KR102511511B1 - Apparatus and method for predicting recovery prognosis of motor function using clinical data from stroke patients - Google Patents

Abstract

The present invention relates to an apparatus and method for predicting the prognosis of motor function recovery using clinical data of a stroke patient. According to the present invention, in an apparatus for predicting the prognosis of recovery of motor function in a stroke patient, body data related to a plurality of patients for each patient, and multiple types of motor function scores related to the affected limb of the patient observed within a reference period from the date of onset Prediction using a data collection unit that collects clinical data including stroke type, the collected physical data and clinical data as input data, and motor function scores of the set type observed after a set period elapses from the onset date as output data A model learning unit that trains a model, a data acquisition unit that acquires body data and currently observed clinical data for the patient to be analyzed, and applies the acquired body data and clinical data to the predictive model for a set period from Provided is an apparatus for predicting the prognosis of recovery of motor function of a stroke patient including a predictor for predicting the motor function score of the set type for the patient subject to analysis at a future time point after the passage of time.
According to the present invention, it is possible to predict the recovery prognosis of the patient's motor function at a future time in advance by applying the current clinical data collected in correspondence with the patient to the pre-learned predictive model.

Description

Apparatus and method for predicting recovery prognosis of motor function using clinical data from stroke patients}

The present invention relates to an apparatus and method for predicting the recovery of motor function using clinical data of a stroke patient, and more particularly, to a stroke that can predict the prognosis of recovery of motor function of a patient at a future point in time only with the patient's current clinical data It relates to a device and method for predicting the prognosis of recovery of motor function of a patient.

This invention is a patent applied for as part of research/service conducted with the support of the 2020 Medi Challenger Fostering Project supported by Gyeongsan City, and related matters are as follows.

[Assignment identification number] 220C000565

[Name of Department] Gyeongsan City

[Research management institution] Yeungnam University Industry-University Cooperation Foundation

[Research Project Name] Research/Service

[Research Project Title] Development of artificial intelligence to predict recovery of swallowing function in stroke patients using medical big data: development based on image data

[Organizer] Daegu-Gyeongbuk Advanced Medical Industry Promotion Foundation

[Research Period] 2020.05.01~2020.12.31

Stroke is the most common neurological disease, and the majority of stroke patients complain of motor dysfunction. These motor dysfunctions not only have an absolute impact on the quality of life of stroke patients, but also cause psychological atrophy and weakening of the will to treat patients, which can lead to aggravation of the condition.

However, in actual clinical practice, there are many cases in which medical staff cannot accurately evaluate the prognosis of stroke patients' motor function, and this makes it difficult to plan a rehabilitation program for each patient.

The background technology of the present invention is disclosed in Korean Patent Publication No. 10-2019-0132710 (published on November 29, 2019).

The present invention provides an apparatus and method for predicting the prognosis of motor function recovery using clinical data of a stroke patient, which can predict the patient's motor function recovery prognosis for a future time point in advance by applying the patient's current clinical data to a pre-learned predictive model. It aims to provide

The present invention, in an apparatus for predicting the prognosis of recovery of motor function in a stroke patient, includes body data related to the patient for each of a plurality of patients, multiple types of motor function scores related to the affected limb of the patient observed within a reference period from the date of onset, Prediction model using a data collection unit that collects clinical data including stroke type, the collected physical data and clinical data as input data, and motor function scores of a set type observed after a set period elapses from the onset date as output data A model learning unit that learns, a data acquisition unit that acquires physical data and currently observed clinical data for the patient to be analyzed, and a set period of time elapsed from the present by applying the obtained physical data and clinical data to the prediction model. Provided is an apparatus for predicting the prognosis of recovery of motor function of a stroke patient including a predictor for predicting the motor function score of the set type for the patient to be analyzed at a future point in time.

In addition, the apparatus for predicting the prognosis of recovery of motor function of the stroke patient may further include a classification unit that classifies whether or not recovery of the corresponding motor function is possible by comparing the predicted motor function score with a threshold value.

Also, the reference period may be within 30 days, and the set period may be 6 months or longer.

In addition, the clinical data may further include the presence or absence of evoked potentials of the abductor and tibialis anterior muscles.

In addition, the plurality of types of motor function scores used as input data of the prediction model include a joint score (Medical Research Council (MRC)), a gait score (Functional Ambulation Category (FAC)), and an upper extremity motion score (MBC; Modified Brunnstrom classification). and motor function scores of a set type used as output data of the prediction model may include the gait score (FAC) and the upper extremity motion score (MBC).

In addition, the apparatus for predicting the prognosis of recovery of motor function of the stroke patient compares the gait score and the upper limb movement score with preset threshold values, and the gait function and upper limb movement function of the patient subject to analysis for a future time point after a set period has elapsed. It may further include a classification unit that individually classifies whether recovery is possible.

In addition, the predictive model may be learned through a deep neural network (DNN) algorithm.

In addition, the present invention, in the apparatus for predicting the prognosis of recovery of motor function of a stroke patient, using the apparatus for predicting the recovery of motor function, the body data related to the patient for each of a plurality of patients, the affected limb of the patient observed within the reference period from the date of onset Collecting clinical data including multiple types of related motor function scores and stroke types, using the collected physical data and clinical data as input data, and calculating motor function scores of the set types observed after a set period of time from the date of onset Learning a predictive model using output data, obtaining physical data and currently observed clinical data for the patient to be analyzed, and applying the acquired physical data and clinical data to the predictive model to set from the present Provided is a method for predicting the prognosis of recovery of motor function of a stroke patient, comprising the step of predicting the motor function score of the set type for the patient subject to analysis at a future time point after the lapse of the period.

According to the present invention, it is possible to predict the recovery prognosis of the patient's motor function at a future time in advance by applying the current clinical data collected in correspondence with the patient to the pre-learned predictive model.

In addition, based on the predicted prognosis of motor function recovery, it enables medical staff to more efficiently plan future treatment/rehabilitation programs for each patient and prepare for worsening symptoms, as well as suppress disease progression and efficiently manage disease. let it do

1 is a diagram showing the configuration of an apparatus for predicting the prognosis of recovery of motor function using clinical data of a stroke patient according to an embodiment of the present invention.
FIG. 2 is a diagram briefly explaining functions and data flow of each unit of FIG. 1 .
FIG. 3 is a diagram illustrating a method for predicting the prognosis of motor function recovery in a stroke patient using the apparatus of FIG. 1 .
4 is a diagram showing detailed factors used in learning a predictive model according to an embodiment of the present invention by way of example.
5 is a diagram comparing the performance results of the three algorithms used.

Then, with reference to the accompanying drawings, an embodiment of the present invention will be described in detail so that those skilled in the art can easily practice it. However, the present invention may be implemented in many different forms and is not limited to the embodiments described herein. And in order to clearly explain the present invention in the drawings, parts irrelevant to the description are omitted, and similar reference numerals are attached to similar parts throughout the specification.

Throughout the specification, when a part is said to be "connected" to another part, this includes not only the case where it is "directly connected" but also the case where it is "electrically connected" with another element interposed therebetween. . In addition, when a certain component is said to "include", this means that it may further include other components without excluding other components unless otherwise stated.

The present invention is an apparatus and method for predicting the recovery of motor function using clinical data of a stroke patient, which can predict the prognosis of the recovery of motor function of the patient for a future time point by applying the current clinical data of the patient to a pre-learned predictive model suggest a technique

1 is a diagram showing the configuration of an apparatus for predicting the prognosis of recovery of motor function using clinical data of a stroke patient according to an embodiment of the present invention, and FIG. 2 is a diagram briefly explaining the functions of each unit and data flow in FIG. 1 .

Referring to FIGS. 1 and 2 , an apparatus 100 (hereinafter referred to as a prediction apparatus) predicting the recovery of motor function of a stroke patient according to an embodiment of the present invention includes a data collection unit 110, a model learning unit 120, data It includes an acquisition unit 130, a prediction unit 140, a classification unit 150, and an output unit 160.

Here, the operation of each unit 110 to 160 and data flow between each unit may be controlled by a controller (not shown).

The data collection unit 110 includes physical data related to the patient for each of a plurality of patients, multiple types of motor function scores related to the affected limb of the patient observed within a reference period (eg, 7 days, 30 days) from the onset date, stroke type Collect clinical data including Here, the reference period may correspond to a period within 30 days.

The data collection unit 110 may build big data by collecting basic data including physical and clinical data of a plurality of patients (stroke patients). The data collection unit 110 may collect basic data of a patient from an EMR that manages and stores medical information of a patient or a hospital server that works with the EMR.

EMR, an electronic medical record system, may be included in a hospital server or operated in conjunction with a hospital server. The EMR or hospital server can manage the patient's medical information (medical information DB) history including the patient's personal information, body information, medical records for the past several months or years, and various examination information on a time-by-time basis. In this way, basic data for each patient can be collected through EMR, hospital server, etc., and from this, big data necessary for predictive model learning can be built.

The data collection unit 110 may collect basic data for patients whose clinical data is collected within 7 to 30 days (early stage) between the date of onset of stroke and the date of hospitalization (or the previous day).

Here, of course, the selection of basic data is for those who have been diagnosed with stroke for the first time, those who are 20 years old or older (excluding minors), those whose period between the date of onset and the date of hospitalization is within 7 to 30 days, and those whose clinical data were collected during that time, Those with an MBC score within 7 to 30 days from the onset of the set value (e.g. 4) and a FAC score below the set value (e.g. 3), serious complications such as pneumonia or heart-related problems from the beginning of the onset to the last assessment It can be performed for those who have not had this, thereby increasing the efficiency, accuracy and validity of the predictive model.

Basic data for each patient collected in an embodiment of the present invention may include physical data and clinical data. The body data may include the patient's age, gender, and the like. Clinical data included the patient's stroke type (infarction/bleeding), multiple types of motor function scores related to the affected limbs (shoulder abductors, elbow flexors, hand flexors, hand extensors, hip flexors, knee extensors, ankle dorsiflexors), and abducted knuckles. The presence or absence of the exercise evoked potential of the muscle and the tibialis anterior muscle may be included.

Here, the plural types of motor function scores related to the affected limb include a joint score (Medical Research Council (MRC)), a gait score (Functional Ambulation Category (FAC)), and an upper limb movement score (MBC; Modified Brunnstrom classification).

The evaluation criteria of each motor function score (MRC, FAC, MBC) are shown in Tables 1 to 3 below. The higher the evaluation score, the better the motor function. These motor function scores may be evaluated by medical staff such as doctors and rehabilitation therapists and then stored in a hospital server or an electronic medical record system.

score MRC 5 Able to move full range of motion (ROM) against gravity and strong resistance (normal) 4 Full ROM against gravity and moderate resistance 3 Against Gravity Complete ROM 2 Full ROM when gravity is not acting One Weak muscle contraction but no joint movement 0 no muscle contraction

score FAC 0 unable to walk One Requires constant support from one person 2 Requires intermittent support from 1 person 3 Oral supervision required 4 Assistance is required on steps, ramps, and uneven floors 5 Able to walk independently anywhere

score MBC One Inability to move fingers voluntarily 2 Can move fingers voluntarily 3 Affected hand can close voluntarily but cannot open 4 Cards can be grasped with the inner surfaces of thumb and index finger, fingers slightly spread 5 Can lift and hold glass and can spread fingers 6 Can catch and throw a ball near normal and can button and unbutton a shirt

The data collection unit 110 may additionally collect and store motor function scores of a set type observed after a set period (eg, 6 months, 1 year) from the onset date for model learning. In this case, the motor function score of the set type corresponds to the gait score (FAC) and the upper extremity motion score (MBC) excluding the joint score (MRC). The setting period may be selected from a period of 6 months or longer.

The patient-specific data collected by the data collection unit 110 is used as learning data for building a model for predicting the patient's motor function recovery prognosis. Among the collected data, the initial data collected within 30 days between the date of onset and the date of hospitalization are used as input data when learning the predictive model, and data after 6 months are used as output data. The predictive model learns the relationship between these input data and output data based on machine learning.

Specifically, the model learning unit 120 takes the patient's body data and clinical data previously collected as input data, and sets motor function scores of the set type observed in the patient after a set period (eg, 6 months) has elapsed from the onset date. A predictive model is trained using (FAC, MBC) as output data.

The model learning unit 120 may generate a predictive model for predicting a prognosis of motor function recovery of stroke patients through artificial intelligence-based running analysis.

Various techniques may be used to generate the predictive model, and deep neural network (DNN), logistic regression, and random forest algorithms may be used.

The embodiment of the present invention uses a deep neural network (DNN) as a representative example. Deep learning is a set of machine learning algorithms that attempt a high level of abstraction through a combination of several nonlinear transformation methods. There are various types of neural networks in deep learning. The embodiment of the present invention uses a deep neural network, and implements a predictive model for motor recovery results after 6 months of a stroke patient using the initial clinical data of stroke patients and data after 6 months did

The input data of the prediction model are the patient's physical data (age, gender, etc.), stroke type, motor function scores (MRC, FAC, MBC) related to the affected limb observed within 30 days between the date of stroke onset and the date of hospitalization, and the abductor abductor and the presence or absence of the motor evoked potential of the tibialis anterior muscle. In addition, the output data includes the patient's motor function scores (FAC, MBC) observed 6 months after the onset date.

The predictive model can learn the relationship between the patient's initial clinical data observed at the beginning of stroke onset and correspondingly observed motor function scores (FAC, MBC) 6 months later through machine learning. That is, the predictive model is trained to derive the patient's gait score and upper extremity motion score as outputs for the next 6 months, which are predicted from the inputs of the patient's basic body data and initial clinical data.

The model learning unit 120 can continuously analyze and update the relationship between a plurality of factors (variables) collected for each patient and output factors predicted in response to machine learning based on supervised learning, and can modify and update the weights in the deep neural network. can be optimized. Of course, in the process, learning may be performed until the error falls below the threshold value.

When the prediction model is completed through learning, simply by applying the current observation data for any patient subject to analysis (e.g., patients with early stroke, patients with suspected stroke) to the model, the gait score and Upper extremity movement scores can be predicted.

To this end, the data acquisition unit 130 acquires body data and currently observed clinical data of the patient to be analyzed. The data acquisition unit 130 may acquire data of a patient to be analyzed in association with an EMR or a hospital server or system, and may receive parameters related to the patient from a manager such as a medical staff through an input interface. In addition, the data acquisition unit 130 may be network-connected to the EMR to search, inquire, and load medical information including physical information and clinical information about the patient to be analyzed.

The prediction unit 140 applies the physical data and clinical data acquired in correspondence with the patient to be analyzed to the prediction model, and the motor function score ( MBC, FAC) are predicted and provided.

For example, the prediction unit 140 applies each factor collected corresponding to the patient to be analyzed at the current time point as an input parameter to the predictive model to derive and provide MBC and FAC scores of the expected patient in the next 6 months. .

Therefore, according to the predictive model according to an embodiment of the present invention, the motor function score after 6 months is predicted and provided in response to the current collection data of the patient to be analyzed, and through this, the deterioration and improvement of the motor function of the upper and lower limbs from the present Alternatively, the maintenance situation can be predicted in advance.

Here, the model learning unit 120 and the prediction unit 140 may be included or embedded in a processor as shown in FIG. 2 . These processors can be loaded with prediction models using artificial intelligence, provide motor function prognosis prediction results, and can be connected to external platforms and medical information systems to update and update artificial intelligence learning data and prediction models.

In addition, the processor can embed artificial intelligence algorithms and perform calculations based on machine learning. For example, the processor may use a machine learning algorithm based on a highly multilayered artificial neural network for pre-training of the algorithm. The processor can predict the patient's motor function score expected in the future in response to the patient's input data, which is currently the subject of analysis, based on pre-learned weights through an artificial intelligence algorithm.

The classification unit 150 compares the predicted motor function score with a threshold value and classifies whether recovery is possible for the corresponding type of motor function.

Specifically, the classification unit 150 compares the gait score (FAC) and the upper extremity motion score (MBC) with preset threshold values, and the gait function and upper extremity motion function of the patient to be analyzed for a future time point after the set period has elapsed. The possibility of recovery is individually classified.

For example, if the predicted gait score (FAC) 6 months after the onset of stroke is 4 points or more, the patient's lower extremity motor function recovery possibility is classified as high, and if it is less than 4 points, the recovery probability is classified as low. In addition, if the expected upper limb movement score (MBC) after 6 months was 5 points or more, the patient's prognosis for recovery of upper limb motor function after 6 months was classified as good, and if the score was less than 5 points, the patient's prognosis for recovery of upper limb motor function was poor. can be classified.

The output unit 160 outputs and provides a motor function score prediction result derived through the prediction model and a recovery possibility classification result based thereon through a display or the like.

In this way, the prediction device 100 executes a process related to predicting the prognosis of recovery of motor function and provides a corresponding result, thereby enabling the patient's diagnosis, counseling, prescription, treatment, and efficient prognosis management.

As described above, the present invention is based on a predictive model that learns the motor function score of the affected limb observed after the initial clinical data of the stroke patient and the set period, and can be used for future time points with only a few current clinical data for the patient to be analyzed. It is possible to conveniently predict and provide a prognosis for recovery of the patient's motor function in advance.

The prediction device 100 according to an embodiment of the present invention may be a server itself that predicts the recovery prognosis of a stroke patient's motor function, or a terminal device (PC, desktop, smartphone, laptop, pad, etc.) It may correspond to an application implemented on the user terminal 200 . Accordingly, each terminal may be connected to the prediction device 100 through a network in a state in which a related application (application program) is executed, and may be provided with a related service.

FIG. 3 is a diagram illustrating a method for predicting the prognosis of motor function recovery in a stroke patient using the apparatus of FIG. 1 .

First, the data collection unit 110 provides physical data related to the patient for each of a plurality of patients (stroke patients) and multiple types of motor function scores related to the patient's affected limb observed within a reference period (eg, 30 days) from the date of onset. (MRC, FAC, MBC), clinical data including stroke type are collected (S310).

In addition, the data collection unit 110 also collects motor function scores (FAC, MBC) of a set type observed from the patient after a set period (eg, 6 months) from the onset date. The data collection unit 110 builds big data for model learning based on this.

Next, the model learning unit 120 uses a processor based on an artificial intelligence algorithm, uses the collected body data and clinical data for each patient as input data, and motor function scores (FAC, MBC) observed after a set period of time from the date of onset. ) as output data, the predictive model is trained based on the training data set (S320).

At this time, the model learning unit 120 may learn the model using a deep neural network (DNN), and the reliability of the model may be increased through learning of big data for related patients.

After learning is completed, a predicted value of a motor function score at a future time point may be provided in response to current factors for the patient to be analyzed.

Specifically, the data acquisition unit 130 acquires body data and currently observed clinical data of the patient to be analyzed (S330), and transfers them to the prediction unit 140. The data acquisition unit 130 may obtain or receive related data by interworking with at least one of a hospital server, an EMR, and a medical staff side terminal device.

The prediction unit 140 derives a motor function score (FAC, MBC) for the patient to be analyzed at a future time point after a set period elapses from the present by applying the body data and clinical data received in correspondence with the patient to be analyzed to the predictive model. Do (S340).

Thereafter, the classification unit 150 compares the predicted motor function score with a threshold value to classify whether or not recovery is possible for the corresponding type of motor function (S350).

For example, if the FAC is greater than or equal to a critical value (eg, 4 points), it can be determined that the recovery prognosis of lower extremity motor function for the patient to be analyzed after 6 months is good, and if it is less than that, it can be determined that the prognosis is bad. In addition, depending on whether the MBC is higher than or equal to a threshold value (eg, 5 points), the probability of recovery of the patient's upper extremity motor function after 6 months may be classified and provided. Of course, it is also possible to classify the recovery prognosis into multiple stages (high, medium, low, etc.) by assigning multiple thresholds to each case.

Thereafter, the output unit 160 may output the motor function score prediction result and the recovery possibility classification result based thereon through a display or the like (S360).

The following describes the results of comparing prediction performance according to the type of machine learning algorithm used to generate the predictive model.

First of all, FIG. 4 is a diagram showing detailed factors used in learning a predictive model according to an embodiment of the present invention by way of example. And, Figure 5 is a diagram comparing the performance results of the three algorithms used.

The three algorithms include a deep neural network (DNN), logistic regression, and random forest algorithm, and performance results for each of these algorithms can be confirmed through FIG. 5 .

All three algorithms were implemented as a python program and trained using TensorFlow 2.3.0 and scikit-learn toolkit 0.23.2. The data set was divided into 70% training data set, 21% validation data set, and 9% test data set. For statistical analysis, receiver operating characteristic curve (ROC curve) analysis was performed, and area under the curve (AUC) was calculated using scikit-learn.

In FIG. 5, the left and right sides represent prediction performance data for MBC scores and FAC scores, respectively. Comparing validation AUC values by algorithm, in the case of MBC score prediction, DNN was 0.996, logistic regression was 0.874, and random forest was 0.882. In the case of FAC score prediction, DNN was 0.882, logistic regression was 0.765, and random forest was 0.802. appear. From the results of FIG. 5, the DNN, random forest, and logistic regression algorithms showed good performance in the order, and in particular, the case where the prediction model was constructed using the DNN algorithm showed the best performance.

According to the present invention as described above, it is possible to predict the patient's motor function recovery prognosis for a future time point in advance by applying the current clinical data collected corresponding to the patient to the pre-learned predictive model.

In addition, based on the predicted prognosis of motor function recovery, it enables medical staff to more efficiently plan future treatment/rehabilitation programs for each patient and prepare for worsening symptoms, as well as suppress disease progression and efficiently manage disease. let it do

Although the present invention has been described with reference to the embodiments shown in the drawings, this is merely exemplary, and those skilled in the art will understand that various modifications and equivalent other embodiments are possible therefrom. Therefore, the true technical scope of protection of the present invention should be determined by the technical spirit of the appended claims.

100: prognosis prediction device for recovery of motor function
110: data collection unit 120: model learning unit
130: data acquisition unit 140: prediction unit
150: classification unit 160: output unit

Claims (14)

In the apparatus for predicting the prognosis of motor function recovery in stroke patients,
A data collection unit that collects physical data related to a plurality of patients for each patient, clinical data including multiple types of motor function scores related to the patient's affected limb and stroke type observed within a reference period from the date of onset;
a model learning unit learning a predictive model by taking the collected physical data and clinical data as input data and using motor function scores of a set type observed after a set period of time from the onset date as output data;
a data acquisition unit that acquires physical data and currently observed clinical data of the patient to be analyzed;
a prediction unit that applies the obtained physical data and clinical data to the predictive model to predict a motor function score of the set type for the patient to be analyzed at a future time point after a set period elapses from the present; and
A classification unit for comparing the predicted motor function score with a threshold value and classifying whether or not recovery is possible for the corresponding type of motor function;
The plurality of types of motor function scores used as input data of the prediction model,
Including joint score (MRC; Medical Research Council), gait score (FAC; Functional Ambulation Category) and upper limb movement score (MBC; Modified Brunnstrom classification),
The motor function score of the setting type used as the output data of the predictive model is,
An apparatus for predicting prognosis of motor function recovery in stroke patients, including the gait score (FAC) and the upper extremity motion score (MBC). delete The method of claim 1,
The reference period is within 30 days, and the set period is a prognosis prediction device for recovery of motor function in stroke patients of 6 months or more. The method of claim 1,
The clinical data,
A device for predicting the prognosis of motor function recovery in stroke patients, further including the presence or absence of motor evoked potentials of the abductor abductor and tibialis anterior muscle. delete The method of claim 1,
The classification unit,
Restoration of motor function of a stroke patient by comparing the gait score and upper extremity movement score with each preset threshold value to individually classify whether or not the possibility of recovery of the gait function and upper limb movement function of the patient subject to analysis for a future time point after the lapse of the set period has elapsed Prognostic prediction device. The method of claim 1,
The predictive model,
A device for predicting the prognosis of motor function recovery in stroke patients learned through a deep neural network (DNN) algorithm. In the method for predicting the prognosis of motor function recovery using a prognosis prediction device for motor function recovery in stroke patients,
Collecting physical data related to the patient for each of a plurality of patients, clinical data including multiple types of motor function scores and stroke types related to the affected limb of the patient observed within a reference period from the date of onset;
learning a predictive model using the collected physical data and clinical data as input data and motor function scores of a set type observed after a set period elapses from the onset date as output data;
Acquiring physical data and currently observed clinical data for the patient to be analyzed;
applying the acquired physical data and clinical data to the predictive model to predict a motor function score of the set type for a patient to be analyzed at a future time point after a set period elapses from the present; and
Comparing the predicted motor function score with a threshold value to classify whether recovery is possible for the corresponding type of motor function,
The plurality of types of motor function scores used as input data of the prediction model,
Including joint score (MRC; Medical Research Council), gait score (FAC; Functional Ambulation Category) and upper limb movement score (MBC; Modified Brunnstrom classification),
The motor function score of the setting type used as the output data of the predictive model is,
A method for predicting the prognosis of motor function recovery in stroke patients, including the gait score (FAC) and the upper extremity motion score (MBC). delete The method of claim 8,
The reference period is within 30 days, and the set period is a method for predicting the prognosis of recovery of motor function in stroke patients of 6 months or more. The method of claim 8,
The clinical data,
A method for predicting the prognosis of motor function recovery, further including the presence or absence of the motor evoked potential of the abductor abductor and tibialis anterior muscle. delete The method of claim 8,
The step of classifying whether or not there is a possibility of recovery for the corresponding type of motor function,
Restoration of motor function of a stroke patient by comparing the gait score and upper extremity movement score with each preset threshold value to individually classify whether or not the possibility of recovery of the gait function and upper limb movement function of the patient subject to analysis for a future time point after the lapse of the set period has elapsed Prognosis prediction method. The method of claim 8,
The predictive model,
A method for predicting the prognosis of motor function recovery in stroke patients learned through a deep neural network (DNN) algorithm.

Images