CN113925480B - Coronary heart disease patient bleeding risk assessment method based on machine learning - Google Patents

Abstract

Embodiments of the present disclosure provide a coronary heart disease patient bleeding risk assessment method, apparatus, device and computer-readable storage medium based on machine learning. The method comprises acquiring a PPG measurement signal of a coronary heart disease patient; carrying out segmentation processing on the PPG measurement signal to obtain M signal segments; m is a positive integer greater than or equal to 1; performing signal quality evaluation on the M signal segments through a preset algorithm to determine signal segments with qualified quality; extracting X characteristics from the signal segments with qualified quality, and inputting the characteristics into a bleeding risk evaluation model to obtain a bleeding risk score of a coronary heart disease patient; x is a positive integer of 1 or more. In this way, a bleeding risk assessment for patients with coronary heart disease is achieved.

Description

Bleeding risk assessment method for patients with coronary heart disease based on machine learning

technical field

Embodiments of the present disclosure generally relate to the field of signal monitoring, and more specifically, relate to a machine learning-based bleeding risk assessment method, device, device, and computer-readable storage medium for patients with coronary heart disease.

Background technique

According to the report of the World Health Organization, cardiovascular diseases accounted for 17.9% of various diseases, and 1 million people died of cardiovascular diseases in 2016, of which patients with coronary artery disease (Coronary Artery Disease) accounted for the main proportion. For patients with coronary heart disease, drug therapy (such as antiplatelet therapy) is the main treatment. During antithrombotic therapy in patients with coronary heart disease, bleeding events are the focus of attention. The occurrence of a bleeding event may lead to treatment interruption, long-term disability, and even death of the patient. Therefore, it is of great significance to conduct bleeding risk assessment research in patients with coronary heart disease to find sensitive factors related to prognosis.

Photoplethysmography (Photoplethysmography, PPG) is a low-cost and non-invasive detection technique that can be applied to the assessment of the cardiovascular system. Changes in the state of the cardiovascular system can be assessed by morphological features of the PPG waveform, such as aortic stiffness, blood volume detection, etc.

However, there are no studies evaluating the risk of bleeding in patients with coronary heart disease by PPG.

Contents of the invention

According to an embodiment of the present disclosure, a machine learning-based bleeding risk assessment scheme for patients with coronary heart disease is provided.

In the first aspect of the present disclosure, a machine learning-based bleeding risk assessment method for patients with coronary heart disease is provided. The method includes:

Obtaining PPG measurement signals of patients with coronary heart disease;

Carry out segment processing to described PPG measurement signal, obtain N signal segments; N is the positive integer greater than or equal to 1;

Carrying out signal quality evaluation on the N signal segments through a preset algorithm, and determining signal segments with qualified quality;

X features are extracted from the signal segments with qualified quality and input into the bleeding risk assessment model to obtain the bleeding risk score of patients with coronary heart disease; X is a positive integer greater than or equal to 1.

Further, the segmenting the PPG measurement signal to obtain N signal segments includes:

Remove baseline drift and high-frequency noise of the PPG measurement signal through a Butterworth band-pass filter;

Based on the waveform of the PPG measurement signal, segment processing is performed on the PPG measurement signal to obtain N signal segments.

Further, performing signal quality evaluation on the N signal segments through a preset algorithm, and determining a signal segment with qualified quality includes:

Resampling each of the N signal segments; wherein the resampling length is the median of the N signal segment lengths;

Substituting the resampled segmented signals into preset formulas, if the calculation result is greater than the preset threshold, the current segmented signal is a qualified signal.

Further, the bleeding risk assessment model is trained through the following steps:

Generate a training sample set, wherein the training sample includes a feature vector corresponding to a PPG signal with labeling information, wherein the labeling information is a bleeding condition, and the bleeding is marked as 1, and the non-bleeding is marked as 0;

Using the samples in the training sample set to train the bleeding risk assessment model, the feature vector corresponding to the PPG signal is used as input, and the bleeding condition is output, when the unity rate of the output bleeding condition and the marked bleeding condition meets a preset threshold , complete the training of the bleeding risk assessment model.

Further, the eigenvector corresponding to the PPG signal includes:

The collected PPG signal is analyzed, and feature vectors corresponding to the PPG signal are extracted from time domain, frequency domain and wavelet packet decomposition respectively.

Further, said analyzing the collected PPG signal includes:

The PPG signal, the first derivative of the PPG signal and the second derivative of the PPG signal are analyzed.

Further, the XGBoost algorithm is used to train the bleeding risk assessment model.

In the second aspect of the present disclosure, a device for assessing bleeding risk in patients with coronary heart disease based on machine learning is provided. The unit includes:

An acquisition module, configured to acquire PPG measurement signals of patients with coronary heart disease;

A processing module, configured to perform segmentation processing on the PPG measurement signal to obtain N signal segments; N is a positive integer greater than or equal to 1;

An evaluation module, configured to evaluate the signal quality of the N signal segments through a preset algorithm, and determine a signal segment with qualified quality;

The scoring module is used to extract X features from the qualified signal segment and input them into the bleeding risk assessment model to obtain a bleeding risk score; X is a positive integer greater than or equal to 1.

In a third aspect of the present disclosure, an electronic device is provided. The electronic device includes: a memory and a processor, where a computer program is stored in the memory, and the processor implements the method as described above when executing the program.

In a fourth aspect of the present disclosure, there is provided a computer-readable storage medium on which a computer program is stored, and when the program is executed by a processor, the method according to the first aspect of the present disclosure is implemented.

The bleeding risk assessment method for patients with coronary heart disease based on machine learning provided in the embodiment of the present application obtains the PPG measurement signal; performs segmentation processing on the PPG measurement signal to obtain N signal segments; N is a positive integer greater than or equal to 1 ; Carry out signal quality assessment on the N signal segments by a preset algorithm, and determine the qualified signal segments; extract X time-domain features from the qualified signal segments, and input them into the bleeding risk assessment model Among them, the bleeding risk score of patients with coronary heart disease is obtained; X is a positive integer greater than or equal to 1, which realizes the bleeding risk assessment of CAD patients.

It should be understood that what is described in the Summary of the Invention is not intended to limit the key or important features of the embodiments of the present disclosure, nor is it intended to limit the scope of the present disclosure. Other features of the present disclosure will be readily understood through the following description.

Description of drawings

The above and other features, advantages and aspects of the various embodiments of the present disclosure will become more apparent with reference to the following detailed description when taken in conjunction with the accompanying drawings. In the drawings, identical or similar reference numerals denote identical or similar elements, wherein:

FIG. 1 shows a schematic diagram of an exemplary operating environment in which embodiments of the present disclosure can be implemented;

2 shows a flow chart of a machine learning-based bleeding risk assessment method for patients with coronary heart disease according to an embodiment of the present disclosure;

Fig. 3 shows a schematic diagram of PPG, VPG and APG signals according to an embodiment of the present disclosure;

4 shows a schematic diagram of a 10-fold cross-validation ROC curve and an average ROC curve of an XGBoost model according to an embodiment of the present disclosure;

FIG. 5 shows a schematic diagram of analyzing the characteristics of the XGBoost model by the SHAP framework according to an embodiment of the present disclosure;

FIG. 6 shows a block diagram of a device for assessing bleeding risk in patients with coronary heart disease based on machine learning according to an embodiment of the present disclosure;

FIG. 7 shows a block diagram of an exemplary electronic device capable of implementing embodiments of the present disclosure.

Detailed ways

In order to make the purpose, technical solutions and advantages of the embodiments of the present disclosure clearer, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings in the embodiments of the present disclosure. Obviously, the described embodiments It is a part of the embodiments of the present disclosure, but not all of them. Based on the embodiments in the present disclosure, all other embodiments obtained by persons of ordinary skill in the art without creative efforts fall within the protection scope of the present disclosure.

In addition, the term "and/or" in this article is only an association relationship describing associated objects, which means that there may be three relationships, for example, A and/or B, which may mean: A exists alone, A and B exist at the same time, There are three cases of B alone. In addition, the character "/" in this article generally indicates that the contextual objects are an "or" relationship.

FIG. 1 shows an exemplary system architecture 100 of an embodiment of the method and apparatus for assessing bleeding risk in patients with coronary heart disease based on machine learning of the present application.

As shown in FIG. 1 , a system architecture 100 may include terminal devices 101 , 102 , 103 , a network 104 and a server 105 . The network 104 is used as a medium for providing communication links between the terminal devices 101 , 102 , 103 and the server 105 . Network 104 may include various connection types, such as wires, wireless communication links, or fiber optic cables, among others.

Users can use terminal devices 101 , 102 , 103 to interact with server 105 via network 104 to receive or send messages and the like. Various communication client applications, such as model training applications, video recognition applications, web browser applications, and social platform software, can be installed on the terminal devices 101, 102, and 103.

The terminal devices 101, 102, and 103 may be hardware or software. When the terminal devices 101, 102, 103 are hardware, they can be various electronic devices with display screens, including but not limited to smart phones, tablet computers, e-book readers, MP3 players (Moving Picture Experts Group Audio Layer III, Moving Picture Experts Compression Standard Audio Layer 3), MP4 (Moving Picture Experts Group Audio Layer IV, Moving Picture Experts Group Audio Layer 4) Players, PPG Measuring Devices, Laptop Portable Computers and Desktop Computers, etc. When the terminal devices 101, 102, 103 are software, they can be installed in the electronic devices listed above. It may be implemented as multiple software or software modules (for example, multiple software or software modules for providing distributed services), or as a single software or software module. No specific limitation is made here.

When the terminals 101, 102, and 103 are hardware, video acquisition equipment may also be installed on them. The video capture device may be various devices capable of capturing video, such as a camera, a sensor, and the like. Users can use the video capture devices on the terminals 101, 102, 103 to capture videos.

The server 105 may be a server that provides various services, such as a background server that processes data displayed on the terminal devices 101 , 102 , 103 . The background server can perform analysis and other processing on the received data, and can feed back the processing results (eg evaluation results) to the terminal device.

It should be noted that the server may be hardware or software. When the server is hardware, it can be implemented as a distributed server cluster composed of multiple servers, or as a single server. When the server is software, it can be implemented as multiple software or software modules (for example, multiple software or software modules for providing distributed services), or can be implemented as a single software or software module. No specific limitation is made here.

It should be understood that the numbers of terminal devices, networks and servers in Fig. 1 are only illustrative. According to the implementation needs, there can be any number of terminal devices, networks and servers. In particular, in the case where the target data does not need to be obtained from a remote location, the above-mentioned system architecture may not include a network, but only include terminal devices or servers.

As shown in FIG. 2 , it is a flow chart of a method for assessing bleeding risk in patients with coronary heart disease based on machine learning according to an embodiment of the present application. As can be seen from Fig. 2, the bleeding risk assessment method for patients with coronary heart disease based on machine learning in this embodiment includes the following steps:

S210, acquiring a PPG measurement signal of a patient with coronary heart disease.

In some embodiments, the executive body (such as the server shown in FIG. 1 ) used for the machine learning-based bleeding risk assessment method for patients with coronary heart disease can acquire the PPG measurement signal through wired or wireless connection.

In some embodiments, the above-mentioned executive body can obtain the PPG measurement signal sent by the electronic device (such as the terminal device shown in FIG. patient) PPG data, for example, optical heart rate sensor, etc.

In some embodiments, when the PPG measurement device cannot upload data alone, such as some finger-end PPG measurement devices, the PPG measurement device can also send the measurement results to the mobile device (mobile phone, etc.) connected to it, through the mobile The device uploads the relevant PPG measurement information to the server shown in Figure 1.

In some embodiments, for the accuracy of measurement, PPG measurement data can be collected through various light intensities (luminous intensity); the collection time is usually 20s-60s; it can be set according to the actual application scene, and no further limitation is made on this .

In some embodiments, the signal waveform of the PPG is as shown in Figure 3, where VPG is the first derivative of PPG and APG is the second derivative of PPG in Figure 3; T and Y represent the time ms and the amplitude of the corresponding point.

It should be noted that the PPG measuring device (electronic device) in the present disclosure is generally a portable device equipped with a PPG sensor, which can be used in a home environment.

S220. Perform segmentation processing on the PPG measurement signal to obtain M signal segments; M is a positive integer greater than or equal to 1.

In some embodiments, before segmenting the PPG measurement signal, the PPG measurement signal needs to be preprocessed;

Specifically, the baseline drift and high-frequency noise of the PPG measurement signal can be removed by using a Butterworth band-pass filter with cut-off frequencies set at 0.2 and 20 Hz.

In some embodiments, according to the waveform period of the PPG measurement signal, the preprocessed PPG measurement signal is segmented to obtain M signal segments, which can be recorded as S={S1, S2,...,SM}; Where M is a positive integer greater than or equal to 1; S represents a set of signal segments.

S230. Perform signal quality evaluation on the M signal segments by using a preset algorithm, and determine signal segments with qualified quality.

Considering the noise that may be generated by the user during the measurement process, some segments in S may be damaged. Therefore, it is necessary to detect the signal quality of each segment, that is, to evaluate the signal quality of each segment.

In some embodiments, each of the M signal segments is resampled, that is, each segment in the set S is resampled to have the same length, which can be written as RS={RS ₁ , RS ₂ ,…,RS _M }; Among them, the sampling length is the median of all segment lengths;

Substitute the segmented signals in S into the following formulas respectively, if the calculation conditions are met, it is a segmented signal with qualified quality;

S _valid ＝{S _i ∣S _i ∈S,RS _i ∈RS,r(RS _i ,RS _T )>0.9}

Wherein, the RS _T is the average value of the RS;

The r is the Pearson correlation coefficient, and the value range is -1 to 1;

The 0.9 is a preset threshold, which is obtained from a large number of experiments; the preset threshold can also be set according to manual experience, big data analysis and/or actual application scenarios.

S240, extracting X features from the qualified signal segment and inputting them into the bleeding risk assessment model to obtain the bleeding risk score of the coronary heart disease patient; X is a positive integer greater than or equal to 1.

In some embodiments, X features are extracted from the qualified signal segment, for example, 30 features are extracted, refer to Table 1; T and Y represent the time ms and the amplitude of the corresponding point, refer to FIG. 3 .

Table 1

In some embodiments, for measurement accuracy, each waveform (signal segment) needs to be normalized before feature extraction, and normalized to a range of 0-1.

In some embodiments, the bleeding risk assessment model can be trained through the following steps:

Generate a training sample set, wherein the training sample includes a feature vector corresponding to a PPG signal with labeling information, wherein the labeling information is a bleeding condition, and the bleeding is marked as 1, and the non-bleeding is marked as 0;

Using the samples in the training sample set to train the bleeding risk assessment model, the feature vector corresponding to the PPG signal is used as input, and the bleeding condition is output, when the unity rate of the output bleeding condition and the marked bleeding condition meets a preset threshold , complete the training of the bleeding risk assessment model; the threshold can be set according to the actual application scenario;

Further, based on the output, the bleeding risk score of patients with coronary heart disease can be calculated through manual experience, big data analysis and/or threshold setting methods, the higher the score, the greater the bleeding risk;

Wherein, the eigenvector corresponding to the PPG signal also includes the eigenvector of the first order derivative (VPG) and the second order derivative (APG) of PPG; the synchronous PPG, VPG, APG signals are as shown in Figure 3;

The PPG signal may be a PPG signal collected through big data, such as a PPG signal collected on a large scale within two years;

Analyze the collected PPG signal to obtain the corresponding eigenvector of the PPG signal, that is, analyze the PPG signal, the first-order derivative of the PPG signal and the second-order derivative of the PPG signal, and decompose ( Extract 30-dimensional feature vectors from energy features) (can be set according to actual application scenarios), refer to Table 1;

Wherein, the frequency domain can be determined by the following method:

The PPG signal is composed of rich frequency components. Welch algorithm can be used to calculate the power spectral density of the PPG signal, and the position of each harmonic can be determined by detecting the extreme points, and the second to sixth harmonics can be divided by the fundamental frequency ( The power normalization processing of the first harmonic) determines the frequency domain characteristics, that is, the frequency domain H1～H5 in Table 1;

Through a large number of research experiments, 99% of the energy in the PPG signal is concentrated in the range of 1-10Hz. Therefore, in this disclosure, only the frequency components within 10Hz are studied and extracted;

Specifically, the sampling rate of the PPG signal is 500 Hz. After 8 times of wavelet packet decomposition, the bandwidth of each subband is 0.977 Hz, and 10 components are reserved as E1-E10. The total energy of the PPG signal is expressed as Eall.

In this disclosure, the collected data of 1683 patients with coronary heart disease were collected and analyzed, among which 114 patients had at least one positive event (bleeding), and the records without demographics and poor signal quality were deleted. Obtain the demographic characteristics of patients as shown in Table 2;

Table 2

Among them, there are 10 characteristics with significant statistical differences between the negative group and the positive group, see Table 3:

table 3

See Table 3, the Td of the positive group is smaller than that of the negative group, but the RI is larger. From the perspective of the geometric characteristics of the PPG waveform, it is possible that the diastolic wave delay in the positive group increases the width of the diastolic wave, and Td and Rarea tend to decrease; H1-H5 in the frequency domain had statistical differences between the negative group and the positive group, and the normalized 5 harmonics gradually decreased; the total energy Eall calculated by wavelet packet decomposition was smaller in the positive group.

In some embodiments, based on the collected PPG data of patients with coronary heart disease, LR (Logistic Regression), SVR (Support Vector Regression), RF (Random Forest), and XGBoost algorithms are respectively used to train and obtain a bleeding analysis and evaluation model. 90% of the data is used to train the classification model (training set), and the remaining 10% of the data is used for testing (test set). After 10-fold cross-validation and grid search for training and evaluation, the optimal hyperparameters obtained by grid search are used After 10-fold cross-validation, the average AUC (model evaluation index) of each model is obtained, as shown in Table 4. The sensitivity and specificity performance of each model is shown in Table 5. After comparison, XGBoost performed the best among multiple algorithm models, with an average AUC of 0.762, sensitivity and specificity of 0.679 and 0.714, which were significantly higher than other models. The 10-fold cross-validation ROC curve, average ROC curve and AUC of the XGBoost model are shown in Figure 4.

Table 4

table 5

Further, the SHAP (SHapley Additive interpretation) framework is used to analyze the characteristics of the XGBoost model, as shown in Figure 5, which shows 20 features; the SHAP value indicates the contribution of the feature to the target (negative: 0, positive: 1) . The eigenvalues are represented by different colors, that is, the larger the eigenvalue, the redder the color, and the smaller the eigenvalue, the bluer the color (in Figure 5, the lower gray value is blue, and the higher gray value is red ); for example, with the increase of H4, E10, SI and other eigenvalues, the SHAP value tends to be less than 0; that is, the model is more inclined to determine the current sample as the target 0; on the contrary, as the RI, VDae and other values decrease , the SHAP value tends to be greater than 0.

In addition, the sum of the absolute values of the SHAP values reflects the importance of this feature, therefore, according to this criterion, H4 can be considered as the most important feature. The above observations are also in good agreement with the results in Table 3; for example, for the negative group, Td, Rarea, H4, H1, H5, H2, Eall are consistent, that is, when the eigenvalue increases, the SHAP value tends to be negative, The RI value behaves just the opposite.

In summary, in this disclosure, the XGBoost model can be used to train the bleeding risk assessment model; the SHAP framework is used to describe the bleeding risk assessment model.

In some embodiments, the extracted features are input into the trained bleeding risk assessment model to determine the bleeding risk score of the current patient;

Further, the bleeding risk score is sent to the PPG measurement device and/or the terminal device as shown in FIG. 1 to form a closed loop and help patients realize self-health management.

According to the embodiments of the present disclosure, the following technical effects are achieved:

The present invention is different from other bleeding risk assessment studies for patients with coronary heart disease. The portable PPG device is used independently of the clinical environment, and the user can use the device to evaluate possible bleeding events in the home environment. The system can remotely help remind CAD patients to pay enough attention or assist doctors to choose a treatment plan.

It should be noted that for the foregoing method embodiments, for the sake of simple description, they are all expressed as a series of action combinations, but those skilled in the art should know that the present disclosure is not limited by the described action sequence. Because of this disclosure, certain steps may be performed in other orders or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily required by the present disclosure.

The above is the introduction of the method embodiments, and the solution of the present disclosure will be further described through the device embodiments below.

Fig. 6 shows a block diagram of an apparatus 600 for assessing bleeding risk of patients with coronary heart disease based on machine learning according to an embodiment of the present disclosure. As shown in Figure 6, the device 600 includes:

An acquisition module 610, configured to acquire a PPG measurement signal of a patient with coronary heart disease;

The processing module 620 is configured to perform segmentation processing on the PPG measurement signal to obtain M signal segments; M is a positive integer greater than or equal to 1;

An evaluation module 630, configured to evaluate the signal quality of the M signal segments through a preset algorithm, and determine a signal segment with qualified quality;

The scoring module 640 is configured to extract X features from the qualified signal segment and input them into the bleeding risk assessment model to obtain a bleeding risk score; X is a positive integer greater than or equal to 1.

Those skilled in the art can clearly understand that for the convenience and brevity of the description, the specific working process of the described modules can refer to the corresponding process in the foregoing method embodiments, which will not be repeated here.

FIG. 7 shows a schematic block diagram of an electronic device 700 that can be used to implement embodiments of the present disclosure. The device 700 may be used to implement at least one of the message system 104 and the message arrival rate determination system 106 of FIG. 1 . As shown, the device 700 includes a central processing unit (CPU) 701 that can be programmed according to computer program instructions stored in a read-only memory (ROM) 702 or loaded from a storage unit 708 into a random-access memory (RAM) 703 program instructions to perform various appropriate actions and processes. In the RAM 703, various programs and data necessary for the operation of the device 700 can also be stored. The CPU 701 , ROM 702 , and RAM 703 are connected to each other via a bus 704 . An input/output (I/O) interface 705 is also connected to the bus 704 .

Multiple components in the device 700 are connected to the I/O interface 705, including: an input unit 706, such as a keyboard, a mouse, etc.; an output unit 707, such as various types of displays, speakers, etc.; a storage unit 708, such as a magnetic disk, an optical disk, etc. ; and a communication unit 709, such as a network card, a modem, a wireless communication transceiver, and the like. The communication unit 709 allows the device 700 to exchange information/data with other devices over a computer network such as the Internet and/or various telecommunication networks.

The processing unit 701 executes the various methods and processes described above. For example, in some embodiments, the methods may be implemented as a computer software program tangibly embodied on a machine-readable medium, such as storage unit 708 . In some embodiments, part or all of the computer program may be loaded and/or installed on the device 700 via the ROM 702 and/or the communication unit 709 . When a computer program is loaded into RAM 703 and executed by CPU 701, one or more steps of the methods described above may be performed. Alternatively, in other embodiments, the CPU 701 may be configured to execute the method in any other suitable manner (eg, by means of firmware).

The functions described herein above may be performed at least in part by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: field programmable gate array (FPGA), application specific integrated circuit (ASIC), application specific standard product (ASSP), system on a chip (SOC), load programmable logic device (CPLD), etc.

Program codes for implementing the methods of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special purpose computer, or other programmable data processing devices, so that the program codes, when executed by the processor or controller, make the functions/functions specified in the flow diagrams and/or block diagrams Action is implemented. The program code may execute entirely on the machine, partly on the machine, as a stand-alone software package partly on the machine and partly on a remote machine or entirely on the remote machine or server.

In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media would include one or more wire-based electrical connections, portable computer discs, hard drives, random access memory (RAM), read only memory (ROM), erasable programmable read only memory (EPROM or flash memory), optical fiber, compact disk read only memory (CD-ROM), optical storage, magnetic storage, or any suitable combination of the foregoing.

In addition, while operations are depicted in a particular order, this should be understood to require that such operations be performed in the particular order shown, or in sequential order, or that all illustrated operations should be performed to achieve desirable results. Under certain circumstances, multitasking and parallel processing may be advantageous. Likewise, while the above discussion contains several specific implementation details, these should not be construed as limitations on the scope of the disclosure. Certain features that are described in the context of separate embodiments can also be implemented in combination in a single implementation. Conversely, various features that are described in the context of a single implementation can also be implemented in multiple implementations separately or in any suitable subcombination.

Although the subject matter has been described in language specific to structural features and/or methodological acts, it is to be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or acts described above. Rather, the specific features and acts described above are merely example forms of implementing the claims.

Claims (5)

1. A bleeding risk assessment device for patients with coronary heart disease based on machine learning, characterized in that it comprises: The acquisition module is used to acquire the PPG measurement signal of the coronary heart disease patient in real time based on the preset acquisition time; A processing module, configured to remove baseline drift and high-frequency noise of the PPG measurement signal through a Butterworth band-pass filter; Based on the waveform of the PPG measurement signal, the PPG measurement signal is segmented to obtain M signal segments; M is a positive integer greater than or equal to 1; An evaluation module, configured to resample each of the M signal segments, resampling each segment in the set S to the same length, denoted as RS={RS ₁ ,RS ₂ ,..., RS _M }; wherein, the resampling length is the median of the M signal segment lengths; Substitute the resampled segmented signal into the following formula, if the calculation result is greater than the preset threshold, the current segmented signal is a qualified signal: S _valid ＝{S _i ∣S _i ∈S,RS _i ∈RS,r(RS _i ,SR _T )>0.9} Wherein, the _SRT is the average value of RS; The r is the Pearson correlation coefficient, the value range is -1～1; The scoring module is used to extract X features from the quality-qualified signal segments and input them into the bleeding risk assessment model to obtain the bleeding risk score of patients with coronary heart disease; X is a positive integer greater than or equal to 1. 2. The device according to claim 1, wherein the bleeding risk assessment model is trained through the following steps: Generate a training sample set, wherein the training sample includes a feature vector corresponding to the PPG signal with labeling information, wherein the labeling information is bleeding status, bleeding is marked as 1, and non-bleeding is marked as 0; Using the samples in the training sample set to train the bleeding risk assessment model, the feature vector corresponding to the PPG signal is used as input, and the bleeding condition is output, when the unity rate of the output bleeding condition and the marked bleeding condition satisfies a preset threshold , complete the training of the bleeding risk assessment model. 3. The device according to claim 2, wherein the eigenvector corresponding to the PPG signal comprises: The collected PPG signal is analyzed, and feature vectors corresponding to the PPG signal are extracted from time domain, frequency domain and wavelet packet decomposition respectively. 4. The device according to claim 3, wherein said analyzing the collected PPG signal comprises: The PPG signal, the first derivative of the PPG signal and the second derivative of the PPG signal are analyzed. 5 . The device according to claim 4 , wherein the bleeding risk assessment model is trained using an XGBoost algorithm.

Images