CN115153562A - Prediction method and device based on electroencephalogram, electronic equipment and storage medium - Google Patents

Abstract

The invention discloses a prediction method and device based on electroencephalogram, electronic equipment and a storage medium. The method comprises the following steps: acquiring electroencephalograms of ischemic stroke patients within a preset time period and human body information corresponding to the electroencephalograms; analyzing an electroencephalogram of a patient with ischemic stroke to obtain electroencephalogram indexes corresponding to the electroencephalogram, wherein the electroencephalogram indexes comprise RAP, RDP, DAR, DTABR and pdBSI; inputting human body information corresponding to the electroencephalogram indexes into the index model to be referred to obtain the indexes to be referred to in the electroencephalogram indexes; and matching a reference range in the standard database according to the value of the index to be referred, predicting the brain function recovery degree of the ischemic stroke patient within a preset interval according to the reference range, and displaying the brain function recovery degree as reference data to a doctor. The invention overcomes the problem that the doctor can not diagnose quickly because the CT and MRI acquire images for a long time, and provides other reference data except the CT and MRI brain images for the evaluation and treatment of the brain function of the patient after healing.

Description

Prediction method and device based on electroencephalogram, electronic equipment and storage medium

Technical Field

The embodiment of the invention relates to computer technology, in particular to a prediction method and device based on electroencephalogram, an electronic device and a storage medium.

Background

Ischemic Stroke (IS) IS the most common type of pathogenesis of neurological diseases. During the onset of IS, it IS possible to progress from ischemia to infarction in only a few minutes, and brain tissue becomes deficient in brain function to various degrees after IS onset, and irreversible brain damage occurs in severe cases.

At present, the technologies such as CT and MRI are mostly adopted to obtain the brain image of a patient, and a doctor judges the disease severity of an IS patient and the recovery degree after subsequent healing through the brain image, but the CT and MRI technologies have a long time period for obtaining the image, so that the doctor cannot quickly diagnose the state of an illness of the patient, the delay of the state of an illness can be caused, and the patient IS dead.

In addition, for the IS patients, doctors need to evaluate the brain injury after the IS patients are cured to determine the subsequent treatment scheme, and in the evaluation process, images acquired by CT and MRI technologies need to be combined to give professional evaluation results, so that the doctors provide guidance suggestions for the next step of recovery treatment of the patients. However, the appointment time of the CT and MRI techniques is long, and the time for acquiring the images is long, which may result in low efficiency of the doctor in seeing and examining.

Disclosure of Invention

The invention provides a prediction method, a prediction device, electronic equipment and a storage medium based on electroencephalogram, which are used for predicting the brain function condition of an ischemic stroke patient within a certain period of time.

In a first aspect, an embodiment of the present invention provides an electroencephalogram-based prediction method, where the method includes:

collecting electroencephalograms of ischemic stroke patients in a preset time period and human body information corresponding to the electroencephalograms;

analyzing the electroencephalogram of the ischemic stroke patient to obtain electroencephalogram indexes corresponding to the electroencephalogram, wherein the electroencephalogram indexes comprise RAP, RDP, DAR, DTABR and pdBSI;

inputting the human body information corresponding to the electroencephalogram index into a model of an index to be referred to obtain the index to be referred to in the electroencephalogram index;

and matching the brain function recovery degree corresponding to the reference range in the standard database according to the value of the index to be referred to obtain the predicted brain function recovery degree of the ischemic stroke patient within the preset interval, and displaying the predicted brain function recovery degree as reference data to a doctor.

Further, after the acquiring the electroencephalogram of the ischemic stroke patient within the preset time period, the method further comprises:

determining an electroencephalogram of the ischemic stroke patient within a second preset time period;

and performing independent component analysis on the electroencephalogram of the ischemic stroke patient in the second preset time period, and removing an eye movement component of the electroencephalogram of the ischemic stroke patient in the second preset time period. Further, the analyzing the electroencephalogram of the ischemic stroke patient to obtain an electroencephalogram index corresponding to the electroencephalogram includes:

determining the power of each frequency band in the electrical signal of each electrode according to the electrical signal of each electrode in the electroencephalogram;

and determining the average power value of each frequency band corresponding to the electroencephalogram according to the power of each frequency band in the electric signal of each electrode, and calculating the electroencephalogram index corresponding to the electroencephalogram.

Further, the acquiring the electroencephalogram of the ischemic stroke patient within the preset time period comprises the following steps:

and acquiring an electroencephalogram of the human body in a second preset period in the waking state in a preset period after the onset of a disease.

Further, the to-be-referenced index model is obtained in the following manner:

acquiring predicted electroencephalogram indexes of a historical ischemic stroke patient to be trained and human body information of the historical ischemic stroke patient;

and training a deep learning classification network for preset times according to the predicted electroencephalogram index of the historical ischemic stroke patient to be trained and the human body information of the historical ischemic stroke patient until the deep learning classification network is converged to obtain the index model to be referred.

Further, the obtaining mode of the prediction electroencephalogram index of the historical ischemic stroke patient to be trained is as follows:

carrying out statistical analysis on the electroencephalogram indexes of the historical ischemic stroke patients and the electroencephalogram indexes of the contrast group by using an inter-group contrast mode matched with human body information to obtain electroencephalogram indexes with statistical significance as electroencephalogram indexes to be predicted;

grouping different brain function degrees corresponding to the prediction result of the electroencephalogram index to be predicted and the post-healing evaluation result of the historical ischemic stroke patient according to the different brain function degrees;

obtaining a mapping parameter corresponding to the to-be-predicted index according to the prediction result of the to-be-predicted electroencephalogram index and the number of different groups of people of the post-healing evaluation result of the historical ischemic stroke patient;

and drawing an ROC (optimum characteristic) graph according to the drawing parameters to obtain a prediction probability value corresponding to the index to be predicted, and determining the prediction electroencephalogram index according to the area under the curve of the prediction probability value.

Further, the reference range in the standard database is obtained as follows:

determining a prognosis score of the historical ischemic stroke patient in a preset interval according to a scoring rule, wherein the prognosis score predicts the brain function recovery degree of the historical ischemic stroke patient in the preset interval;

and generating a reference range corresponding to each brain function recovery degree according to the brain function recovery degree of the historical ischemic stroke patient and the pre-measured electroencephalogram indexes of the historical ischemic stroke patient in a preset interval, and storing the reference range in the standard database.

In a second aspect, an embodiment of the present invention provides an electroencephalogram-based prediction apparatus, including:

the image acquisition module is used for acquiring electroencephalograms of ischemic stroke patients within a preset time period and human body information corresponding to the electroencephalograms;

the image analysis module is used for analyzing the electroencephalogram of the ischemic stroke patient to obtain electroencephalogram indexes corresponding to the electroencephalogram, wherein the electroencephalogram indexes comprise RAP, RDP, DAR, DTABR and pdBSI;

the index determining module is used for inputting the human body information corresponding to the electroencephalogram index into an index model to be referred to obtain an index to be referred in the electroencephalogram index;

and the degree determining module is used for matching the brain function recovery degree corresponding to the reference range in the standard database according to the value of the index to be referred to, so as to obtain the predicted brain function recovery degree of the ischemic stroke patient within the preset interval.

In a third aspect, an embodiment of the present invention provides an electronic device, including:

one or more processors;

a storage device to store one or more programs,

when executed by the one or more processors, cause the one or more processors to implement the electroencephalogram-based prediction method.

In a fourth aspect, embodiments of the present invention provide a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements an electroencephalogram-based prediction method as described.

In the embodiment of the invention, electroencephalograms of ischemic stroke patients in a preset time period and human body information corresponding to the electroencephalograms are collected; analyzing an electroencephalogram of a patient with ischemic stroke to obtain electroencephalogram indexes corresponding to the electroencephalogram, wherein the electroencephalogram indexes comprise RAP, RDP, DAR, DTABR and pdBSI; inputting human body information corresponding to the electroencephalogram indexes into the index model to be referred to obtain indexes to be referred to in the electroencephalogram indexes; and matching a reference range in the standard database according to the value of the index to be referred, predicting the brain function recovery degree of the ischemic stroke patient within a preset interval according to the reference range, and displaying the predicted brain function recovery degree as reference data to a doctor. The embodiment of the invention solves the problem that a doctor cannot quickly diagnose the state of an illness of a patient due to long time period for acquiring images by CT and MRI technologies, and provides other reference data except CT and MRI brain images for the evaluation and treatment of the brain function of the patient with acute ischemic stroke, so that the doctor can know the state of the illness and the recovery condition of the patient more comprehensively and quickly.

Drawings

FIG. 1 is a schematic flow chart diagram of an electroencephalogram-based prediction method provided in accordance with an embodiment of the present invention;

FIG. 2 is another schematic flow diagram of an electroencephalogram-based prediction method provided by an embodiment of the present invention;

FIG. 3 is a schematic diagram of an electroencephalogram-based predictive device according to an embodiment of the present invention;

fig. 4 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.

Detailed Description

The present invention will be described in further detail with reference to the accompanying drawings and examples. It is to be understood that the specific embodiments described herein are merely illustrative of the invention and are not limiting of the invention. It should be further noted that, for the convenience of description, only some structures related to the present invention are shown in the drawings, not all of them.

Fig. 1 is a schematic flow chart of an electroencephalogram-based prediction method according to an embodiment of the present invention, which is applicable to a case where it is necessary to evaluate the degree of recovery of brain function of an ischemic stroke patient, and which can be performed by an electroencephalogram-based prediction apparatus, which can be integrated in a server. As shown in fig. 1, the method comprises the steps of:

step 110, collecting electroencephalograms of ischemic stroke patients in a preset time period and human body information corresponding to the electroencephalograms;

for example, the electroencephalogram of the ischemic stroke patient may be a waveform of the electrical brain signal of the ischemic stroke patient acquired by the electroencephalogram acquisition device over time, and the waveform may include waves of all frequencies within a frequency band range corresponding to the acquisition device. The brain electricity collecting device can be an electrode cap, and can also be an electronic device consisting of a plurality of dispersed electrode beams and used for collecting electric signals of the brain of a human body; when the electroencephalogram acquisition equipment is an electrode cap, an ischemic stroke patient can wear the electrode cap to acquire an electroencephalogram signal diagram within a preset time period. The preset time period can be understood as a time period in which the electroencephalogram signal of the ischemic stroke patient in the preset state changes obviously according to actual requirements or experimental data, the medical reference value of the electroencephalogram of the ischemic stroke patient collected in the preset time period is higher, namely, the electroencephalogram change chart of the ischemic stroke patient in the attack state can be monitored, for example, the attack time of the ischemic stroke patient is within 72 hours. The electroencephalogram corresponding human body information can be understood as human body characteristic information of the ischemic stroke patient, such as: age, sex, past medical history (hypertension, diabetes, hyperlipidemia), smoking history, infarct site, treatment and medication conditions, etc.

In the specific implementation, electroencephalograms of patients with ischemic stroke in a preset time period can be collected, electroencephalograms with sampling rate of 512HZ are collected by adopting an electroencephalogram cap, the collection frequency range can be set to be 1-100Hz, the collection time is 10 minutes, and the preset time period can be set to be 72 hours. The collecting equipment can also be a Nicolet EEG digital electroencephalograph, the electrodes of the electroencephalogram equipment are disc-shaped Ag/AgCl electrodes and are placed according to an international 10-20 system, the number of the electrodes of the electroencephalogram equipment can be more than or equal to 19, and the impedance between a human body and the electrodes is less than 10k omega; when the electroencephalogram of the ischemic stroke patient in the preset time period is collected, the ischemic stroke patient is in a quiet, awake and eye-closing state, and the name of the ischemic stroke patient needs to be inquired every 2 minutes to determine whether the ischemic stroke patient is awake or not so as to evaluate the awakening state.

Step 120, analyzing an electroencephalogram of the ischemic stroke patient to obtain electroencephalogram indexes corresponding to the electroencephalogram, wherein the electroencephalogram indexes comprise RAP, RDP, DAR, DTABR and pdBSI;

illustratively, the brain electrical index may be a signal feature extracted from the brain electrical signal. The electroencephalogram signal may be a signal of any one of the electrodes recorded in the electroencephalogram, or may be a signal of a part of or all of the electrodes recorded in the electroencephalogram. The signal characteristics can be the power of the electroencephalogram signal in different frequency ranges in the frequency domain, and can also be other magnitudes such as the amplitude, the phase and the like of the electroencephalogram signal in different frequency ranges in the frequency domain. The electroencephalogram frequency spectrum can be characterized by power within a specific frequency range, an electroencephalogram which is not diseased by a patient only has the characteristics of one electroencephalogram frequency spectrum, and the characteristic number of the electroencephalogram frequency spectrum can be changed when the patient is diseased, so that the characteristics of the electroencephalogram frequency spectrum become important indexes for clinical diagnosis and doctor research. The frequency can be the number of times of the periodic change of the electroencephalogram signal in unit time and is used for describing the frequency degree of periodic motion, the frequency domain can be the electroencephalogram signal power or other signal magnitude value of the electroencephalogram signal converted from the time domain to the frequency domain through a time-frequency conversion method, the time domain can be the change relation of the electroencephalogram signal to time, the conversion method can be Fourier transform, and the Fourier transform decomposes the electroencephalogram signal into a frequency spectrum so as to display the amplitude corresponding to the frequency. When some features in the frequency domain are extracted from the electroencephalogram signal, the electroencephalogram signal needs to satisfy a certain condition. The condition can be that certain signal quality is achieved, the signal quality can be obtained by processing the electroencephalogram signals, the processing of the electroencephalogram signals can be achieved through software, and the software can be either analyzer or eeglab and the like.

In the specific implementation, an electroencephalogram with less artifact in a first preset time period is selected from electroencephalogram records of a patient, the electroencephalogram with less artifact in the selected first preset time period is filtered, the electroencephalogram with less artifact in the first preset time period is used for extracting electroencephalograms in a required range from the electroencephalograms with less artifact in the first preset time period, and noise and power frequency interference are filtered. And looking up the electro-ocular artifacts in the electroencephalogram after filtering, removing the electro-ocular artifacts from the electroencephalogram with more electro-ocular artifacts, and extracting the power of the electroencephalogram in different frequency ranges in a frequency domain. The power extraction in different frequency ranges is performed on the signals of the electrodes corresponding to the identified electroencephalogram. The 120s electroencephalogram signal can be averagely divided into 2s segments (1024 data points), 50% overlap is performed, then Fast Fourier Transform (FFT) is applied to calculate the power of each lead of 4 frequency bands (delta (1-4 Hz), theta (5-7 Hz), alpha (8-13 Hz) and beta (14-30 Hz)), and then the power of all leads in each frequency band is averaged to obtain the average power value after all leads in each frequency band are averaged. And calculating to obtain 5 electroencephalogram indexes RAP, RDP, DAR, DTABR and pdBSI based on the lead power of each frequency band and the average power value of all leads of each frequency band after averaging.

Step 130, inputting human body information corresponding to the electroencephalogram indexes into a model of indexes to be referenced to obtain indexes to be referenced in the electroencephalogram indexes;

for example, the to-be-referenced index model may be understood as a model for training a historical ischemic stroke patient in a historical database through a machine learning algorithm, and is used for identifying the to-be-referenced index for predicting the brain function degree corresponding to the electroencephalogram according to the human body information.

In the specific implementation, 5 electroencephalogram indexes RAP, RDP, DAR, DTABR and pdBSI are calculated based on the lead power of each frequency band and the average power value obtained by averaging all leads of each frequency band. The human body information of the ischemic stroke patient corresponding to the 5 electroencephalogram indexes RAP, RDP, DAR, DTABR and pdBSI is input into the index model to be referred, and the index model to be referred identifies the index to be referred in the electroencephalogram indexes corresponding to the human body information of the ischemic stroke patient. The body information of the ischemic stroke patient can be body information of the ischemic stroke patient, such as: age, sex, past history (hypertension, diabetes, hyperlipidemia), smoking history, infarct site, treatment and medication.

And 140, matching the brain function recovery degree corresponding to the reference range in the standard database according to the value of the index to be referred to obtain the predicted brain function recovery degree of the ischemic stroke patient within a preset interval, and displaying the predicted brain function recovery degree as reference data to a doctor.

For example, the standard database may be understood as a database storing the brain function recovery degrees of the ischemic stroke patients within the preset interval and the corresponding reference range to be referred to, and the brain function recovery degrees of the ischemic stroke patients within the preset interval and the corresponding reference range to be referred to are stored in association according to the difference in the brain function recovery degrees of the ischemic stroke patients within the preset interval, which may be in the form of a comparison table or a data chain. The reference range and the brain function recovery degree of the ischemic stroke patient in the preset interval are stored in the standard database in a correlation mode, and the number of people with the reference range is correspondingly determined.

In the specific implementation, the index to be referred in the electroencephalogram index corresponding to the human body information of the ischemic stroke patient can be identified according to the index model to be referred, the brain function recovery degree corresponding to the reference range where the index to be referred is located is matched in the standard database, and the predicted brain function recovery degree of the ischemic stroke patient within the preset interval is obtained. The predicted brain function recovery degree is displayed to a doctor as reference data, so that the doctor can conveniently carry out subsequent diagnosis and treatment.

According to the embodiment of the invention, electroencephalograms of ischemic stroke patients in a preset time period and human body information corresponding to the electroencephalograms are collected; analyzing an electroencephalogram of a patient with ischemic stroke to obtain electroencephalogram indexes corresponding to the electroencephalogram, wherein the electroencephalogram indexes comprise RAP, RDP, DAR, DTABR and pdBSI; inputting human body information corresponding to the electroencephalogram indexes into the index model to be referred to obtain indexes to be referred to in the electroencephalogram indexes; and matching a reference range in the standard database according to the value of the index to be referred, predicting the brain function recovery degree of the ischemic stroke patient within a preset interval according to the reference range, and displaying the predicted brain function recovery degree as reference data to a doctor. The embodiment of the invention solves the problem that a doctor cannot quickly diagnose the condition of a patient due to long time period for acquiring images by CT and MRI technologies, and provides other reference data except CT and MRI brain images for the evaluation and treatment of the brain function of the patient with acute ischemic stroke after healing, so that the doctor can more comprehensively and quickly know the condition and recovery condition of the patient.

The electroencephalogram-based prediction method provided by the embodiment of the present invention is further described below, and as shown in fig. 2, the method may specifically include the following steps:

step 210, collecting electroencephalograms of ischemic stroke patients in a first preset time period and human body information corresponding to the electroencephalograms;

illustratively, the primary cause of disability IS neurological disease, which IS also the second leading cause of death worldwide, with Ischemic Stroke (IS) being the most common type of morbidity. During ischemic stroke, brain tissues are subjected to irreversible damage caused by ischemia and hypoxia, so that a series of brain dysfunction is caused, and functional defects with different degrees are left. Despite advances in stroke prevention, treatment, rehabilitation, etc., stroke burden is expected to continue to rise, leaving up to 70% of stroke patients with moderate and severe dysfunction. The progression from ischemia to infarction is a dynamic, rapidly progressing process that changes irreversibly within minutes to hours. Early diagnosis of IS therefore of vital importance, it affects patient treatment, rehabilitation, quality of life and medical costs after IS both short-term and long-term. Although imaging modalities such as CT and MRI can perform detailed anatomical examinations of the brain and assess IS severity. But all have the limitations that in ischemic stroke, CT has a 6-8 hour time window; MRI, while capable of identifying stroke within 30 minutes, is less available, often with longer appointments, and cannot be applied in a timely manner. Electroencephalography (EEG) can reflect changes in cerebral blood flow and metabolism in a few seconds, as evidenced by increased slow wave activity and decreased fast wave activity. Although EEG has poor spatial resolution, its high temporal resolution helps to assess instantaneous brain function quickly. Because of its superior temporal resolution and high sensitivity to ischemic hypoxia, several studies have shown that electroencephalography can be used in AIS assisted diagnosis, disease monitoring, and prognosis. At present, a clinician judges the prognosis only according to the clinical expression in the acute stage and the neurological function score and combines the neural image in the morbidity and the high-risk factors of a patient, and objective quantitative indexes are lacked. The result of the electroencephalogram indexes makes up the defects, and is more objective, visual, quantitative and scientific than the factors.

The first preset time period means that within 72 hours after the onset of the disease, 72 hours are supported by literature, and electroencephalograms within 72 hours are better related to prognosis.

The electroencephalogram signal is sent to the input part of an electroencephalogram machine through the electroencephalogram electrode arranged on the recording part, and is amplified by an amplifier to form an electroencephalogram. The contact between the electrode and the scalp affects the amount of electrical resistance between the electrode and the skin. The smaller the skin resistance of the electrode, the smaller the introduced alternating current interference, and the higher and more stable the waveform quality, which is generally less than 10K ohm. The electroencephalogram signal frequency is low, the electroencephalogram corresponding to the general electroencephalogram is in the range of 1-100Hz, and high-frequency interference except the electroencephalogram signal frequency needs to be filtered.

In the specific implementation, a Nicolet EEG digital electroencephalograph is used for collecting an electroencephalogram of an acute ischemic stroke patient within 72 hours of being in a waking state within 3 months after the onset of a disease. When the electroencephalogram acquisition equipment is used for acquiring electroencephalograms, an electroencephalogram cap with the sampling rate of 512HZ is adopted for acquisition, electroencephalograms with the acquisition time of 10 minutes and corresponding electroencephalograms are in the range of 1-100Hz, the impedance between a human body and electrodes is less than 10k omega, and the number of the acquisition electrodes is equal to 19.

And step 220, determining an electroencephalogram of the human body in a second preset time period in the waking state in the first preset time period after the occurrence of the disease.

Further, after the electroencephalogram of the ischemic stroke patient within the second predetermined time period, the method further includes:

and performing independent component analysis on the electroencephalogram of the ischemic stroke patient in the second preset time period, and removing an eye movement component of the electroencephalogram of the ischemic stroke patient in the second preset time period.

The independent component analysis method is illustratively a signal processing method that utilizes high order statistics, which is a linear transformation that minimizes statistical dependence between components and identifies individual components in the signal. ICA has been applied to electroencephalogram signal processing, and blink artifact is common noise in an electroencephalogram signal acquisition process and seriously influences the extraction of useful information of the electroencephalogram signal, so that eye movement is considered as noise in the electroencephalogram signal, people adopt a plurality of methods for inhibiting the influence of the noise for a long time, and an independent component analysis method can identify eye movement components in the signal. The reference electrode of the electroencephalography data can be reset to the common average reference. The index of the reference electrode recorded in the electroencephalogram is a voltage value, which is a potential difference between two electrodes, so that one electrode is a reference electrode in addition to a single electrode attached to a brain portion during the recording of the electroencephalogram. The common average reference method is to eliminate the error influence caused by the change of the original reference electrode and average the sum of all electrode signals to obtain an inherent value;

in specific implementation, the second preset time period refers to a time range of 120 seconds of the electroencephalogram signal with less artifact randomly selected within 72 hours after the onset of the disease. The electroencephalogram over the second preset period of time may be analyzed using the independent component analysis method option in the analyzer software. According to the characteristics that eye movement signals are distributed at the front end of a scalp topographic map (generally right ahead), small squares (relatively centralized), randomly distributed, high low-frequency energy in a power spectrogram and relatively early component sequencing, after independent component analysis, independent components are visually identified, and components considered as eye movement are selected according to experience and removed.

Step 230, analyzing the electroencephalogram of the ischemic stroke patient to obtain electroencephalogram indexes corresponding to the electroencephalogram, wherein the electroencephalogram indexes comprise RAP, RDP, DAR, DTABR and pdBSI;

further, the analyzing the electroencephalogram of the ischemic stroke patient to obtain an electroencephalogram index corresponding to the electroencephalogram includes:

determining the power of each frequency band in the electrical signal of each electrode according to the electrical signal of each electrode in the electroencephalogram;

and determining the average power value of each frequency band corresponding to the electroencephalogram according to the power of each frequency band in the electric signal of each electrode, and calculating the electroencephalogram index corresponding to the electroencephalogram.

For example, electroencephalograms require an experienced professional to interpret the image, and may lead to different conclusions. The Quantitative electroencephalogram (qEEG) is used for converting electroencephalogram amplitude changing along with time into electroencephalogram power changing along with frequency through Fourier transform and other modes to obtain absolute power values of delta (1-4 Hz), theta (5-7 Hz), alpha (8-13 Hz) and beta (14-30 Hz) within a certain time. The ratio of the absolute power value of different frequencies to the total power value of each frequency band is the relative power ratio of different frequency bands. The relative power ratio can quantitatively reflect the distribution, proportion and amplitude change of brain waves of each frequency band. Compared with the common electroencephalogram, QEEG is not influenced by subjective factors, and the result is more accurate and quantitative. Electroencephalogram indices commonly used today include the relative alpha power Ratio (RAP), the relative delta power Ratio (RDP), the delta to alpha power ratio (DAR), (delta + theta) to (alpha + beta) power ratio (DTABR), asymmetry index (BSI), and the like.

BSI reflects bilateral brain symmetry, high sensitivity, ranging from 0 (fully symmetric) to 1 (fully asymmetric), and is associated with AIS clinical severity. Sheorajpanday et al 2009 introduced a pairwise derived asymmetric index (pdBSI), which is a natural extension of BSI, as an indicator for monitoring AIS patient condition. Compared to BSI, pdBSI is an asymmetric parameter that assesses the power spectrum of homologous channels, rather than global asymmetry, increases the sensitivity of monitoring, reliably distinguishes patients with stroke and transient ischemic attacks or control subjects, and is significantly correlated with clinical severity and recent ischemic volume. Due to its superior temporal resolution and high sensitivity to ischemia and hypoxia, several studies have shown that quantitative electroencephalography can be used for AIS-assisted diagnosis, disease monitoring, and prognostic prognosis.

DTABR is a mark of whole brain function, is sensitive to delta activity in an infarct core area, can reduce alpha activity in areas such as an ischemic penumbra and is sensitive to theta activity increase, and therefore the accuracy of identifying AIS by DTABR is higher. DTABR reflects not only δ activity but also θ activity and α activity, possibly reflecting conditions such as edema of brain tissue.

DAR is apparently related to DTABR, quantifying the overall signal strength for delta and alpha activity. Some studies have shown that there is a correlation between delta power and alpha power and their associated power and 30-day functional outcomes during the acute phase, and it is likely that predictions can be made for AIS patients. DAR has higher discrimination ability in distinguishing AIS and predicting functional prognosis. DAR quantifies delta and alpha activity. δ is most strongly correlated with CBF and metabolism. Beta is easily polluted by myoelectric artifacts, theta alone is unreliable as a pathophysiological index after stroke, and faster theta (6-8 Hz) can be slow alpha in background waves, so DAR has higher discrimination capability in AIS diagnosis.

Early studies found that gradual attenuation of alpha wave activity was associated with gray matter ischemia, parallel to the progression of infarction from apical cortex to sensorimotor cortex in the subacute phase, and that unlike delta wave activity, recovery of alpha wave activity was better in the late phase of infarction, and preservation of its activity could reflect neuronal survival following infarction. RAP was strongly negatively correlated with the 30 day NIHSS score.

The infarct core zone presents delta activity, an irreversible lesion, quantified by DAR and RDP, and therefore the higher accuracy BSI can reflect the symmetry of the left and right hemisphere brain electricity, quantifying the difference in the average spectral power of each hemisphere.

In a specific implementation, the absolute power of each lead of each frequency band is determined according to the signal of each electrode corresponding to the electroencephalogram, and for an ischemic stroke patient, 5 electroencephalogram indexes RAP, RDP, DAR, DTABR and pdBSI are calculated based on the lead power of each frequency band and the average power value obtained by averaging all leads of each frequency band, that is, 5 indexes of each patient can obtain a value respectively.

The name relative alpha power ratio RAP is calculated based on the following formula:

the name relative delta power ratio RDP is calculated based on the following formula:

the name delta to alpha power ratio DAR is calculated based on the following formula:

the name (δ + θ) to (α + β) power ratio DTABR is calculated based on the following formula:

wherein, delta, theta, alpha and beta respectively represent average power values after all leads of delta (1-4 Hz), theta (5-7 Hz), alpha (8-13 Hz) and beta (14-30 Hz) frequency bands are averaged, and the power values are directly analyzed by an analyzer.

Calculating an asymmetric index pdBSI derived by name pair based on the following formula

Wherein Lij and Rij respectively represent the power spectral densities of i frequencies of j channels of left and right hemispheres calculated based on FFT, and then take values according to frequencies i =1,2.. N, N =4, i =1 representing a frequency range of 1-3Hz, i =2 representing a frequency range of 1-7Hz, i =3 representing a frequency range of 1-13Hz, and i =4 representing a frequency range of 1-25 Hz; j =1, 2.. Said, M =8, M represents 8 pairs of homologous channels, for example, F3 and F4 are one pair of homologous channels.

Wherein α, β, θ and δ represent average power values averaged over all leads of the α, β, θ and δ bands, respectively.

Step 240, inputting the human body information corresponding to the electroencephalogram indexes into a model of indexes to be referenced to obtain indexes to be referenced in the electroencephalogram indexes;

further, the to-be-referenced index model is obtained in the following manner:

acquiring predicted electroencephalogram indexes of a historical ischemic stroke patient to be trained and human body information of the historical ischemic stroke patient;

and training a deep learning classification network for preset times according to the predicted electroencephalogram index of the historical ischemic stroke patient to be trained and the human body information of the historical ischemic stroke patient until the deep learning classification network is converged to obtain the index model to be referred.

Illustratively, the acquired human body information of the historical ischemic stroke patients to be trained has the characteristics of meeting range standards and not meeting screening conditions, human body clinical data including the conditions of age, sex, past medical history (hypertension, diabetes and hyperlipidemia), smoking history and the like are recorded, and AIS patients collect infarct positions, treatment and medication conditions of the patients.

In a specific implementation, the human body information of the historical ischemic stroke patient includes:

all patients met AIS diagnostic criteria established by the neurological division of the chinese medical society in 2018; first onset, and within 72 hours CT or MRI demonstrated pre-circulating stroke; the stroke scale (NHISS) score of the national institute of health is more than or equal to 4 points; not receiving thrombolytic therapy; independent behavioral capacity was obtained prior to admission. Hypertension, diabetes and smoking history all accord with the human body information of the history database.

In addition, the human body information that the historical ischemic stroke patient should not have includes: people with severe complications of the heart, liver, kidney and digestive system, disturbance of consciousness or dementia; other diseases causing brain electrical changes exist, such as metabolic encephalopathy, epilepsy, cerebral hemorrhage, tumor, encephalitis, stroke and the like; taking drugs that may affect the electroencephalogram, such as antiepileptic drugs, barbiturates, lithium, tricyclic drugs, or antipsychotic drugs.

The prediction electroencephalogram indexes of the historical ischemic stroke patients to be trained are acquired as RAP, RDP, DAR, DTABR and pdBSI, wherein each prediction electroencephalogram index of each patient in the historical ischemic stroke patients has one value, and therefore each patient in the historical ischemic stroke patients has five indexes, namely five values.

Illustratively, the predicted electroencephalogram index of the historical ischemic stroke patient to be trained and the human body information of the historical ischemic stroke patient train the deep learning classification network for a preset number of times until the deep learning classification network converges. The deep learning classification network can be a two-classification logistic regression analysis, and can also be other deep learning classification methods.

In specific implementation, the deep learning classification network may use binary logistic regression analysis, and when using binary logistic regression analysis, the age and NHISS scores are used as covariates, and a certain electroencephalogram index (such as DTABR) is used as a variable and is input into binary logistic regression analysis of SPSS software together, and is output as a constant term, an age coefficient, a NHISS score coefficient, and a certain predicted electroencephalogram index coefficient, a table (which refers to a table with mapping parameters obtained by summarizing the number of people of two groups of good prognosis and bad prognosis predicted by logistic regression analysis and the number of people of the good prognosis and bad prognosis predicted prognosis according to the mRS score) in a certain prediction electroencephalogram index prognosis model, and a significant difference P value.

According to the deep learning classification network, namely, the coefficient and constant term of each prediction EEG index obtained by the binary logistic regression analysis, a prediction model of the functional prognosis of 4 indexes of the acute ischemic stroke patient is constructed, and each index can establish a prognosis model, namely P = 1/(1 + e) ^-Y ) Y = constant + coefficient + age + coefficient + NHISS + coefficient a certain electroencephalogram index to be predicted. The respective prognosis models for the respective indices are then stored for invocation.

Further, the obtaining mode of the prediction electroencephalogram index of the historical ischemic stroke patient to be trained is as follows:

carrying out statistical analysis on various indexes of the poor prognosis group and the good prognosis group of the historical ischemic stroke patient by using an inter-group comparison mode matched with human body information to obtain an electroencephalogram index with statistical significance as an electroencephalogram index to be predicted;

grouping different brain function degrees corresponding to the prediction result of the electroencephalogram index to be predicted and the prognosis evaluation result of the historical ischemic stroke patient according to the different brain function degrees;

obtaining a drawing parameter corresponding to the to-be-predicted index according to the number of people with different groups of prediction results of the to-be-predicted electroencephalogram index and the prediction evaluation results of the historical ischemic stroke patients;

and drawing an ROC (rock characteristic) graph according to the drawing parameters to obtain the area under the curve corresponding to the index to be predicted, and determining the predicted electroencephalogram index according to the area under the curve.

Illustratively, binary logistic regression is one of the most classical methods of machine learning, which takes as input a feature vector and outputs a probability, i.e., a value between 0 and 1, and classifies samples according to the probability, including binary and multi-class problems, wherein the binary logistic regression model outputs a value representing the probability that a sample belongs to one class, and the binary logistic regression analysis yields the respective number of samples of the two classes with the two classes of features as input.

Dividing the patients into a poor prognosis group and a good prognosis group according to the mRS score, comparing clinical data of the good prognosis group and the poor prognosis group between groups, and adjusting the variables with statistical significance as known influence factors. And (3) after the influence factors are adjusted, constructing a prediction model of the acute ischemic stroke patient functional prognosis by adopting binary logistic regression analysis on the electroencephalogram index, and obtaining an index with statistical significance as the reference index after the acute-stage prediction is poor.

In a specific implementation, various indexes of a poor prognosis group and a good prognosis group of the historical ischemic stroke patient are subjected to statistical analysis by using an inter-group comparison mode matched with human body information, and all statistical treatments can be carried out by using SPSS 22.0.

Wherein, the clinical data sex, hypertension, diabetes, hyperlipemia, smoking history and mRS score in the human body information are counting data, and the age, NHISS score and ASPECTS score in the human body information are metering data. Counting data is represented by frequency or percentage, and comparison between groups is carried out by chi-square test or Fisher exact probability method; the measured data is expressed by mean plus or minus standard deviation when conforming to normal distribution, the independent sample t test is used for comparing between groups, the independent sample rank and test are used for comparing between groups when not conforming to normal distribution and expressed by median (interquartile distance).

According to the mRS score of the patients after 3 months, the patients are divided into two groups, namely a good prognosis group and a bad prognosis group, wherein the patients with mRS > =3 points are classified into the bad prognosis group, and the patients with mRS < =2 points are classified into the good prognosis group.

First, clinical data of good prognosis and poor prognosis groups of stroke patients are compared among groups, the clinical data comprise sex, age, hypertension, diabetes, hyperlipidemia, smoking history, ASPECT score and NHISS score, and the result shows that the age and the NHISS score have statistical difference, so the age and the NHISS score are known influencing factors.

And then, taking the age and the NHISS scores as covariates, inputting a certain electroencephalogram index (such as DTABR) as a variable together, carrying out two-classification logistic regression analysis, and outputting a corresponding constant item, an age coefficient, an NHISS score coefficient, an electroencephalogram index (such as DTABR) coefficient, a table (the number of people in two groups of good prognosis and bad prognosis predicted by logistic regression analysis and a table with a drawing parameter obtained by summarizing the number of people in the two groups of good prognosis and bad prognosis according to the mRS scores) and a significant difference P value. Therefore, five indexes of DTABR, DAR, RDP, RAP and pdBSI can be subjected to five logistic regression analyses, and corresponding constant terms, age coefficients, NHISS score coefficients, index coefficients, tables (tables with mapping parameters, which are formed by summarizing two groups of people with good prognosis and poor prognosis through logistic regression analysis and two groups of people with good prognosis and people with poor prognosis according to mRS scores) and significant difference P values in the prognosis models of the five indexes are obtained. Therefore, five indexes including DTABR, DAR, RDP, RAP and pdBSI can generate five prognosis models.

And selecting an index with the significance difference value p <0.05 as an index to be predicted of the functional prognosis of the patient after 3 months according to the significance difference value p obtained by the logistic regression analysis. Therefore, according to the significance difference value p obtained by the logistic regression analysis, DTABR, DAR, RDP and RAP are determined as indexes to be predicted (p is less than 0.05).

For example, to avoid multiple collinearity of electroencephalogram indices, i.e., the possibility of inter-influence between indices, all models are constructed using a qEEG index, i.e., any electroencephalogram index, and known influencing factors.

A binary classification problem, examples can be classified into Positive classes (Positive) or Negative classes (Negative), but in practice, four cases occur in classification, and four indexes, namely True classes (TP), false Negative classes (FN), false Positive classes (FP), and True Negative classes (TN), can be obtained for the four cases. Wherein, the value of the real class is a positive class in a certain example, and the prediction result is the number of the positive classes; the number of the false negative classes is predicted as the number of the positive classes in a certain example; the number of false positive classes is predicted as the number of negative classes in a certain example; the value of the true negative class is a negative class in one example, and the prediction result is the number of the negative classes. After four values of TP, FN, FP, TN were obtained, the sensitivity and specificity could be calculated. Wherein, the sensitivity is the proportion of correctly judging actual sick people as true positive, and the ability of detecting sick people in a certain test is measured, and the calculation formula is as follows: sensitivity = true rate (TPR) = TP/(TP + FN); the specificity is the proportion of correctly judging actual patients as true negative, and the ability of correctly judging the patients without disease in the test is measured, and the calculation formula is as follows: specificity = 1-False Positive Rate (FPR) = TN/(FP + TN).

And then drawing an ROC curve according to the sensitivity and the specificity. Wherein, the ROC curve is a characteristic curve (receiver operating characteristic curve) of the operation of the testee, is a comprehensive index reflecting continuous variables of sensitivity and specificity, is a mutual relation of sensitivity and specificity disclosed by a mapping method, a series of sensitivity and specificity are calculated by setting the continuous variables into a plurality of different critical values, then the sensitivity is used as a vertical coordinate, and (1-specificity) is used as a horizontal coordinate to draw a curve, and the larger the area below the curve is, the higher the diagnosis accuracy is. Wherein, the area under the curve has lower accuracy when being 0.5-0.7, has certain accuracy when being 0.7-0.9, and has higher accuracy when being more than 0.9.

In the specific implementation, 4 indexes of DTABR, DAR, RDP and RAP are determined as the electroencephalogram indexes to be predicted of the functional prognosis of the patient after 3 months according to the significant difference value p obtained by the logistic regression analysis.

Grouping according to different brain function degrees corresponding to the prediction result of the electroencephalogram index to be predicted and the real prognosis evaluation result of the historical ischemic stroke patient, drawing respective ROC (rock characteristic curve) curves of the indexes according to the grouping condition, and drawing one ROC curve for each index of the 4 electroencephalogram indexes to be predicted.

Firstly, the different brain function degrees corresponding to the prediction result of the electroencephalogram index to be predicted and the prognosis evaluation result of the historical ischemic stroke patient are divided into two groups of good prognosis and bad prognosis. Obtaining the real number of people with good prognosis and poor prognosis according to the mRS score, and obtaining the number of people with good prognosis and poor prognosis and parameters in a table (the number of people with good prognosis and poor prognosis predicted by logic regression analysis and the number of people with good prognosis and poor prognosis summarized according to the mRS score), namely the mapping parameters corresponding to the index to be predicted comprise TP, FN, FP and TN.

TP represents the number of patients with poor actual prognosis and poor prognosis predicted by a certain electroencephalogram index in binary logistic regression analysis, FN represents the number of patients with poor actual prognosis and good prognosis predicted by a certain electroencephalogram index in binary logistic regression analysis, FP represents the number of patients with good actual prognosis and poor prognosis predicted by a certain electroencephalogram index in binary logistic regression analysis, and TN represents the number of patients with good actual prognosis and good prognosis predicted by a certain electroencephalogram index in binary logistic regression analysis.

Next, an ROC graph is plotted based on the four values of the plotting parameters TP, FN, FP, TN. The process comprises the following steps: firstly, the sensitivity and specificity of a certain electroencephalogram index are calculated according to four values of drawing parameters TP, FN, FP and TN. Here, the sensitivity and the specificity were calculated using the formulas of sensitivity = true rate (TPR) = TP/(TP + FN), and specificity = 1-False Positive Rate (FPR) = TN/(FP + TN). Then, a plurality of different critical values are set for the continuous variables, so that a series of sensitivity and specificity are calculated, and then the sensitivity is used as an ordinate and the (1-specificity) is used as an abscissa to draw a curve, so that an ROC curve graph of a certain electroencephalogram index is obtained.

Therefore, 4 ROC graphs of the indexes to be referred to DTABR, DAR, RDP and RAP are obtained, wherein one ROC graph corresponds to one index. And then calculating the Area under the curve (AUC) of the respective corresponding ROC curve of each index to determine the electroencephalogram index for more accurately predicting the functional prognosis. The result shows that the area under the curve of DAR and RDP is larger, so that the classification accuracy of DAR and RDP is higher, so that the prediction electroencephalogram indexes are determined to be DAR and RDP, and the prediction models of the functional prognosis corresponding to DAR and RDP are stored, and the functional prognosis model of DAR indexes is P =1/[1+ e- ^{(-25.466 Bic 0.227X age + 0.827X NHISS score + 1.936X DAR)} ]The functional prognosis model of RDP index is P =1/[1+ e- ^{(-23.226+0.174X age + 1.005X NHISS score + 14.003X LnRDP)} ]。

And step 250, according to the brain function recovery degree corresponding to the reference range in the value matching standard database of the index to be referred, obtaining the predicted brain function recovery degree of the ischemic stroke patient within the preset interval.

Further, the reference range in the standard database is obtained as follows:

determining a prognosis score of the historical ischemic stroke patient in a preset interval according to a scoring rule, wherein the prognosis score predicts the brain function recovery degree of the historical ischemic stroke patient in the preset interval;

and generating a reference range corresponding to each brain function recovery degree according to the brain function recovery degree of the historical ischemic stroke patient and the pre-measured electroencephalogram indexes of the historical ischemic stroke patient in a preset interval, and storing the reference range in the standard database.

Illustratively, the sensitivity and specificity of all cut-off values are derived from the reference index prediction function prognostic analysis ROC curve that analyzes the electroencephalogram. And subtracting 1 from the sum of the sensitivity and the specificity to obtain a calculation formula of the Jordan index, and obtaining the Jordan indexes of all the critical values. The Johnson index represents the total ability of a screening method for finding real patients and non-patients, the larger the index is, the better the screening experiment effect is, and the larger the authenticity is, so the critical value of the maximum value of the Johnson index is selected as a diagnosis point, and the sensitivity and the specificity of the model are calculated to be the sensitivity and the specificity of the model.

In specific implementation, the reference indexes of the electroencephalogram are DAR and RDP, ROC curves and prognosis models of DAR and RDP are analyzed, the Jordan indexes of all critical values are obtained according to the sensitivity and the specificity of all critical values, the critical value with the maximum value of the Jordan indexes is selected as a diagnosis point, and the sensitivity and the specificity are calculated to be the sensitivity and the specificity of the model. Therefore, the optimal diagnosis point of each of the DAR index and the RDP index can be calculated.

And calculating the probability predicted value P of the DAR index. The AUC of the probability predicted value P obtained according to the ROC curve is 0.921, the sensitivity is 75.0%, the specificity is 97.5%, and the optimal diagnosis boundary point of the DAR index obtained through calculation is 0.540.

And calculating the probability predicted value P of the RDP index. The AUC of the probability predicted value P obtained according to the ROC curve is 0.910, the sensitivity is 78.6%, the specificity is 87.5%, and the optimal diagnosis boundary point of the RDP index obtained by calculation is 0.483.

Therefore, when the P of the DAR index is more than or equal to 0.540, the possibility of predicting the poor prognosis of the patient is high; when the P of the RDP index is more than or equal to 0.483, the possibility of predicting poor prognosis of the patient is high. Matching each index of the patient to the prognosis model of the patient with the historical ischemic stroke, calculating the P values of the DAR and RDP indexes of the patient, and judging the prognosis condition of the patient according to the reference range of the respective P values of the DAR and RDP indexes. And storing the reference range corresponding to each brain function recovery degree generated according to the brain function recovery degree of the historical ischemic stroke patient in the preset interval and the predicted electroencephalogram index of the historical ischemic stroke patient in the standard database.

According to the embodiment of the invention, electroencephalograms of ischemic stroke patients in a preset time period and human body information corresponding to the electroencephalograms are collected; analyzing an electroencephalogram of a patient with ischemic stroke to obtain electroencephalogram indexes corresponding to the electroencephalogram, wherein the electroencephalogram indexes comprise RAP, RDP, DAR, DTABR and pdBSI; inputting human body information corresponding to the electroencephalogram indexes into the index model to be referred to obtain the indexes to be referred to in the electroencephalogram indexes; and matching a reference range in the standard database according to the value of the index to be referred, predicting the brain function recovery degree of the ischemic stroke patient within a preset interval according to the reference range, and displaying the predicted brain function recovery degree as reference data to a doctor. The embodiment of the invention solves the problem that a doctor cannot quickly diagnose the condition of a patient due to long time period for acquiring images by CT and MRI technologies, and provides other reference data except CT and MRI brain images for the evaluation and treatment of the brain function of the patient with acute ischemic stroke after healing, so that the doctor can more comprehensively and quickly know the condition and recovery condition of the patient.

Fig. 3 is a schematic structural diagram of an electroencephalogram-based prediction apparatus according to an embodiment of the present invention, and as shown in fig. 3, the apparatus may include:

the image acquisition module 310 is used for acquiring an electroencephalogram of the ischemic stroke patient within a preset time period;

the image analysis module 320 is configured to analyze the electroencephalogram to obtain electroencephalogram indexes corresponding to the electroencephalogram, where the electroencephalogram indexes include RAP, RDP, DAR, DTABR, and pdBSI;

the range determining module 330 is configured to perform statistical processing on an index by using the SPSS to obtain a to-be-referenced index corresponding to the electroencephalogram index, and perform a two-class logistic regression analysis on the reference index to obtain a reference range of each reference index corresponding to the electroencephalogram index;

the degree determining module 340 is configured to predict a brain function recovery degree of the ischemic stroke patient within a preset interval according to the electroencephalogram index and the reference range of each reference index corresponding to the electroencephalogram index.

In one embodiment, before the image analysis module 320 analyzes the electroencephalogram, the method further includes:

determining an electroencephalogram within a second preset time period;

and carrying out independent component analysis on the electroencephalogram in the second preset time period, and removing eye movement components of the electroencephalogram in the second preset time period.

In one embodiment, the image analysis module 320 analyzes the electroencephalogram, including:

determining the power of each lead of each frequency band according to the signal of each electrode corresponding to the electroencephalogram;

determining an average power value after all leads of each frequency band are averaged according to the power of each lead of each frequency band; and obtaining the electroencephalogram index based on the lead power of each frequency band and the average power value after all leads of each frequency band are averaged.

In one embodiment, the image capturing module 310 captures an electroencephalogram of an ischemic stroke patient within a preset time period, and includes:

collecting electroencephalograms of a patient with ischemic stroke in a waking state in a first preset time period within a preset time period after the onset of a disease;

when the electroencephalogram acquisition equipment acquires electroencephalograms, the impedance between a human body and the electrodes is less than 10k omega, the number of the acquisition electrodes is more than or equal to 19, and the electroencephalograms corresponding to the electroencephalograms are within the range of 1-100 Hz.

In one embodiment, the range determination module 330 statistically processes the index using SPSS, including:

screening reference human body information of the same historical data item from a historical database according to the human body information corresponding to the electroencephalogram, wherein the electroencephalograms of the electroencephalograms with the same processing method are stored in the historical database;

and dividing the reference human body information into a health group and a patient group according to the health state, performing chi-square test according to the human body information and the electroencephalogram corresponding to the health group and the human body information and the electroencephalogram corresponding to the patient group, and determining the index to be referenced with the contrasting significance.

In an embodiment, the range determining module 330 performs a two-class logistic regression analysis on the reference index to obtain a reference range corresponding to the reference value index, including:

performing two-classification logistic regression analysis on the reference index to obtain the reference index;

and analyzing the reference index to obtain a reference range corresponding to the reference index.

In an embodiment, the predicting of the brain function recovery degree of the ischemic stroke patient in the preset interval by the degree determining module 340 according to the electroencephalogram index and the reference range of each reference index corresponding to the electroencephalogram index includes:

mapping the poor prognosis degree of the electroencephalogram index corresponding to the ischemic stroke patient according to the position of the numerical value of the electroencephalogram index in the reference range of each reference index corresponding to the electroencephalogram index;

obtaining a probability predicted value P according to a reference index prediction function prognosis analysis ROC curve for analyzing the electroencephalogram;

according to the embodiment of the invention, electroencephalograms of patients with ischemic stroke in a preset time period can be collected; analyzing the electroencephalogram to obtain electroencephalogram indexes corresponding to the electroencephalogram, wherein the electroencephalogram indexes comprise RAP, RDP, DAR, DTABR and pdBSI; carrying out statistical processing on the indexes by using the SPSS to obtain indexes to be referenced corresponding to the electroencephalogram indexes, and carrying out two-classification logic regression analysis on the reference indexes to obtain reference ranges of all the reference indexes corresponding to the electroencephalogram indexes; and predicting the brain function recovery degree of the ischemic stroke patient within a preset interval according to the electroencephalogram index and the reference range of each reference index corresponding to the electroencephalogram index. The embodiment of the invention solves the problems that the assessment of the severity of ischemic stroke by adopting technologies such as CT, MRI and the like is long in reserved time and cannot be applied in time, can well realize the early diagnosis of the acute ischemic stroke patient, has high accuracy and realizes the prognosis of the functions of the acute ischemic stroke patient.

Fig. 4 is a schematic structural diagram of an electronic device according to an embodiment of the present invention. FIG. 4 illustrates a block diagram of an exemplary electronic device 12 suitable for use in implementing embodiments of the present invention. The electronic device 12 shown in fig. 4 is only an example and should not bring any limitation to the function and the scope of use of the embodiment of the present invention.

As shown in fig. 4, electronic device 12 is in the form of a general purpose computing device. The components of electronic device 12 may include, but are not limited to: one or more processors or processing units 16, a system memory 28, and a bus 18 that couples various system components including the system memory 28 and the processing unit 16.

Bus 18 represents one or more of any of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, and a processor or local bus using any of a variety of bus architectures. By way of example, such architectures include, but are not limited to, industry Standard Architecture (ISA) bus, micro-channel architecture (MAC) bus, enhanced ISA bus, video Electronics Standards Association (VESA) local bus, and Peripheral Component Interconnect (PCI) bus.

Electronic device 12 typically includes a variety of computer system readable media. Such media may be any available media that is accessible by electronic device 12 and includes both volatile and nonvolatile media, removable and non-removable media.

The system memory 28 may include computer system readable media in the form of volatile memory, such as Random Access Memory (RAM) 30 and/or cache memory 32. Electronic device 12 may further include other removable/non-removable, volatile/nonvolatile computer system storage media. By way of example only, storage system 34 may be used to read from and write to non-removable, nonvolatile magnetic media (not shown in FIG. 4, commonly referred to as a "hard drive"). Although not shown in FIG. 4, a magnetic disk drive for reading from and writing to a removable, nonvolatile magnetic disk (e.g., a "floppy disk") and an optical disk drive for reading from or writing to a removable, nonvolatile optical disk (e.g., a CD-ROM, DVD-ROM, or other optical media) may be provided. In these cases, each drive may be connected to bus 18 by one or more data media interfaces. Memory 28 may include at least one program product having a set (e.g., at least one) of program modules that are configured to carry out the functions of embodiments of the invention.

A program/utility 40 having a set (at least one) of program modules 42 may be stored, for example, in memory 28, such program modules 42 including, but not limited to, an operating system, one or more application programs, other program modules, and program data, each of which examples or some combination thereof may comprise an implementation of a network environment. Program modules 42 generally carry out the functions and/or methodologies of the described embodiments of the invention.

Electronic device 12 may also communicate with one or more external devices 14 (e.g., keyboard, pointing device, display 24, etc.), one or more devices that enable a user to interact with electronic device 12, and/or any device (e.g., network card, modem, etc.) that enables electronic device 12 to communicate with one or more other computing devices. Such communication may be through an input/output (I/O) interface 22. Also, the electronic device 12 may communicate with one or more networks (e.g., a Local Area Network (LAN), a Wide Area Network (WAN), and/or a public network, such as the Internet) via the network adapter 20. As shown, the network adapter 20 communicates with the other modules of the electronic device 12 over the bus 18. It should be understood that although not shown in the figures, other hardware and/or software modules may be used in conjunction with electronic device 12, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems, to name a few.

The processing unit 16 executes various functional applications and data processing by executing programs stored in the system memory 28, for example, implementing an electroencephalogram-based prediction method provided by an embodiment of the present invention, the method including:

acquiring an electroencephalogram of a patient with ischemic stroke within a preset time period;

analyzing the electroencephalogram to obtain electroencephalogram indexes corresponding to the electroencephalogram, wherein the electroencephalogram indexes comprise RAP, RDP, DAR, DTABR and pdBSI;

carrying out statistical processing on indexes by using SPSS to obtain indexes to be referenced corresponding to the electroencephalogram indexes, and carrying out two-classification logistic regression analysis on the reference indexes to obtain reference ranges of all the reference indexes corresponding to the electroencephalogram indexes;

and predicting the brain function recovery degree of the ischemic stroke patient within a preset interval according to the electroencephalogram index and the reference range of each reference index corresponding to the electroencephalogram index.

Embodiments of the present invention also provide a computer-readable storage medium having stored thereon a computer program, which when executed by a processor, implements an electroencephalogram-based prediction method according to embodiments of the present invention, the method including:

collecting electroencephalograms of ischemic stroke patients within a preset time period;

analyzing the electroencephalogram to obtain electroencephalogram indexes corresponding to the electroencephalogram, wherein the electroencephalogram indexes comprise RAP, RDP, DAR, DTABR and pdBSI;

carrying out statistical processing on indexes by using SPSS to obtain indexes to be referenced corresponding to the electroencephalogram indexes, and carrying out two-classification logistic regression analysis on the reference indexes to obtain reference ranges of all the reference indexes corresponding to the electroencephalogram indexes;

and predicting the brain function recovery degree of the ischemic stroke patient within a preset interval according to the electroencephalogram index and the reference range of each reference index corresponding to the electroencephalogram index.

Computer storage media for embodiments of the invention may employ any combination of one or more computer-readable media. The computer readable medium may be a computer readable signal medium or a computer readable storage medium. A computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples (a non-exhaustive list) of the computer readable storage medium would include the following: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the context of this document, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.

A computer readable signal medium may include a propagated data signal with computer readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A computer readable signal medium may be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device.

Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

Computer program code for carrying out operations for aspects of the present invention may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, smalltalk, C + + or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet service provider).

It is to be noted that the foregoing is only illustrative of the preferred embodiments of the present invention and the technical principles employed. It will be understood by those skilled in the art that the present invention is not limited to the particular embodiments described herein, but is capable of various obvious modifications, rearrangements and substitutions as will now become apparent to those skilled in the art without departing from the scope of the invention. Therefore, although the present invention has been described in some detail by the above embodiments, the invention is not limited to the above embodiments, but may include other equivalent embodiments without departing from the spirit of the invention, and the scope of the invention is determined by the scope of the appended claims.

Claims (10)

1. An electroencephalogram-based prediction method, comprising:

collecting electroencephalograms of ischemic stroke patients in a first preset time period and human body information corresponding to the electroencephalograms;

analyzing the electroencephalogram of the ischemic stroke patient to obtain electroencephalogram indexes corresponding to the electroencephalogram, wherein the electroencephalogram indexes comprise RAP, RDP, DAR, DTABR and pdBSI;

inputting the human body information corresponding to the electroencephalogram indexes into a model of indexes to be referred to obtain the indexes to be referred to in the electroencephalogram indexes;

and according to the brain function recovery degree corresponding to the reference range in the value matching standard database of the index to be referred, obtaining the predicted brain function recovery degree of the ischemic stroke patient within a preset interval, and showing the predicted brain function recovery degree as reference data to a doctor.

2. The method as claimed in claim 1, further comprising, after said acquiring the electroencephalogram of the ischemic stroke patient for the first preset period of time:

determining an electroencephalogram of an ischemic stroke patient in a second preset time period when a human body is in an awake state in a first preset time period after the occurrence of a disease;

and performing independent component analysis on the electroencephalogram of the ischemic stroke patient in the second preset time period, and removing an eye movement component of the electroencephalogram of the ischemic stroke patient in the second preset time period.

3. The method according to claim 1, wherein the analyzing the electroencephalogram of the ischemic stroke patient to obtain an electroencephalogram index corresponding to the electroencephalogram comprises:

determining the power of each frequency band in the electrical signal of each electrode according to the electrical signal of each electrode in the electroencephalogram;

and determining the average power value of each frequency band corresponding to the electroencephalogram according to the power of each frequency band in the electric signal of each electrode, and calculating the electroencephalogram index corresponding to the electroencephalogram.

4. The method as claimed in claim 1, wherein the collecting of electroencephalography of ischemic stroke patients for a preset period of time comprises:

and acquiring electroencephalograms of the human body in a second preset time period in a waking state within a preset time period after the occurrence of a disease.

5. The method according to claim 1, characterized in that the index model to be referenced is obtained as follows:

acquiring predicted electroencephalogram indexes of a historical ischemic stroke patient to be trained and human body information of the historical ischemic stroke patient;

and training a deep learning classification network for preset times according to the predicted electroencephalogram index of the historical ischemic stroke patient to be trained and the human body information of the historical ischemic stroke patient until the deep learning classification network is converged to obtain the index model to be referred.

6. The method of claim 5, wherein the predicted EEG index of the patient with the historical ischemic stroke to be trained is obtained as follows:

carrying out statistical analysis on different groups of electroencephalogram indexes corresponding to different brain function degrees of the prognosis evaluation result of the historical ischemic stroke patient by using an inter-group comparison mode matched with human body information to obtain electroencephalogram indexes with statistical significance as electroencephalogram indexes to be predicted;

grouping different brain function degrees corresponding to the prediction result of the electroencephalogram index to be predicted and the prognosis evaluation result of the historical ischemic stroke patient according to the different brain function degrees;

obtaining a mapping parameter corresponding to the index to be predicted according to the number of people with different groups of prediction results of the electroencephalogram index to be predicted and the prognosis evaluation results of the historical ischemic stroke patients;

and drawing an ROC (rock characteristic) graph according to the drawing parameters to obtain the area under the curve corresponding to the index to be predicted, and determining the predicted electroencephalogram index according to the area under the curve.

7. The method of claim 1, wherein the reference range in the standard database is obtained by the following method:

determining a prognosis score of the historical ischemic stroke patient in a preset interval according to a scoring rule, wherein the prognosis score predicts the brain function recovery degree of the historical ischemic stroke patient in the preset interval;

and generating a reference range corresponding to each brain function recovery degree according to the brain function recovery degree of the historical ischemic stroke patient and the predicted electroencephalogram index of the historical ischemic stroke patient within a preset interval, and storing the reference range in the standard database.

8. An electroencephalogram-based prediction apparatus, comprising:

the image acquisition module is used for acquiring electroencephalograms of ischemic stroke patients within a preset time period and human body information corresponding to the electroencephalograms;

the image analysis module is used for analyzing the electroencephalogram of the ischemic stroke patient to obtain electroencephalogram indexes corresponding to the electroencephalogram, and the electroencephalogram indexes comprise RAP, RDP, DAR, DTABR and pdBSI;

the index determining module is used for inputting the human body information corresponding to the electroencephalogram index into a model of an index to be referred to obtain the index to be referred in the electroencephalogram index;

and the degree determining module is used for matching a reference range in a standard database according to the value of the index to be referred, and predicting the brain function recovery degree of the ischemic stroke patient in a preset interval according to the reference range.

9. An electronic device, characterized in that the electronic device comprises:

one or more processors;

a storage device for storing one or more programs,

the one or more programs, when executed by the one or more processors, cause the one or more processors to implement the electroencephalogram-based prediction method of any of claims 1 to 7.

10. A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the electroencephalogram-based prediction method according to any one of claims 1 to 7.

Images