Method and device for assessing disorders of consciousness based on heart rate variability features under standardized musical stimulation

Abstract

This invention discloses a method and apparatus for assessing consciousness disorders based on heart rate variability characteristics under standardized music stimulation. The method includes: acquiring electrocardiogram (ECG) signals, audio signals, and a first synchronization signal and a second synchronization signal; performing time synchronization processing to determine ECG signal segments during music intervention; performing first and second preprocessing on the ECG signal segments to obtain a target heartbeat interval sequence; calculating heart rate variability characteristics on the target heartbeat interval sequence to obtain a heart rate variability feature set; performing principal component analysis on the feature set to obtain a dimensionality-reduced principal component score vector; then constructing a consciousness disorder assessment model using a supervised learning regression algorithm, and inputting the dimensionality-reduced principal component score vector into the pre-trained consciousness disorder assessment model for evaluation. This invention achieves ECG data-driven consciousness disorder assessment, reducing reliance on clinical physician assessment and enhancing the objectivity of the assessment. This invention can be applied in the field of ECG data application technology.

Description

Technical Field

[0001] This invention relates to the field of electrocardiogram data application technology, and in particular to a method and apparatus for assessing consciousness disorders based on heart rate variability characteristics under standardized music stimulation. Background Technology

[0002] Disorders of Consciousness (DOC) are diseases caused by severe damage to the central nervous system, such as the cerebral cortex, leading to weakened or even lost cognitive abilities. Based on the severity of the disturbance, patients with DOC can be further classified into Vegetative State (VS / Unresponsive Wakefulness Syndrome, UWS) and Minimally Consciousness State (MCS). Traditional assessment methods use the Behavioral Assessment Scale-Responsive (CRS-R), which assesses patients based on their perceptual and behavioral response characteristics (such as auditory, visual, motor, oral motor / speech function, communication, and alertness). This scale relies on the integrity of the peripheral and central nervous systems related to various sensory functions and motor control, making it difficult to accurately and reliably assess the patient's level of consciousness. Furthermore, this assessment method depends on the assessor's experience; for subtle patient manifestations, different assessors may yield different results, resulting in low accuracy.

[0003] In summary, the assessment of consciousness disorders based on objective physiological indicators and the technical problems existing in this assessment method still need to be solved. Summary of the Invention

[0004] The present invention provides a method and apparatus for assessing consciousness disorders based on heart rate variability characteristics under standardized music stimulation, which effectively improves accuracy.

[0005] On one hand, embodiments of the present invention provide a method for assessing consciousness disorders based on heart rate variability characteristics under standardized musical stimulation, including the following steps:

[0006] The first electrocardiogram signal and the first audio signal during standardized music stimulation of the patient to be evaluated were acquired, and the first synchronization signal and the second synchronization signal were recorded simultaneously with the two signals, respectively.

[0007] Time synchronization processing is performed based on the first synchronization signal and the second synchronization signal to determine the first timestamp of the first ECG signal segment corresponding to the first ECG signal during the standardized music stimulation, and the first ECG signal segment is extracted from the first ECG signal based on the first timestamp.

[0008] The first ECG signal segment is preprocessed to obtain a first target ECG signal with a high signal-to-noise ratio.

[0009] The first target electrocardiogram signal is subjected to a second preprocessing to obtain the first target heartbeat interval sequence;

[0010] Calculate the first heart rate variability feature and perform feature filtering processing on the first target heartbeat interval sequence;

[0011] The first heart rate variability feature is input into the pre-trained principal component analysis model, and the first principal component component after dimensionality reduction is output.

[0012] The first principal component is input into the pre-trained consciousness impairment assessment model, and the quantitative assessment value of the consciousness impairment level is output.

[0013] The pre-trained principal component analysis model and the pre-trained consciousness impairment assessment model are obtained through the following steps:

[0014] The experimental data and clinical behavior scores of multiple patients with impaired consciousness under the same standardized music stimulation were obtained. The experimental data included a second electrocardiogram signal, a second audio signal, and a third synchronization signal and a fourth synchronization signal recorded simultaneously with the two signals, respectively.

[0015] Time synchronization processing is performed based on the third synchronization signal and the fourth synchronization signal to determine the second timestamp of the second ECG signal segment corresponding to the second ECG signal during the standardized music stimulation, and the second ECG signal segment is extracted from the second ECG signal based on the second timestamp.

[0016] The second ECG signal segment is subjected to a first preprocessing to obtain a second target ECG signal with a high signal-to-noise ratio;

[0017] The second target electrocardiogram signal is subjected to a second preprocessing to obtain the second target heartbeat interval sequence;

[0018] Calculate the second heart rate variability feature and perform feature filtering on the second target heartbeat interval sequence;

[0019] The principal component analysis model is pre-trained using the second heart rate variability feature, and the second principal component component after dimensionality reduction is output through the pre-trained principal component analysis model.

[0020] Based on the second principal component, a predictive model for the patient's clinical behavior score is constructed using a supervised learning regression algorithm, and the pre-trained consciousness impairment assessment model is generated by combining the clinical behavior score.

[0021] In some embodiments, the step of performing time synchronization processing based on a first synchronization signal and a second synchronization signal to determine a first timestamp of a first ECG signal segment corresponding to the first ECG signal during the standardized music stimulation, and extracting the first ECG signal segment from the first ECG signal based on the first timestamp, includes:

[0022] Perform cross-correlation analysis on the first synchronization signal and the second synchronization signal, and calculate the first time shift value corresponding to the maximum cross-correlation coefficient between the two synchronization signals. The first time shift value is used as the first relative time difference between the recording clocks of the two devices.

[0023] Based on the first audio signal, determine the duration between the start point of the first audio signal and the start point of the music stimulus recording;

[0024] The duration from the start of the first electrocardiogram signal to the start of the music stimulation recording is calculated based on the first relative time difference between the recording clocks of the two devices and the duration from the start of the first audio signal to the start of the music stimulation recording.

[0025] The starting point of music stimulation recording in the first target electrocardiogram signal is determined based on the duration between the starting point of the first electrocardiogram signal and the starting point of music stimulation recording.

[0026] The first ECG signal segment is determined based on the preset duration of music stimulation and the starting point of music stimulation recording in the first target ECG signal.

[0027] In some embodiments, the first preprocessing of the first ECG signal segment to obtain a first target ECG signal with a high signal-to-noise ratio includes:

[0028] The first ECG signal segment is subjected to bandpass filtering to obtain a first target ECG signal with a high signal-to-noise ratio.

[0029] In some embodiments, the second preprocessing of the first target electrocardiogram signal to obtain the first target heartbeat interval sequence includes:

[0030] The QRS complex wave detection algorithm was used to locate the peak of the filtered first ECG signal segment and determine the R wave peak corresponding to each cardiac cycle.

[0031] Based on the multiple R-wave peaks, a peak timestamp sequence is generated;

[0032] Based on the peak timestamp sequence, an initial heartbeat interval sequence defined by the R-wave peak is generated;

[0033] The initial heartbeat interval sequence is corrected using a preset peak correction algorithm to obtain the first target heartbeat interval sequence.

[0034] In some embodiments, the calculation of the second heart rate variability feature and the feature filtering process for the second target heart rate interval sequence include:

[0035] The second target heart rate interval sequence is subjected to time-domain features, frequency-domain features, and nonlinear features to obtain the second heart rate variability feature set;

[0036] The significance test of inter-group differences was performed on the second heart rate variability feature set to obtain the second feature subset whose inter-group differences meet the preset requirements.

[0037] In some embodiments, the step of pre-training the principal component analysis model using the second heart rate variability feature and outputting the dimensionality-reduced second principal component components through the pre-trained principal component analysis model includes:

[0038] Principal component analysis is performed on the second feature subset to extract principal components whose cumulative variance is greater than or equal to a preset threshold, the principal component analysis model is constructed and the principal component feature vectors are saved.

[0039] The second principal component components are obtained by reducing the dimensionality of the second feature subset using the principal component feature vector.

[0040] In some embodiments, the supervised learning regression algorithm includes a linear regression equation and a power-law-based nonlinear regression equation. The step of constructing a predictive model for patient clinical behavior scores using the supervised learning regression algorithm based on the second principal component components, and generating the pre-trained consciousness impairment assessment model by combining the clinical behavior scores, includes:

[0041] Based on the linear regression equation, construct the first regression model;

[0042] Based on the power-based nonlinear regression equation, a second regression model is constructed;

[0043] The second principal component is input into the first regression model to train the first regression model.

[0044] The second principal component is input into the second regression model to train the second regression model;

[0045] If the prediction error of the first regression model after training is less than the prediction error of the second regression model after training, then the first regression model after training is used as the consciousness disorder assessment model; otherwise, the second regression model after training is used as the consciousness disorder assessment model.

[0046] In some embodiments, the expression for the linear regression equation is:

[0047] ;

[0048] In the formula, The score for the first coma recovery scale, M represents the number of principal components. and These are all regression parameters that need to be fitted in the linear regression equation. For the m-th principal component feature, This refers to the residual part of the model;

[0049] The expression for the power-law-based nonlinear regression equation is as follows:

[0050] ;

[0051] In the formula, For the second coma recovery scale score, M represents the number of principal components. , and These are all regression parameters that need to be fitted in the nonlinear regression equation. For the m-th principal component feature, This represents the residual part of the model.

[0052] On the other hand, embodiments of the present invention provide a device for assessing disorders of consciousness based on heart rate variability characteristics under standardized music stimulation, comprising:

[0053] The signal acquisition module is used to acquire the first electrocardiogram signal and the first audio signal during standardized music stimulation of the patient to be evaluated, and to obtain the first synchronization signal and the second synchronization signal recorded simultaneously with the two signals, respectively.

[0054] The data processing module is used to perform time synchronization processing based on the first synchronization signal and the second synchronization signal, determine the first timestamp of the first ECG signal segment corresponding to the first ECG signal during the standardized music stimulation, and extract the first ECG signal segment from the first ECG signal based on the first timestamp.

[0055] The preprocessing module is used to perform a first preprocessing on the first ECG signal segment to obtain a first target ECG signal with a high signal-to-noise ratio; and to perform a second preprocessing on the first target ECG signal to obtain a first target heartbeat interval sequence.

[0056] The calculation module is used to calculate the first heart rate variability feature and perform feature filtering processing on the first target heartbeat interval sequence;

[0057] The principal component analysis module is used to input the first heart rate variability feature into the pre-trained principal component analysis model and output the first principal component component after dimensionality reduction.

[0058] The model recognition module is used to input the first principal component into the pre-trained consciousness disorder assessment model and output a quantitative assessment value of the level of consciousness disorder.

[0059] The pre-trained principal component analysis model and the pre-trained consciousness impairment assessment model are obtained through the following steps:

[0060] The experimental data and clinical behavior scores of multiple patients with impaired consciousness under the same standardized music stimulation were obtained. The experimental data included a second electrocardiogram signal, a second audio signal, and a third synchronization signal and a fourth synchronization signal recorded simultaneously with the two signals, respectively.

[0061] Time synchronization processing is performed based on the third synchronization signal and the fourth synchronization signal to determine the second timestamp of the second ECG signal segment corresponding to the second ECG signal during the standardized music stimulation, and the second ECG signal segment is extracted from the second ECG signal based on the second timestamp.

[0062] The second ECG signal segment is subjected to a first preprocessing to obtain a second target ECG signal with a high signal-to-noise ratio;

[0063] The second target electrocardiogram signal is subjected to a second preprocessing to obtain the second target heartbeat interval sequence;

[0064] Calculate the second heart rate variability feature and perform feature filtering on the second target heartbeat interval sequence;

[0065] The principal component analysis model is pre-trained using the second heart rate variability feature, and the second principal component component after dimensionality reduction is output through the pre-trained principal component analysis model.

[0066] Based on the second principal component, a predictive model for the patient's clinical behavior score is constructed using a supervised learning regression algorithm, and the pre-trained consciousness impairment assessment model is generated by combining the clinical behavior score.

[0067] On the other hand, embodiments of the present invention provide a computer device, including:

[0068] At least one processor;

[0069] At least one memory for storing at least one program;

[0070] When the at least one program is executed by the at least one processor, the at least one processor performs the method described above.

[0071] On the other hand, embodiments of the present invention provide a data acquisition device, including:

[0072] ECG signal amplifier, used for amplifying and converting ECG signals from analog to digital;

[0073] ECG acquisition processor and memory are used to process the recording (digital-to-analog conversion and data storage) of ECG signals and synchronization signals.

[0074] A recording microphone, used for capturing audio signals;

[0075] Audio acquisition processor and memory, used to process the recording of audio signals and synchronization signals;

[0076] A music player used to play standardized music.

[0077] The embodiments of this application include at least the following beneficial effects: This application provides a method and apparatus for assessing consciousness disorders based on heart rate variability characteristics under standardized music stimulation. The embodiments of this application first acquire the continuous first electrocardiogram signal of the patient to be assessed during standardized music stimulation, then perform data preprocessing and feature extraction to obtain the first heart rate variability characteristics, input the first heart rate variability characteristics into a pre-trained principal component analysis model, output the first principal component score vector after dimensionality reduction, and finally input the first principal component score vector into a pre-trained consciousness disorder assessment model to obtain the consciousness disorder assessment result, thereby realizing the assessment of consciousness disorders and improving accuracy. The pre-trained principal component analysis (PCA) model and the pre-trained consciousness impairment assessment model record a second electrocardiogram (ECG) signal and a second audio signal using ECG and audio acquisition devices, respectively, to obtain a third synchronization signal and a fourth synchronization signal recorded simultaneously with the two signals. Time synchronization processing is performed based on the third and fourth synchronization signals to determine the second timestamp of the second ECG signal segment during music intervention, and the second ECG signal segment is extracted based on the second timestamp. The second ECG signal segment undergoes a first preprocessing step to obtain a high signal-to-noise ratio second target ECG signal. The second target ECG signal undergoes a second preprocessing step to obtain a second target heartbeat interval sequence. The second heart rate variability feature is calculated and feature selection is performed on the second target heartbeat interval sequence. The PCA model is pre-trained using the second heart rate variability feature, and the dimensionality-reduced second principal component components are output from the pre-trained PCA model. Based on the second principal component components, a supervised learning regression algorithm is used to construct a predictive model for patient clinical behavior scoring, thereby improving model accuracy.

[0078] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the description and the drawings. Attached Figure Description

[0079] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0080] Figure 1 This is a flowchart of a method for assessing consciousness disorders based on heart rate variability characteristics under standardized music stimulation, according to an embodiment of the present invention.

[0081] Figure 2 This is a flowchart illustrating an assessment model for obtaining a consciousness disorder according to an embodiment of the present invention;

[0082] Figure 3 This is a schematic diagram of an electrocardiogram (ECG) signal acquisition process according to an embodiment of the present invention;

[0083] Figure 4 This is a signal acquisition structure diagram according to an embodiment of the present invention;

[0084] Figure 5 This is a schematic diagram of a time synchronization process according to an embodiment of the present invention;

[0085] Figure 6 This is a schematic diagram of the structure of a consciousness disorder assessment device based on the characteristics of heart rate variability under standardized music stimulation, according to an embodiment of the present invention.

[0086] Figure 7 This is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. Detailed Implementation

[0087] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of this application and are not intended to limit it. In the following description, when referring to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with those of this application; they are merely examples of apparatuses and methods consistent with some aspects of the embodiments of this application as detailed in the appended claims.

[0088] It is understood that the terms “first,” “second,” etc., used in this application may be used herein to describe various concepts, but unless otherwise stated, these concepts are not limited by these terms. These terms are only used to distinguish one concept from another. For example, without departing from the scope of the embodiments of this application, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the words “if,” “when,” or “in response to a determination” as used herein may be interpreted as “when…” or “when…” or “in response to a determination.”

[0089] As used in this application, the terms "at least one", "multiple", "each", "any", etc., "at least one" includes one, two or more, "multiple" includes two or more, "each" refers to each of the corresponding multiples, and "any" refers to any one of the multiples.

[0090] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0091] Before providing a detailed description of the embodiments of this application, some of the nouns and terms involved in the embodiments of this application will be explained first. The nouns and terms involved in the embodiments of this application are subject to the following interpretations.

[0092] Disorders of Consciousness (DOC): refers to a state of loss of consciousness caused by brain injury from various causes, including coma, vegetative state (VS), and minimally conscious state (MCS).

[0093] Electrocardiography (ECG) is a technique that uses an electrocardiograph to record the changes in electrical activity of the heart during each cardiac cycle from the body surface.

[0094] R-wave peak: refers to the highest amplitude point of the QRS complex in a surface electrocardiogram (ECG), corresponding to the peak moment of cardiac electrical activity conduction from the atrioventricular node to the ventricles. Its peak position marks the starting point of ventricular contraction and is the most prominent and stable characteristic point in the electrocardiogram signal.

[0095] Heart rate variability (HRV) refers to the variation in the difference between successive heartbeat cycles. It is calculated by the timestamp sequence of R wave peaks in continuous heartbeats. It contains information on the regulation of the cardiovascular system by neurohumoral factors and is the core indicator for evaluating consciousness disorders in this invention.

[0096] TSM (therapist-selected music): refers to music selected by a music therapist.

[0097] Coma Recovery Scale-Revised (CRS-R): A behavioral assessment scale used clinically to diagnose the degree of altered consciousness.

[0098] In related technologies, impaired consciousness is a disease caused by severe damage to the central nervous system, such as the cerebral cortex, leading to weakened or even lost cognitive abilities. Based on the severity of the impaired consciousness, patients can be further classified into vegetative state / unresponsive wakefulness syndrome (UWS) and minimally conscious state (MCS). Currently, the CRS-R is the gold standard for assessing the severity of impaired consciousness. This scale assesses patients based on their perceptual and behavioral response characteristics (auditory, visual, motor, oral motor / speech function, communication, and alertness). However, because this method relies on the integrity of the peripheral and central nervous systems related to various perceptual functions and motor control, it cannot accurately and reliably assess the patient's level of consciousness. Furthermore, this assessment method depends on the assessor's experience; for subtle patient manifestations, different assessors may yield different results. Therefore, assessing impaired consciousness based on objective physiological indicators is a crucial problem that urgently needs to be addressed in clinical practice.

[0099] Currently, researchers have developed electrophysiological biomarkers based on electroencephalography (EEG) and neuroimaging methods for assessing disorders of consciousness based on fMRI and PET technologies. However, EEG signal measurement suffers from low signal-to-noise ratio, inconsistent electrode placement among subjects, and discomfort due to the need for conductive gel; while imaging techniques are limited by high cost and low temporal resolution. Therefore, with its advantages of high signal-to-noise ratio, low measurement cost, and real-time measurement, electrocardiography (ECG) signals are more feasible and promising than other objective physiological indicators. Heart rate variability (HRV) extracted from ECG signals reflects the changes in the interbeat interval regulated by the dynamic activation and inhibition of the sympathetic and parasympathetic nervous systems in the autonomic nervous system. Cardiac activity and cerebral cortex activity influence each other through the heart-brain connection; therefore, HRV has been widely used in many studies to monitor acquired brain injury, emotions, and behaviors. Structurally rich music has the potential to stimulate both internal mental activity and external physical movement, and past studies have shown that music selected by music therapists can better activate the autonomic nervous system function of patients with disorientation of consciousness (DOC), resulting in a stronger HRV response. On the other hand, using the HRV index to assess the level of consciousness in DOC patients requires ensuring that all DOC patients receive the same musical stimulation.

[0100] In view of this, this invention uses heart rate variability (HRV) characteristics under standardized music conditions as a basis, relies on machine learning methods, and combines the patient's Coma Recovery Scale-Revised (CRS-R) score to build a regression model, thereby achieving an assessment of the level of consciousness impairment based on objective physiological indicators. This invention measures the HRV response of DOC patients under standardized music stimulation conditions. The standardized music stimulation is selected by five music therapists through multiple rounds of voting from a library of 50 relaxing music pieces. This invention verifies that, compared to quiet conditions, standardized music stimulation conditions are more conducive to evoking differences in heart rate variability indicators among patients with different levels of DOC, and these differences are more conducive to improving the accuracy of predicting the patient's total CRS-R score in the subsequent regression model.

[0101] The method for assessing consciousness disorders based on heart rate variability characteristics under standardized music stimulation provided in this application relates to the field of electrocardiogram (ECG) data application technology. The data acquisition module of this method can be applied to a terminal and the software running on that terminal, while the data processing and model recognition modules can be applied to a server and the software running on that server. In some embodiments, the terminal can be a smartphone, tablet, laptop, desktop computer, smart speaker, smartwatch, or in-vehicle terminal, but is not limited to these. The server can be configured as an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The server can also be a node server in a blockchain network. The software can be an application capable of acquiring, processing, and analyzing ECG signals, but is not limited to the above forms.

[0102] This application can be used in a wide variety of general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0103] The embodiments of this application will be explained in detail below with reference to the accompanying drawings:

[0104] Figure 1 This is an optional flowchart of a method for assessing consciousness disorders based on heart rate variability characteristics under standardized music stimulation, provided in an embodiment of this application. Figure 1 The method may include, but is not limited to, steps S101 to S107.

[0105] Step S101: Acquire the first electrocardiogram signal and the first audio signal during standardized music stimulation of the patient to be evaluated, and obtain the first synchronization signal and the second synchronization signal recorded simultaneously with the two signals respectively;

[0106] Step S102: Perform time synchronization processing based on the first synchronization signal and the second synchronization signal to determine the first timestamp of the first ECG signal segment corresponding to the first ECG signal during the standardized music stimulation period, and extract the first ECG signal segment from the first ECG signal based on the first timestamp.

[0107] Step S103: Perform a first preprocessing on the first ECG signal segment to obtain a first target ECG signal with a high signal-to-noise ratio;

[0108] Step S104: Perform a second preprocessing on the first target ECG signal to obtain the first target heartbeat interval sequence;

[0109] Step S105: Calculate the first heart rate variability features and perform feature screening for the first target heartbeat interval sequence;

[0110] Step S106: Input the first heart rate variability feature into the pre-trained principal component analysis model and output the first principal component component after dimensionality reduction.

[0111] Step S107: Input the first principal component into the pre-trained consciousness impairment assessment model and output a quantitative assessment value of the consciousness impairment level.

[0112] In some embodiments, in steps S101 to S104, standardized music stimulation can be applied to the new patient to be evaluated to obtain the ECG signal to be evaluated as the first ECG signal. Then, the first ECG signal is preprocessed to extract the heart rate variability (HRV) features during the music stimulation period as the first HRV features. Principal component analysis is then performed on the selected first HRV features to calculate the k-dimensional principal component score components as the first principal component components. Finally, the first principal component feature vector is input into the pre-trained consciousness disorder assessment model to obtain the consciousness disorder assessment results including the new sample, which is the CRS-R score prediction value of the consciousness disorder level of DOC patients. This enables objective and quantitative analysis to achieve consciousness disorder assessment, thereby improving accuracy.

[0113] In some embodiments, such as Figure 2 As shown, the specific implementation process of obtaining the pre-trained principal component analysis model and the pre-trained consciousness disorder assessment model may include, but is not limited to, steps S201 to S207.

[0114] Step S201: Obtain experimental data and clinical behavior scores of multiple patients with impaired consciousness under the same standardized music stimulation. The experimental data includes the second electrocardiogram signal, the second audio signal, and the third and fourth synchronization signals recorded simultaneously with the two signals, respectively.

[0115] Step S202: Perform time synchronization processing based on the third synchronization signal and the fourth synchronization signal to determine the second timestamp of the second ECG signal segment corresponding to the second ECG signal during the standardized music stimulation period, and extract the second ECG signal segment from the second ECG signal based on the second timestamp.

[0116] Step S203: Perform a first preprocessing on the second ECG signal segment to obtain a second target ECG signal with a high signal-to-noise ratio;

[0117] Step S204: Perform a second preprocessing on the second target ECG signal to obtain the second target heartbeat interval sequence;

[0118] Step S205: Calculate the second heart rate variability features and perform feature screening for the second target heartbeat interval sequence;

[0119] Step S206: Use the second heart rate variability feature to pre-train the principal component analysis model, and output the dimensionality-reduced second principal component components through the pre-trained principal component analysis model;

[0120] Step S207: Based on the second principal component, construct a predictive model for the patient's clinical behavior score using a supervised learning regression algorithm, and generate the pre-trained consciousness impairment assessment model by combining the clinical behavior score.

[0121] In some embodiments, in step S201, the original electrocardiogram (ECG) signals (second ECG signals) and second audio signals of multiple patients with impaired consciousness under the same standardized music stimulation can be acquired, and a third synchronization signal and a fourth synchronization signal recorded simultaneously with the two signals can be obtained. Exemplarily, the acquisition process of the second ECG signal is as follows: Figure 3 As shown, the patient can be placed in a quiet environment for 3 minutes, followed by 4 minutes of musical stimulation, with the entire acquisition process lasting no more than 15 minutes. Specifically, the acquisition process for the first electrocardiogram signal of the patient to be evaluated can also be carried out using... Figure 3 The process is shown below.

[0122] During subject preparation, non-sleep DOC patients (including MCS and VS patients classified by CRS-R) should be placed in a quiet and comfortable environment. Patients should assume a standard semi-recumbent position (e.g., back support angle adjusted to 30-45 degrees), ensuring stable and relaxed posture. Ambient noise should be kept low (recommended <40dB) to avoid external interference. Baseline resting-state recording should then be performed. After correctly connecting and calibrating the ECG recording equipment, a pre-set duration of resting-state ECG signal acquisition should be conducted. A duration of 3 minutes is recommended. During this stage, patients should remain quiet, avoiding active communication or external stimuli, and a stable baseline physiological state should be recorded.

[0123] Standardized music stimulation is then administered. The stimulation source can be the song with the highest number of votes selected by five music therapists from a library of 50 relaxation music tracks through multiple rounds of voting; this is the TSM (Traditional Music Therapy). All patients use this standardized music stimulation as their stimulation source. The music duration is approximately 4 minutes. The music is played through standard audio playback equipment (frequency response range 50-20kHz, signal-to-noise ratio ≥80dB, rated power ≥40W). The speaker should be positioned 1-1.5 meters directly in front of the patient, at ear level. The playback volume must be strictly calibrated and fixed, using a sound level meter to measure the sound pressure level at the patient's ear position as 65±5 dBSPL (A-weighted) to ensure consistent stimulation intensity. After baseline recording, TSM playback begins uninterrupted. Throughout the music stimulation, the patient's ECG signals are continuously recorded. ECG recording stops after TSM playback ends. ECG signal quality must be monitored in real time during acquisition. If severe motion artifacts or lead slippage occur, the entire procedure should be repeated (or the recording should be restarted from baseline rest) once the patient's condition has stabilized to ensure that valid data are obtained.

[0124] Meanwhile, the intervals between the start of ECG acquisition and the start of baseline resting recording, the end of baseline resting recording and the start of music stimulation, and the end of music stimulation and the end of ECG acquisition should theoretically not affect data analysis, and there are no fixed requirements for their duration. However, because DOC patients (especially severely ill patients) have limited tolerance, excessively long acquisition times can easily lead to patients falling asleep due to fatigue or affecting signal quality due to agitation. Therefore, the music stimulation phase should begin as soon as possible after baseline resting recording, and the total ECG measurement time should be controlled within 15 minutes.

[0125] In signal acquisition, the signal acquisition structure diagram is as follows: Figure 4 As shown, a single-lead (I-lead) method can be used to acquire electrocardiogram (ECG) signals. The positive (+) electrode is placed on the patient's left arm, the negative (-) electrode on the patient's right arm, and the reference / ground electrode on the right lower limb. An ECG signal amplifier 401 and an ECG acquisition host 402 are used to amplify, filter (bandpass filter 0.5-100Hz), perform analog-to-digital conversion (ADC), and store the raw ECG signal. A recommended ECG sampling rate is 1000Hz. An audio playback device 403 (such as a Bluetooth speaker) can be used to play standardized musical stimuli (TSM). The audio acquisition host 404, in conjunction with a recording microphone 405, records the actual played audio signal (including TSM content) at an audio sampling rate of 44.1kHz. Simultaneously, a synchronization signal from the ECG acquisition host is input to the audio acquisition host via a line, recording synchronously with the audio signal to record the actual output audio waveform for accurate automatic detection of the start and end times of the musical stimulus.

[0126] In the synchronization signal generation process, at the start of signal acquisition, a high-precision, fixed-frequency periodic electrical signal (synchronization signal source) is generated by the ECG acquisition host (or an independent master clock / signal generator). This synchronization signal source is simultaneously output to the following two channels: Channel A (recorded by the ECG acquisition host): As an additional analog input channel, it is sampled and stored by the ECG acquisition host in sync with the ECG signal (recorded as...). Channel B (recorded on the audio acquisition host): Sampled and stored by the audio acquisition host synchronously with the recorded audio signal via the line-in port (recorded as...). ).

[0127] In some embodiments, in step S102, time synchronization processing is performed based on the first synchronization signal and the second synchronization signal to determine the first timestamp of the first ECG signal segment corresponding to the first ECG signal during the standardized music stimulation period, and the first ECG signal segment is extracted from the first ECG signal based on the first timestamp, which may include, but is not limited to, the following steps:

[0128] Cross-correlation analysis is performed on the first synchronization signal and the second synchronization signal to calculate the first time shift value corresponding to the maximum cross-correlation coefficient between the two synchronization signals. The first time shift value is used as the first relative time difference between the recording clocks of the two devices.

[0129] Based on the first audio signal, determine the duration between the start point of the first audio signal and the start point of the music stimulus recording;

[0130] The duration from the start of the first electrocardiogram signal to the start of the music stimulation recording is calculated based on the first relative time difference between the clocks recorded by the two devices and the duration from the start of the first audio signal to the start of the music stimulation recording.

[0131] The starting point for music stimulation recording in the first target ECG signal is determined based on the duration between the starting point of the first ECG signal and the starting point of the music stimulation recording.

[0132] The first ECG signal segment is determined based on the preset duration of music stimulation and the starting point of music stimulation recording in the first target ECG signal.

[0133] In some embodiments, in step S202, time synchronization processing is performed based on the third synchronization signal and the fourth synchronization signal to determine the second timestamp of the second ECG signal segment corresponding to the second ECG signal during the standardized music stimulation, and the second ECG signal segment is extracted from the second ECG signal based on the second timestamp, which may include, but is not limited to, the following steps:

[0134] Cross-correlation analysis is performed on the third and fourth synchronization signals to calculate the second time shift value corresponding to the maximum cross-correlation coefficient between the two synchronization signals. The second time shift value is used as the second relative time difference between the recording clocks of the two devices.

[0135] Based on the second audio signal, determine the duration between the start point of the second audio signal and the start point of the music stimulus recording;

[0136] The duration from the start of the second electrocardiogram signal to the start of the music stimulation recording is calculated based on the first relative time difference between the clocks recorded by the two devices and the duration from the start of the first audio signal to the start of the music stimulation recording.

[0137] The starting point for music stimulation recording in the second target ECG signal is determined based on the duration between the starting point of the second ECG signal and the starting point of the music stimulation recording.

[0138] The second ECG signal segment is determined based on the preset duration of music stimulation and the starting point of music stimulation recording in the second target ECG signal.

[0139] In some embodiments, during the specific processing of the first and second ECG signals, as well as the first and second audio signals, the sampling rates of the ECG and audio stimulation signals are different, and the clocks of the sampling devices are also different. This is to ensure that the ECG signal is synchronized with the audio stimulation event (especially the start time of the TSM, i.e., ...). Precise time alignment can be achieved using hardware-triggered synchronization signals. A schematic diagram of time synchronization processing is shown below. Figure 5 As shown, cross-correlation analysis can be performed on the two synchronization signals first to calculate the time shift value (Lag) corresponding to the maximum cross-correlation coefficient between the two synchronization signals. This time shift value is used as the relative time difference between the recording clocks of the two devices. According to... Figure 5 As shown, the relative time difference can be defined as: In the formula, The relative time difference The start time of the audio signal. The starting point of the electrocardiogram (ECG) signal is then determined. Based on the audio signal, the duration between the start of the audio signal and the start of the music stimulus recording is then determined. For example, the actual start time of the TSM music playback can be determined by manually listening to the audio signal recorded by the audio acquisition host. The duration between the start of the audio signal and the start of the music stimulus recording is determined. Based on the relative time difference between the recording clocks of the two devices and the duration between the start of the audio signal and the start of the music stimulus recording, the duration between the start of the ECG signal and the start of the music stimulus recording is calculated. The formula for calculating the duration between the start of the ECG signal and the start of the music stimulus recording is as follows: In the formula, The duration between the start of the electrocardiogram signal and the start of the music stimulus recording. This refers to the duration between the start of the audio signal and the start of the music stimulus recording. Then, based on the duration between the start of the ECG signal and the start of the music stimulus recording, the start point for music stimulus recording in the target ECG signal is determined. Finally, based on the preset duration of the music stimulus and the starting point of the music stimulus recording in the target ECG signal, the ECG signal segment is determined, and then... This allows for the precise extraction of the period of musical stimulation from electrocardiogram (ECG) recording data. to ECG segments (i.e., electrocardiogram signal segments) of +4 minutes were used for subsequent HRV analysis.

[0140] In some embodiments, step S103 involves performing a first preprocessing on the first ECG signal segment to obtain a first target ECG signal with a high signal-to-noise ratio, which may include, but is not limited to, the following steps:

[0141] The first ECG signal segment was bandpass filtered to obtain a first target ECG signal with a high signal-to-noise ratio.

[0142] In some embodiments, step S203, performing a first preprocessing on the second ECG signal segment to obtain a second target ECG signal with a high signal-to-noise ratio, may include, but is not limited to, the following steps:

[0143] The second ECG signal segment was bandpass filtered to obtain a second target ECG signal with a high signal-to-noise ratio.

[0144] Specifically, in this embodiment, a high-pass filter and a low-pass filter can be used to filter the ECG signal segment. For example, a zero-phase high-pass filter (cutoff frequency: -3dB, 0.5Hz) and a low-pass filter (cutoff frequency: -3dB, 35Hz) can be used to filter the ECG signal segment, aiming to effectively suppress baseline drift (low-frequency interference <0.5Hz), power line interference (50Hz), and high-frequency electromyographic noise (>35Hz), thereby significantly improving signal quality and signal-to-noise ratio (SNR).

[0145] In some embodiments, step S104 involves performing a second preprocessing on the first target electrocardiogram signal to obtain a first target heartbeat interval sequence, which may include, but is not limited to, the following steps:

[0146] The QRS complex wave detection algorithm was used to locate the peak of the first filtered ECG signal segment and determine the R wave peak corresponding to each cardiac cycle.

[0147] Generate a peak timestamp sequence based on multiple R-wave peaks;

[0148] Based on the peak timestamp sequence, generate the initial heartbeat interval sequence defined by the R-wave peak;

[0149] The initial heartbeat interval sequence was corrected using a preset peak correction algorithm to obtain the first target heartbeat interval sequence.

[0150] In some embodiments, step S204 involves performing a second preprocessing on the second target electrocardiogram signal to obtain a second target heartbeat interval sequence, which may include, but is not limited to, the following steps:

[0151] The QRS complex wave detection algorithm was used to locate the peak of the filtered second ECG signal segment and determine the R wave peak corresponding to each cardiac cycle.

[0152] Generate a peak timestamp sequence based on multiple R-wave peaks;

[0153] Based on the peak timestamp sequence, generate the initial heartbeat interval sequence defined by the R-wave peak;

[0154] The initial heartbeat interval sequence was corrected using a preset peak correction algorithm to obtain the second target heartbeat interval sequence.

[0155] In some embodiments, a robust QRS complex detection algorithm (Pan-Tompkins algorithm) is used to accurately locate the R-wave peak corresponding to each cardiac cycle on the filtered ECG signal segment, and the output is a precise timestamp sequence of the R-wave peak value. Then, based on the peak timestamp sequence, an initial heartbeat interval sequence is generated, and a preset peak correction algorithm is used to correct the initial heartbeat interval sequence to obtain the target heartbeat interval sequence. For example, the time interval between every two adjacent peak timestamps can be used as a peak interval to generate the initial heartbeat interval sequence. The detected RR interval sequence (i.e., the initial heartbeat interval sequence) is visualized and manually checked, or automated abnormal beat detection and correction algorithms (such as threshold-based elimination or interpolation) are applied to remove or correct artifacts caused by detection errors or ectopic beats (such as premature ventricular contractions) to generate a high-quality RR interval sequence (i.e., the first heartbeat interval sequence). This serves as the basis for subsequent HRV analysis.

[0156] In some embodiments, step S105, calculating the first heart rate variability feature and performing feature filtering on the first target heartbeat interval sequence, may include, but is not limited to, the following steps:

[0157] The first heart rate variability feature set is obtained by extracting time-domain features, frequency-domain features, and nonlinear features from the first target heartbeat interval sequence;

[0158] A significance test of inter-group differences was performed on the first heart rate variability feature set to obtain the first feature subset whose inter-group differences meet the preset requirements.

[0159] In some embodiments, step S205, calculating the second heart rate variability feature and performing feature filtering on the first target heartbeat interval sequence, may include, but is not limited to, the following steps:

[0160] The second target heart rate variability feature set is obtained by extracting time-domain features, frequency-domain features, and nonlinear features from the second target heart rate interval sequence.

[0161] A significance test was performed on the second heart rate variability feature set between groups to obtain the second feature subset whose differences between groups met the preset requirements.

[0162] In this embodiment, after obtaining the feature subset, the principal component components with reduced dimensionality are obtained based on the feature subset analysis.

[0163] In some embodiments, step S206, using the second heart rate variability feature to pre-train the principal component analysis model, and outputting the dimensionality-reduced second principal component components through the pre-trained principal component analysis model, may include, but is not limited to, the following steps:

[0164] Principal component analysis is performed on the second feature subset to extract principal components whose cumulative variance is greater than or equal to a preset threshold. A principal component analysis model is constructed and the principal component feature vectors are saved.

[0165] The second principal component components are obtained by reducing the dimensionality of the second feature subset using the principal component eigenvectors.

[0166] Specifically, in this embodiment, after constructing the principal component analysis model, the process of inputting the first heart rate variability feature into the pre-trained principal component analysis model and outputting the dimensionality-reduced first principal component component can be achieved by using the principal component feature vector obtained from the pre-trained principal component model to reduce the dimensionality of the first feature subset corresponding to the first heart rate variability feature to obtain the first principal component component.

[0167] In some embodiments, feature extraction can be performed on the target heart rate interval sequence to obtain time-domain features, frequency-domain features, and nonlinear features, and a heart rate variability feature set can be obtained. The time-domain features may include mean RR interval (MeanRR), standard deviation of RR interval (SDNN), root mean square of the difference between adjacent RR intervals (RMSSD), percentage of adjacent RR interval differences greater than 50ms (pNN50), etc. The frequency-domain features can be extracted by estimating the power spectral density (PSD) of the RR interval sequence (or instantaneous heart rate sequence) (using the Lomb-Scargle periodogram method or autoregressive model method) and calculating the absolute power (ms²), relative power (nu), and LF / HF ratio of the preset frequency band (low frequency LF: 0.04-0.15Hz, high frequency HF: 0.15-0.4Hz). The nonlinear features may include Poincaré plot indices (SD1, SD2), detrended volatility analysis (DFA) scaling indices (ɑ1, ɑ2), sample entropy (SampEn), etc. Then, a significance test for inter-group differences is performed on the heart rate variability feature set to obtain a subset of features whose inter-group differences meet the preset requirements. For example, on the training dataset, a significance test for inter-group differences can be performed on all extracted HRV features (e.g., the Mann-Whitney U test comparing the MCS group and the VS / UWS group), retaining the feature subset Fsig that shows significant differences (p<0.05) between groups. Principal component analysis is then performed on the feature subset, and multiple principal components are sorted according to the feature subset to obtain the initial principal component sequence. Finally, principal components with cumulative variance greater than or equal to a preset threshold are selected from the initial principal component sequence as principal component components. The principal component sequence is the set of the first few principal components in the initial principal component sequence, representing the minimum number required for the cumulative explained variance to reach the preset threshold. For example, the first k principal components, representing the minimum number required for the cumulative explained variance to reach the preset threshold (e.g., ≥95%), can be selected to obtain the first principal component sequence. This is used to perform dimensionality reduction transformation on heart rate variability features to obtain principal component components. Specifically, in the regression modeling process of this embodiment, the dimensionality-reduced principal component score vector can be used as the input feature vector for subsequent regression modeling, thereby effectively reducing feature dimensionality, eliminating multicollinearity, and retaining the main variation information of the data.

[0168] In some embodiments, in step S207, a predictive model for the patient's clinical behavior score is constructed using a supervised learning regression algorithm based on the second principal component, and the pre-trained consciousness impairment assessment model is generated by combining the clinical behavior score. This may include, but is not limited to, the following steps:

[0169] Based on the linear regression equation, construct the first regression model;

[0170] A second regression model is constructed based on a power-law-based nonlinear regression equation.

[0171] The second principal component is input into the first regression model to train the first regression model.

[0172] The second principal component is input into the second regression model to train the second regression model;

[0173] If the prediction error of the first regression model after training is smaller than the prediction error of the second regression model after training, then the first regression model after training is used as the model for assessing consciousness disorders; otherwise, the second regression model after training is used as the model for assessing consciousness disorders.

[0174] In some embodiments, the supervised learning regression algorithm includes a linear regression equation and a power-law-based nonlinear regression equation. A mapping model from principal component features to CRS-R (Coma Recovery Scale) scores can be established first using the supervised learning regression algorithm. A first regression model can be constructed based on the linear regression equation, and a second regression model can be constructed based on the power-law-based nonlinear regression equation. The expression for the linear regression equation is as follows: In the formula, The score for the first coma recovery scale, M represents the number of principal components. and These are all regression parameters that need to be fitted in the linear regression equation. For the m-th principal component feature, The residual part of the model follows... Distribution. The expression for the nonlinear regression equation based on the power exponent is: In the formula, For the second coma recovery scale score, M represents the number of principal components. , and These are all regression parameters that need to be fitted in the nonlinear regression equation. For the m-th principal component feature, The residual part of the model follows... Distribution. Then, the second principal component is input into the first regression model to train the first regression model, and the second principal component is input into the second regression model to train the second regression model. The first and second regression models are fitted using the training data. The model hyperparameters (for linear models) are optimized through cross-validation (e.g., k-fold cross-validation). , Nonlinear models , , The goal is to minimize the prediction error. If the prediction error of the first regression model after training is less than the prediction error of the second regression model after training, then the first regression model after training is used as the model for assessing consciousness disorders; otherwise, the second regression model after training is used as the model for assessing consciousness disorders. For example, the performance of the trained first and second regression models can be evaluated on independent validation or test sets. Objective regression performance metrics are used to quantify the error between the predicted CRS-R score and the actual CRS-R score, obtaining the prediction error, such as mean absolute error (MAE), root mean square error (RMSE), coefficient of determination (R²), etc. The regression model with the smallest prediction error is used as the model for assessing consciousness disorders.

[0175] The beneficial effects of implementing the embodiments of the present invention include: This application provides a method and apparatus for assessing consciousness disorders based on heart rate variability characteristics under standardized music stimulation. In the embodiments of this application, the continuous electrocardiogram signal of the patient to be assessed during standardized music stimulation is collected, and then data preprocessing and feature extraction are performed to obtain heart rate variability characteristics. The heart rate variability characteristics are input into a pre-trained principal component analysis model, and the dimensionality-reduced principal component score vector is output. Finally, the principal component score vector is input into a pre-trained consciousness disorder assessment model to obtain the consciousness disorder assessment result, thereby realizing the assessment of consciousness disorders and improving the accuracy. The pre-trained principal component analysis model and the pre-trained consciousness impairment assessment model are obtained through the following steps: ECG and audio acquisition devices record a second ECG signal and a second audio signal, respectively, and obtain a third synchronization signal and a fourth synchronization signal recorded simultaneously with the two signals; time synchronization processing is performed based on the third and fourth synchronization signals to determine the second timestamp of the second ECG signal segment during music intervention, and the second ECG signal segment is extracted based on the second timestamp; a first preprocessing is performed on the second ECG signal segment to obtain a high signal-to-noise ratio second target ECG signal; a second preprocessing is performed on the second target ECG signal to obtain a second target heartbeat interval sequence; a second heart rate variability feature is calculated and feature screening is performed on the second target heartbeat interval sequence, and the dimensionality-reduced second principal component components and the pre-trained principal component analysis model are obtained through principal component analysis; based on the second principal component components, a predictive model for patient clinical behavior scoring is constructed using a supervised learning regression algorithm.

[0176] like Figure 6 As shown, this embodiment of the invention also provides a device for assessing consciousness disorders based on heart rate variability characteristics under standardized music stimulation, comprising:

[0177] The signal acquisition module 601 is used to acquire the first electrocardiogram signal of the patient to be evaluated and the first audio signal during standardized music stimulation, and to obtain the first synchronization signal and the second synchronization signal recorded simultaneously with the two signals, respectively.

[0178] The data processing module 602 is used to perform time synchronization processing based on the first synchronization signal and the second synchronization signal, determine the first timestamp of the first electrocardiogram signal segment corresponding to the first electrocardiogram signal during the standardized music stimulation, and extract the first electrocardiogram signal segment from the first electrocardiogram signal based on the first timestamp.

[0179] The preprocessing module 603 is used to perform a first preprocessing on the first ECG signal segment to obtain a first target ECG signal with a high signal-to-noise ratio; and to perform a second preprocessing on the first target ECG signal to obtain a first target heartbeat interval sequence.

[0180] Calculation module 604 is used to calculate the first heart rate variability features and perform feature filtering processing on the first target heartbeat interval sequence;

[0181] Principal component analysis module 605 is used to input the first heart rate variability feature into the pre-trained principal component analysis model and output the first principal component component after dimensionality reduction.

[0182] The model recognition module 606 is used to input the first principal component into the pre-trained consciousness disorder assessment model and output a quantitative assessment value of the level of consciousness disorder.

[0183] The pre-trained principal component analysis model and the pre-trained consciousness impairment assessment model are obtained through the following steps:

[0184] The experimental data and clinical behavior scores of multiple patients with impaired consciousness under the same standardized music stimulation were obtained. The experimental data included a second electrocardiogram signal, a second audio signal, and a third synchronization signal and a fourth synchronization signal recorded simultaneously with the two signals, respectively.

[0185] Time synchronization processing is performed based on the third and fourth synchronization signals to determine the second timestamp of the second ECG signal segment corresponding to the standardized music stimulation period, and the second ECG signal segment is extracted from the second ECG signal based on the second timestamp.

[0186] The second ECG signal segment is subjected to a first preprocessing to obtain a second target ECG signal with a high signal-to-noise ratio;

[0187] The second target ECG signal is subjected to a second preprocessing to obtain the second target heartbeat interval sequence;

[0188] Calculate the second heart rate variability feature and perform feature filtering on the second target heartbeat interval sequence;

[0189] The principal component analysis model is pre-trained using the second heart rate variability feature, and the second principal component component after dimensionality reduction is output through the pre-trained principal component analysis model.

[0190] Based on the second principal component, a predictive model for the patient's clinical behavior score is constructed using a supervised learning regression algorithm, and the pre-trained consciousness impairment assessment model is generated by combining the clinical behavior score.

[0191] The content of the above method embodiments is applicable to the device embodiments. The specific functions implemented by the device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0192] like Figure 7 As shown, embodiments of the present invention also provide a computer device, including:

[0193] At least one processor 701;

[0194] At least one memory 702 is used to store at least one program;

[0195] When at least one program is executed by at least one processor, the at least one processor performs the method described above.

[0196] The content of the above method embodiments is applicable to the device embodiments. The specific functions implemented by the device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0197] This invention also provides a data acquisition device, comprising:

[0198] ECG signal amplifier, used for amplifying and converting ECG signals from analog to digital;

[0199] ECG acquisition processor and memory are used to process the recording (digital-to-analog conversion and data storage) of ECG signals and synchronization signals.

[0200] A recording microphone, used for capturing audio signals;

[0201] Audio acquisition processor and memory, used to process the recording of audio signals and synchronization signals;

[0202] A music player used to play standardized music.

[0203] The content of the above method embodiments is applicable to the device embodiments. The specific functions implemented by the device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0204] The embodiments described in this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this application are also applicable to similar technical problems.

[0205] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of this application, and may include more or fewer steps than shown, or combine certain steps, or different steps.

[0206] The preferred embodiments of the present application have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of the present application. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and substance of the embodiments of the present application shall be within the scope of the claims of the present application.

Claims

1. A method for assessing a disturbance of consciousness based on features of heart rate variability under standardized musical stimulation, characterized in that, Includes the following steps: The first electrocardiogram signal and the first audio signal during standardized music stimulation of the patient to be evaluated were acquired, and the first synchronization signal and the second synchronization signal were recorded simultaneously with the two signals, respectively. Time synchronization processing is performed based on the first synchronization signal and the second synchronization signal to determine the first timestamp of the first ECG signal segment corresponding to the first ECG signal during the standardized music stimulation, and the first ECG signal segment is extracted from the first ECG signal based on the first timestamp. The first ECG signal segment is preprocessed to obtain a first target ECG signal with a high signal-to-noise ratio. The first target electrocardiogram signal is subjected to a second preprocessing to obtain the first target heartbeat interval sequence; Calculate the first heart rate variability feature and perform feature filtering processing on the first target heartbeat interval sequence; The first heart rate variability feature is input into the pre-trained principal component analysis model, and the first principal component component after dimensionality reduction is output. The first principal component is input into the pre-trained consciousness impairment assessment model, and the quantitative assessment value of the consciousness impairment level is output. The pre-trained principal component analysis model and the pre-trained consciousness impairment assessment model are obtained through the following steps: The experimental data and clinical behavior scores of multiple patients with impaired consciousness under the same standardized music stimulation were obtained. The experimental data included a second electrocardiogram signal, a second audio signal, and a third synchronization signal and a fourth synchronization signal recorded simultaneously with the two signals, respectively. Time synchronization processing is performed based on the third synchronization signal and the fourth synchronization signal to determine the second timestamp of the second ECG signal segment corresponding to the second ECG signal during the standardized music stimulation, and the second ECG signal segment is extracted from the second ECG signal based on the second timestamp. The second ECG signal segment is subjected to a first preprocessing to obtain a second target ECG signal with a high signal-to-noise ratio; The second target electrocardiogram signal is subjected to a second preprocessing to obtain the second target heartbeat interval sequence; Calculate the second heart rate variability feature and perform feature filtering on the second target heartbeat interval sequence; The principal component analysis model is pre-trained using the second heart rate variability feature, and the second principal component component after dimensionality reduction is output through the pre-trained principal component analysis model. Based on the second principal component, a predictive model for the patient's clinical behavior score is constructed using a supervised learning regression algorithm, and the pre-trained consciousness impairment assessment model is generated by combining the clinical behavior score.

2. The method of claim 1, wherein, The step of performing time synchronization processing based on the first synchronization signal and the second synchronization signal to determine the first timestamp of the first ECG signal segment corresponding to the first ECG signal during the standardized music stimulation, and extracting the first ECG signal segment from the first ECG signal based on the first timestamp, includes: Perform cross-correlation analysis on the first synchronization signal and the second synchronization signal, and calculate the first time shift value corresponding to the maximum cross-correlation coefficient between the two synchronization signals. The first time shift value is used as the first relative time difference between the recording clocks of the two devices. Based on the first audio signal, determine the duration between the start point of the first audio signal and the start point of the music stimulus recording; The duration from the start of the first electrocardiogram signal to the start of the music stimulation recording is calculated based on the first relative time difference between the recording clocks of the two devices and the duration from the start of the first audio signal to the start of the music stimulation recording. The starting point of music stimulation recording in the first target electrocardiogram signal is determined based on the duration between the starting point of the first electrocardiogram signal and the starting point of music stimulation recording. The first ECG signal segment is determined based on the preset duration of music stimulation and the starting point of music stimulation recording in the first target ECG signal.

3. The method of claim 1, wherein, The first preprocessing of the first ECG signal segment to obtain a high signal-to-noise ratio first target ECG signal includes: The first ECG signal segment is subjected to bandpass filtering to obtain a first target ECG signal with a high signal-to-noise ratio.

4. The method of claim 1, wherein, The step of performing a second preprocessing on the first target electrocardiogram signal to obtain the first target heartbeat interval sequence includes: The QRS complex wave detection algorithm was used to locate the peak of the filtered first ECG signal segment and determine the R wave peak corresponding to each cardiac cycle. Based on the multiple R-wave peaks, a peak timestamp sequence is generated; Based on the peak timestamp sequence, an initial heartbeat interval sequence defined by the R-wave peak is generated; The initial heartbeat interval sequence is corrected using a preset peak correction algorithm to obtain the first target heartbeat interval sequence.

5. The method of claim 1, wherein, The calculation of the second heart rate variability feature and feature filtering process for the second target heart rate interval sequence includes: The second target heart rate interval sequence is subjected to time-domain features, frequency-domain features, and nonlinear features to obtain the second heart rate variability feature set; The significance test of inter-group differences was performed on the second heart rate variability feature set to obtain the second feature subset whose inter-group differences meet the preset requirements.

6. The method of claim 5, wherein, The step of pre-training the principal component analysis model using the second heart rate variability feature and outputting the dimensionality-reduced second principal component components through the pre-trained principal component analysis model includes: Principal component analysis is performed on the second feature subset to extract principal components whose cumulative variance is greater than or equal to a preset threshold, the principal component analysis model is constructed and the principal component feature vectors are saved. The second principal component components are obtained by reducing the dimensionality of the second feature subset using the principal component feature vector.

7. The method of claim 1, wherein, The supervised learning regression algorithm includes linear regression equations and power-law-based nonlinear regression equations. The step of constructing a predictive model for patient clinical behavior scores using the supervised learning regression algorithm based on the second principal component components, and generating the pre-trained consciousness impairment assessment model by combining the clinical behavior scores, includes: Based on the linear regression equation, construct the first regression model; Based on the power-based nonlinear regression equation, a second regression model is constructed; The second principal component is input into the first regression model to train the first regression model. The second principal component is input into the second regression model to train the second regression model; If the prediction error of the first regression model after training is less than the prediction error of the second regression model after training, then the first regression model after training is used as the consciousness disorder assessment model; otherwise, the second regression model after training is used as the consciousness disorder assessment model.

8. The method of claim 7, wherein, The expression for the linear regression equation is as follows: ； In the formula, The score for the first coma recovery scale, M represents the number of principal components. and These are all regression parameters that need to be fitted in the linear regression equation. For the m-th principal component feature, This refers to the residual part of the model; The expression for the power-law-based nonlinear regression equation is as follows: ； In the formula, For the second coma recovery scale score, M represents the number of principal components. , and These are all regression parameters that need to be fitted in the nonlinear regression equation. For the m-th principal component feature, This represents the residual part of the model.

9. Apparatus for assessing disorders of consciousness based on features of heart rate variability under standardized musical stimulation, characterized in that, include: The signal acquisition module is used to acquire the first electrocardiogram signal and the first audio signal during standardized music stimulation of the patient to be evaluated, and to obtain the first synchronization signal and the second synchronization signal recorded simultaneously with the two signals, respectively. The data processing module is used to perform time synchronization processing based on the first synchronization signal and the second synchronization signal, determine the first timestamp of the first ECG signal segment corresponding to the first ECG signal during the standardized music stimulation, and extract the first ECG signal segment from the first ECG signal based on the first timestamp. The preprocessing module is used to perform a first preprocessing on the first ECG signal segment to obtain a first target ECG signal with a high signal-to-noise ratio; The first target electrocardiogram signal is subjected to a second preprocessing to obtain the first target heartbeat interval sequence; The calculation module is used to calculate the first heart rate variability feature and perform feature filtering processing on the first target heartbeat interval sequence; The principal component analysis module is used to input the first heart rate variability feature into the pre-trained principal component analysis model and output the first principal component component after dimensionality reduction. The model recognition module is used to input the first principal component into the pre-trained consciousness disorder assessment model and output a quantitative assessment value of the level of consciousness disorder. The pre-trained principal component analysis model and the pre-trained consciousness impairment assessment model are obtained through the following steps: The experimental data and clinical behavior scores of multiple patients with impaired consciousness under the same standardized music stimulation were obtained. The experimental data included a second electrocardiogram signal, a second audio signal, and a third synchronization signal and a fourth synchronization signal recorded simultaneously with the two signals, respectively. Time synchronization processing is performed based on the third synchronization signal and the fourth synchronization signal to determine the second timestamp of the second ECG signal segment corresponding to the second ECG signal during the standardized music stimulation, and the second ECG signal segment is extracted from the second ECG signal based on the second timestamp. The second ECG signal segment is subjected to a first preprocessing to obtain a second target ECG signal with a high signal-to-noise ratio; The second target electrocardiogram signal is subjected to a second preprocessing to obtain the second target heartbeat interval sequence; Calculate the second heart rate variability feature and perform feature filtering on the second target heartbeat interval sequence; The principal component analysis model is pre-trained using the second heart rate variability feature, and the second principal component component after dimensionality reduction is output through the pre-trained principal component analysis model. Based on the second principal component, a predictive model for the patient's clinical behavior score is constructed using a supervised learning regression algorithm, and the pre-trained consciousness impairment assessment model is generated by combining the clinical behavior score.

10. A computer apparatus, comprising: include: At least one processor; At least one memory for storing at least one program; When the at least one program is executed by the at least one processor, the at least one processor is caused to implement the method recited in any one of claims 1-8.

Images