 0.902 | 0.455 | 0.746 |
K-neighborhood algorithm | 0.888 | 0.967 | 0.809 | 0.959 |
Lightweight gradient lifting model | 0.981 | 0.982 | 0.981 | 0.997 |
Multilayer perceptron | 0.814 | 0.873 | 0.755 | 0.895 |
Extreme gradient boost | 0.860 | 0.967 | 0.753 | 0.968 |
Table 2: probability of emergency call back again within three days
Method | Accuracy rate | Sensitivity of the probe | Degree of specificity | AUC |
Logistic regression analysis | 0.678 | 0.858 | 0.497 | 0.702 |
Random forest algorithm | 0.911 | 0.912 | 0.909 | 0.975 |
Support vector machine | 0.687 | 0.939 | 0.263 | 0.687 |
K-neighborhood algorithm | 0.823 | 0.934 | 0.712 | 0.915 |
Lightweight gradient lifting model | 0.922 | 0.941 | 0.902 | 0.981 |
Multilayer perceptron | 0.695 | 0.855 | 0.535 | 0.732 |
Extreme gradient boost | 0.738 | 0.949 | 0.527 | 0.861 |
Table 3: probability of emergency treatment within fourteen days
Method | Accuracy rate | Sensitivity of the probe | Degree of specificity | AUC |
Logistic regression analysis | 0.622 | 0.748 | 0.496 | 0.648 |
Random forest algorithm | 0.867 | 0.869 | 0.865 | 0.945 |
Support vector machine | 0.582 | 0.752 | 0.413 | 0.613 |
K-neighborhood algorithm | 0.775 | 0.912 | 0.639 | 0.872 |
Lightweight gradient lifting model | 0.879 | 0.867 | 0.890 | 0.947 |
Multilayer perceptron | 0.632 | 0.739 | 0.524 | 0.662 |
Extreme gradient boost | 0.670 | 0.901 | 0.439 | 0.785 |
Table 4: chance of hospitalization
Method | Accuracy rate | Sensitivity of the probe | Degree of specificity | AUC |
Logistic regression analysis | 0.748 | 0.695 | 0.800 | 0.815 |
Random forest algorithm | 0.782 | 0.785 | 0.780 | 0.862 |
Support vector machine | 0.737 | 0.621 | 0.853 | 0.801 |
K-neighborhood algorithm | 0.727 | 0.680 | 0.774 | 0.796 |
Lightweight gradient lifting model | 0.759 | 0.746 | 0.772 | 0.842 |
Multilayer perceptron | 0.755 | 0.707 | 0.804 | 0.819 |
Extreme gradient boost | 0.756 | 0.815 | 0.697 | 0.839 |
Table 5: probability of complicated pneumonia
Method | Accuracy rate | Sensitivity of the probe | Degree of specificity | AUC |
Logistic regression analysis | 0.653 | 0.533 | 0.774 | 0.714 |
Random forest algorithm | 0.775 | 0.789 | 0.761 | 0.864 |
Support vector machine | 0.644 | 0.553 | 0.736 | 0.700 |
K-neighborhood algorithm | 0.723 | 0.812 | 0.634 | 0.790 |
Lightweight classGradient lifting model | 0.770 | 0.738 | 0.801 | 0.843 |
Multilayer perceptron | 0.656 | 0.582 | 0.730 | 0.717 |
Extreme gradient boost | 0.685 | 0.799 | 0.570 | 0.772 |
Table 6: chance of complicated sepsis or shock
Method | Accuracy rate | Sensitivity of the probe | Degree of specificity | AUC |
Logistic regression analysis | 0.761 | 0.693 | 0.830 | 0.846 |
Random forest algorithm | 0.962 | 0.962 | 0.963 | 0.994 |
Support vector machine | 0.728 | 0.647 | 0.809 | 0.810 |
K-neighborhood algorithm | 0.902 | 0.976 | 0.828 | 0.969 |
Lightweight gradient lifting model | 0.965 | 0.959 | 0.971 | 0.992 |
Multilayer perceptron | 0.796 | 0.794 | 0.798 | 0.867 |
Extreme gradient boost | 0.872 | 0.899 | 0.846 | 0.939 |
Table 7: chance of transferring to intensive care unit
Method | Accuracy rate | Sensitivity of the probe | Degree of specificity | AUC |
Logistic regression analysis | 0.798 | 0.739 | 0.857 | 0.874 |
Random forest algorithm | 0.983 | 0.993 | 0.974 | 0.999 |
Support vector machine | 0.800 | 0.736 | 0.864 | 0.881 |
K-neighborhood algorithm | 0.946 | 0.989 | 0.903 | 0.977 |
Lightweight gradient lifting model | 0.987 | 0.996 | 0.977 | 0.999 |
Multilayer perceptron | 0.929 | 0.965 | 0.893 | 0.974 |
Extreme gradient boost | 0.921 | 0.971 | 0.870 | 0.972 |
Table 8: probability of death
Method | Accuracy rate | Sensitivity of the probe | Degree of specificity | AUC |
Logistic regression analysis | 0.783 | 0.730 | 0.836 | 0.866 |
Random forest algorithm | 0.972 | 0.985 | 0.958 | 0.997 |
Support vector machine | 0.772 | 0.712 | 0.832 | 0.851 |
K-neighborhood algorithm | 0.923 | 0.987 | 0.860 | 0.968 |
Lightweight gradient lifting model | 0.980 | 0.981 | 0.979 | 0.998 |
Multilayer perceptron | 0.869 | 0.933 | 0.804 | 0.936 |
Extreme gradient boost | 0.905 | 0.954 | 0.857 | 0.966 |
Step of obtaining a prediction model
And (3) obtaining a prediction model C for subsequent AI prediction according to the model training, wherein the prediction model C comprises eight prediction models, namely a hospital transfer probability C1, a complication pneumonia probability C2, a complication septicemia or shock probability C3, a hospital transfer ward probability C4, a death probability C5, a one-day emergency return probability C6, a three-day emergency return probability C7 and a fourteen-day emergency return probability C8.
Steps for establishing network services
After obtaining the prediction model C, the service interface 12 of the medical information system, the characteristic value capturing service program 13 and the disease condition predicting service program 14 required for the AI prediction are established as a program product, and an application program is loaded into a computer as the server 1 through the program product. When the server 1 is installed, the server 1 is connected to a medical information system 2(HIS) of a hospital and the medical database 3.
Referring to fig. 3 and 4, when the patient is seen, the physician enters relevant medical data into a medical record interface 21 of the medical information system 2 in the hospital, the medical data is stored in the relevant medical database 3, and a connection instruction 22 for predicting the advanced influenza disease is provided on the medical record interface 21. When the physician determines that the patient is an elderly flu patient, the physician can click the connection command 22, and the medical information system 2 calls the medical information system service interface 12, at this time, the medical information system service interface 12 is displayed on the physician's computer, and the characteristic value extraction service program 13 and the disease condition prediction service program 14 are executed, the characteristic value extraction service program 13 extracts the medical characteristic value related to the characteristic variable a from the medical data of the elderly flu patient from the related medical database 3, the disease condition prediction service program 14 predicts the medical data according to the medical characteristic value by using the prediction model C, and transmits the prediction result back to the medical information system service interface 12, and the prediction result can be displayed on the medical information system service interface 12 in a visual graph and digital parallel manner, the prediction result can help doctors to know the possible disease development of the old flu patient, so that the follow-up treatment is facilitated, and the cure rate of the old flu patient is improved.
The disease condition prediction service program 14 may select one of a Random Forest algorithm B1(Random Forest), a Support Vector Machine B2 (SVM), a K-Neighbor algorithm B3(K near Neighbor, KNN), a Multilayer Perceptron B4 (MLP), a lightweight Gradient boost model B5(Light Gradient boost Machine, LightGBM), a limit Gradient boost B6 (extrement Gradient boost, XGBoost), and a Logistic Regression B7(Logistic Regression) for prediction. In order to improve the prediction accuracy, all algorithms can be used for prediction at the same time, and the most same prediction result is returned.
In the embodiment of the invention, 84 common advanced influenza-meeting historical data are tested in 2019 in 1 month to 3 months by a Qimei hospital college, the satisfaction degree is filled by a doctor in the test result, 51 common advanced influenza-meeting historical data are recycled, the score is graded by a 5-point method, the average satisfaction degree of the doctor is 4.6 points, and the feasibility of the system is displayed.
Referring to fig. 1 and 5, when an elderly flu patient is diagnosed, a physiological monitoring device 4, such as a sphygmomanometer, an oximeter, etc., may be used to acquire medical data of the elderly flu patient at any time and upload the medical data to the related medical database 3, the characteristic value extraction service 13 continuously or periodically extracts related medical characteristic values from the medical database 3, and the disease condition prediction service 14 continuously predicts the medical characteristic values, thereby facilitating real-time correct treatment and reducing complications.
While the operation, use and efficacy of the present invention have been described in connection with the above embodiments, it should be understood that they are merely illustrative of the preferred embodiments of the invention and that various changes and modifications can be made without departing from the spirit and scope of the invention.
Claims (13)
1. A method for establishing a system for predicting an advanced influenza disease state, the method comprising: acquiring medical data of influenza: retrieving flu medical data related to flu from a raw data of a medical database; and (3) carrying out model training by AI learning: screening out the aged flu patients with the ages of more than 65 years to be diagnosed from the flu medical data, and excluding the aged flu patients with cardiac and pulmonary function stop before hospital arrival; further cleaning and converting the medical influenza data to obtain a plurality of characteristic variables to enter a big data database, and performing model training by using AI according to the characteristic variables, wherein the characteristic variables comprise vital signs, medical history, patient action state and blood test values, and the characteristic variables comprise: vital signs include respiratory rate and coma index; the medical history comprises hypertension, coronary artery disease and malignant tumor; the patient is in bed; the blood test values include white blood cell count, rod-shaped nuclear granulocytes, hemoglobin and C-reactive protein; a step of obtaining a prediction model: training according to the model to obtain a prediction model, wherein the prediction model comprises hospitalization transfer probability, concurrent pneumonia probability, concurrent septicemia or shock probability, intensive care unit transfer probability and death probability; establishing a network service: providing a medical information system service interface, a characteristic value acquisition service program and a disease condition prediction service program; the medical information system service interface is connected to a medical information system, the medical information system calls the medical information system service interface, the characteristic value acquisition service program acquires a medical characteristic value of the elderly flu patient related to the characteristic variable from the medical database, the illness state prediction service program predicts by the prediction model according to the medical characteristic value, and a prediction result is transmitted back to the medical information system service interface.
2. The method of claim 1, wherein the washout transformation of the influenza medical data modifies those of the influenza medical data that do not conform to the standard data type to conform to the standard data type.
3. The method of claim 2, wherein the data that does not conform to the standard data type includes incomplete data, mixed content, repeated data, data that is input without error detection, incorrect format, null values, or different reporting units with different detection methods, or a combination thereof.
4. The method of claim 1, wherein the AI learning algorithm uses one of a random forest algorithm, a support vector machine, a K-neighborhood algorithm, a multi-level sensor, a lightweight gradient boosting model, extreme gradient boosting, and logistic regression analysis.
5. The method of claim 4, wherein during AI learning, the influenza medical data is divided into a training set and a testing set, and seventy percent of the influenza medical data is used for the training set and thirty percent is used for the testing set, and the testing set is used for verification.
6. The method of claim 1, wherein the prediction model further comprises a probability of a second emergency return within one day, a probability of a second emergency return within three days, and a probability of a second emergency return within fourteen days.
7. The method of claim 1, wherein the prediction result is displayed in parallel in the service interface of the health care information system as a visual graph and a number.
8. An aged influenza disease prediction system established using the method of establishing an aged influenza disease prediction system according to any one of claims 1 to 7, comprising: the server is characterized in that the server is provided with the big data database, the server is connected with the medical information system, the server and the medical information system are jointly connected with the medical database, the server provides a service interface of the medical information system to the medical information system, and executes the characteristic value capturing service program and the illness state predicting service program.
9. The system for predicting an advanced influenza condition as set forth in claim 8, wherein a physiological monitor is connected to the medical database.
10. A program product for loading an application program into a computer to create the advanced influenza condition prediction system of claim 8.
11. A method of using the system for predicting an advanced influenza disease according to claim 8, the method comprising: calling the medical information system service interface by the medical information system; the characteristic value capturing service program captures the medical characteristic value of the elderly influenza patient from the medical database; the disease condition prediction service program predicts by the prediction model according to the medical characteristic value; and transmitting the prediction result back to the medical information system service interface.
12. The method of claim 11, wherein the disease prediction service uses a plurality of different algorithms to predict the disease and returns the most similar prediction results.
13. The method of claim 11, wherein the medical characteristic value of the elderly influenza patient is obtained at any time by a physiological monitoring device, so that the disease prediction service program can continuously predict the disease.
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
TW109130621 | 2020-09-07 | ||
TW109130621A TWI751683B (en) | 2020-09-07 | 2020-09-07 | Pathological condition prediction system for elderly flu patients, a program product thereof, and a method for establishing and using the same |
Publications (1)
Publication Number | Publication Date |
---|---|
CN114242259A true CN114242259A (en) | 2022-03-25 |
Family
ID=80742929
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202111038166.9A Pending CN114242259A (en) | 2020-09-07 | 2021-09-06 | Advanced influenza condition prediction system, program product and establishing and using method thereof |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN114242259A (en) |
TW (1) | TWI751683B (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114974565A (en) * | 2022-05-17 | 2022-08-30 | 上海市第四人民医院 | Refined management and control method and system for diagnosis and treatment stages of new coronary patients |
WO2024073826A1 (en) * | 2022-10-06 | 2024-04-11 | Sociedade Beneficente Israelita Brasileira Hospital Albert Einstein | Method of training an artificial intelligence–based model for predicting the clinical outcome of an individual and method of predicting the clinical outcome of an indivdual using such artificial intelligence–based model |
Family Cites Families (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US10332637B2 (en) * | 2013-02-15 | 2019-06-25 | Battelle Memorial Institute | Use of web-based symptom checker data to predict incidence of a disease or disorder |
EP3433614A4 (en) * | 2016-03-23 | 2019-12-11 | Peach Intellihealth, Inc. | Use of clinical parameters for the prediction of sirs |
US20210249138A1 (en) * | 2018-06-18 | 2021-08-12 | Nec Corporation | Disease risk prediction device, disease risk prediction method, and disease risk prediction program |
-
2020
- 2020-09-07 TW TW109130621A patent/TWI751683B/en active
-
2021
- 2021-09-06 CN CN202111038166.9A patent/CN114242259A/en active Pending
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114974565A (en) * | 2022-05-17 | 2022-08-30 | 上海市第四人民医院 | Refined management and control method and system for diagnosis and treatment stages of new coronary patients |
WO2024073826A1 (en) * | 2022-10-06 | 2024-04-11 | Sociedade Beneficente Israelita Brasileira Hospital Albert Einstein | Method of training an artificial intelligence–based model for predicting the clinical outcome of an individual and method of predicting the clinical outcome of an indivdual using such artificial intelligence–based model |
Also Published As
Publication number | Publication date |
---|---|
TW202211258A (en) | 2022-03-16 |
TWI751683B (en) | 2022-01-01 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2021120936A1 (en) | Chronic disease prediction system based on multi-task learning model | |
US11037070B2 (en) | Diagnostic test planning using machine learning techniques | |
CN111710420B (en) | Complication onset risk prediction method, system, terminal and storage medium based on electronic medical record big data | |
Maghari | COVID-19 Detection in X-ray Images using CNN Algorithm | |
CN110348019B (en) | Medical entity vector conversion method based on attention mechanism | |
Hasan et al. | Heart disease diagnosis system based on multi-layer perceptron neural network and support vector machine | |
CN114242259A (en) | Advanced influenza condition prediction system, program product and establishing and using method thereof | |
Gao et al. | Chest X-ray image analysis and classification for COVID-19 pneumonia detection using Deep CNN | |
WO2021143780A1 (en) | Multi-label learning-based post-physical examination chronic disease prognosis system | |
CN107145715B (en) | Clinical medicine intelligence discriminating gear based on electing algorithm | |
Pal et al. | Deep learning techniques for prediction and diagnosis of diabetes mellitus | |
Al Ahdal et al. | Monitoring cardiovascular problems in heart patients using machine learning | |
Rehman et al. | Notice of Retraction: Infrared Sensing Based Non-Invasive Initial Diagnosis of Chronic Liver Disease Using Ensemble Learning | |
CN112908452A (en) | Event data modeling | |
Hasan et al. | FPGA design and hardware implementation of heart disease diagnosis system based on NVG-RAM classifier | |
CN116895372A (en) | Intelligent first-aid grading system based on large-scale language model and meta-learning | |
Schlosser et al. | Visual acuity prediction on real-life patient data using a machine learning based multistage system | |
Sharma et al. | Classification of heart disease from MRI images using convolutional neural network | |
Bahar et al. | Model Structure of Fetal Health Status Prediction | |
CN114550896A (en) | Dizziness patient emergency pre-examination triage decision-making method, device and model based on artificial neural network | |
TW202143248A (en) | System and method of biomedical data prediction risk including a data collection unit, a data processing unit and a judgment unit | |
Capellán-Martín et al. | A Lightweight, Rapid and Efficient Deep Convolutional Network for Chest X-Ray Tuberculosis Detection | |
Mandava | MDensNet201-IDRSRNet: Efficient cardiovascular disease prediction system using hybrid deep learning | |
Shaikh et al. | Automated classification of pneumonia from chest x-ray images using deep transfer learning efficientnet-b0 model | |
WO2021256784A1 (en) | Method and device for predicting dementia by using dementia risk factor according to sex of patient |
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 |