KR20170140757A - A clinical decision support ensemble system and the clinical decision support method by using the same - Google Patents

Abstract

The present invention relates to a clinical decision support ensemble system and a clinical decision support method using the same. Ensemble prediction is performed by integrating the clinical prediction result of a patient through machine learning received from a plurality of external medical institutions. So, it is possible to predict the current state of the patient as well as the progression of a disease of the patient in the future and support rapid and accurate clinical decision-making about the medical care of a medical person.

Description

TECHNICAL FIELD [0001] The present invention relates to a clinical decision support ensemble system and a clinical decision support method using the same.

The present invention relates to a clinical decision support ensemble system and a method for supporting clinical decision making using the system. More particularly, the present invention relates to a system for integrating clinical prediction results of a patient's disease predicted by at least one medical institution, To provide a guideline for the diagnosis and treatment of the patient to the medical practitioner by carrying out the patient-customized analysis reflecting the patient's medical condition, thereby minimizing the misdiagnosis by cross-checking the judgment of the medical practitioner with the above guidelines when determining the diagnosis or treatment policy for the user The present invention relates to a system and a method for supporting quick and correct decision-making in a series of processes related to medical treatment for accurate prediction, diagnosis, treatment, prevention, and future management of a patient's disease.

The conventional way of understanding and treating the condition of a patient suffering from a disease or a disease depends on the experience, knowledge, career, or information of the individual medical practitioner who treats the patient, and a simple structure Respectively.

Such a structure has a problem that the patient's diversion to a specific hospital may increase due to the difference in the medical care based on the experience, knowledge, career or information of each medical person, and thus the polarization of the hospital becomes worse.

In addition, due to the rapid development of medical technology and information processing technology, data used in medical fields such as development of various medical devices and medical records are being digitized, and the amount of information about systematic and accumulated medical research is explosively increasing. Also, due to the spread of personal health information collection devices linked to smart devices, clinical information including personal health records is gradually becoming big data.

In recent years, new and various types of diseases and diseases are occurring due to global warming due to development of industrial technology, increase of risk factors for human body, changes in lifestyle and lifestyle.

Due to the changes in the medical environment, the clinical information necessary for the medical consultation of the medical staff is becoming diverse and complex. This is because, as in the prior art, when a patient is diagnosed and treated depending on individual medical personnel, medical accidents due to misdiagnosis may occur, resulting in the patient suffering fatal injury to the body or mind.

In other words, the modern medical environment is no longer able to construct accurate diagnosis or treatment strategies for patients depending on the experience or information of individual medical personnel, and has limitations in diagnosis and treatment of new diseases.

In order to solve the above problems, the Clinical Decision Support System (CDSS) has been developed based on the patient's clinical information to help decision-makers in diagnosis and treatment of patients' diseases. .

In general, the clinical decision support system is a system that programs medical guidelines based on rules (precisely, IF-THEN rules) and informs the physician of the results of the patient's condition guidelines. There are DUR (Drug Utilization Review) drug relief services that set constraints on a large number of complex drugs to avoid mutual interaction.

Based on the clinical information of the patient, the clinical decision support system supports decision making such as selection of a medical person, selection of a treatment method, or selection of a treatment drug for the patient, minimizes the risk to the patient's misdiagnosis, It is possible to provide an objective medical service to the patient by reducing the disparity of the medical care for each medical person.

Recently, the incidence of chronic diseases such as cardiovascular diseases or diabetes has been greatly increased. The chronic diseases have a high risk of emergency death due to acute exacerbation, so that the diagnosis and prediction of an emergency situation and early diagnosis of complications, Management is very important.

However, since the existing clinical decision support system provides functions such as drug prescription and medication alert and provides results only within a predetermined rule, rather than providing a guide for predicting or early diagnosis of disease progress, There is a limit to the comprehensive management of patients such as prediction, diagnosis, treatment, or prevention of a patient's disease, and to form an accurate diagnosis or treatment strategy according to the progress of the disease in the future.

In addition, since the existing clinical decision support system is based on the result of repeatedly learning the case of the same patient, it can not provide a reliable diagnosis and treatment method for a new patient suffering from the same disease.

In addition, the clinical information necessary for the patient's medical care or treatment is generated, scaled and managed in the respective data format of each institution centering on the institutions providing medical services such as a hospital or a clinic, and the respective medical institutions We have constructed and operate a systematic clinical decision support system. There is a problem in that valuable and newly discovered information can not be effectively utilized for a specific disease due to the lack of interoperability between the medical institutions.

Therefore, in the present invention, based on the clinical information generated, scaled and managed by each institute through interlocking with a plurality of medical institutions using the ensemble prediction system, clinical prediction results for similar cases of specific patients performed for each institution And provides a guideline for the diagnosis and treatment of the patient to the medical practitioner by performing the patient-customized analysis reflecting the integrated clinical prediction result and the characteristic of the specific patient, thereby providing a medical treatment, a treatment strategy, a prescription, And to provide a system and method for supporting rapid and accurate decision-making in a series of medical procedures related to overall medical activities such as prevention and future management.

In addition, the present invention collects biometric information data of a patient with an IoT device and continuously monitors and manages the disease of the patient to predict a risk condition for a disease of a high-risk group such as cardiovascular disease, And to provide a system and a method for preventing the occurrence of a problem.

Next, a brief description will be given of the prior arts that exist in the technical field of the present invention, and technical matters which the present invention intends to differentiate from the prior arts will be described.

First, Korea Patent No. 1558142 (Oct. 20, 2015) refers to a decision support system for medical staff, and it provides various data such as patient's test result, vital sign, patient's appeal, The present invention relates to a system for enabling appropriate treatment of a patient and a treatment plan to be applied to decision rule rules concerning treatments or treatments.

Korean Patent Laid-Open Publication No. 2009-0000196 (2009.01.07.) Relates to a clinical decision support apparatus using home care data and medical care information, and includes measurement data transmitted through a home health network and various sensors, The present invention relates to an automated clinical decision support apparatus capable of prompt medical treatment and emergency treatment using the patient's medical care data managed in the hospital.

The above prior arts use only the patient information collected by each hospital to identify only the patient condition at a specific point in time and establish a diagnosis and treatment plan of the patient based on this, And are unable to provide reliable clinical decision-making information.

On the other hand, according to the present invention, a common medical institution that performs clinical decision making for diagnosing and treating a patient through a common ensemble prediction system requests at least one or more external medical institutions for a clinical prediction result on the condition of the patient, Each of the medical institutions provides a clinical prediction result independently predicted based on clinical information generated, accumulated, and managed by itself, to the specific medical institution, and the specific medical institution provides the clinical prediction result from the plurality of medical institutions Diagnosis, treatment, prevention, and future management of the patient's condition based on the clinical prediction result and the clinical information of the patient, thereby providing the effect that the medical institution can make quick and accurate decision-making have.

The present invention has been made in order to solve the above problems, and it is an object of the present invention to integrate the clinical prediction results of patients performed for each medical institution through interlocking between a plurality of medical institutions, Diagnosis and prevention of the condition of the patient, prediction of an emergency situation, and the like, by predicting the diagnosis and future trend of the patient and providing the necessary guidelines for decision making about the medical care action of the medical person, And to provide a system and a method for supporting accurate and prompt decision-making by medical personnel on effective management of disease in the future.

In addition, the present invention collects biometric information data of a patient such as a lifelog through an IoT device, continuously monitors and manages the patient's condition, and predicts and prevents the occurrence of disease deterioration and an emergency situation in a high-risk patient group And a method thereof.

The clinical decision support ensemble system according to an embodiment of the present invention integrates clinical prediction results provided from a plurality of medical institutions, a medical information provider, or a combination of them to provide ensemble prediction, thereby providing clinical decision information .

And is performed on the basis of the results of clinical prediction by machine learning using big data for self-medical information learning from a plurality of medical institutions.

The ensemble prediction includes a machine learning-based clinical prediction result on the big data for self-medical information learning of a specific medical institution and a machine learning-based clinical prediction result on the big data for medical information learning of each medical institution from at least one external medical institution To support clinical decision making.

Further, the clinical decision information may be a decision tree, a neural network, a Naive Bayes, or a combination thereof according to a rule-based knowledge base, or according to individual, disease, Or by a combination of these.

The machine learning also includes an artificial intelligence technique that performs learning, reasoning, prediction, or a combination thereof for multidimensional analysis of big data including medical records, life logs, or combinations thereof for patient-centered personalized care .

Further, the machine learning is characterized by outputting numerical information of high reliability extracted or integrated through quantification of big data for medical information learning learning, as a result of clinical prediction.

In addition, the clinical decision support ensemble system includes a clinical decision support system that provides clinical decision information according to a user's clinical decision request, and an ensemble prediction system that provides a clinical prediction result at the request of the clinical decision support system The ensemble prediction system is characterized by performing ensemble prediction for decision making by integrating clinical prediction results provided from a plurality of medical institutions, medical information providing organizations, or a combination of them.

The clinical decision support ensemble system may further include a machine learning engine that includes deep learning to perform learning using the big data from the clinical information database.

In addition, the clinical information database includes hospital clinical information of an individual hospital and life log information of an outpatient.

Further, the machine learning engine abstracts characteristics of medical information from big data including EMR, PHR, medical image, lifelog information, or a combination thereof, extracts a clinical prediction model by learning the big data, And the reliability of the clinical prediction result is improved by predicting the situation prematurely.

The ensemble prediction system includes an interface for making a prediction request to an ensemble prediction system existing in a plurality of external medical institutions and receiving prediction results from an external medical institution and a collaboration hospital management unit for managing hospital information for cooperation with the plurality of external medical institutions. And a control unit.

Further, the organization is characterized by comprising the ensemble predicting system, a machine learning engine, and big data for medical information learning.

In addition, the clinical decision support ensemble system collects biometric information data of a patient including life log information of a patient, continuously monitors and manages the state of the patient, and provides the biometric information data to the IoT device.

According to another aspect of the present invention, there is provided a method for supporting clinical decision making, including: performing ensemble prediction by integrating a plurality of clinical prediction results; and providing clinical decision information with improved accuracy through ensemble prediction .

Further, the ensemble prediction is performed based on a result of clinical prediction by machine learning using big data for self-medical information learning from a plurality of medical institutions.

The ensemble prediction includes a machine learning-based clinical prediction result on the big data for self-medical information learning of a specific medical institution and a machine learning-based clinical prediction result on the big data for medical information learning of each medical institution from at least one external medical institution To support clinical decision making.

The clinical decision information may also be a decision tree, a neural network, a Naive Bays, or a combination thereof based on a rule-based knowledge base, or by individual, disease, Or by a combination of these.

The machine learning also includes an artificial intelligence technique that performs learning, reasoning, prediction, or a combination thereof for multidimensional analysis of big data, including medical records, lifelog information, or combinations thereof for patient- .

In addition, the machine learning is characterized by outputting high reliability numerical information extracted or integrated through quantization of big data for medical information learning learning, as a result of clinical prediction.

Providing clinical decision information according to a user's clinical decision making request through a clinical decision support system and providing a clinical prediction result at the request of the clinical decision support system through an ensemble prediction system, The ensemble prediction system is characterized by performing ensemble prediction for decision making by integrating clinical prediction results provided from a plurality of medical institutions, a medical information providing agency, or an organization including a combination thereof.

Further, the method for supporting clinical decision support may further include performing learning using big data from a clinical information database through a machine learning engine.

In addition, the clinical information database includes hospital clinical information of an individual hospital and life log information of an outpatient.

The machine learning engine abstracts the characteristics of the medical information from the big data including the EMR, the PHR, the medical image, the life log or a combination thereof, extracts the clinical prediction model by learning the big data, The reliability of the clinical prediction result is improved.

The ensemble prediction system includes an interface for making a prediction request to an ensemble prediction system existing in a plurality of external medical institutions and receiving prediction results from an external medical institution and a collaboration hospital management unit for managing hospital information for cooperation with the plurality of external medical institutions. And a control unit.

Further, the organization is characterized by comprising the ensemble predicting system, a machine learning engine, and big data for medical information learning.

Also, the clinical decision support method collects biometric information data of a patient including a life log of a patient, continuously monitors and manages the patient's condition, and provides the biometric information data to the IoT device.

According to another aspect of the present invention, there is provided a method for supporting clinical decision support, comprising the steps of: receiving prediction data predicted by another medical institution or a medical information providing agency by inputting learning data provided by a specific medical institution or a medical information providing agency; Extracting ensemble learning data in combination with a prediction result predicted by a specific medical institution or a medical information providing organization, and performing ensemble learning using the extracted ensemble learning data.

In addition, when the prediction result is provided, the specific medical institution or the medical information providing agency generates the self-learning data and provides the learning data to the other medical institution or the medical information providing agency. In the other medical institutions or the medical information providing agency, And providing a prediction result obtained by using the predictive model as the input of the learning data to the specific medical institution or the medical information providing agency so that the specific medical institution or the medical information providing agency is provided with the prediction result .

The ensemble learning may further include performing pre-learning using prediction results of the other medical institutions or medical information providers, and performing fine adjustment learning by further using disease information to the prediction results.

In addition, a prediction result predicted by inputting a self-prediction result predicted by using a clinical data of a specific medical institution or a medical information provider and a clinical data of the specific patient provided from another medical institution or a medical information provider And performing ensemble prediction using the integrated prediction result.

In addition, the clinical decision support ensemble system according to another embodiment of the present invention receives prediction data predicted by another medical institution or a medical information provider by inputting learning data provided by a specific medical institution or a medical information provider, Extracting the ensemble learning data in combination with the predicted results predicted by the specific medical institution or the medical information providing agency, and performing ensemble learning using the extracted ensemble learning data.

The present invention relates to a clinical decision support ensemble system and a decision support method using the system. The system includes a common ensemble prediction system, which integrates the clinical prediction results of specific patients provided for each medical institution, Diagnosis and future trends of the patient's condition reflecting the integrated clinical prediction results and the characteristics of the patient are predicted, and the prediction, diagnosis, prevention, treatment strategy, emergency alert for the emergency condition, prescription and future Effective medical management, and the like, to the medical practitioner, thereby helping the medical practitioner to make quick and accurate decisions about the medical care performed.

1 is a conceptual diagram for explaining a schematic concept of a clinical decision support ensemble system according to an embodiment of the present invention.
2 is a block diagram illustrating a process of providing clinical decision information according to an embodiment of the present invention.
3 is a block diagram illustrating a configuration of a clinical decision support ensemble system according to an embodiment of the present invention.
4 is a diagram for explaining a process of performing machine learning for each medical institution through a machine learning engine according to an embodiment of the present invention.
FIG. 5 is a diagram for explaining an ensemble learning process for estimating an ensemble in an ensemble prediction system according to another embodiment of the present invention. Referring to FIG.
6 is a view for explaining a process of outputting ensemble prediction results of a specific patient through ensemble learning and providing clinical decision information to a user according to another embodiment of the present invention.
7 is a block diagram illustrating a configuration of a clinical decision support ensemble system according to another embodiment of the present invention.
8 is a flowchart illustrating a procedure for generating clinical decision information according to an embodiment of the present invention and providing the generated clinical decision information to a terminal of a medical person.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. Like reference symbols in the drawings denote like elements.

1 is a conceptual diagram for explaining a schematic concept of a clinical decision support ensemble system according to an embodiment of the present invention.

FIG. 1 is a conceptual diagram for schematically explaining a clinical decision support ensemble system according to an embodiment of the present invention. FIG. 2 is a block diagram for explaining a process of providing clinical decision information according to an embodiment of the present invention. .

1 and 2, the clinical decision support ensemble system 100 is constructed in an institution including a plurality of medical institutions, a medical information provider, or a combination thereof, And can interoperate with each other through the decision support ensemble system 100.

The clinical decision support ensemble system 100 basically includes a clinical decision support system 110 for providing clinical decision information to a user who treats the patient (that is, a medical person), a clinical information database 130 A machine learning engine 140 for generating a clinical prediction model for each disease by performing machine learning on the basis of the medical information learning big data stored in the medical information learning learning unit 140 and outputting the clinical prediction result of the patient based on the generated clinical prediction model And an ensemble predicting system (120).

In addition, a medical person of a medical institution that treats a specific patient to accurately diagnose and perform a series of medical actions such as diagnosis of a disease of a patient, selection of a treatment method, and selection of a therapeutic drug is provided to the clinical decision support system 110, Request the decision information.

When the clinical decision support system 110 receives a request for clinical decision information from the medical care provider, the clinical decision support system 110 requests the ensemble prediction result of the patient to the ensemble prediction system 120. The ensemble prediction system 120 ) Requests a plurality of external medical institutions through the communication network to each of the patient's clinical prediction results. At this time, the ensemble prediction system 120 provides the clinical information of the patient to the external medical institution, and each of the external medical institutions performs clinical prediction of the patient based on the received clinical information of the patient .

Hereinafter, the medical institution is used as a concept including a medical institution such as a hospital or a clinic, and a medical information provider such as a NHIC or a HIRA.

On the other hand, the clinical information of the patient provided to the plurality of external medical institutions includes personal information such as the patient's name, address, and resident registration number, which is not shown. The patient's age, disease type, An electronic medical record (EMR), a medical image, a life log, or a combination of these, including a combination of the above and the like, a prescription or a combination thereof, and the clinical information may vary depending on the disease of the patient.

The clinical prediction also predicts the current state of the patient's disease and the progress of the disease progression in the future.

When the clinical decision support ensemble system 100 receives the clinical prediction result from another medical institution, the clinical decision support ensemble system 100 inputs the clinical information of the received patient to the clinical prediction model generated through the machine learning engine 140 , The clinical prediction of the patient is performed, and the clinical prediction result is output.

On the other hand, the clinical prediction is performed for each medical institution that receives the request for the clinical prediction result, and each of the medical institutions performs the clinical prediction based on the clinical information generated, scaled, stored and managed by itself, To the medical institution requesting the clinical prediction. The clinical information generated, accumulated, stored and managed for each of the above-mentioned medical institutions is time-series big data as medical information for each patient including PHR and hospital clinical information collected from a plurality of patients.

The PHR and hospital clinical information will be described in detail with reference to FIG.

Also, the machine learning engine 140 carries out machine learning based on its own clinical information in each of the medical institutions to extract the clinical prediction model. In addition, the machine learning engine 140 may perform machine learning by including public cohort big data on public clinical information provided by a medical information providing institution such as NHIC or NHIC, in addition to its own clinical information in each of the medical institutions , Which can extract more accurate and reliable clinical prediction models.

Also, the machine learning engine 140 may be implemented by applying different machine learning techniques according to the characteristics of each medical institution, such as a special field or an application range.

On the other hand, in the case where one machine learning technique is applied to the machine learning engine 140, since it is not possible to derive an optimal result in all cases, it is possible to use a combination of a plurality of machine learning techniques, By extracting the prediction model, more accurate and reliable clinical prediction results can be output.

Further, the clinical prediction model extracted by the machine learning engine 140 is input to the clinical information of the specific patient received from the ensemble prediction system 120 of the medical institution requesting the clinical prediction result, And outputs the clinical prediction result. That is, the clinical prediction result is information output based on a case similar to the clinical information, using the clinical information of the patient, which is numerical information predicting the progress of the specific patient.

Also, the ensemble prediction system 120 transmits the output clinical result to the ensemble prediction system 120 of the requested medical institution. At this time, according to the clinical prediction result, it is possible to transmit, together with the clinical prediction result, the guidelines of the treatment or prescription suggested by the medical institution or the opinion of the expert on the disease.

Further, the ensemble prediction system 120 receiving the clinical prediction results from the plurality of external medical institutions integrates the received plurality of clinical prediction results, performs decision ensemble prediction, and outputs the performed ensemble prediction results to the clinical And provides it to the decision support system 110.

Wherein the ensemble prediction includes at least one clinical prediction result received from at least one external medical institution based on the clinical information of the patient and a clinical prediction result output from the specific medical institution requesting the clinical prediction result, Means the prediction of the current state of the patient and the progress of the disease in the future reflecting the characteristics of the patient for clinical decision making. This process is more effective than performing a clinical prediction result for a specific patient in a single medical institution. In particular, it is possible to provide a more accurate and improved reliability of the patient's current And the progress of disease progression can be predicted. In addition, it is possible to utilize valuable new information about the disease found in the external medical institution, and thus a reliable ensemble prediction result can be derived.

Also, the accuracy of the machine learning engine 140 configured in the clinical decision support ensemble system 100 of each of the foreign bodies is improved as the number of big data to be machine-learned in the machine learning engine 140 increases.

Further, the clinical decision support system 110 provides clinical decision information to the healthcare provider using the prediction result received from the ensemble prediction system 120. [

Further, the clinical decision information includes a diagnosis of the current state of the patient and a progress of the future patient state, a configuration for an optimal treatment method such as surgery or drug prescription, a prescription, a warning for prescription, or a combination thereof . This provides a guideline for the current state and the future state of the patient when the medical staff performs a medical operation such as an operation for diagnosing and treating a disease based on clinical information collected from the patient, By supporting decision-making on accurate medical actions, misdiagnosis of the medical person is minimized, and the patient can receive objective medical treatment.

In addition, the clinical decision support ensemble system 100 receives the biometric information of the outpatient from the IoT device provided in the outpatient, monitors the outpatient state, and determines the state of the outpatient based on the biometric information Based on the results of the ensemble prediction of the outpatients, the detection and prediction diagnosis of the state change of the outpatient can be made so that the early detection of the emergency situation and the quick response thereof are possible, and the clinical prediction result is provided to the IoT device.

Further, the IoT device includes a biosensor and a wireless communication terminal, and the biosensor is a wearable device attached to the body of the outpatient, and the blood pressure, blood sugar, hormone, electrocardiogram, exercise Information such as information or a cholesterol level, and life log information including various biometric information of the outpatient.

In addition, the biosensor transmits the measured various biometric information of the outpatient to the clinical decision support ensemble system 100 in real time or periodically in cooperation with the mobile service. Accordingly, the clinical decision support ensemble system 100 can periodically or real-time monitor and manage the state of the patient, perform ensemble prediction based on life log information including the biometric information, And the like can be predicted and prevented from occurring, and the predicted ensemble prediction result can be provided to the wireless communication terminal.

FIG. 3 is a block diagram illustrating a configuration of a clinical decision support ensemble system 100 according to an embodiment of the present invention.

3, the clinical decision support ensemble system 100 includes a clinical information system 150 for collecting life log information of outpatients and clinical information of an inpatient or an inpatient, A clinical learning DB system 130 for storing and managing the clinical information, a machine learning engine 140 for performing deep layer learning on the stored life log and clinical information, an ensemble prediction system 120 for performing decision ensemble prediction And a clinical decision support system 110 for generating clinical decision information of a medical care action for a patient based on the performed ensemble prediction and providing a medical care provider.

In addition, the outpatient includes a wearable bio-sensor to measure, collect and store life log information including biometric information and activity information of the patient in real time or periodically, and to transmit the collected life log information to the clinical information system (150) in real time or periodically, and the transmission is performed through a mobile service interlocked with the biosensor.

Meanwhile, the mobile service provides the patient with a healing platform 200, which can transmit the life log to the clinical information system 150 through the healing platform 200.

Also, the healing platform 200 may store the personal health record (PHR) of the individual patient in a personal repository so that the individual can manage the initiative. On the other hand, the personal repository may be a local repository or a private cloud repository.

Also, the personal health record includes the life log information and the electronic medical record stored in the medical institution, and the outpatient transmits the electronic medical record through the healing platform 200 from the medical institution, Can be provided. Accordingly, the outpatient manages the personal health record of the patient, and can continuously receive the medical services provided by the plurality of medical institutions without duplication.

Meanwhile, the life log information may be collected not only from an outpatient but also from an inpatient or an inpatient.

Also, when the collected clinical information and the life log information are collected from the same patient, the granular information system 150 can map and store the collected clinical information and the life log information for each patient.

In addition, the clinical information system 150 may include an order communication system (OCS) (not shown), a picture archiving and communication system (PACS) An electronic medical record (EMR) (not shown) (not shown), and a laboratory information system (LIS) (not shown) (not shown).

The OCS is a system in which a medical person diagnoses a patient and then transmits a prescription issued to a corresponding department through a local communication network to organically connect a doctor's prescription or a nurse's nursing behavior to each department , Reduce miscommunication between doctors and nurses and other related departments, and reduce mistakes in prescribing delivery. Accordingly, the clinical information system 150 maps information (OCS information) about the prescription of the medical care person by patient and stores the information in the clinical clinical information database 132.

In addition, the PACS is a medical image comprehensive management system that acquires medical image information (PACS information) of a patient from a medical imaging device and then performs storage, management and transmission through a high-speed network. So that the medical image information can be efficiently managed. Accordingly, the clinical information system 150 maps the medical image information of the patient to each patient, and stores the medical image information in the clinical clinical information database 132.

In addition, the EMR is a system for describing all items related to a patient's disease in a computerized electronic medical record. The electronic medical record includes all the matters related to the patient's disease, , And the electronic medical record serves as a basis for providing continuous and consistent medical service to the patient. That is, when the electronic medical record is digitized and the patient is consulted, the user can immediately use the electronic medical record through the terminal of the medical person, thereby solving the problem due to the waiting time. Further, convenience for the patient and reliability of the medical institution can be increased, Can be provided. Accordingly, the clinical information system 150 maps the electronic medical records according to patients and stores them in the hospital clinical information database 132.

The LIS is a system that automates inspection equipment provided in a medical institution, and provides information promptly and accurately to a patient and a healthcare provider as a result of inspection through the inspection equipment. Such inspection result information may be recorded in the electronic medical record or independently stored. Accordingly, the clinical information system 150 maps the test result information to each patient and stores the information in the clinical clinical information database 132.

Also, the clinical information DB system 130 includes a clinical information database (not shown), and the clinical information database includes a patient life log for storing and managing life log information collected from the outpatient, A database 131, a hospital clinical information database 132 for storing and managing hospital clinical information collected from the clinical information system 150, and the collected patient life log and hospital clinical information to the machine learning engine 140, And a medical data learning big data database 133 for storing and managing big data for medical information learning converted into a common format so that it can be easily learned.

Meanwhile, the Big Data for medical information learning is medical information including hospital clinical information including EMR, PACS, OCS and LIS, life log information, or a combination thereof, Data. Also, the machine learning engine 140 may be configured to generate the medical education information on the basis of a numerical value (age range, biometric data, diabetes, electrocardiogram), a treatment process, a prescription, a preference (presence or absence of smoking, And the like are automatically extracted and abstracted to perform learning.

The life log database 131 stores life log information of outpatients collected in real time or periodically by the clinical information system 150.

The hospital clinical information database 132 also stores OCS information (prescription information), PACS information (medical image information), electronic medical records and LIS information (test result information) for each patient from the clinical information system 150, , And the LIS information may be described in the electronic medical record.

The stored life log information and the hospital clinical information are mapped and stored for each patient.

Also, the clinical information DB system 130 may be configured to store the life log and the hospital clinical information in a common medical format so that the machine learning engine 140 can integrate and learn the stored life log information and the hospital clinical information. Generates big data for information learning and stores it in the big data database 133 for medical information learning.

On the other hand, the common format is the HL7 standard, such as VMR (virtual medical record), eVMR (extendible VMR), or arden syntax, which allows other medical institutions to use the common format. Of course, the common format is not limited to the above-described VMR, eVMR, or arden syntax, but can be any format as long as it is a standard format data format commonly used by the plurality of medical care periods.

Also, the Big Data for medical information learning can be converted and managed into a format of the medical institution, and the clinical prediction result received from the plurality of external medical institutions can be converted into the format of the medical institution and used.

Also, the machine learning engine 140 mechanically learns the stored medical information learning big data through deep learning, extracts a clinical prediction model, which is a multi-layered perceived network with high reliability, based on the machine learning learning data, Based on the medical data learning big data inputted thereto, it is possible to determine the medical characteristics of each medical condition (for example, the age, the numerical value of the biometric data, the diabetic value, the electrocardiogram value, the treatment process, the prescription, Presence or absence), a treatment process, a change in a patient's state, and the like) are automatically extracted and abstracted, and machine learning is repeatedly performed. Thus, the clinical prediction model performs a clinical prediction on the future progress of the disease, and outputs a clinical prediction result. That is, the machine learning engine 140 abstracts the characteristics of the clinical information, which is the medical information including the electronic medical record (EMR), the PHR including the life log information, the medical image, or a combination thereof, The data is elaborated and the clinical prediction model is extracted to improve the reliability of the clinical prediction results by early prediction of future progress and risk of the disease or disease of a particular patient.

The machine learning also includes artificial intelligence techniques that perform learning, reasoning, prediction, or a combination thereof for multidimensional analysis of large data including medical records, lifelog information, or combinations thereof for patient-centered personalized care, And to output the results of the clinical prediction.

Further, the machine learning engine 140 can generate a more accurate and reliable clinical prediction model as the quantity of the medical information learning big data increases, and when the medical information learning big data is added, the machine learning engine 140 140 may automatically update the clinical prediction model so that it can repeatedly perform machine learning every time the medical information learning big data is added and output more accurate clinical prediction results.

Also, the machine learning performed by the machine learning engine 140 can output a clinical prediction result at a high speed using a parallel cluster, and a high reliability numerical information As a clinical prediction result.

Also, the medical information learning big data is the clearing information big data having the time-series characteristic, and the patient's personal information such as name, resident number or address is deleted.

In addition, the machine learning engine 140 can learn by each disease (for example, diabetes, cardiovascular disease) or can simultaneously learn by combining a plurality of diseases (complications of diabetes and diabetes, cardiovascular diseases and other chronic diseases, etc.) have. This is because the patient has multiple diseases at the same time as the patient has one disease. For example, most people with diabetes have complications of diabetes at the same time.

Accordingly, the clinical prediction model is generated through at least one of the machine learning engines 140, and may be variously generated according to the expertise and characteristics of each medical institution. For example, if a particular medical institution is a cancer specialist, a clinical prediction model can be generated according to the type of cancer.

In addition, the input of the generated clinical prediction model is clinical information from which personal information of a specific patient to be treated is deleted, and the output of the clinical prediction model is a clinical prediction result of the future progress of the disease of the specific patient.

The ensemble prediction system 120 includes a decision ensemble prediction unit 121 for performing ensemble prediction for decision making, a collaborative hospital management unit 122 for managing hospital information for collaborating with a plurality of external medical institutions, And a CDSS interface unit for requesting the clinical prediction result of the patient to the ensemble predicting system 120 constructed in the external financial institution and receiving the same, and a CDSS interface unit for communicating with the clinical decision support system 110 .

In addition, the decision ensemble predicting unit 121 may use the clinical prediction model generated through the machine learning engine 140 to perform a clinical prediction of the patient when receiving a clinical prediction request of a specific patient from a plurality of external medical institutions Print the result.

In addition, the decision ensemble predicting unit 121 transmits the output of the clinical prediction to the medical institution requested for the clinical prediction, and the clinical clinical information and the life log, which generate, scale and manage the clinical prediction result, And outputted based on the transformed medical information learning big data.

In addition, the decision ensemble predicting unit 121 may perform a clinical prediction request for a specific patient with a plurality of external medical institutions. At this time, the patient's personal information may provide the deleted clinical information, .

If the requested clinical prediction result is received from the plurality of external medical institutions, the decision ensemble predicting unit 121 integrates the received plurality of clinical prediction results and self-output clinical prediction results, Performs ensemble prediction for decision making based on clinical information.

The ensemble prediction is a prediction of the future progress of the disease of the specific patient. The ensemble prediction combines the clinical prediction result output by itself and the clinical prediction result received from the plurality of medical institutions to output a more accurate and reliable ensemble prediction result . It is possible to acquire new information about the disease found in the plurality of external medical institutions and to integrate a plurality of prediction results to output an improved accuracy and a reliable ensemble prediction result. That is, the decision ensemble predicting unit 121 predicts the results of the clinical prediction based on the machine learning using the big data for the self-medical information learning of the specific medical institution and the machine learning for the big data for medical information learning of each medical institution from at least one external medical institution Based clinical prediction results, performs ensemble prediction, and provides the ensemble prediction results to the clinical decision system 110 to support the decision making of the medical person having improved accuracy and reliability.

The plurality of external medical institutions may transmit the clinical prediction result output by itself to the medical institution having the clinical prediction request, together with the opinions of the medical professional of the medical institution, and may transmit the clinical prediction request to the medical institution The medical practitioner synthesizes the opinions of the above-mentioned medical professionals and provides accurate medical services to the patients.

Further, the decision ensemble predicting unit 121 transmits the output decision ensemble prediction to the clinical decision support system 110, and the clinical decision support system 110 performs the decision making And to generate clinical decision information so that the patient can

The collaborative hospital management unit 122 stores, updates, deletes, or manages hospital information for collaboration between the medical institution and a plurality of external medical institutions. The hospital information is managed by each medical institution, Network address, location, size, or any combination thereof.

Also, the ensemble prediction system 120 requests a clinical prediction result for a specific patient through the other institution interface or receives the clinical prediction result.

The other organization interface may be configured with a plurality of virtual interfaces to directly perform communication with the respective external medical institutions, or may be configured with a single communication interface to perform communication with the plurality of medical institutions.

The CDSS interface is an interface for transmitting and receiving data between the ensemble prediction system 120 and the clinical decision support system 110.

In addition, the clinical decision support system 110 may be configured to perform a diagnosis, a treatment strategy, a prescription and prevention of the patient based on the ensemble prediction result received from the ensemble prediction system 120, And generates and provides clinical decision information, which is an overall guideline on the management method.

In addition, the clinical decision information is preferably generated by a rule-based knowledge base, but may also be generated by a non-knowledge based knowledge base.

In addition, the clinical decision information may be classified into a decision tree, a neutral network, a Naive Bays, or a combination thereof based on the rule-based knowledge base, the individual, the disease, , ≪ / RTI > or a combination thereof.

Meanwhile, the rule knowledge base generates the decision information based on a specific rule set by an operator who operates the clinical decision support ensemble system 100, and the rule can be set differently according to the operator.

4 is a diagram for explaining a process of performing machine learning for each medical institution through a machine learning engine according to an embodiment of the present invention.

As shown in Fig. 4, machine learning for each medical institution is performed through a machine learning engine 140 provided for each medical institution.

The process for performing the machine learning first extracts individual clinical information data from the clinical information DB 132 of the medical institution to generate learning data for machine learning.

The generated learning data is stored in the medical data learning large database 133 for each hospital, and becomes input data for machine learning.

The learning data may also include numerical values for individual health characteristics such as blood glucose, hemoglobin, HMG, ALANine Transaminase (ALT), and disease information (e.g., presence or absence of disease) .

Further, since the learning data is generated based on the clinical data of the patient stored and stored for each medical institution, it is natural that personal information (e.g., address, resident registration number, telephone number, etc.) of each patient is deleted and generated.

On the other hand, the learning data may be generated based on not only clinical data of patients accumulated for each medical institution but also public cohort clinical data provided by the NHIC or the HIRA, or may be data provided by other medical institutions or medical information providers have.

Next, we design a deep learning network structure for disease prediction in the hospital. The structure of the deep learning network may be designed differently depending on the characteristics of each hospital such as the specialty fields of each hospital (e.g., cancer specialist hospital, internal medicine, pediatrics, etc.) and main patient type.

Next, the learning data is input to the designed deep learning network so that the deep learning network can learn the learning data.

The final output of the deep learning network is also the result of a clinical prediction that predicts future disease progression for certain diseases (eg, hypertension, diabetes, cardiovascular, etc.).

Hereinafter, a process of performing ensemble learning for ensemble prediction according to another embodiment of the present invention, outputting ensemble clinical prediction results for a specific patient using the same, and providing clinical decision information including the results 5 and FIG.

FIG. 5 is a diagram illustrating a procedure for performing ensemble learning for estimating an entanglement in a clinical decision support ensemble system according to another embodiment of the present invention.

In addition, the clinical decision support ensemble system 1001 can generate an ensemble clinical prediction result through machine learning as performed in each hospital.

5, the clinical decision support ensemble system 1001, which performs ensemble learning to generate an ensemble clinical prediction result for a specific patient, first obtains a plurality of patients from the hospital clinical information DB 1321, And extracts clinical information data to generate learning data for each patient.

The generated learning data for each patient is stored in the ensemble learning data DB 1331.

The learning data may be composed of a matrix as shown in Fig. 5, or may be composed of a numeric vector, and include numerical values and disease information (may be indicated by a class or label) by health characteristics.

Next, the clinical decision support ensemble system 1001 transmits the generated learning data for each patient to a plurality of other medical institutions or medical information providing agencies to request a prediction result of the learning data.

At this time, the learning data (vector or matrix form) transmitted is a numerical value according to the health characteristic excluding the disease information, and the patient's personal information is deleted and replaced with another identifier (e.g., ID).

Next, each of the medical institutions or the medical information institutions that have received the learning data from the clinical decision support ensemble system 1001, learns by learning data of each hospital through a deep learning machine learning process for each medical institution or medical information institution And inputs the received learning data to the deep learning network (i.e., prediction model) to output the prediction result data.

Then, each of the medical institutions or the medical information agency transmits the output prediction result data to the clinical decision support ensemble system 1001 that requested the prediction result.

On the other hand, the prediction result data transmitted to each medical institution or the medical information provider is an output value of a prediction model for each medical institution or medical information provider, and can be configured in the form of a numeric vector. In addition, the numerical vector may be represented by a disease-specific risk (probability vector) and may be an abstract numerical vector that does not map 1: 1 with a specific disease due to the nature of deep learning (machine learning).

Next, the clinical decision support ensemble system 1001, which receives prediction results from the respective medical institutions and the medical information providing agencies, integrates the prediction results of itself and the prediction results received from the respective medical institutions or the medical information providing agencies And based on this, the final ensemble learning data is extracted and stored in the ensemble learning data DB 1331.

The generated ensemble learning data includes numerical values and disease information for each patient's health characteristics, and prediction results for each hospital.

Next, the clinical decision support ensemble system 1001 designs an ensemble deep learning network structure for ensemble learning.

On the other hand, the design of the ensemble deep learning network structure is based on the input data dimension, the output data dimension, the number of layers of the deep learning network, the number of nodes in each deep learning network layer, Min-batch size, Acting function type, and the like).

Next, the ensemble predicting system 1201 performs pre-training on the ensemble deep learning network using ensemble learning data.

The data input to the ensemble deep learning network for the pre-learning is prediction results for each patient included in the ensemble learning data for each medical institution or medical information provider.

The preliminary learning is to learn the weights in the ensemble deep learning network in advance according to the characteristics of data using data without disease information and to prevent the ensemble deep learning network from overfitting due to learning data do.

On the other hand, the pre-learning may be performed using a CD (Contribute Divergence) technique for each layer of the ensemble deep learning network based on a Restricted Bolzmann Machine (RBM), which is one of the machine learning methods. However, the present invention is not limited to RBM and CD.

Next, the clinical decision support ensemble system 1001 performs fine-tuning on the ensemble deep learning network using the ensemble learning data.

The data input for the fine adjustment learning is the analysis result of each medical institution or medical information provider according to each patient and the disease information according to each patient.

The fine adjustment learning is performed to re-adjust the weights of the ensemble deep learning network that has been pre-learned to match the characteristics of the input data so as to perform prediction of the disease.

5, the ensemble learning process is described separately from the hospital A and the hospital B. However, the ensemble learning process may be performed in the hospital A or the hospital B and may be performed in another hospital (e.g., the hospital C) .

FIG. 5 shows the ensemble learning based on the prediction results of two hospitals (hospitals A and B), but it is obvious that the number of hospitals (that is, medical institutions or medical information providers) Do.

6 is a view for explaining a process of outputting ensemble prediction results of a specific patient through ensemble learning and providing clinical decision information to a user according to another embodiment of the present invention.

As shown in FIG. 6, the process of outputting the ensemble prediction result of a specific patient through ensemble learning and providing the clinical decision information to the user includes first transmitting clinical data for a specific patient to another medical institution or a medical information provider And requests a prediction result.

The clinical data for the specific patient transmitted to the other medical institutions or the medical information providing institution is data in which the disease information and the patient's personal information are deleted.

Further, each of the medical institutions or the medical information providing organizations, which have received the clinical data for the specific patient, transmits the clinical data of the received specific patient to each of the medical institutions or the medical information providing organizations, To output a prediction result for the patient and transmits the prediction result to the requested clinical decision support ensemble system 1001. [

On the other hand, the predicted results for each medical institution or medical information provider may be a risk (probability) value for each type of disease.

Next, the clinical decision support ensemble system 1001 integrates predicted self-predicted results using the prediction results received from the plurality of medical institutions and the clinical data for the specific patient, and inputs them to the ensemble deep learning network And outputs the ensemble prediction result.

In addition, the clinical decision support ensemble system 1001 may combine the prediction results received from a plurality of medical institutions or medical information providers with the self-prediction results to form a single vector, and then input to the ensemble deep learning network So that the ensemble prediction result can be output.

Also, the end result of an ensemble deep learning network can be composed of disease-specific risk (probability vector). This can vary depending on how the deep learning network is configured. For example, if a deep learning network and an ensemble deep learning network are established for each disease, the final clinical decision information provides only the risk (probability value) for the disease.

Next, the clinical decision support ensemble system 1001 transmits the clinical decision information including the final result to the terminal of the user (medical person) so that the user can use the clinical decision information.

7 is a block diagram illustrating a configuration of a clinical decision support ensemble system according to another embodiment of the present invention.

7, the clinical decision support ensemble system 100 according to the present invention is described in detail with reference to FIGS. 1 to 3. The clinical decision support ensemble system 100, The system 110 is configured as a part of a hospital information system of a medical institution (e.g., a hospital), and the ensemble prediction system 120 can be configured in the form of a platform or a server on the Internet.

Meanwhile, the hospital information system refers to a system for managing overall hospital information such as finance or accounting of the medical institution including the clinical information system 150.

Also, the machine learning engine 1401 of the medical institution learns using the hospital big data managed by the hospital information system, and the hospital big data includes PHR, life log information, EMR, medical image, or a combination thereof . That is, the machine learning engine 1401 collects and learns PHR, lifelog information, EMR, medical images, etc. for a plurality of patients, thereby generating at least one or more clinical prediction models, Of the patient.

When the healthcare provider requests the clinical decision support system 110 for a specific patient, the hospital information system may communicate with the ensemble prediction system 120 on the Internet via a communication interface The clinical information is deleted and the clinical prediction result is requested.

When the requested clinical prediction result is received from the ensemble prediction system 120, the machine learning engine 1401 converts the requested clinical prediction result into a common format of data used by the medical institution through the data conversion unit 160, The learning engine 1401 provides the clinical decision support system 110 with the clinical prediction results of the ensemble prediction system 120 received from the data conversion unit 160 and the self-output clinical prediction results.

In addition, the clinical decision support system 110 provides clinical decision information to the healthcare provider based on the provided clinical prediction results.

The ensemble prediction system 120 on the Internet may include a standard-based data conversion unit 170, a decision ensemble predicting unit 121, and a machine learning engine 1402.

The ensemble prediction system 120 receives the request for the clinical prediction result from the hospital information system and the clinical information of the patient, and transmits the received clinical information to the standard-based data converter 170 in the standard format Conversion. The standard format is the VMR, eVMR or Arden syntax of the HL7 standard.

The decision ensemble predicting unit 121 provides the converted clinical information of the patient to the machine learning engine 1402. The machine learning engine 1402 provides the clinical prediction information of the patient to the extracted clinical prediction model And outputs the clinical prediction result to the decision ensemble predicting unit 121.

Meanwhile, the machine learning engine 1402 learns based on public cohort big data to extract a clinical prediction model, and the public cohort big data means public clinical information periodically provided from a public medical institution. Through this, the machine learning engine 1402 can learn a lot of public clinical information, and can output more accurate and reliable clinical prediction results. The description of the learning has been described above with reference to FIG. 3, and a detailed description thereof will be omitted.

Also, the decision ensemble predicting unit 121 performs ensemble prediction based on the clinical information of the patient and the clinical prediction result received from the medical institution, and provides the ensemble prediction result to the medical institution requesting the prediction result do.

On the other hand, when the ensemble prediction system 120 receives a request for a clinical prediction result from a specific medical institution, the ensemble prediction system 120 can request a plurality of external medical institutions except the specific medical institution for a clinical prediction result of the patient, And concurrently outputting the clinical prediction results based on the public cohort big data, thereby performing decision ensemble prediction.

As described with reference to FIG. 7, the clinical decision support ensemble system 100 of the present invention can be implemented in various forms, and can be applied to various fields requiring decision-making as well as medical decision-making.

8 is a flowchart illustrating a procedure for generating clinical decision information according to an embodiment of the present invention and providing the generated clinical decision information to a terminal of a medical person.

As shown in FIG. 8, the procedure for generating clinical decision information for a specific patient and providing clinical decision information to the medical care person includes, first, when there is a request for decision information on a specific patient from the medical care provider (S110 , And requests the ensemble prediction system 120 for the ensemble prediction result through the clinical decision support system 110 (S120).

The clinical decision support system 110 provides the clinician with clinical decision information, which is a guideline for a series of medical actions, such as diagnosis, prescription or configuration of treatment methods for a particular patient, and the clinical decision support ensemble Based on a rule-based knowledge base or a non-rule-based knowledge base, depending on the operator who operates the system 100.

Next, the patient's clinical prediction result is requested and received through a plurality of external medical institutions through the ensemble prediction system 120 (S130).

On the other hand, when requesting the clinical prediction result to the plurality of external medical institutions, the patient's personal information is provided with the deleted clinical information, and each of the external medical institutions generates a clinical prediction result for the patient itself, And transmits the prediction result to the requesting medical institution.

The clinical prediction result is output through a clinical prediction model obtained by extracting big data for medical information learning from the machine learning engine 140 of the clinical decision ensemble supporting system 100 built in each of the medical institutions.

Next, the clinical prediction result of the patient is output using the clinical prediction model generated through the machine learning engine 140 (S140).

Meanwhile, the result of the clinical prediction in the step S130 means a result of the clinical prediction generated in the plurality of external medical institutions. The result of the clinical prediction in the step S140 is a result of the clinical prediction generated in the corresponding medical institution requesting the clinical prediction result from the external medical institution. it means.

Next, the ensemble prediction system 120 integrates the received plurality of clinical prediction results and the self-generated clinical prediction results based on the clinical information of the patient to perform ensemble prediction (S150).

The ensemble prediction is performed by interlocking between a plurality of medical institutions, integrating the clinical prediction result of the patient generated by the plurality of medical institutions and the clinical prediction result generated by the medical institution, and based on the result, To accurately predict the progress of the patient, and to allow the medical person to perform the medical care action for the patient accurately and promptly.

Next, the ensemble prediction system 120 transmits the ensemble prediction result to the clinical decision support system 110 (S160). The clinical decision support system 110 receives the ensemble prediction result And transmits the generated clinical decision information to the terminal of the healthcare provider (S170).

The clinical decision information is a guideline that includes information on diagnosis, prediction, prescription, treatment strategy, future management method, or a combination of the current state of the patient and progress of the disease in the future.

As described above, the clinical decision support ensemble system of the present invention performs clinical prediction for a specific patient in cooperation with a plurality of medical institutions, and thereby, based on the improved accuracy and reliability, And supports clinical decision on diagnosis, prescription, treatment strategy, or future management method of the predicted current state of the specific patient and progress of the disease in the future, thereby performing quick and accurate medical action of the medical person There is an effect to be able to.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is to be understood that the invention is not limited to the disclosed exemplary embodiments. .

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is to be understood that the invention is not limited to the disclosed exemplary embodiments, but, on the contrary, It will be understood by those skilled in the art that various changes and modifications may be made without departing from the spirit and scope of the present invention.

100, 1001: Clinical decision support ensemble system
110: Clinical decision support system
120: Ensemble prediction system 121: Decision ensemble prediction unit
122: Collaboration hospital management part 130: Clinical information DB system
131: patient life log DB 132, 1321: hospital clinical information DB
133: Big data DB for medical information learning 1331: Ensemble learning data DB
140, 1401, 1402: machine learning engine
150: clinical information system 160: data conversion unit
170: Standards Based Data Changer

Claims (34)

Wherein the clinical decision support information is provided by integrating the clinical prediction results provided from a plurality of medical institutions, a medical information providing agency, or a combination thereof, and performing ensemble prediction. The method according to claim 1,
In the ensemble prediction,
Wherein the clinical decision support ensemble system is performed on the basis of a clinical prediction result by machine learning using the big data for self-medical information learning from a plurality of medical institutions. The method of claim 2,
In the ensemble prediction,
Based on the results of clinical learning based on machine learning for big medical data for self -
Wherein the clinical decision support ensemble system comprises at least one external medical institution to integrate the machine learning based clinical prediction results of the medical information learning big data of each medical institution to support clinical decision making. The method according to claim 1,
The clinical decision information comprises:
Based on a rule-based knowledge base;
A decision tree, a neural network, a Naive Bayes, or a combination thereof depending on the individual, the disease, or a combination thereof; or
Lt; RTI ID = 0.0 > a < / RTI > combination of these. The method of claim 2,
The machine-
Clinical decision making characterized by including artificial intelligence techniques that perform learning, reasoning, prediction, or a combination thereof for multidimensional analysis of big data including medical records, life logs, or combinations thereof for patient- Supported ensemble system. The method of claim 2,
The machine-
Wherein the clinical decision support ensemble system is characterized by outputting numerical information of high reliability extracted or integrated through high-speedization using a parallel cluster and quantification of big data for medical information learning as a clinical prediction result. A clinical decision support system that provides clinical decision information in response to a user's clinical decision request; And
And an ensemble prediction system that provides a clinical prediction result at the request of the clinical decision support system,
Wherein the ensemble prediction system comprises an ensemble prediction for decision making by integrating clinical prediction results provided from a plurality of medical institutions, a medical information providing agency, or a combination thereof. The method of claim 7,
Wherein the clinical decision support ensemble system comprises:
And a machine learning engine that includes deep learning to perform learning using the big data from the clinical information database. The method of claim 8,
Wherein the clinical information database includes hospital clinical information of an individual hospital and lifelog information of an outpatient. The method of claim 8,
Wherein the machine learning engine comprises:
Abstracts the characteristics of medical information from big data including EMR, PHR, medical image, life log information, or a combination thereof, extracts a prediction model by learning the big data, And the reliability of the prediction result is improved. The method of claim 7,
The ensemble prediction system includes:
An interface for making a prediction request to an ensemble prediction system existing in a plurality of external medical institutions and receiving a prediction result from an external medical institution; And
And a collaborative hospital management unit for managing hospital information for collaborating with the plurality of external medical institutions. The method of claim 7,
The above-
The ensemble prediction system, the machine learning engine, and the medical data learning big data. The method of claim 7,
Wherein the clinical decision support ensemble system comprises:
Wherein the bio-information data of the patient including the life log of the patient is collected, and the condition of the patient is continuously monitored and managed to be provided as an IoT device. Performing ensemble prediction by integrating a plurality of clinical prediction results; And
And providing clinical decision information with improved accuracy through the ensemble prediction. 15. The method of claim 14,
In the ensemble prediction,
Wherein the clinical decision support method is performed based on a clinical prediction result by machine learning using a plurality of medical institutions' own big data for medical information learning. 16. The method of claim 15,
In the ensemble prediction,
Based on the results of clinical learning based on machine learning for big medical data for self -
The method comprising the steps of: integrating a machine learning-based clinical prediction result of the medical information learning large data of each medical institution from at least one external medical institution to support clinical decision making. 15. The method of claim 14,
The clinical decision information comprises:
Based on a rule-based knowledge base;
A decision tree, a neural network, Naive Bays, or a combination thereof depending on the individual, the disease, or a combination thereof; or
Or a combination thereof. ≪ RTI ID = 0.0 > 11. < / RTI > 16. The method of claim 15,
The machine-
Clinical decision making characterized by including artificial intelligence techniques that perform learning, reasoning, prediction, or a combination thereof for multidimensional analysis of big data including medical records, life logs, or combinations thereof for patient- How to Apply. 16. The method of claim 15,
The machine-
And outputting the high reliability numerical information extracted or integrated through the quantification of the big data for the medical information learning as the clinical prediction result. Providing clinical decision information according to a user's clinical decision request through a clinical decision support system; And
And providing a clinical prediction result at the request of the clinical decision support system through an ensemble prediction system,
Wherein the ensemble prediction system integrates the clinical prediction results provided from a plurality of medical institutions, a medical information provider, or a combination thereof to perform ensemble prediction for the decision. The method of claim 20,
Wherein the clinical decision support method comprises:
And performing learning using the big data from the clinical information database through the machine learning engine. 23. The method of claim 21,
Wherein the clinical information database includes hospital clinical information of individual hospitals and life log information of outpatients. 23. The method of claim 21,
Wherein the machine learning engine comprises:
Abstracts the characteristics of medical information from big data including EMR, PHR, medical image, life log, or a combination thereof, extracts a clinical prediction model by learning the big data, And the reliability of the prediction result is improved. The method of claim 20,
The ensemble prediction system includes:
An interface for making a prediction request to an ensemble prediction system existing in a plurality of external medical institutions and receiving a prediction result from an external medical institution; And
And a collaborative hospital management unit for managing hospital information for collaborating with the plurality of external medical institutions. The method of claim 20,
The above-
The ensemble prediction system, the machine learning engine, and the medical data learning big data. The method of claim 20,
Wherein the clinical decision support method comprises:
Wherein the bio-information data of the patient including the life log information of the patient is collected, and the state of the patient is continuously monitored and managed to provide the bio-information data to the IoT device. The predicted result predicted by another medical institution or the medical information providing institution is inputted by inputting the learning data provided by the specific medical institution or the medical information providing institution and integrated with the prediction result predicted by the specific medical institution or the medical information providing institution, Extracting the ensemble learning data, and performing ensemble learning using the extracted ensemble learning data. 28. The method of claim 27,
To receive the prediction result,
When a specific medical institution or a medical information providing agency generates learning data and provides the learning data to another medical institution or a medical information providing agency,
The other medical institution or the medical information providing agency provides a prediction result obtained by inputting the learning data to the specific medical institution or the medical information providing institution by utilizing a prediction model held by the other medical institution or the medical information providing agency, The method comprising the steps of: (a) providing an ensemble learning function to a user; 28. The method of claim 27,
In the ensemble learning,
The pre-learning is performed using the prediction results of the other medical institutions or the medical information provider,
And performing fine adjustment learning by further using disease information in the prediction result. The prediction result obtained by inputting the self-prediction result predicted by the specific medical institution or the medical information provider using the clinical data of the specific patient and the clinical data of the specific patient provided from the other medical institution or the medical information provider And performing ensemble prediction using the integrated prediction result. The method for assisting clinical decision making according to ensemble prediction. The predicted result predicted by another medical institution or the medical information providing institution is inputted by inputting the learning data provided by the specific medical institution or the medical information providing institution and integrated with the prediction result predicted by the specific medical institution or the medical information providing institution, Extracting ensemble learning data, and performing ensemble learning using the extracted ensemble learning data. 2. The ensemble system according to claim 1, 32. The method of claim 31,
To receive the prediction result,
If the learning data is generated by a specific medical institution or a medical information providing agency and the learning data is provided to another medical institution or a medical information providing agency,
The other medical institution or the medical information providing agency provides a prediction result obtained by inputting the learning data to the specific medical institution or the medical information providing institution by utilizing a prediction model held by the other medical institution or the medical information providing agency, Wherein the ensemble learning support system comprises: 32. The method of claim 31,
In the ensemble learning,
The pre-learning is performed using the prediction results of the other medical institutions or the medical information provider,
And performing fine adjustment learning by further using disease information in the prediction result. The ensemble system for ensuring clinical decision support by ensemble learning. The prediction result obtained by inputting the self-prediction result predicted by the specific medical institution or the medical information provider using the clinical data of the specific patient and the clinical data of the specific patient provided from the other medical institution or the medical information provider And performing ensemble prediction using the integrated prediction result. The ensemble system according to claim 1,

Images