WO2019225910A1

WO2019225910A1 - Method and system for predicting and analyzing stroke severity using nihss

Info

Publication number: WO2019225910A1
Application number: PCT/KR2019/005922
Authority: WO
Inventors: 유재학; 박홍규; 김다미; 홍승희; 박세진; 김선진; 유성규
Original assignee: 한국표준과학연구원; 한국전자통신연구원
Priority date: 2018-05-21
Filing date: 2019-05-17
Publication date: 2019-11-28
Also published as: KR20190132710A

Abstract

A method for prediction and in-depth and accurate analysis of stroke severity according to an embodiment of the present invention may include a step of performing prediction and in-depth analysis of stroke severity on the basis of real-time NIHSS data received by a patient device and a medical data server, and health medical examination data received by the medical data server, and include a comprehensive medical service function for supporting a patient in getting alarm service or prompt treatment depending on stroke severity and risk.

Description

[Correction by Rule 26 09.07.2019] Method and system for predicting and analyzing stroke severity using NIHSS

The present invention relates to a method and system for predicting and analyzing stroke severity using NIHSS, and more particularly, to assess initial damage of stroke using machine learning and data mining based on NIHSS data of a patient, and to further analyze A method and system that can be performed.

According to a recent survey, there were 275,895 deaths in South Korea in 2015, including malignant neoplasms (76,855), heart diseases (28,326), and cerebrovascular diseases (24,455). have. In particular, the mortality rate of cerebrovascular disease in Koreans is 48 per 100,000, followed by malignant neoplasms and heart disease. The number of deaths from cerebrovascular disease in Korea has been on the decline since 2005, but still ranks second among the causes of death from single diseases caused by malignant tumors. In particular, the mortality rate due to cardiovascular disease and cerebrovascular disease continues to increase in elderly people over 60 years old.

Stroke is a fatal disease that, depending on its extent, causes dysfunction in adults and the elderly, difficulties in social or economic activity. Stroke can vary greatly depending on the patient's symptom pattern and the illness involved. Stroke patients should accurately assess the level of disability and induce rehabilitation or medical center visits. In addition, due to the variety of symptoms and the location of strokes, it is difficult to reliably assess stroke and neurological damage. In particular, early detection of stroke in persons without a past history is important, but those who have a past history of stroke are nine times more likely to relapse than those without a history, so that patients with strokes are continuously monitored and visited by doctors and doctors in a short period of time. There is an urgent need for techniques to support diagnosis and treatment.

Recent studies have made many ongoing efforts to apply NIHSS technology to new solutions for adaptively evaluating early disease disorders to prevent stroke recurrence and to address the major risk factors for stroke. Since the Mathew scale was first published in 1972, methods for assessing disability in stroke patients include the European Stroke Scale, the Canadian Neurological Scale, the Scandinavian Stroke Scale, and the National Institutes of Health's NIHSS. Among the various methods discussed above, NIHSS, developed by Brott et al., Is a widely used tool to provide quantitative criteria for post-stroke disorders. The NIHSS is a health stroke measure that adds up the scores from each function of the neurological tests to provide an overall stroke disorder score. The NIHSS, which takes 6.6 minutes per patient, consists of 14 features. NIHSS is widely used worldwide and is a proven tool with extensive research into the reliability and validity of tests and retests. As such, although NIHSS is a tool for collectively evaluating the initial damage of each stroke patient, it does not provide the analytical results for evaluating the initial stroke disorder.

In addition, the risk of stroke is reported to be caused by the interaction of various risk factors, not a single factor. Accordingly, new methods for evaluating the risk factors of each stroke and for predicting disease or early detection are emerging. Recently, studies on stroke prediction using various statistical methods and machine learning methods for risk factors have been conducted. Kannel et al. Have developed a logistic regression model that is related to this, and the ratio risk suggested by Cox or Weibull's model. Model-based studies have been reported. However, previous studies based on these risk factors have been problematic in predicting the risk of stroke in Korean elderly. Therefore, there is a need for developing a classification and prediction model of stroke risk suitable for Koreans.

The present invention is to solve the problems described above, and to build a stroke severity prediction model suitable for Koreans by analyzing the early damage and severity of stroke more deeply and accurately using machine learning and data mining based on NIHSS data It is an object to provide a method and system that can be.

In addition, to provide a method and system for providing a comprehensive medical service to provide patients and service users in real time information about the analyzed stroke, and to provide prompt notification when a stroke is serious, so that the treatment can be promptly treated. There is this.

In addition, by analyzing the analytic models of various machine learning and data mining applied according to an embodiment of the present invention, it provides a method and system that can contribute to the research and development for more accurate and comprehensive stroke disease analysis and prediction There is a purpose.

In accordance with an embodiment of the present invention, a method for predicting and analyzing stroke severity using the National Institutes of Health Stroke Scale (NIHSS) includes receiving real-time NIHSS data of a patient from a patient device at a medical data server, and at the medical data server. Filtering and storing the received real-time NIHSS data, and the factors for predicting and analyzing stroke severity using machine learning and data mining based on the filtered and stored real-time NIHSS data and the medical examination data of the patient included in the medical data server. Determining them in the medical data server, receiving the factors for the stroke severity determined by the stroke severity analysis server, generating a stroke severity prediction model for the patient, and based on the stroke severity prediction model generated by the stroke severity analysis server To the patient Predicting the severity of stroke, and may comprise any one of a plurality of classification categories of the status of the patient in stroke severity.

In the method for predicting and analyzing stroke severity according to an embodiment of the present invention, machine learning and data mining may include algorithms using a decision tree technique.

Factors for predicting and analyzing stroke severity determined by an algorithm in the method for predicting and analyzing stroke severity according to an embodiment of the present invention include: Level of Consiciousness (LOC) and LOC questions. LOC commands, facial palsy, right arm movements, left arm movements, right leg movements, left leg movements, limb ataxia, best language, Dysarthria, loss of cognition and inattention, and age.

A method for predicting and analyzing stroke severity according to an embodiment of the present invention includes predicting stroke severity using machine learning and data mining based on a prediction result of stroke severity and classification into one of a plurality of categories. And analyzing, at the stroke severity analysis server, rules and meanings of execution mechanisms for the analysis.

In accordance with an embodiment of the present invention, a method for predicting and analyzing stroke severity includes visualizing a prediction result of stroke severity for a patient and a result classified into any one of a plurality of categories in a stroke severity analysis server. Providing an alert notification to the patient device and the clinician device if the patient falls into a predetermined category according to the step of providing the at least one of the stroke monitoring server and the classification into any one of the plurality of categories. It may further include.

Stroke severity prediction and analysis system using the National Institutes of Health Stroke Scale (NIHSS) according to an embodiment of the present invention, a patient device equipped with an application for the prediction and analysis of stroke severity, real-time for the patient from the patient device Stroke using machine learning and data mining based on a communication unit for receiving NIHSS data, a memory unit for filtering and storing the received real-time NIHSS data, and the filtered and stored real-time NIHSS data and the medical examination data of the patient included in the memory unit A medical data server including a stroke severity factor determining unit for determining factors for prediction and analysis of severity, a prediction model generator and a generated stroke that receive factors for the determined stroke severity and generate a stroke severity prediction model for a patient Severity Prediction Model The second may comprise a stroke severity analysis server including a data analysis for classification of any one of a plurality of categories on the state of the prediction, and the Stroke patients the severity of the patient's stroke severity.

In the stroke severity prediction and analysis system according to an embodiment of the present invention, machine learning and data mining may include algorithms using a decision tree technique.

Factors for predicting and analyzing stroke severity determined by an algorithm in a stroke severity prediction and analysis system according to an embodiment of the present invention include a level of consiciousness (LOC), a question about a consciousness level (LOC questions), LOC commands, facial palsy, right arm movement, left arm movement, right leg movement, left leg movement, limb ataxia, best language, oral Dysarthria, extinction and inattention, and age may be included.

In the stroke severity prediction and analysis system according to an embodiment of the present invention, the stroke severity analysis server uses machine learning and data mining based on a prediction result of stroke severity and classification into one of a plurality of categories for a patient. The mechanism analysis unit may further include analyzing a rule and a meaning of an execution mechanism for predicting and analyzing stroke severity.

In the stroke severity prediction and analysis system according to an embodiment of the present invention, the stroke severity analysis server visualizes a prediction result of stroke severity and a result classified into any one of a plurality of categories for a patient, thereby providing a patient device, a clinical device, and a stroke. It includes a visualization data providing unit provided to at least one of the monitoring server, if the patient falls into a predetermined category according to the result of classification into any one of the plurality of categories, the stroke monitoring server alerts the patient device and the clinician device Can transmit

Meanwhile, according to an embodiment of the present invention, a computer-readable recording medium having recorded thereon a program for executing the above-described method on a computer can be provided.

According to a method and system for predicting and analyzing stroke severity, which is provided as an embodiment of the present invention, by analyzing the initial damage and severity of stroke using machine learning and data mining based on NIHSS data, By reducing the number of features), the prediction and classification speed can be further improved, and the reliability and accuracy of stroke severity prediction results can be improved.

In addition, since the NIHSS data, medical examination data and emergency patient data can be comprehensively analyzed, it is possible to construct a stroke severity prediction model suitable for Koreans unlike the prior art.

In addition, by providing information on the analyzed stroke in real time to patients and service users, and can provide a notification service for prompt treatment in the event of a severe stroke, continuous management and treatment of the patient's stroke disease can be achieved Can provide comprehensive medical services.

In addition, more accurate stroke severity analysis can be achieved through semantic analysis of machine learning and data mining used in stroke severity analysis, and by obtaining reliable data related to stroke disease analysis, Can contribute to research and development for prediction.

1 is a flowchart illustrating a first embodiment of a method for predicting and analyzing stroke severity using the National Institutes of Health Stroke Scale (NIHSS) according to an embodiment of the present invention.

Figure 2 shows an example of the results using the machine learning algorithm and data mining to determine factors for the prediction and analysis of stroke severity according to an embodiment of the present invention.

3 is a flowchart illustrating a second embodiment of a method for predicting and analyzing stroke severity using NIHSS according to an embodiment of the present invention.

Figure 4 shows an example of the results of confirming the accuracy of the prediction and analysis of stroke severity by the machine learning algorithm and data mining according to an embodiment of the present invention.

5 is a flowchart illustrating a third embodiment of a method for predicting and analyzing stroke severity using NIHSS according to an embodiment of the present invention.

6 is a block diagram showing a system for predicting and analyzing stroke severity using NIHSS according to an embodiment of the present invention.

7 shows an example of a system for predicting and analyzing stroke severity using NIHSS according to an embodiment of the present invention.

8 illustrates an example of a development environment and a structure of a system for predicting and analyzing stroke severity using NIHSS according to an embodiment of the present invention.

Terms used herein will be briefly described and the present invention will be described in detail.

The terms used in the present invention have been selected as widely used general terms as possible in consideration of the functions in the present invention, but this may vary according to the intention or precedent of the person skilled in the art, the emergence of new technologies and the like. In addition, in certain cases, there is also a term arbitrarily selected by the applicant, in which case the meaning will be described in detail in the description of the invention. Therefore, the terms used in the present invention should be defined based on the meanings of the terms and the contents throughout the present invention, rather than the names of the simple terms.

When any part of the specification is to "include" any component, this means that it may further include other components, except to exclude other components unless otherwise stated. In addition, the terms "... unit", "module", etc. described in the specification mean a unit for processing at least one function or operation, which may be implemented in hardware or software or a combination of hardware and software. .

Data included in the table and the like described as an embodiment of the present invention is collected at the Chungnam National University Hospital Emergency Medical Center, data collected from the patient within three days from the date of stroke diagnosis. The collection period is from 2015 to 2017 and includes NIHSS data for a total of 287 stroke patients. Of 287 patients, 286 patients had an emergency blood pressure test and a biosignal test, 284 had an emergency blood test, 261 had an emergency chemical test, 281 had a coagulation test, and 161 had a urine test. The NIHSS data used according to one embodiment of the present invention was used. Includes final experimental data for 227 patients except 16 patients with no stroke symptoms and 44 patients with no value or upper limit of NIHSS characteristics. Patients subject to the test according to an embodiment of the present invention are 65 or older, male 117, female 110, and the average age is 75.30 years for males and 78.48 years for females.

DETAILED DESCRIPTION Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings so that those skilled in the art may easily implement the present invention. As those skilled in the art would realize, the described embodiments may be modified in various different ways, all without departing from the spirit or scope of the present invention. In the drawings, parts irrelevant to the description are omitted in order to clearly describe the present invention, and like reference numerals designate like parts throughout the specification.

Hereinafter, the present invention will be described in detail with reference to the accompanying drawings.

Referring to FIG. 1, a method for predicting and analyzing stroke severity using the National Institutes of Health Stroke Scale (NIHSS) according to an embodiment of the present invention may include real-time NIHSS data of a patient from a patient device 100. Receiving in step (S100) (S100), filtering and storing the real-time NIHSS data received from the medical data server (200) (S200), the filtered and stored real-time NIHSS data and the patient included in the medical data server 200 Determining factors for the prediction and analysis of stroke severity in the medical data server 200 using machine learning and data mining based on the medical examination data (S300), the stroke severity determined in the stroke severity analysis server 300 Receiving the factors for, generating a stroke severity prediction model for the patient (S400) and stroke severity analysis server (300) Predicting stroke severity for the patient based on the stroke severity prediction model generated in the step), and classifying the patient's condition into any one of a plurality of categories of stroke severity (S500).

According to an embodiment of the present invention, NIHSS data refers to information previously measured by a clinician during the stroke diagnosis of the patient, the patient may proceed to the NIHSS test through the application provided in the patient device 100, See the history of real-time NIHSS data and its contents. In this case, the NIHSS test may include a language test, a questionnaire test, an upper limb and a lower limb muscle test, and the patient device 100 may receive information about each test from the medical data server 200. In addition, the patient may transmit (S100) real-time NIHSS data to the medical data server 200 through an application provided in the patient device 100, which will be understood as a first step in a method for predicting and analyzing stroke severity. Can be.

The real-time NIHSS data according to an embodiment of the present invention specifically includes Level of Consiciousness (LOC), LOC questions, LOC commands, and best gaze. ), Visual, facial palsy, right arm movement, left arm movement, right leg movement, left leg movement, limb ataxia, sensory, best language This may include data on dysarthria, loss of cognition and inattention. That is, such real-time NIHSS data refers to information measured from a patient according to a measurement scenario predefined by a clinician, and each of the above-described NIHSS characteristics (factors), such as consciousness level, questions about consciousness level, and instructions about consciousness level, It can be evaluated according to the patient's response, and information about the evaluation result for each NIHSS characteristic (factor) can be included in the real-time NIHSS data.

Referring to FIG. 1, after receiving (S100) real-time NIHSS data of a patient from the patient device 100 (S100) according to an embodiment of the present invention, the medical data server 200 receives the received data. The filtering and storing of the real-time NIHSS data may be performed (S200). In other words, the memory unit 220 of the medical data server 200 may store and simultaneously filter real-time NIHSS data received from the patient device 100. In the filtering process, real-time NIHSS data for each patient may be allocated to each patient. Can be categorized into private databases. When real-time NIHSS data is classified in a personal database assigned to each patient, if there is no personal database for the patient, a new personal database for the patient may be created to classify the real-time NIHSS data, and if there is a personal database for the patient. The real-time NIHSS data can be categorized into a personal database assigned to each patient.

In addition, the personal database allocated to each patient may store not only real-time NIHSS data for each patient but also health examination data for each patient. Health examination data for each patient may include information on the sex, age, history of diseases other than stroke, etc., such health examination data may be collected from national institutions, hospitals and the like.

In addition, in the filtering process of the real-time NIHSS data, a score may be assigned to each item data of the real-time NIHSS data (for example, data relating to each of the above-described NIHSS characteristics (factors)), and the score of each item data is a predetermined coefficient. Can be used and analyzed. In this case, the predetermined coefficient may be a Spearman coefficient, and referring to [Table 1] below, it is possible to confirm a result in which the Spearman correlation coefficient ρ for each NIHSS characteristic (factor) is statistically verified. That is, Table 1 shows the results of the Spearman correlation coefficients for each NIHSS characteristic (factor) analyzed statistically based on the data of the patients collected according to an embodiment of the present invention.

TABLE 1

As described above, after real-time NIHSS data and health examination data are classified and stored in each patient's personal database, each patient's data is stored in the memory unit 220 of the medical data server 200 to predict and analyze the stroke severity of the patient. Can be transferred to and stored in an integrated database. In this case, data transfer from the patient's personal database to the integrated database may be performed through the shared authorization module. In other words, when the authorization signal is received from the patient device 100, the shared authorization module may transfer real-time NIHSS data, health examination data, and the like stored in the patient's personal database based on the received authorization signal to the integrated database. The data may be stored in an integrated database and used for the prediction and analysis of the severity of the cerebrum, which will be described later.

When real-time NIHSS data is filtered and stored in an integrated database included in the memory unit 220 of the medical data server 200 according to an embodiment of the present invention, the stroke severity factor determination unit 230 of the medical data server 200 is stored. In S300, factors for predicting and analyzing stroke severity may be determined (S300), in which machine learning and data mining may be used. Machine learning and data mining are intended to accurately predict and analyze stroke severity through more in-depth analysis of NIHSS data. Machine learning and data mining in the method for predicting and analyzing stroke severity according to an embodiment of the present invention are preferred. The algorithm may include algorithms based on decision tree techniques, but may also include naive bayes algorithms, random forest algorithms, artificial neural network (ANN) algorithms, and logistic regression. a regression algorithm and a multi-class Surpport Vector Machine (SVM).

The algorithm by the decision tree technique included in the machine learning and data mining according to an embodiment of the present invention may be a C4.5 algorithm. The decision tree technique is a representative supervised learning method used in machine learning and represents a model for dealing with classification problems. The C4.5 algorithm, which is one of the algorithms for generating such a decision tree, can perform complementary functions such as numerical attribute processing, missing value processing, and weighting of the number of branches for attribute selection. Each node in the tree generated by the C4.5 algorithm can be associated with a series of cases, and a series of cases can be assigned weights to account for unknown attribute values. For example, if property a is a discrete value and Y ₁ , Y ₂ , ..., Y _S is a subset of Y consisting of cases with known values that are distinct for property a, The information gain of the characteristic a may be estimated as shown in Equation 1 below.

[Equation 1]

As shown in Equation 1, the estimated information gain is an entropy function. When discrete attributes are selected at the ancestor node, it is easy to see that the information gain and information gain ratio is zero, so the C4.5 algorithm does not estimate the information gain of these attributes. Referring to [Equation 1] and [Equation 2] below, the partial information of Y (Split info (Y)) represents potential information generated by dividing into n subsets, while the information gain is generated from the same subset. It can be seen that the information related to the classification is measured.

[Equation 2]

By using the C4.5 algorithm, analysis using information gain is possible, but the information gain ratio for each segmentation information of Y ₁ , Y ₂ , ..., Y _S can be additionally considered. The information gain ratio may be estimated through Equation 3 below.

[Equation 3]

The information gain ratio indicates the ratio of information generated by the useful segmentation and can be used for classification and analysis for decision tree generation.

As described above, the C4.5 algorithm can be used to generate a decision tree using information gain and information gain ratio, which allows prediction and analysis of stroke severity based on NIHSS data and patient health examination data. You can determine the factors for Using the C4.5 algorithm's information gain and information gain ratio, it is possible to determine only the factors necessary for the prediction and analysis of stroke severity from NIHSS data and patient health examination data, which can speed up the prediction and analysis of stroke severity. And accuracy can also be improved.

Referring to FIG. 2, when the C4.5 algorithm is used as a machine learning algorithm and data mining, factors may be determined and classified only as necessary for predicting and analyzing stroke severity, and the classified and generated decision trees may be used. The severity of stroke of a patient can be predicted and the condition of the patient can be easily classified. At this time, the factors for predicting and analyzing stroke severity determined by the algorithm include Level of Consiciousness (LOC), LOC questions, LOC commands, and facial paralysis. (facial palsy), right arm, left arm, right leg, left leg, limb ataxia, best language, dysarthria, loss of cognition and inattention and inattention) and age.

For example, the real-time NIHSS data and the characteristics (factors) included in the patient's health examination data include consciousness levels, questions about consciousness levels, instructions about consciousness levels, optimal gaze, vision, facial paralysis, right arm movements, left There are 18 types of arm movements, right leg movements, left leg movements, limb ataxia, sensations, best language skills, oral dysfunction, loss of cognition and carelessness, gender and age. In this case, only 13 characteristics (factors) of the above 18 may be used, and thus, the prediction and analysis speed may be considerably improved compared to using all 18 characteristics (factors) data when predicting and analyzing stroke severity. Accuracy can also be significantly improved.

Table 2 below shows the results of a summary of the rules for classification and analysis of stroke severity that can be obtained from decision trees created with factors determined using the C4.5 algorithm. Referring to Table 2, it is easier to see how stroke severity is classified and analyzed according to the patient's NIHSS data and health examination data.

TABLE 2

Referring to FIG. 1, when the factors for stroke severity prediction and analysis are determined in the medical data server 200 (S200), the predictive model generator 310 of the stroke severity analysis server 300 is determined. In step S400, a stroke severity prediction model for the patient may be generated. The stroke severity prediction model may be generated for each patient individually, and a decision tree may be generated based on the factors determined through the aforementioned C4.5 algorithm when generating the stroke severity prediction model. For example, referring to FIG. 2 and Table 2, an example of a decision tree generated in a prediction model can be confirmed, and the severity of stroke prediction and analysis through the relationship between the factors constituting the decision tree can be found. You can check the progress.

Referring to FIG. 1, when the stroke severity prediction model is generated, the stroke severity prediction model is predicted by the data analysis unit 320 of the stroke severity analysis server 300 to predict the severity of stroke of the patient, and the patient's state is plural. Classifying into any one of the categories (S500) may be performed. At this time, the plurality of categories may be four categories of mild stroke, moderate stroke, moderate-to-severe stroke, and severe stroke. For example, referring to rule 1 of FIG. 2 and Table 2, it is determined that the score assigned to the best verbal ability among the factors for predicting and analyzing stroke severity is greater than 1 and the age is 84 years or less. In this case, the data analyzer 320 may predict and classify the patient's condition as a moderate stroke.

That is, in the above-described predictive model generation process (S400), it is possible to select and determine factors for predicting and analyzing stroke severity by using machine learning and data mining. When the predictive model is generated, information about each corresponding factor is obtained. Since the data analysis unit 320 predicts and classifies the stroke severity of the patient into one of four categories, the patient's condition can be analyzed more accurately, and the patient's prognosis is accurately predicted. You can ensure that the treatment is appropriate for your condition.

Referring to FIG. 3, the method for predicting and analyzing stroke severity according to an embodiment of the present invention includes machine learning and data mining based on a prediction result of stroke severity for a patient and a result classified into one of a plurality of categories. Analyzing the rules and meaning of the execution mechanism for the prediction and analysis of stroke severity in the stroke severity analysis server 300 may further include the step (S610).

The above-described analysis step (S610) is for analyzing and utilizing the accuracy and the rules and meanings of the machine learning and data mining execution mechanisms used in the prediction and classification of stroke severity, and the overall diagnosis and To obtain high quality data for use in research on therapy. For example, when selecting and determining factors for stroke severity prediction and analysis using the C4.5 algorithm, and generating a decision tree based on the determined factors, each factor through the mechanism analyzer 330. By analyzing the relationship between the rules and the meaning of the execution mechanism for predicting and analyzing stroke severity, the results as shown in [Table 2] can be derived.

In order to derive the analysis results as shown in FIG. 4, a total of 227 stroke severity data were used, and 182 stroke severity data corresponding to 80% were randomly extracted to generate a machine learning model. In the actual test, the test was performed on 45 subjects, the remaining 20% of which was not used to generate machine learning models in all subjects.

Recall and precision were used as measurements to measure the accuracy of the prediction and analysis of the proposed method in accordance with one embodiment of the present invention. Through machine learning algorithms and data mining, data correctly categorized into a given class is represented as True Positive (TP), and misidentified data is represented as False Positive (FP) and incorrectly belonged to another class. Data of the classified class may be represented as False Negatives (FNs). Accordingly, recall and precision and accuracy of the entire operating system can be estimated as shown in Equation 4 below.

[Equation 4]

Referring to FIG. 4, as a result of analyzing the accuracy of stroke severity prediction and classification using various machine learning and data mining using Equation 4, C4.5 algorithm, random forest algorithm, and artificial neural network (ANN) ) Algorithms can predict and classify stroke severity with high accuracy.

In addition, referring to Table 3 below, as a result of classifying the patient's condition into a plurality of categories of stroke severity through the C4.5 algorithm according to an embodiment of the present invention, the accuracy of the entire system is 91.11%. It is measured, and it can be seen that the accuracy of analysis of the moderate-to-severe stroke and the severe storke is 100%. That is, when the stroke severity is predicted and analyzed using the C4.5 algorithm, the stroke severity of a severely severe patient can be determined and classified very accurately.

TABLE 3

Referring to FIG. 5, the method for predicting and analyzing stroke severity according to an embodiment of the present invention may include: classifying a result of the stroke severity analysis server 300 into one of a plurality of categories and a prediction result of a stroke severity for a patient. The patient corresponds to a predetermined category according to the result of visualizing and classifying it into at least one of the patient device 100, the clinician device 500, and the stroke monitoring server 400 (S620) and the classification into one of a plurality of categories. If so, the stroke monitoring server 400 may further include the step (S700) of transmitting a warning notification to the patient device 100 and the clinician device (500).

Through such a result providing step (S620) can be provided so that the patient and the clinician can continuously check the stroke status (including severity) of the patient. In this case, the results predicted and analyzed by the visualization data provider 340 of the stroke severity analysis server 300 are graphs and charts on at least one of the patient device 100, the clinician device 500, and the stroke monitoring server 400. And may be visualized and the like.

In addition, when the patient's condition corresponds to a predetermined category (eg, a moderate-to-severe stroke category) through the notification transmission step (S700), the stroke monitoring server 400 may perform in real time. According to the confirmed result, a warning notification may be transmitted to the patient device 100 and the clinician device 500 so that the patient and the clinician may quickly visit a medical center, a hospital, etc. to receive diagnosis and treatment. In this case, when the stroke monitoring server 400 transmits an alert notification, the notification information is repeatedly repeated for a predetermined number of times (eg, 10 times) through the screen and the speaker of the patient device 100 and the clinician device 500. The signal can be output.

6 is a block diagram illustrating a system for predicting and analyzing stroke severity using NIHSS according to an embodiment of the present invention, and FIG. 7 is a diagram for predicting and analyzing stroke severity using NIHSS according to an embodiment of the present invention. For example.

6 and 7, a system for predicting and analyzing stroke severity using the National Institutes of Health Stroke Scale (NIHSS) according to an embodiment of the present invention includes a patient equipped with an application for predicting and analyzing stroke severity. Device 100, communication unit 210 for receiving real-time NIHSS data for the patient from the patient device 100, memory unit 220 for filtering and storing the received real-time NIHSS data and filtered and stored real-time NIHSS data and memory A medical data server including a stroke severity factor determining unit 230 that determines factors for predicting and analyzing stroke severity using machine learning and data mining based on the health examination data of the patient included in the unit 220 ( 200) and predictions that receive the factors for the determined stroke severity and generate a stroke severity prediction model for the patient. A model analyzer 310 and a data analyzer 320 to predict stroke severity for a patient based on the generated stroke severity prediction model and classify the patient's condition into any one of a plurality of categories regarding stroke severity. Stroke severity analysis server 300 may be included.

6 and 7, in the stroke severity prediction and analysis system according to an embodiment of the present invention, the stroke severity analysis server 300 classifies one of a plurality of categories and a prediction result of stroke severity for a patient. The mechanism analysis unit 330 may further include analyzing a rule and a meaning of an execution mechanism for predicting and analyzing stroke severity based on the result of the machine learning and data mining.

6 and 7, in the stroke severity prediction and analysis system according to an embodiment of the present invention, the stroke severity analysis server 300 classifies one of a plurality of categories and a prediction result of stroke severity for a patient. And a visualization data providing unit 340 which visualizes the results and provides them to at least one of the patient device 100, the clinician device 500, and the stroke monitoring server 400, and classifies the result into one of a plurality of categories. Accordingly, when the patient corresponds to a predetermined category, the stroke monitoring server 400 may transmit an alert notification to the patient device 100 and the clinician device 500.

Referring to FIG. 8, the client 10 running the application for severity prediction and analysis may receive information for the NIHSS test from the API server host 20. The NIHSS test may include language tests, questionnaire tests, upper limbs, and the like. Muscle test, etc. Information for the NIHSS test received from the API server host 20 may be updated at predetermined time intervals, and the predetermined time interval may be set through the client 10. In addition, the client 10 running the application for severity prediction and analysis may perform a function of managing a schedule for diagnosing or treating a stroke of a patient and receiving and outputting a warning notification. For example, when a schedule for severity prediction and analysis is set by an application user (patient), a push service according to the schedule setting may be provided through the client 10. The push service referred to herein refers to a service for outputting a schedule set by a user (patient) through a message or a sound.

The NIHSS test data stored through the client 10 may be stored in a personal database through the cloud of the medical data server 200. When a signal is applied from the client 10 to the shared authorization module, the data stored in the personal database is integrated into the integrated database. Can be delivered. The data stored through this process may be managed through the API server host 20 and used for the prediction and analysis of the stroke severity of the patient.

Referring to FIG. 8, a machine learning infrastructure 30 for predicting and analyzing stroke severity of a patient may be identified based on information received from the API server host 20. When the stroke severity prediction and analysis through machine learning and data mining is completed through the above-described process in the machine learning infrastructure 30, the stroke severity prediction and analysis result is provided to the client 10 through the API server host 20. Can be. In this case, the stroke severity prediction and analysis result may be provided when a request for the prediction and analysis result is received from the client 10 at the API server host 20.

Referring to FIG. 8, the development environment of a system according to an embodiment of the present invention is not limited to an architecture such as a RESTful API and Spring framework 4.0, etc. described as an example of implementing the same, and a development environment optimized to implement the present system. Various architectures, frameworks, etc. may be used according to the present invention.

With regard to the system according to an embodiment of the present invention, the above-described method may be applied. Therefore, in the context of the system, the description of the same content as the above-described method is omitted.

Meanwhile, according to an embodiment of the present invention, a computer-readable recording medium having recorded thereon a program for executing the above-described method on a computer can be provided. In other words, the above-described method can be written as a program that can be executed in a computer, and can be implemented in a general-purpose digital computer for operating the program using a computer readable medium. In addition, the structure of the data used in the above-described method can be recorded on the computer-readable medium through various means. A recording medium for recording an executable computer program or code for performing various methods of the present invention should not be understood to include temporary objects, such as carrier waves or signals. The computer readable medium may include a storage medium such as a magnetic storage medium (eg, a ROM, a floppy disk, a hard disk, etc.), an optical reading medium (eg, a CD-ROM, a DVD, etc.).

The foregoing description of the present invention is intended for illustration, and it will be understood by those skilled in the art that the present invention may be easily modified in other specific forms without changing the technical spirit or essential features of the present invention. will be. Therefore, it should be understood that the embodiments described above are exemplary in all respects and not restrictive. For example, each component described as a single type may be implemented in a distributed manner, and similarly, components described as distributed may be implemented in a combined form.

The scope of the present invention is shown by the following claims rather than the detailed description, and all changes or modifications derived from the meaning and scope of the claims and their equivalents should be construed as being included in the scope of the present invention. .

Claims

In the method of predicting and analyzing stroke severity using the National Institutes of Health Stroke Scale (NIHSS),

Receiving at the medical data server real-time NIHSS data for the patient from a patient device;

Filtering and storing the received real-time NIHSS data at the medical data server;

The medical data server determines factors for predicting and analyzing the stroke severity using machine learning and data mining based on the filtered and stored real-time NIHSS data and the medical examination data of the patient included in the medical data server. step;

Receiving the factors for the determined stroke severity in the stroke severity analysis server, generating a stroke severity prediction model for the patient; And

Predicting the stroke severity for the patient based on the generated stroke severity prediction model in the stroke severity analysis server, and classifying the patient's condition into any one of a plurality of categories related to the stroke severity. Characterized by stroke stroke prediction and analysis method.
The method of claim 1,

The machine learning and data mining includes stroke severity prediction and analysis method comprising a decision tree (Decision tree) algorithm.
The method of claim 2,

Factors for predicting and analyzing stroke severity determined by the algorithm include:

Level of Consiciousness (LOC), LOC questions, LOC commands, facial palsy, right arm movement, left arm movement, right leg movement Predicting stroke severity, including left leg exercise, limb ataxia, best language, dysarthria, loss of cognition and inattention, and age Analytical Method.
The method of claim 1,

The stroke severity and the rules of the execution mechanism for the prediction and analysis of stroke severity using the machine learning and data mining based on the prediction result of the stroke severity for the patient and the results classified into any one of the plurality of categories Stroke severity prediction and analysis method further comprising the step of analyzing in the analysis server.
The method of claim 1,

Visualizing, by the stroke severity analysis server, a prediction result of stroke severity for the patient and a result classified into one of the plurality of categories and providing the result to at least one of the patient device, a clinician device, and a stroke monitoring server; And

If the patient corresponds to a predetermined category according to a result classified into any one of the plurality of categories, the stroke monitoring server further comprises the step of transmitting an alert notification to the patient device and the clinician device Stroke severity prediction and analysis method.
In a system for predicting and analyzing stroke severity using the National Institutes of Health Stroke Scale (NIHSS),

A patient device equipped with an application for predicting and analyzing the stroke severity;

A communication unit for receiving real-time NIHSS data for the patient from the patient device; A memory unit for filtering and storing the received real-time NIHSS data; And a stroke severity factor determining unit configured to determine factors for predicting and analyzing the stroke severity using machine learning and data mining based on the filtered and stored real-time NIHSS data and the medical examination data of the patient included in the memory unit. A medical data server comprising; And

A predictive model generator configured to receive the factors for the determined stroke severity and generate a stroke severity prediction model for the patient; And a stroke severity analysis server including a data analysis unit for predicting stroke severity for the patient based on the generated stroke severity prediction model and classifying the patient's condition into any one of a plurality of categories related to the stroke severity. Stroke severity prediction and analysis system comprising a.
The method of claim 6,

The machine learning and data mining include stroke severity prediction and analysis system, characterized in that the algorithm by a decision tree (Decision tree) technique.
The method of claim 7, wherein

Factors for predicting and analyzing stroke severity determined by the algorithm include:

Level of Consiciousness (LOC), LOC questions, LOC commands, facial palsy, right arm movement, left arm movement, right leg movement Predicting stroke severity, including left leg exercise, limb ataxia, best language, dysarthria, loss of cognition and inattention, and age Analysis system.
The method of claim 6,

The stroke severity analysis server,

Mechanism to analyze the rules and meaning of the execution mechanism for the prediction and analysis of stroke severity using the machine learning and data mining based on the prediction result of stroke severity for the patient and the result classified into any one of the plurality of categories Stroke severity prediction and analysis system further comprising an analysis unit.
The method of claim 6,

The stroke severity analysis server may visualize a prediction result of stroke severity for the patient and a result classified into any one of the plurality of categories to provide visualization data providing unit to at least one of the patient device, a clinician device, and a stroke monitoring server. Include,

Predicting and analyzing stroke severity, wherein the stroke monitoring server transmits an alert notification to the patient device and a clinician device when the patient corresponds to a predetermined category according to a result classified into any one of the plurality of categories. system.
A computer-readable recording medium having recorded thereon a program for implementing the method of any one of claims 1 to 5.