A digital signal processing-aided modeling method for biological signal analysis

Abstract

A method and apparatus for performing a biosignal analysis task using a set of models, comprising: receiving an input biosignal; receiving information identifying a biosignal analysis task to be performed in association with the input biosignal; selecting a waveform model and a digital signal processing (DSP) model; identifying first-type features and second-type features of the input biosignal; selecting an analysis model and performing the biosignal analysis task using that analysis model.

Description

[0001] Cross-references to related applications

[0002] This application claims priority to U.S. Patent Application No. 16 / 562,576, filed September 6, 2019, with the United States Patent and Trademark Office, the disclosure of which is incorporated herein by reference in its entirety. Background Technology

[0003] Electrocardiogram (ECG) signals are among the most common waveform biosignals that help doctors diagnose many heart diseases, including atrial fibrillation, myocardial infarction, and acute coronary syndrome (ACS). Approximately 300 million ECGs are recorded annually. Traditional methods for ECG analysis tend to use digital signal processing algorithms, such as wavelet transforms, to compute features from the ECG signal.

[0004] However, such methods are not comprehensive, and therefore, using only these methods is often insufficient to distinguish multiple types of arrhythmias. Recent methods employ deep neural networks, such as convolutional neural networks (CNNs) and recurrent neural networks (RNNs), and provide improved accuracy for multi-class classification tasks based on ECG signals. Summary of the Invention

[0005] According to some possible implementations, a method for performing a biosignal analysis task using a set of models includes: receiving an input biosignal; receiving identification information that identifies a biosignal analysis task to be performed in association with the input biosignal; selecting, based on the identification information identifying the biosignal analysis task to be performed, a waveform model from a set of waveform models for identifying a first type of feature of the input biosignal; selecting, based on the identification information identifying the biosignal analysis task to be performed, a DSP model from a set of digital signal processing (DSP) models for identifying a second type of feature of the input biosignal; using the waveform model to identify the first type of feature of the input biosignal; using the DSP model to identify the second type of feature of the input biosignal; selecting, based on the first type of feature and the second type of feature, an analysis model from a set of analysis models for performing the biosignal analysis task associated with the input biosignal; and using the analysis model to perform the biosignal analysis task based on the first type of feature and the second type of feature.

[0006] According to some possible implementations, an apparatus for performing a biosignal analysis task using a set of models includes: at least one memory configured to store program code; and at least one processor configured to read the program code and operate according to the instructions of the program code, the program code including: receiving code configured to cause the at least one processor to receive an input biosignal and receive identification information, the identification information identifying a biosignal analysis task to be performed in association with the input biosignal; and first selection code configured to cause the at least one processor to select, based on the identification information identifying the biosignal analysis task to be performed, a waveform model from a set of waveform models for identifying a first type of feature of the input biosignal, and a waveform model based on the identification information identifying the biosignal analysis task to be performed. The signal analysis task includes identification information, selecting a DSP model from a set of digital signal processing (DSP) models for identifying a second type of feature of the input biological signal; identification code, configured to cause at least one processor to identify a first type of feature of the input biological signal using a waveform model, and to identify a second type of feature of the input biological signal using a DSP model; a second selection code, configured to cause at least one processor to select an analysis model from a set of analysis models based on the first type of feature and the second type of feature for performing a biosignal analysis task associated with the input biological signal; and execution code, configured to cause at least one processor to perform the biosignal analysis task using the analysis model based on the first type of feature and the second type of feature.

[0007] According to some possible implementations, a non-transitory computer-readable medium stores instructions, the instructions comprising one or more instructions, which, when executed by one or more processors of a device, cause the one or more processors to: receive an input biological signal; receive identification information that identifies a biosignal analysis task to be performed in association with the input biological signal; based on the identification information identifying the biosignal analysis task to be performed, select from a set of waveform models a waveform model for identifying a first type of feature of the input biological signal; based on the identification information identifying the biosignal analysis task to be performed, select from a set of digital signal processing (DSP) models a DSP model for identifying a second type of feature of the input biological signal; use the waveform model to identify the first type of feature of the input biological signal; use the DSP model to identify the second type of feature of the input biological signal; based on the first type of feature and the second type of feature, select from a set of analysis models an analysis model for performing the biosignal analysis task associated with the input biological signal; and use the analysis model to perform the biosignal analysis task based on the first type of feature and the second type of feature. Attached Figure Description

[0008] Figure 1 This is a schematic diagram of the example implementation described in this article;

[0009] Figure 2 This is a diagram of an example environment in which the systems and / or methods described herein can be implemented;

[0010] Figure 3 yes Figure 2 A diagram of example components of one or more devices; and

[0011] Figure 4 This is a flowchart of an example process for performing a biosignal analysis task using a set of models. Detailed Implementation

[0012] The digital signal processing (DSP)-assisted modeling method disclosed herein is configured for waveform biological signal analysis. Biological signals are any signals generated by a living organism that can be continuously measured and detected. For example, biological signals can be measured by electrocardiography (ECG), electroencephalography (EEG), electromyography (EMG), actiography (MMG), electrooculography (EOG), geoskin conductance (GSR), and magnetoencephalography (MEG).

[0013] This disclosure allows physicians, researchers, and other healthcare practitioners to apply artificial intelligence (AI) algorithms in symptom and disease classification, computer-aided diagnosis, bedside alerts, and patient monitoring. The waveform biosignals used for modeling the process can originate from a single or multiple sources. For example, a standard dynamic ECG report contains signals from 12 different leads, requiring 10 electrodes in contact with the body. When using ECG signals for diagnosis, cardiologists and physicians can consider two levels of signal patterns. The first is the intracardiac pattern, which acquires the signal variation with a single heartbeat. The second is the intercardiac pattern, which measures the alternation of shapes between multiple heartbeats.

[0014] For decades, various types of traditional DSP algorithms have been applied to ECG diagnosis. These algorithms provide understandable statistical / mathematical features, but require domain-specific knowledge and costly, time-consuming DSP and feature engineering processes. This disclosure provides an artificial intelligence framework for ECG analysis that takes ECG signals and raw DSP features as input and provides possible analysis results and rationale for making decisions. For example, a comprehensive multi-lead ECG signal and corresponding DSP results can be used, and the final result of the framework can be a diagnosis (e.g., anxiety about premature atrial contractions) and suspicious heartbeats (e.g., 3, 6, etc.). Specifically, according to embodiments, this disclosure provides a DSP feature extraction module, a waveform feature extraction module, and a comprehensive task-specific analysis module.

[0015] While existing methods in cardiology offer promising results in terms of accuracy, several challenges and limitations remain. For example, machine learning and deep learning offer performance advantages; however, they lack the ability to help physicians and patients understand the rationale behind a diagnosis. Researchers and cardiologists may not be able to directly utilize the model to explain where, how, or why it made the final decision.

[0016] DSP features offer thousands of alternatives, and finding a balance between model complexity and performance is an exhausting process. Complex models theoretically offer better performance results because they can capture more variance in the feature space. However, complex models may also require additional computation time and introduce computational complexity.

[0017] In short, according to some methods, cardiologists may not be able to fully utilize the model in practice and may not be able to make comprehensive decisions based on the complexity of the model.

[0018] This disclosure allows for multiple data analysis tasks with various types of features. Features can be extracted using different feature extraction modules. According to an embodiment, a first type of feature can be waveform features embedded in a digital readout sequence of a biological signal. These features can help identify symptoms and diseases associated with signal changes, such as T-wave inversion, premature beats, sinus rhythm, etc.

[0019] According to the embodiment, the second type is DSP features that can be obtained by different types of signal processing algorithms (e.g., Fast Fourier Transform, Wavelet Transform, Filters, etc.). For the final analysis module, both types of extracted features are collected. Therefore, the proposed analysis framework can be widely applied to various types of analysis tasks.

[0020] As mentioned above, the framework disclosed herein comprises two main modules: 1) a set of feature extraction modules (e.g., waveform and DSP) that analyze input information and extract high-level data features. For each extraction module, an attention mechanism is optional to model hidden relationships and correlations in the data outcomes; 2) an analysis module that performs specific tasks such as clustering, classification, and prediction, and then provides the framework's ultimate goal.

[0021] Reference Figure 1The feature extraction module receives preprocessed data as input and generates feature vectors as output. For example, as shown in the figure, waveform biosignals 110 are provided to models from the extraction model pool 120. Specifically, waveform biosignals 110 are provided to DSP model 130 and waveform model 140. Regarding the models, the extraction module can be: 1) signal processing algorithms, such as filtering, fast Fourier transform, and wavelet transform for DSP features; and 2) machine learning methods, such as support vector machines (SVM), random forests (RF), or deep learning models such as CNNs and RNNs.

[0022] The parameters of each module can be trained and / or selected individually. According to embodiments, an attention mechanism can be optionally attached to each module to model dependencies and relationships between inputs and outputs, as well as among multiple outputs. For example, if an ECG signal is diagnosed as atrial premature beat anxiety, the likelihood that the same signal will not be simultaneously diagnosed as sinus rhythm is very high; if the signal exhibits an invariant pattern across all heartbeats, the probability of being diagnosed as a normal ECG and sinus rhythm is relatively high. This dependency can be modeled as an RNN model, Bayesian network, etc., which provides decision-making based on input and output information.

[0023] According to an embodiment, extracted features are collected from multiple modules with various types of input. For example, these features are post-processed using batch normalization, instance normalization, etc., to provide a final feature set for analysis.

[0024] like Figure 1 As shown, analysis model 150 is selected from task model pool 160. In this disclosure, the final module is a biosignal analysis module, which receives the extracted features and produces final results, such as classification results, outlier alerts, and predictive diagnoses. Figure 1 As shown, a task-specific module pool can be a collection of different models configured for various tasks related to biosignals. For example, a task-specific module pool may include several statistical process control algorithms for biosignal detection and alerting, several predictive and classifier models for computer-aided diagnosis, and some statistical tools for calculating common pathological states. Based on the desired output, the analysis module can deploy appropriate tools from the pool to complete the end-to-end framework and achieve the final goal.

[0025] The training framework proposed in this disclosure is designed as an end-to-end framework. Compared with existing methods for biosignal analysis models, this disclosure can extract features from multiple perspectives simultaneously. More specifically, as shown in the figure, this disclosure can extract waveform features and DSP features. In actual diagnosis, physicians should consider multi-perspective, heterogeneous, and even hierarchical structural features to make a comprehensive conclusion.

[0026] This disclosure also incorporates features from multiple inputs in different forms. Another advantage is that this disclosure helps researchers and physicians better understand the correlation between biosignals and analysis results (e.g., diagnosis). Furthermore, this disclosure can be extended to other architectures by applying optional dependency networks based on existing instances to add attention to different extraction modules.

[0027] According to some embodiments, the extraction module can use a combination of several algorithms and structures, such as a combination of RNN and CNN, or a combination of RNN and SVM. Due to the flexibility in defining "features" in machine learning, the precise implementation of the extraction module can vary.

[0028] The feature extraction approach offers flexibility across multiple perspectives. According to the embodiment, waveform and DSP categories are used. Alternatively, statistical features such as moving averages and autocorrelation are employed for feature extraction. Furthermore, the addition of attention is flexible, meaning that attention can be used or not used in certain modules, and the same or different attention can be applied to different modules depending on the requirements and objectives of feature extraction.

[0029] According to the embodiments, and for the feature extraction module, similar models are configured to share a subset of parameters to take into account the similarity between inputs.

[0030] Through this disclosure, the framework is designed as an end-to-end process, and the entire framework can be optimized and modified simultaneously. An alternative is a stepwise training process, in which the extraction module can be trained separately, for example, using encoder and decoder structures.

[0031] This approach can be extended to other applications with multiple input sources. For example, a model can use ECG and EEG as sources and extract features separately. The analysis model can either accept each feature set individually or combine the two feature sets.

[0032] Based on our chosen extraction module and analysis task, optional feedback mechanisms can be added from the output to the feature extraction module. For example, using an RNN model for multi-symptom diagnosis. To account for dependencies between symptom mentions, a feedback chain can be added from the diagnosed symptoms to the feature extraction module to tell the module how and where it should cluster in subsequent steps.

[0033] Figure 2 This is a diagram of an example environment 200 that can implement the systems and / or methods described herein. (See diagram 200 for example environment 200.) Figure 2 As shown, environment 200 may include user equipment 210, platform 220, and network 230. Devices in environment 200 can be interconnected via wired connections, wireless connections, or a combination of wired and wireless connections.

[0034] User equipment 210 includes one or more devices capable of receiving, generating, storing, processing, and / or providing information associated with platform 220. For example, user equipment 210 may include computing devices (e.g., desktop computers, laptop computers, tablet computers, handheld computers, smart speakers, servers, etc.), mobile phones (e.g., smartphones, cordless phones, etc.), wearable devices (e.g., a pair of smart glasses or a smartwatch), or similar devices. In some implementations, user equipment 210 may receive information from platform 220 and / or send information to platform 220.

[0035] Platform 220 includes one or more devices capable of performing biosignal analysis tasks using a set of models, as described elsewhere herein. In some implementations, platform 220 may include a cloud server or a set of cloud servers. In some implementations, platform 220 may be designed as a modular platform, allowing certain software components to be swapped in or out as needed. Therefore, platform 220 can be easily and / or quickly reconfigured for different applications.

[0036] In some implementations, as shown in the figure, platform 220 may be hosted in a cloud computing environment 222. It should be noted that although the implementations described herein depict platform 220 as hosted in a cloud computing environment 222, in some implementations, platform 220 may not be cloud-based (i.e., it may be implemented outside of a cloud computing environment) or may be partially cloud-based.

[0037] The cloud computing environment 222 includes the environment of the hosting platform 220. The cloud computing environment 222 can provide computing, software, data access, storage, and other services without requiring end users (e.g., user equipment 210) to know the physical location and configuration of the systems and / or devices of the hosting platform 220. As shown in the figure, the cloud computing environment 222 may include a group of computing resources 224 (this group of computing resources is collectively referred to as "computing resources 224", and a single computing resource is referred to as "computing resource 224").

[0038] Computing resource 224 includes one or more personal computers, workstations, server devices, or other types of computing and / or communication devices. In some implementations, computing resource 224 may control platform 220. Cloud resources may include computing instances running in computing resource 224, storage devices provided in computing resource 224, data transmission devices provided by computing resource 224, etc. In some implementations, computing resource 224 may communicate with other computing resources 224 via wired connections, wireless connections, or a combination of wired and wireless connections.

[0039] Further as Figure 2As shown, computing resources 224 include a set of cloud resources, such as one or more applications (“APP”) 224-1, one or more virtual machines (“VM”) 224-2, virtualized storage (“VS”) 224-3, one or more hypervisors (“HYP”) 224-4, etc.

[0040] Application 224-1 includes one or more software applications that can be provided to or accessed by user equipment 210 and / or sensor device 220. Application 224-1 eliminates the need to install and run software applications on user equipment 210. For example, application 224-1 may include software associated with platform 220 and / or any other software that can be provided through cloud computing environment 222. In some implementations, an application 224-1 may send information to / receive information from one or more other applications 224-1 via virtual machine 224-2.

[0041] Virtual machine 224-2 includes a software implementation of a machine (e.g., a computer) that runs programs, similar to a physical machine. Depending on the extent and purpose of its correspondence to any actual machine, virtual machine 224-2 can be a system virtual machine or a process virtual machine. A system virtual machine can provide a complete system platform supporting the operation of a full operating system (“OS”). A process virtual machine can run a single program and can support a single process. In some implementations, virtual machine 224-2 can run on behalf of a user (e.g., user device 210) and manage the infrastructure of the cloud computing environment 222, such as data management, synchronization, or long-duration data transfer.

[0042] Virtualized storage 224-3 includes one or more storage systems and / or one or more devices that utilize virtualization technology within the storage system or device of computing resource 224. In some implementations, the type of virtualization in the storage system environment may include block virtualization and file virtualization. Block virtualization may refer to the abstraction (or separation) of logical storage from physical storage, enabling access to the storage system regardless of physical storage or heterogeneous architecture. Separation allows storage system administrators flexibility in how they manage end-user storage. File virtualization eliminates the dependency between data accessed at the file level and the location where the file is physically stored. This allows for optimization of storage usage, server consolidation, and / or interference-free file migration performance.

[0043] Hypervisor 224-4 provides hardware virtualization technology, which allows multiple operating systems (e.g., "guest operating systems") to run simultaneously on a host computer, such as computing resource 224. Hypervisor 224-4 presents a virtual operating platform to the guest operating systems and manages their operation. Multiple instances of each operating system can share virtualized hardware resources.

[0044] Network 230 includes one or more wired and / or wireless networks. For example, network 230 may include cellular networks (e.g., fifth-generation (5G) networks, long-term evolution (LTE) networks, third-generation (3G) networks, code division multiple access (CDMA) networks, etc.), public land mobile networks (PLMN), local area networks (LAN), wide area networks (WAN), metropolitan area networks (MAN), telephone networks (e.g., public switched telephone network (PSTN)), private networks, self-organizing networks, intranets, the Internet, fiber-optic networks, etc., and / or combinations of these or other types of networks.

[0045] Figure 2 The number and arrangement of devices and networks shown are provided as examples. In practice, additional devices and / or networks, fewer devices and / or networks, different devices and / or networks, or networks may exist. Figure 2 The devices and / or networks shown are arranged differently. Furthermore, Figure 2 The two or more devices shown can be implemented within a single device, or Figure 2 The single device shown can be implemented as multiple distributed devices. Alternatively, a group of devices in environment 200 (e.g., one or more devices) can perform one or more functions described as being performed by another group of devices in environment 200.

[0046] Figure 3 This is a diagram of example components of device 300. Device 300 may correspond to user device 210 and / or platform 220. Figure 3 As shown, device 300 may include bus 310, processor 320, memory 330, storage component 340, input component 350, output component 360 and communication interface 370.

[0047] Bus 310 includes components that allow communication between components of device 300. Processor 320 is implemented in hardware, firmware, or a combination of hardware and software. Processor 320 is a central processing unit (CPU), graphics processing unit (GPU), accelerated processing unit (APU), microprocessor, microcontroller, digital signal processor (DSP), field-programmable gate array (FPGA), application-specific integrated circuit (ASIC), or another type of processing component. In some implementations, processor 320 includes one or more processors that can be programmed to perform functions. Memory 330 includes random access memory (RAM), read-only memory (ROM), and / or another type of dynamic or static storage device (e.g., flash memory, magnetic memory, and / or optical memory) that stores information and / or instructions for use by processor 320.

[0048] Storage component 340 stores information and / or software related to the operation and use of device 300. For example, storage component 340 may include hard disks (e.g., magnetic disks, optical disks, magneto-optical disks, and / or solid-state drives), optical discs (CDs), digital versatile discs (DVDs), floppy disks, cassette tapes, magnetic tapes, and / or other types of non-transitory computer-readable media, and corresponding drives.

[0049] Input component 350 includes components that allow device 300 to receive information, such as information input via user input (e.g., a touchscreen display, keyboard, keypad, mouse, button, switch, and / or microphone). Alternatively, input component 350 may include sensors for sensing information (e.g., a Global Positioning System (GPS) component, accelerometer, gyroscope, and / or actuator). Output component 360 includes components that provide output information from device 300 (e.g., a display, speaker, and / or one or more light-emitting diodes (LEDs)).

[0050] The communication interface 370 includes transceiver-like components (e.g., a transceiver and / or separate receiver and transmitter) that enable the device 300 to communicate with other devices, such as via a wired connection, a wireless connection, or a combination of wired and wireless connections. The communication interface 370 allows the device 300 to receive information from and / or provide information to another device. For example, the communication interface 370 may include an Ethernet interface, an optical interface, a coaxial interface, an infrared interface, a radio frequency (RF) interface, a universal serial bus (USB) interface, a Wi-Fi interface, a cellular network interface, etc.

[0051] Device 300 can perform one or more processes described herein. Device 300 can perform these processes in response to processor 320 executing software instructions stored in a non-transitory computer-readable medium such as memory 330 and / or storage component 340. In this document, a computer-readable medium is defined as a non-transitory memory device. A memory device includes memory space within a single physical storage device or memory space distributed across multiple physical storage devices.

[0052] Software instructions can be read into memory 330 and / or storage component 340 from another computer-readable medium via communication interface 370, or from another device. When executed, the software instructions stored in memory 330 and / or storage component 340 cause processor 320 to perform one or more processes described herein. Alternatively, hardwired circuitry may be used in place of or in combination with software instructions to perform one or more processes described herein. Therefore, the implementations described herein are not limited to any particular combination of hardware circuitry and software.

[0053] Figure 3 The number and arrangement of components shown are provided as examples. In practice, device 300 may include additional components, fewer components, different components, or components with... Figure 3 The components shown are arranged differently. Alternatively, a group of components of device 300 (e.g., one or more components) may perform one or more functions described as being performed by another group of components of device 300.

[0054] Figure 4 This is a flowchart of an example process 400 for performing a biosignal analysis task using a set of models. In some implementations, this can be performed by platform 220. Figure 4 One or more process blocks. In some implementations, these blocks may be executed by another device or group of devices (e.g., user equipment 210) that is separate from or includes platform 220. Figure 4 One or more process blocks.

[0055] like Figure 4 As shown, process 400 may include receiving an input biological signal (block 410). For example, platform 220 may receive the input biological signal from another device, memory, etc.

[0056] Input biological signals can include electrocardiogram (ECG) signals, electroencephalogram (EEG) signals, electromyogram (EMG) signals, myocardiogram (MMG) signals, electrooculogram (EOG) signals, skin conductance response (GSR) signals, magnetoencephalogram (MEG) signals, etc.

[0057] In some implementations, platform 220 can receive a single type of biological signal and perform... Figure 4 Alternatively, platform 220 can receive various types of biological signals (e.g., ECG signals and EEG signals, etc.) and perform [operations]. Figure 4 The operation.

[0058] Further as Figure 4 As shown, process 400 may include receiving identification information that identifies a biosignal analysis task to be performed in association with the input biosignal (box 420). For example, platform 220 may receive information identifying a specific type of biosignal task to be performed using the input biosignal. Platform 220 may receive the information from another device, memory, etc.

[0059] Biosignal analysis tasks may include determining diagnoses of conditions associated with input biosignals, clustering input biosignals, classifying input biosignals, predicting diseases associated with input biosignals, biosignal detection, biosignal alerts, patient detection, etc.

[0060] Further as Figure 4 As shown, process 400 may include selecting a waveform model from a set of waveform models (box 430) for identifying a first type of feature of the input biological signal based on information identifying the biological signal analysis task to be performed.

[0061] In some implementations, platform 220 may access a waveform model pool. For example, a waveform model pool may refer to a set of models that can be used to perform waveform analysis of biological signals. In this case, platform 220 may select a specific waveform model from the waveform model pool to identify a first type of feature of the input biological signal.

[0062] The first type of feature can include any type of feature of the input biological signal that can be detected, analyzed, etc. As an example, the first type of feature can correspond to the waveform of the input biological signal, the amplitude of the input biological signal, the frequency of the input biological signal, etc.

[0063] Platform 220 can identify waveform models based on a selected biosignal analysis task. That is, platform 220 can store information mapping waveform models to biosignal analysis tasks in a data structure. In other words, platform 220 can identify a specific waveform model used to perform a biosignal analysis task based on the selected biosignal analysis task.

[0064] Further as Figure 4 As shown, process 400 may include selecting a DSP model from a set of digital signal processing (DSP) models for identifying a second type of feature of the input biological signal based on information identifying the biological signal analysis task to be performed (box 440).

[0065] In some implementations, platform 220 may access a DSP model pool. For example, a DSP model pool may refer to a set of models that can be used to perform DSP analysis of biological signals. In this case, platform 220 may select a specific DSP model from the DSP model pool to identify a second type of feature of the input biological signal.

[0066] The second type of feature can include any type of feature of the input biological signal that can be detected, analyzed, etc. As an example, the first type of feature can correspond to the waveform, amplitude, frequency, etc. of the input biological signal.

[0067] Platform 220 can identify DSP models based on a selected biosignal analysis task. That is, platform 220 can store information mapping DSP models to biosignal analysis tasks in a data structure. In other words, platform 220 can identify a specific DSP model used to perform a biosignal analysis task based on the selected biosignal analysis task.

[0068] Alternatively, platform 220 may store data structures mapping DSP models and waveform models used to perform specific biosignal analysis tasks. In this way, platform 220 can identify the DSP model and waveform model used to perform the selected biosignal analysis task.

[0069] Further as Figure 4 As shown, process 400 may include using a waveform model to identify a first type of feature of the input biological signal (box 450).

[0070] For example, platform 220 can input an input biological signal into a waveform model and identify a first type of feature based on the output of the waveform model. In some implementations, the waveform model may include a recurrent neural network (RNN), a convolutional neural network (CNN), a support vector machine (SVM), etc. The waveform model can analyze the input biological signal and generate an output corresponding to the first type of feature of the input biological signal.

[0071] Further as Figure 4 As shown, process 400 may include using a DSP model to identify a second type of feature of the input biological signal (box 460).

[0072] For example, platform 220 can input biological signals into a DSP model and identify second-type features based on the output of the DSP model. The DSP model can be configured to perform filtering techniques, fast Fourier transform techniques, wavelet transform techniques, etc. The DSP model can analyze the input biological signal and generate an output corresponding to the second-type features of the input biological signal.

[0073] Further as Figure 4As shown, process 400 may include selecting an analysis model from a set of analysis models based on a first type of feature and a second type of feature for performing a biosignal analysis task associated with the input biosignal (box 470).

[0074] In some implementations, platform 220 may access an analysis model pool. For example, the analysis model pool may refer to a set of models that can be used to perform biosignal analysis tasks. In this case, platform 220 may select a specific analysis model from the analysis model pool to perform the selected biosignal analysis task.

[0075] Platform 220 can identify waveform models based on a selected biosignal analysis task. That is, platform 220 can store information about the mapping analysis model, the biosignal analysis task, the waveform model, and the DSP model in a data structure. In other words, platform 220 can identify a specific waveform model used to perform a biosignal analysis task based on a selected biosignal analysis task, waveform model, and / or DSP model.

[0076] Further as Figure 4 As shown, process 400 may include performing a biosignal analysis task using an analysis model based on a first type of feature and a second type of feature (box 480).

[0077] Although Figure 4 Example blocks of process 400 are shown, but in some implementations, process 400 may include additional blocks, fewer blocks, different blocks, or blocks similar to... Figure 4 The blocks depicted are arranged differently. Alternatively, two or more blocks of process 400 can be executed in parallel.

[0078] The foregoing disclosure provides explanations and descriptions, but is not intended to be exhaustive or to limit the implementation to the precise forms disclosed. Modifications and variations may be made based on the foregoing disclosure, or from practical experience with the implementation.

[0079] As used in this article, the term "component" is intended to be interpreted broadly as hardware, firmware, or a combination of hardware and software.

[0080] Clearly, the systems and / or methods described herein can be implemented in various forms of hardware, firmware, or a combination of hardware and software. The actual dedicated control hardware or software code used to implement these systems and / or methods is not limiting in its implementation. Therefore, no specific software code is referenced in this document to describe the operation and behavior of the systems and / or methods—it should be understood that software and hardware can be designed to implement the systems and / or methods described herein.

[0081] Even with specific combinations of features recited in the claims and / or disclosed in the specification, these combinations are not intended to limit the disclosure of possible implementations. In fact, many of these features can be combined in ways not specifically recited in the claims and / or disclosed in the specification. While each dependent claim listed below may be directly dependent on only one claim, the disclosure of possible implementations includes combinations of each dependent claim with every other claim in the claim set.

[0082] Unless explicitly stated otherwise, elements, actions, or instructions used herein should not be construed as essential or necessary. Furthermore, as used herein, the articles “a” and “an” are intended to include one or more items and are used interchangeably with “one or more.” Additionally, as used herein, the term “set” is intended to include one or more items (e.g., related items, unrelated items, a combination of related and unrelated items, etc.) and is used interchangeably with “one or more.” The term “a” or similar language is used where the intent is to refer to only one item. Furthermore, as used herein, the terms “having,” “possessing,” “containing,” or similar terms are intended to be open-ended terms. Additionally, the phrase “based on” is intended to mean “at least partially based on” unless otherwise explicitly stated.

Claims

1. A method for performing a biosignal analysis task using a set of models, the method being executed by at least one processor and comprising: Receive input biological signals, which are electrocardiogram (ECG) signals, electroencephalogram (EEG) signals, electromyogram (EMG) signals, motor myograph (MMG) signals, electrooculogram (EOG) signals, skin conductance response (GSR) signals, and magnetoencephalogram (MEG) signals. Receive identification information, the identification information being associated with the input biological signal to perform a biosignal analysis task, the biosignal analysis task including at least one of the following: determining a diagnosis of a condition associated with the input biological signal, clustering the input biological signal, classifying the input biological signal, or predicting a disease associated with the input biological signal; Based on the identification information used to identify the biological signal analysis task to be performed, a waveform model for identifying the first type of feature of the input biological signal is selected from a set of waveform models; Based on the identification information used to identify the biosignal analysis task to be performed, a DSP model for identifying the second type of features of the input biosignal is selected from a set of digital signal processing (DSP) models. The waveform model is used to identify the first type of feature of the input biological signal, the first type of feature including the waveform of the input biological signal, the amplitude of the input biological signal, and the frequency of the input biological signal; The second type of feature of the input biological signal is identified using the DSP model; Based on the first type of features and the second type of features, an analysis model is selected from a set of analysis models to perform a biosignal analysis task associated with the input biosignal. The set of analysis models includes multiple statistical process control algorithms for biosignal detection and alarm, multiple prediction models and classifier models for computer-aided diagnosis, and multiple statistical tools for calculating common pathological states. as well as The biosignal analysis task is performed using the analysis model based on the first type of features and the second type of features.

2. The method according to claim 1, wherein, The DSP model is configured to perform at least one of the following techniques: filtering, fast Fourier transform, and wavelet transform.

3. The method according to claim 1, wherein, The biosignal analysis task includes at least one of biosignal detection, biosignal alarm, and patient detection.

4. The method according to claim 1, wherein, The waveform model includes at least one of recurrent neural network (RNN), convolutional neural network (CNN), and support vector machine (SVM).

5. The method of claim 1, further comprising: Receive another type of biological signal; as well as The biosignal analysis task is performed using the other type of biosignal.

6. An apparatus for performing a biosignal analysis task using a set of models, comprising: At least one memory configured to store program code; At least one processor is configured to read the program code and operate according to the instructions of the program code to perform the method according to any one of claims 1 to 5.

7. A computer-readable medium storing instructions, the instructions comprising one or more instructions that, when executed by one or more processors of a device using a set of models to perform a biosignal analysis task, cause the one or more processors to perform the method according to any one of claims 1 to 5.

8. An apparatus for performing a biosignal analysis task using a set of models, comprising: The first receiving module is used to receive input biological signals, which are electrocardiogram (ECG) signals, electroencephalogram (EEG) signals, electromyogram (EMG) signals, motor myograph (MMG) signals, electrooculogram (EOG) signals, skin conductance response (GSR) signals, and magnetoencephalogram (MEG) signals. The second receiving module is used to receive identification information, which is associated with the input biological signal to perform a biological signal analysis task, the biological signal analysis task including at least one of the following: determining a diagnosis of a condition associated with the input biological signal, clustering the input biological signal, classifying the input biological signal, or predicting a disease associated with the input biological signal; The first selection module is used to select a waveform model from a set of waveform models for identifying a first type of feature of the input biological signal based on the identification information for identifying the biological signal analysis task to be performed; The second selection module is used to select a DSP model from a set of digital signal processing (DSP) models for recognizing the second type of features of the input biological signal based on the recognition information used to identify the biological signal analysis task to be performed. A first identification module is used to identify the first type of feature of the input biological signal using the waveform model, wherein the first type of feature includes the waveform of the input biological signal, the amplitude of the input biological signal, and the frequency of the input biological signal; The second identification module is used to identify the second type of feature of the input biological signal using the DSP model; The third selection module is used to select an analysis model from a set of analysis models based on the first type of features and the second type of features, for performing a biosignal analysis task associated with the input biosignal. The set of analysis models includes multiple statistical process control algorithms for biosignal detection and alarm, multiple prediction models and classifier models for computer-aided diagnosis, and multiple statistical tools for calculating common pathological states. as well as An analysis module is used to perform the biosignal analysis task based on the first type of features and the second type of features, using the analysis model.

Images