Single molecule state monitoring methods, devices, storage media, and products

Abstract

The application discloses a single-molecule state monitoring method and device, a storage medium and a product, relates to the technical field of electrochemical detection, and comprises the following steps: through a signal acquisition unit, first signals and second signals are collected in real time in the process that a single molecule changes state, and time-varying data signals are obtained; through an intelligent processing unit, the time-varying data signals are preprocessed, and standardized data is generated; through feature fusion processing on the standardized data, transient features in the first signals and time sequence correlation features of the first signals and the second signals are extracted, and a time sequence feature vector is generated; and according to a dynamic recognition model, the time sequence feature vector is classified, and the state type in which the single molecule is currently located is determined. The application realizes real-time and accurate recognition of the state change process of a single molecule in a complex microenvironment background by synchronously collecting two types of signals and extracting time sequence correlation features of the two types of signals for fusion analysis, and the anti-interference capability of monitoring and the accuracy of state recognition are improved.

Description

Technical Field

[0001] This application relates to the field of electrochemical monitoring technology, and in particular to a method, device, storage medium and product for single-molecule state monitoring. Background Technology

[0002] Single-molecule state monitoring refers to the technology of real-time observation and analysis of the dynamic changes in the configuration and connectivity of a single molecule. It has significant scientific value in fundamental research fields such as elucidating the mechanisms of biomolecular action. Since the state changes of a single molecule are essentially a dynamic process evolving over time, accurate monitoring requires the simultaneous acquisition of signals reflecting both the molecule's own state and the conditions of its microenvironment. Furthermore, the correlation between these two types of signals over time can be used to analyze the patterns of molecular state changes. Currently, signal acquisition technologies based on various physical principles, including optics, electricity, and mechanics, have been developed in this field. These technologies can detect the state of a single molecule or its environmental conditions, generating time-varying signal data. Building upon this foundation, techniques utilizing machine learning or deep learning models for feature extraction and state classification of the acquired signal data have emerged. Some of these techniques can identify specific molecular state patterns from single-type time-series signals.

[0003] However, in traditional techniques, signals reflecting the molecular state and signals reflecting environmental conditions are usually acquired as independent detection objects. This leads to subsequent feature extraction and state recognition being based solely on a single type of time-series signal. When a molecular state undergoes an instantaneous change, its characteristics in a single signal dimension are often extremely weak and easily overlapped with signal disturbances caused by fluctuations in environmental parameters. It is difficult to accurately distinguish between the actual state change and external interference based on a single signal, thus making it impossible to achieve real-time and accurate identification of single-molecule state change processes. Summary of the Invention

[0004] The main objective of this application is to provide a single-molecule state monitoring method, which aims to solve the technical problem that traditional single-molecule monitoring technology is difficult to accurately identify molecular states.

[0005] To achieve the above objectives, this application proposes a single-molecule state monitoring method applied to a state monitoring system. The state monitoring system includes a signal acquisition unit for acquiring a first signal reflecting the state of the single molecule itself and a second signal reflecting the environmental conditions of the single molecule, and an intelligent processing unit electrically connected to the signal acquisition unit. The method includes: The signal acquisition unit acquires the first signal and the second signal in real time during the state change of the single molecule to obtain time-varying data signals. The intelligent processing unit preprocesses the time-varying data signal to generate standardized data. By performing feature fusion processing on the standardized data, transient features in the first signal and temporal correlation features between the first signal and the second signal are extracted to generate a temporal feature vector. Based on the dynamic recognition model, the temporal feature vector is classified to determine the current state type of the single molecule.

[0006] In one embodiment, the intelligent processing unit includes a preprocessing model, a feature fusion processing model, and a dynamic recognition model; Before the step of acquiring the first signal and the second signal in real time to obtain the time-varying data signal, the method further includes: Construct a training dataset, wherein the training dataset includes positive samples and negative samples, the positive samples include signal data of the single-molecule state change process and corresponding environmental condition data, and the negative samples include signal data of background noise; Based on the training dataset, the preprocessing model, the feature fusion processing model, and the dynamic recognition model are iteratively trained until each model meets the corresponding preset model training performance index.

[0007] In one embodiment, the step of acquiring the first signal and the second signal in real time during the state change of the single molecule through the signal acquisition unit to obtain time-varying data signals includes: The first signal is acquired at a preset sampling rate; At the same time as the acquisition of the first signal, the second signal is acquired synchronously; The first signal and the second signal are time-aligned respectively to obtain the time-varying data signal, so that the data at each time point in the time-varying data signal includes the state information of the single molecule and the corresponding environmental condition information.

[0008] In one embodiment, the step of preprocessing the time-varying data signal using the intelligent processing unit to generate standardized data includes: The time-varying data signal is denoised using the preprocessing model in the intelligent processing unit. The time-varying data signal after noise reduction is subjected to pseudo signal removal processing to identify and remove pseudo signals, wherein the pseudo signals include at least one of fluorescent dye blinking, photobleaching and electrode interference. Normalization processing is performed on the time-varying data signal after pseudo-signal removal to adjust the signal components of different magnitudes in the first signal and the second signal to a unified numerical range, thereby obtaining the standardized data.

[0009] In one embodiment, the step of extracting transient features from the first signal and temporal correlation features between the first signal and the second signal by performing feature fusion processing on the standardized data to generate a temporal feature vector includes: The standardized data is input into the feature fusion processing model, which is composed of a convolutional neural network and a long short-term memory network; The first signal in the standardized data is extracted by the convolutional neural network to obtain transient features characterizing the peak value and waveform shape of the first signal. Through the Long Short-Term Memory Network, temporal correlation features are extracted from the first signal and the second signal in the standardized data to obtain continuous change features that characterize the rate of change and duration of change of the first signal, as well as the temporal correspondence between the change trend of the first signal and the change trend of the second signal. The transient features and the continuous change features are fused to generate the temporal feature vector.

[0010] In one embodiment, the step of classifying the temporal feature vector based on a dynamic recognition model to determine the current state type of the single molecule includes: The temporal feature vector is input into the dynamic recognition model; Self-attention weighting is applied to the dependencies between time steps in the time-series feature vector to generate the probability distribution of the single molecule in each preset state type at each time step. Based on the probability distribution, the preset state type with the highest probability is determined as the state type of the single molecule at the corresponding time step; Output the current state type of the single molecule, wherein the preset state type includes at least one of stable configuration state, configurational transition state, and molecular breakage state.

[0011] In one embodiment, after the step of classifying the temporal feature vector according to the dynamic recognition model to determine the current state type of the single molecule, the method further includes: A comprehensive analysis is performed on the time-series feature vector and the environmental condition data corresponding to the second signal to determine whether the single molecule will undergo an unexpected state change event, wherein the unexpected state change event includes unexpected molecule breakage or abnormal configuration change. When the probability of the occurrence of the unexpected state change event reaches a preset warning threshold, a warning signal is generated, and the time point that triggers the warning, the corresponding environmental condition data, and the feature fragment of the time-series feature vector are marked.

[0012] In addition, to achieve the above objectives, this application also proposes an electronic device, the device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the single-molecule state monitoring method as described above.

[0013] In addition, to achieve the above objectives, this application also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the steps of the single-molecule state monitoring method described above.

[0014] In addition, to achieve the above objectives, this application also provides a computer program product, which includes a computer program that, when executed by a processor, implements the steps of the single-molecule state monitoring method described above.

[0015] One or more technical solutions proposed in this application have at least the following technical effects: This application uses the signal acquisition unit to acquire the first signal and the second signal in real time during the state change of the single molecule to obtain time-varying data signals; the intelligent processing unit preprocesses the time-varying data signals to generate standardized data; by performing feature fusion processing on the standardized data, transient features in the first signal and the temporal correlation features between the first signal and the second signal are extracted to generate a temporal feature vector; based on a dynamic recognition model, the temporal feature vector is classified to determine the current state type of the single molecule. In this embodiment, by synchronously and in real-time acquiring a first signal reflecting the state of a single molecule and a second signal reflecting its environmental conditions, and sequentially performing preprocessing, feature fusion processing, and dynamic identification model classification on the acquired time-varying data signals, the final time-series feature vector used for state discrimination not only contains the transient change information of the first signal itself, but also integrates the correlation change law between the first and second signals in the time dimension. This solves the problem in traditional technology that weak instantaneous state changes in a single signal are difficult to effectively distinguish from micro-environmental disturbances, and realizes real-time and accurate identification of the state change process of a single molecule in a complex micro-environment, significantly improving the anti-interference capability and identification accuracy of state monitoring. Attached Figure Description

[0016] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0017] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating an embodiment of the single-molecule state monitoring method of this application. Figure 2 This is a schematic diagram of the internal model structure of the intelligent processing unit provided in Embodiment 1 of the single-molecule state monitoring method of this application; Figure 3 This is a flowchart illustrating Embodiment 2 of the single-molecule state monitoring method of this application. Figure 4 This is a simplified flowchart of the single-molecule state monitoring method of this application; Figure 5 This is a schematic diagram of the device structure of the hardware operating environment involved in the single-molecule state monitoring method in the embodiments of this application.

[0019] Explanation of icon numbers: 1-Preprocessing algorithm module, 2-Feature extraction model, 3-Dynamic recognition model, 4-Early warning model.

[0020] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0021] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0022] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0023] In this embodiment, for ease of description, the following description uses an electronic device as the execution subject.

[0024] Single-molecule state monitoring refers to the technology of real-time observation and analysis of the dynamic changes in the configuration and connectivity of a single molecule. It has significant scientific value in fundamental research fields such as elucidating the mechanisms of biomolecular action. Since the state changes of a single molecule are essentially a dynamic process evolving over time, accurate monitoring requires the simultaneous acquisition of signals reflecting both the molecule's own state and the conditions of its microenvironment. Furthermore, the correlation between these two types of signals over time can be used to analyze the patterns of molecular state changes. Currently, signal acquisition technologies based on various physical principles, including optics, electricity, and mechanics, have been developed in this field. These technologies can detect the state of a single molecule or its environmental conditions, generating time-varying signal data. Building upon this foundation, techniques utilizing machine learning or deep learning models for feature extraction and state classification of the acquired signal data have emerged. Some of these techniques can identify specific molecular state patterns from single-type time-series signals.

[0025] However, in traditional techniques, signals reflecting the molecular state and signals reflecting environmental conditions are usually acquired as independent detection objects. This leads to subsequent feature extraction and state recognition being based solely on a single type of time-series signal. When a molecular state undergoes an instantaneous change, its characteristics in a single signal dimension are often extremely weak and easily overlapped with signal disturbances caused by fluctuations in environmental parameters. It is difficult to accurately distinguish between the actual state change and external interference based on a single signal, thus making it impossible to achieve real-time and accurate identification of single-molecule state change processes.

[0026] This application uses the signal acquisition unit to acquire a first signal and a second signal in real time during the state change of a single molecule to obtain time-varying data signals; the intelligent processing unit preprocesses the time-varying data signals to generate standardized data; the standardized data is then subjected to feature fusion processing to extract transient features from the first signal and temporal correlation features between the first and second signals to generate a temporal feature vector; and the temporal feature vector is classified according to a dynamic recognition model to determine the current state type of the single molecule. In this embodiment, by synchronously and in real-time acquiring a first signal reflecting the state of a single molecule and a second signal reflecting its environmental conditions, and sequentially performing preprocessing, feature fusion processing, and dynamic identification model classification on the acquired time-varying data signals, the final time-series feature vector used for state discrimination not only contains the transient change information of the first signal itself, but also integrates the correlation change law between the first and second signals in the time dimension. This solves the problem in traditional technology that weak instantaneous state changes in a single signal are difficult to effectively distinguish from micro-environmental disturbances, and realizes real-time and accurate identification of the state change process of a single molecule in a complex micro-environment, significantly improving the anti-interference capability and identification accuracy of state monitoring.

[0027] It should be noted that the executing entity in this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone, or an electronic device capable of performing the above functions. The following description uses an electronic device as an example to illustrate this embodiment and the subsequent embodiments.

[0028] Based on this, embodiments of this application provide a method for single-molecule state monitoring, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the single-molecule state monitoring method of this application.

[0029] In this embodiment, the single-molecule state monitoring method is applied to a state monitoring system. The state monitoring system includes a signal acquisition unit for acquiring a first signal reflecting the state of the single molecule itself and a second signal reflecting the environmental conditions of the single molecule, as well as an intelligent processing unit electrically connected to the signal acquisition unit. Please refer to [reference needed]. Figure 2 , Figure 2 This is a schematic diagram of the internal model structure of the intelligent processing unit provided in Embodiment 1 of the single-molecule state monitoring method of this application. The intelligent processing unit has a pre-processing algorithm module 1, a feature extraction model 2, a dynamic recognition model 3, and an early warning model 4. The pre-processing algorithm module 1 can employ an adaptive denoising algorithm to perform denoising processing on the original time-varying data signal. The feature extraction model 2 can employ a fusion model of convolutional neural network and long short-term memory network to extract transient features and temporal correlation features from standardized data. The dynamic recognition model 3 can employ a model combining a recurrent neural network based on a self-attention mechanism and a Transformer architecture to classify temporal feature vectors and determine the current state type of the single molecule. The early warning model 4 can employ a support vector machine classification model to comprehensively judge whether an unexpected state change event will occur by combining temporal feature vectors and environmental condition data.

[0030] The single-molecule state monitoring method includes steps S10 to S40: Step S10: During the process of state change of a single molecule, the first signal and the second signal are acquired in real time through the signal acquisition unit to obtain time-varying data signals. It should be noted that in the field of single-molecule state monitoring, the configurational transitions and molecular breakage of a single molecule are essentially dynamic processes that evolve over time. Traditional techniques typically acquire signals reflecting the molecule's own state and signals reflecting its microenvironment as independent detection objects. These two types of signals lack a synchronous correspondence in the time dimension, leading to subsequent analysis being based on only one type of signal and making it difficult to effectively distinguish between real molecular state changes and microenvironment interference signals. Therefore, this application uses a signal acquisition unit to acquire a first signal and a second signal in real time during the state change of a single molecule, obtaining time-varying data signals. The first signal refers to a signal that can directly or indirectly reflect changes in the configurational state or connection state of the single molecule itself. When using optical monitoring, the first signal can be a fluorescence intensity signal, fluorescence lifetime signal, or energy transfer efficiency signal. When using electrical monitoring, the first signal can be a current intensity signal or current change rate signal acquired through a nanopore detection device or a single-molecule junction detection device. The second signal refers to a signal that reflects the physical or chemical microenvironmental conditions of a single molecule. For example, it can be a temperature signal, pH value signal, ion concentration signal, or a mechanical force signal applied to the single molecule. The signal acquisition unit is a combination of hardware devices that simultaneously acquire the above two types of signals. Its specific configuration can be flexibly configured according to actual monitoring needs. For example, it may include at least one of an optical signal acquisition module, an electrical signal acquisition module, and a microenvironment parameter sensing module. The specific model of each module can be selected according to the requirements of monitoring accuracy and sampling frequency. The first and second signals are acquired synchronously at the same acquisition time. The resulting time-varying data signal is a multi-dimensional data set containing a time dimension generated by the synchronous acquisition process. The data at each time point in this time-varying data signal includes the values ​​of the first and second signals acquired at that time. The two are strictly aligned on the time axis, forming a multi-dimensional data structure with a temporal correspondence.

[0031] Understandably, by simultaneously acquiring the first and second signals in real time during the state change of a single molecule through the signal acquisition unit, the resulting time-varying data signal carries corresponding molecular state information and environmental condition information at each time point. This multi-dimensional data structure provides a data foundation that is not available in existing technologies using single-type signals, enabling subsequent steps to determine the response relationship between single-molecule state changes and their environmental conditions using the temporal correlation characteristics between the first and second signals. This solves the problem of traditional monitoring methods struggling to accurately distinguish between real state changes and microenvironmental interference due to the lack of synchronous correspondence between the two types of signals, and improves the dimensional richness and temporal correlation of the original acquired data.

[0032] In one feasible implementation, the intelligent processing unit includes a preprocessing model, a feature fusion processing model, and a dynamic recognition model; Before the step of acquiring the first and second signals in real time to obtain the time-varying data signal, the following steps are also included: Construct a training dataset, which includes positive samples and negative samples. Positive samples include signal data of single-molecule state change processes and corresponding environmental condition data, while negative samples include signal data of background noise. Based on the training dataset, the preprocessing model, feature fusion processing model, and dynamic recognition model are trained iteratively until each model meets the corresponding preset model training performance indicators.

[0033] It should be noted that in practical single-molecule state monitoring applications, the performance of the preprocessing model, feature fusion model, and dynamic recognition model directly affects the accuracy of the final state recognition result. If a general-purpose model without specific training is used, its recognition accuracy often fails to meet the monitoring requirements in complex microenvironments. On the other hand, training models by collecting large amounts of real-world labeled data for specific monitoring scenarios faces practical difficulties such as the high cost of acquiring single-molecule experimental data and high labeling costs. Therefore, the time-consuming model training process is separated from the real-time monitoring process, with training and optimization of each model completed before formal monitoring begins. The preprocessing model refers to the computational model used to perform preprocessing operations such as denoising, pseudo-signal removal, and normalization on the raw time-varying data signal. It may contain at least one of a denoising algorithm module, a pseudo-signal recognition module, and a normalization module. The feature fusion model refers to the computational model used to extract and fuse transient features and time-series correlation features from standardized data. It may contain a convolutional neural network module and a long short-term memory network module. A dynamic recognition model is a computational model used to classify and determine state types based on input temporal feature vectors. Internally, it may include a self-attention mechanism module and a classification layer module. The training dataset is a collection of data used for offline training of the aforementioned models. The training dataset includes two types of sample data: positive samples and negative samples. Positive samples refer to signal data containing actual state change processes of single molecules and corresponding environmental condition data. These data carry clear state change labels, indicating the type of state change and the time point of occurrence. Negative samples refer to signal data dominated by background noise and lacking effective state change information, used to train the model to learn to distinguish between effective signals and noise interference. Iterative training involves inputting sample data from the training dataset into each model in batches, calculating the model output through forward propagation, comparing the model output with the labeled information to calculate the loss value, and then using the backpropagation algorithm to backpropagate the gradient of the loss value with respect to each model parameter and update the model parameters. This process is repeated until the model converges or reaches a preset performance index. The preset model training performance index for the preprocessed model can be the improvement in the signal-to-noise ratio after denoising or the accuracy of pseudo-signal removal. The preset model training performance metrics for feature fusion processing models can be the completeness of feature extraction or the discriminative power of feature vectors. The preset model training performance metrics for dynamic recognition models can be the accuracy of state type recognition or the recognition latency.

[0034] Understandably, through the aforementioned offline training steps, the preprocessing model, feature fusion model, and dynamic recognition model in the intelligent processing unit can obtain targeted parameter configurations before formal monitoring begins, avoiding insufficient monitoring accuracy due to inadequate model training. Simultaneously, separating the time-consuming model training process from the real-time monitoring process allows the trained model to directly perform forward inference during the real-time monitoring phase without repeated parameter updates, ensuring the processing efficiency of the real-time monitoring stage. Furthermore, constructing the training dataset by combining computer-simulated labeled data with supplementary real data effectively alleviates the constraint of scarce real labeled data in the single-molecule domain on model training, reduces the reliance on extensive manual annotation, and improves the model's generalization ability and applicability under different types of single molecules and experimental conditions.

[0035] In one feasible implementation, the step of acquiring a first signal and a second signal in real time during the state change of a single molecule using a signal acquisition unit to obtain a time-varying data signal includes: Acquire the first signal at a preset sampling rate; At the same time as the acquisition of the first signal, the second signal is acquired synchronously. The first and second signals are time-aligned to obtain time-varying data signals, so that the data at each time point in the time-varying data signals includes the state information of a single molecule and the corresponding environmental condition information.

[0036] It should be noted that in the signal acquisition stage, the sampling rate and the timing alignment accuracy between the two types of signals directly affect the accuracy of subsequent state identification. A sampling rate that is too low will result in the loss of instantaneous state change signals of a single molecule, while a lack of strict timing alignment between the two types of signals will lead to unreliable basic data upon which subsequent temporal correlation feature extraction depends. The preset sampling rate refers to the sampling frequency pre-set based on the dynamic characteristics of the monitored object and the minimum time resolution required to capture the signal. The acquisition of the first signal is performed by the corresponding sensor module in the signal acquisition unit. The sensor converts the physical quantity into an electrical or digital signal that can be processed subsequently, forming a sequence of data points on a discrete time series. Each data point represents the state information of a single molecule at the corresponding sampling time. Synchronous acquisition means that the sampling time of the second signal is consistent with the sampling time of the first signal; that is, under the control of the same clock signal or trigger signal, the first and second signals are sampled simultaneously. The second signal is a microenvironmental parameter signal of a different type from the first signal, and may include at least one of the following: temperature signal, pH value signal, ion concentration signal, or mechanical force signal applied to a single molecule. Time alignment processing refers to the operation of pairing and binding the first and second signal sample values ​​obtained at the same acquisition time according to timestamp information, forming a multidimensional data record with a one-to-one correspondence. Specifically, for each sampling time, the first and second signal values ​​acquired at that time are combined into a data pair and labeled with a timestamp tag for that time. After this processing, each sampling time point in the time-varying data signal simultaneously contains the state information of the individual molecule at that time and the environmental condition information it is in at that time. Among them, the state information is represented by the value of the first signal, and the environmental condition information is represented by the value of the second signal, and there is a clear and traceable temporal correspondence between the two.

[0037] Understandably, the signal acquisition stage can discretely sample the first signal at a preset sampling rate and synchronously acquire the value of the second signal at each sampling moment. Then, time alignment processing binds the two together into a multi-dimensional data record. This process ensures that each point in time in the time-varying data signal carries information from both the state and environment dimensions, and that the temporal correspondence between the two is strict and traceable. The resulting high-quality time-varying data signal provides the foundational data with a strict temporal synchronization relationship for subsequent preprocessing and feature fusion processing stages. This solves the problem of existing technologies where the two types of signals are acquired independently and lack temporal dimension correlation from the source of acquisition, laying the foundation for improving the accuracy of state recognition using temporal correlation features in subsequent steps.

[0038] Step S20: The time-varying data signal is preprocessed by the intelligent processing unit to generate standardized data; It should be noted that the time-varying data signals collected in step S10, in addition to containing effective signals reflecting single-molecule state changes and environmental conditions, also contain various interference components. These interference components mainly include electromagnetic interference from the detection environment and background noise and electronic noise introduced by the electronic thermal motion of the sensor itself; intermittent loss of fluorescence signal caused by fluorescent dye molecules entering the dark state (i.e., fluorescent dye blinking); continuous attenuation of fluorescence signal caused by irreversible photochemical degradation of the fluorescent dye under continuous excitation light irradiation (i.e., photobleaching); and abnormal pulse abrupt changes in current signal caused by electrochemical reactions or contact instability on the electrode surface in the electrical detection device (i.e., electrode interference). If these interference components are directly introduced into the subsequent feature fusion and state recognition stages without processing, they will severely interfere with the model's extraction of true state change features, leading to a decrease in state recognition accuracy. Therefore, a pre-processing model is used in the intelligent processing unit to pre-process the time-varying data signals to generate standardized data that can be directly used by the subsequent model. Preprocessing refers to a series of operations that improve the quality and normalize the acquired raw time-varying data signals. Its purpose is to remove or suppress interference components in the signal and adjust the signal to a uniform and standardized form suitable for subsequent model processing. Standardized data refers to signal data obtained after preprocessing, which has removed noise and spurious signal interference and has a uniform magnitude, while retaining signal characteristics effective for identifying single-molecule states.

[0039] Understandably, by using an intelligent processing unit to sequentially or in combination perform denoising, spurious signal removal, and normalization on the time-varying data signal, background noise, electronic noise, and spurious signal components such as fluorescent dye blinking, photobleaching, and electrode interference mixed in with the original signal are effectively removed. Furthermore, signal components with significantly different magnitudes in the first and second signals are adjusted to a uniform numerical range. The resulting standardized data retains signal characteristics effective for identifying single-molecule states while possessing a regular data format and a uniform magnitude range, providing high-quality, standardized input data for subsequent feature fusion processing.

[0040] In one feasible implementation, the step of preprocessing time-varying data signals to generate standardized data using an intelligent processing unit includes: The time-varying data signal is denoised using the preprocessing model in the intelligent processing unit. The time-varying data signal after noise reduction is subjected to pseudo signal removal processing to identify and remove pseudo signals, including at least one pseudo signal among fluorescent dye blinking, photobleaching and electrode interference. Normalization processing is performed on the time-varying data signal after spurious signal removal, adjusting the signal components of different magnitudes in the first and second signals to a uniform numerical range to obtain standardized data.

[0041] It should be noted that denoising refers to the removal of background noise and electronic noise introduced during the acquisition and transmission of time-varying data signals. Background noise mainly originates from electromagnetic interference in the detection environment and the thermal motion of molecules in the buffer solution, while electronic noise mainly originates from the electronic thermal noise of the sensor itself and circuit noise introduced by the signal amplification circuit. False signal removal refers to the identification and removal of signal components in the time-varying data signal that do not originate from actual single-molecule state changes but have external characteristics similar to valid signals. False signals include at least one of the following: fluorescent dye blinking, photobleaching, and electrode interference. Fluorescent dye blinking refers to the intermittent loss of fluorescence signal caused by the fluorescently labeled molecule entering a momentary non-emission dark state; its characteristic is usually manifested as periodic fluctuations in signal intensity over a short period or a momentary decrease followed by recovery. Photobleaching refers to the continuous decay of fluorescence intensity caused by irreversible photochemical degradation of the fluorescent dye under continuous excitation light irradiation; its characteristic is that the signal intensity decreases monotonically and exponentially over time and is irrecoverable. Electrode interference refers to abnormal current fluctuations caused by electrochemical reactions on the electrode surface or unstable molecular contact with the electrode in electrical detection devices. It is typically characterized by pulse abrupt changes in signal width and amplitude. Specific methods for identifying and eliminating these pseudo-signals include pre-extracting characteristic pattern templates for various pseudo-signals, marking signal segments in the time-varying data signal that match the characteristic pattern templates with a preset matching threshold as pseudo-signals, and then zeroing out or directly removing the marked signal segments from the data sequence. Normalization refers to the operation of mapping signal components with different physical dimensions and numerical ranges to the same or similar numerical ranges.

[0042] Understandably, the process involves first denoising to remove random interference components such as background noise and electronic noise from the signal; then, spurious signal removal to identify and eliminate spurious signal components with specific characteristic patterns; and finally, normalization to adjust signal components with large magnitude differences between the first and second signals to a uniform numerical range. The resulting standardized data retains signal features effective for identifying single-molecule states while possessing a regular data format and a uniform magnitude range, providing high-quality, standardized input data for subsequent feature fusion processing.

[0043] Step S30: By performing feature fusion processing on the standardized data, the transient features in the first signal and the temporal correlation features between the first signal and the second signal are extracted to generate a temporal feature vector; It should be noted that feature fusion processing refers to the operation of simultaneously or sequentially extracting features from standardized data across multiple dimensions and integrating these extracted features to form a unified feature representation. Transient features refer to local features in the first signal that reflect changes in the state of a single molecule at a certain moment or within a very short time window. For example, transient features may include the peak height, peak width, peak-valley shape, waveform morphology at step changes, the time of occurrence of signal abrupt changes, and the magnitude of the abrupt change. These features can capture the instantaneous morphological information of single-molecule state changes at the signal level, providing a basis for distinguishing different types of state change events. Temporal correlation features refer to the correspondence between the changing trend of the first signal over continuous time and the changing trend of the second signal within the same or related time periods. For example, temporal correlation features may include the rate of change of the first signal (the amount of change in signal intensity per unit time), the duration of the first signal (the length of time the signal remains in a certain state or undergoes a certain change), and the temporal correspondence between the trend of change of the first signal and the trend of change of the second signal (e.g., the temporal relationship where the signal feature reflecting configurational change in the first signal strengthens after the temperature rises in the second signal, or the synchronous relationship where the signal feature reflecting molecular breakage appears immediately after mechanical force is applied in the second signal). These features can reflect the dynamic response law between the state change of a single molecule and its environmental conditions, providing information support for understanding the triggering mechanism of state change. The temporal feature vector is a multi-dimensional numerical vector generated by fusing the above transient features and temporal correlation features, which can comprehensively characterize the instantaneous state change of a single molecule and its temporal correlation with environmental conditions. The numerical values ​​of each dimension in this vector encode different aspects of feature information, which together constitute a complete feature description of the current state of a single molecule and its dynamic evolution process.

[0044] Understandably, by performing feature fusion processing on standardized data, transient features reflecting the instantaneous changes in the state of a single molecule are extracted from the first signal, and temporal correlation features reflecting the temporal correspondence between state changes and changes in environmental conditions are extracted from the joint analysis of the first and second signals. These two types of features are then fused to generate a temporal feature vector. This process explicitly transforms the information about single-molecule state changes implicit in the original signal sequence and its correlation with environmental conditions into a multi-dimensional feature representation with rich discriminative power. Compared to existing techniques that extract features from only a single type of signal, the temporal feature vector generated in this step simultaneously contains transient snapshot information of the molecule's own changes and temporal evolution information of the molecule's interaction with the environment, providing a more comprehensive and discriminative feature foundation for subsequent dynamic recognition models to accurately determine the current state type of a single molecule.

[0045] In one feasible implementation, the step of generating a time-series feature vector by performing feature fusion processing on standardized data to extract transient features from the first signal and temporal correlation features between the first and second signals includes: Standardized data is input into a feature fusion processing model composed of a convolutional neural network and a long short-term memory network; Transient features are extracted from the first signal in the standardized data using a convolutional neural network to obtain transient features that characterize the peak value and waveform shape of the first signal. By using a long short-term memory network, temporal correlation features are extracted from the first and second signals in the standardized data to obtain continuous change features that characterize the rate of change and duration of the first signal, as well as the temporal correspondence between the change trend of the first signal and the change trend of the second signal. By fusing transient and continuously changing features, a time-series feature vector is generated.

[0046] It should be noted that the feature fusion processing model refers to a neural network model constructed by combining convolutional neural networks (CNNs) and long short-term memory (LSTM) networks, used to extract and fuse multi-dimensional features from standardized data. A CNN is a feedforward neural network adept at extracting locally invariant features from gridded data. It uses multiple learnable convolutional kernels to slide across the input data along the time dimension, with each kernel responsible for capturing a specific local signal morphology pattern. A LSTM network is a special type of recurrent neural network that, through gating mechanisms such as input gates, forget gates, and output gates, can selectively remember or forget information in a time series, thereby learning and encoding long-range dependencies within the time series. After standardized data is input into the feature fusion processing model, it is executed in parallel along two processing paths. In the first processing path, the CNN receives the first signal time series from the standardized data as input. Its multiple one-dimensional convolutional kernels slide across the first signal along the time dimension, detecting and encoding local morphological patterns such as peak height, peak width, peak-valley shape, and waveform morphology at step changes. After multiple layers of convolution and pooling operations, it outputs a transient feature vector representing the peak values ​​and waveform morphology of the first signal. In the second processing path, the Long Short-Term Memory (LSTM) network simultaneously receives the first and second signal time series from the standardized data as input. At each time step, the network controls the retention rate of information from the previous memory state through a forget gate, the degree to which current input information enters the memory state through an input gate, and the content of information output from the memory state to the current time step through an output gate. Through this gating mechanism, the LSM network learns and encodes the extent of change in the first signal over a given time period (i.e., the rate and duration of change), and the correspondence between the trend of change in the first signal and the trend of change in the second signal on the time axis. For example, the temporal relationship between the state change features in the first signal and the environmental parameter changes in the second signal. Both together constitute the continuous change feature. Fusion refers to the operation of integrating the transient feature vectors and continuously changing feature vectors output from the two processing paths. Fusion can be achieved by directly concatenating the vectors, combining the two feature vectors end-to-end into a higher-dimensional feature vector. Alternatively, a weighted summation method can be used, where the weight parameters of each feature dimension are automatically learned and determined during model training. The fused temporal feature vector contains transient snapshot information and continuous evolution information of single-molecule state changes, as well as temporal correlation information between state changes and changes in environmental conditions.

[0047] Understandably, the feature fusion processing model, composed of a convolutional neural network and a long short-term memory network, leverages the advantages of convolutional neural networks in extracting local morphological features and long short-term memory networks in modeling long-range temporal dependencies. This model extracts transient features reflecting the instantaneous changes in the state of a single molecule and continuous features reflecting the temporal correspondence between state changes and changes in environmental conditions from standardized data in parallel. These two features are then fused into an information-rich and discriminative temporal feature vector. Compared to feature extraction using a single type of network, the fusion model in this implementation can simultaneously capture both transient details and long-range evolutionary patterns in the process of single-molecule state changes.

[0048] Step S40: Based on the dynamic recognition model, classify the temporal feature vectors to determine the current state type of the single molecule.

[0049] It should be noted that the dynamic recognition model refers to a pre-trained computational model used to classify and determine the state type of a single molecule based on the input temporal feature vector. Its internal parameters have been iteratively optimized using a large number of labeled samples, enabling it to identify different state patterns from the temporal feature vector. The dynamic recognition model can handle the dependencies between time steps in the temporal feature vector; that is, when determining the state type at a certain moment, it not only refers to the feature information at that moment but also integrates the feature information of the preceding and following time steps for context-aware judgment. The temporal feature vector, as the input to the dynamic recognition model, is a multi-dimensional numerical vector output by the feature fusion processing model, encoding transient morphological information and continuous evolution information during the state change process of a single molecule. Classification refers to the operation of the dynamic recognition model mapping the input temporal feature vector to a certain category in a preset set of state types. The dynamic recognition model outputs the current state type of the single molecule through calculation of the temporal feature vector. The preset state type refers to a set of state categories predefined according to the possible state change patterns of a single molecule. For example, the preset state type may include at least one of stable configuration state, configurational transition state, and molecular breakage state. A stable configuration state refers to a state in which the configuration of a single molecule remains relatively unchanged or fluctuates only slightly around its equilibrium position within a certain time window. In the first signal, this typically manifests as a stable signal value that fluctuates randomly within a small range around a certain mean. A configurational change state refers to a state in which the configuration of a single molecule is undergoing a identifiable and significant change, encompassing dynamic configurational changes of various magnitudes, from minute twisting to large folding or stretching. In the first signal, this typically manifests as a sustained upward or downward trend in the signal value, or a significant step change within a short period. A molecular breakage state refers to a state in which the intramolecular connections of a single molecule or the connections between the molecule and the electrode break. This encompasses any stage of the entire process, from the initial stage to the complete breakage stage. In the first signal, this typically manifests as a sharp change in the signal value or a sudden interruption of the signal. In practical applications, the specific types and granularity of the preset state types can be flexibly set according to the characteristics of the monitored object and application requirements.

[0050] Understandably, a pre-trained dynamic recognition model classifies temporal feature vectors and leverages the state pattern recognition capabilities learned during offline training to map the complex feature information encoded in the feature vectors into clear state type determination results. Because the dynamic recognition model can perform context-aware comprehensive analysis of the dependencies between time steps in the temporal feature vectors, it can more accurately identify the current state type of a single molecule compared to methods that rely solely on features from a single time step. This reduces unreasonable jumps in state determination along the time axis and erroneous responses to noise, achieving real-time and accurate identification of single-molecule state changes in complex microenvironments.

[0051] In one feasible implementation, the step of classifying the temporal feature vector based on the dynamic recognition model to determine the current state type of a single molecule includes: Input the temporal feature vector into the dynamic recognition model; Self-attention weighting is applied to the dependencies between time steps in the temporal feature vector to generate the probability distribution of a single molecule in each preset state type at each time step. Based on the probability distribution, the preset state type with the highest probability is determined as the state type of the single molecule at the corresponding time step; Output the current state type of a single molecule, wherein the preset state type includes at least one of stable configuration state, configurational transition state, and molecular breakage state.

[0052] It's important to clarify that inputting temporal feature vectors into a dynamic recognition model means using these vectors as input data. The dynamic recognition model is a pre-trained classification model whose internal parameters have been optimized through iterative training on a large number of labeled samples, enabling it to identify different state patterns from the temporal feature vectors. Self-attention weighted processing refers to a mechanism that captures long-range dependencies between time steps by calculating attention weights between any two time steps in the temporal feature vector. Unlike traditional recurrent neural networks that process data sequentially, the self-attention mechanism allows the model to directly establish connections between any two time steps, regardless of their temporal distance. Specifically, for each time step in the temporal feature vector, the model first generates a query vector, a key vector, and a value vector for that time step. The query vector is used for matching with other time steps, the key vector is used to be matched by other time steps, and the value vector represents the actual feature information carried by that time step. Then, by calculating the similarity between the query vector of the current time step and the key vectors of all time steps (including itself), a set of attention weights is obtained. This set of weights reflects the importance of each time step to the current time step. Finally, the value vectors of all time steps are weighted and summed using the attention weights to obtain the context-aware feature representation of the current time step. This representation incorporates global contextual information related to the current time step throughout the entire time series. Based on the context-aware feature representations of each time step, the probability value of a single molecule in each preset state type at each time step is calculated through the classification layer in the dynamic recognition model. After normalizing the probability values ​​using a function, a probability distribution for each time step is formed. Determining the preset state type with the highest probability as the state type of the single molecule at the corresponding time step means selecting the preset state type with the highest probability value in the probability distribution of each time step as the state determination result for that time step. When the probability values ​​of multiple preset state types are relatively close, the state determination results of adjacent time steps can be combined for smoothing, for example, by using majority voting within a sliding window to determine the final state type, in order to reduce the instability of the determination caused by probability fluctuations in individual time steps. Outputting the current state type of a single molecule refers to outputting the judgment result of the dynamic recognition model for real-time display by the feedback module or for further analysis by the early warning module. Preset state types include at least one of stable configuration state, configurational transition state, and molecular breakage state. A stable configuration state refers to a state in which the configuration of a single molecule remains relatively unchanged within a certain time window or fluctuates only slightly around its equilibrium position. In the first signal, it typically exhibits a stable characteristic where the signal value fluctuates randomly within a small range around a certain mean.Configurational transition states refer to the state in which the configuration of a single molecule is undergoing a identifiable and significant change. This can encompass dynamic configurational changes of varying magnitudes, from minute twisting to large folding or stretching. In the first signal, it typically manifests as a sustained upward or downward trend in the signal value, or a significant step change within a short period. Molecular breakage states refer to the state in which the intramolecular connections of a single molecule or the connections between the molecule and the electrode are broken. This can cover any stage of the entire process, from the initial stage to the complete breakage stage. In the first signal, it typically manifests as a sharp change in the signal value or a sudden interruption of the signal.

[0053] Understandably, by employing a dynamic recognition model based on a self-attention mechanism, the dependencies between time steps in the temporal feature vector are explicitly modeled. This allows the model to comprehensively utilize the global contextual information of the entire time series when determining the state type of each time step, rather than relying solely on the isolated features of that time step. This context-aware determination method effectively reduces unreasonable jumps in state determination along the time axis and erroneous responses to noise, improving the accuracy and temporal consistency of state recognition. Furthermore, by outputting a clear probability distribution and selecting the highest probability as the determination result, the state determination process becomes interpretable and traceable. Moreover, by elevating the preset state types to stable configuration states, configurational transition states, and molecular breakage states, this method covers various state change patterns that a single molecule may exhibit, enabling it to adapt to the monitoring needs of different types of single molecules and different experimental conditions, demonstrating strong versatility and adaptability.

[0054] Based on the first embodiment of this application, in the second embodiment of this application, the content that is the same as or similar to that in the first embodiment described above can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 3 , Figure 3 This is a flowchart illustrating Embodiment 2 of the single-molecule state monitoring method of this application. After the step of classifying the time-series feature vectors based on the dynamic identification model to determine the current state type of the single molecule, it also includes steps A10 to A20: Step A10: Perform a comprehensive analysis on the environmental condition data corresponding to the time-series feature vector and the second signal to determine whether an unexpected state change event will occur in a single molecule. The unexpected state change event includes unexpected molecular breakage or abnormal configuration change. Step A20: When the probability of an unexpected state change event reaches a preset warning threshold, a warning signal is generated, and the time point that triggers the warning, the corresponding environmental condition data, and the feature fragments of the time-series feature vector are marked.

[0055] It should be noted that in practical single-molecule monitoring applications, simply determining the current state of a single molecule is often insufficient to meet the requirements of real-time regulation and immediate intervention. For example, in biomolecular reaction monitoring, unexpected breakage of a key intermediate can lead to the failure of the entire experimental process. If early signs of breakage can be detected and an early warning issued before the breakage actually occurs, a window of opportunity for intervention can be gained. Similarly, abnormal changes in single-molecule configuration (such as changes exceeding the normal range or abnormally accelerated rates of change) may also be precursors to uncontrolled experimental conditions or abnormal molecular structures, requiring timely identification and response. Traditional monitoring methods mostly remain at the level of passive identification of the current state, lacking the ability to proactively predict and warn of future abnormal events. Based on this, this implementation method further introduces an early warning step, which achieves early judgment and proactive warning of unexpected state change events by comprehensively analyzing time-series feature vectors and environmental condition data. Comprehensive analysis refers to using the time-series feature vectors generated in the dynamic identification step and the environmental condition data corresponding to the current moment as the basis for analysis, jointly judging from two dimensions whether an unexpected state change event is about to occur in a single molecule. The temporal feature vector contains transient morphological features and continuous evolution features during the single-molecule state change process. These features implicitly contain trend information about the state change, such as whether weak feature patterns indicating pre-fracture appear in the signal, or whether the rate of configurational change is continuously accelerating. Environmental condition data provides the actual parameter values ​​of the microenvironment in which the single molecule is located at the current moment. Changes in these parameters may be triggering factors for unexpected state change events, such as a sudden increase in temperature leading to molecular structural instability, or a pH deviation from the normal range leading to abnormal molecular configuration. Comprehensive analysis can be performed based on a pre-established early warning model. The early warning model has learned the distinguishing boundary between normal and abnormal state change features during the offline training phase and has set corresponding early warning judgment rules accordingly. Unexpected state change events refer to state change events that occur in a single molecule under normal monitoring conditions and do not conform to the expected experimental design or natural behavior pattern, specifically including unexpected molecular fracture and abnormal configurational changes. Unexpected molecular breakage refers to a molecular connection breakage event that occurs under normal monitoring conditions, outside the expected time window, or triggered by abnormal factors. At the signal level, it typically manifests as abnormal precursor features in the first signal, such as a sudden increase in signal fluctuation amplitude or the appearance of intermittent unstable segments in the signal. Abnormal conformational changes refer to changes in the amplitude, rate, or direction of a single molecule's conformation that exceed a preset normal range. At the signal level, it typically manifests as the rate of change in the first signal exceeding a normal threshold, or a significant deviation in the temporal correlation pattern between the first and second signals from the normal correlation pattern. The preset warning threshold is the minimum probability value required to trigger a warning signal, determined by the warning model during offline training based on the statistical distribution of normal and abnormal samples.When the probability of an unexpected state change event calculated by the early warning model reaches or exceeds a certain threshold, it indicates that the urgency of the unexpected event is high and users need to be notified promptly. Generating an early warning signal refers to transmitting a perceptible alert signal through an audible and visual alarm device, or by displaying a pop-up alert window through a feedback module, to deliver early warning information to the user. Marking the time point that triggers the early warning, the corresponding environmental condition data, and the feature segments of the time-series feature vector means that while generating the early warning signal, key information about the event is automatically recorded. This includes the specific time the warning is triggered (accurate to the millisecond level), the environmental condition data reflected by the second signal collected at that time (such as temperature, pH, ion concentration, mechanical force, etc.), and the feature segments in the time-series feature vector that trigger the warning that are related to anomaly detection. This marked information provides traceable data for subsequent analysis of the root causes and triggering mechanisms of unexpected state change events.

[0056] Understandably, based on determining the current state type of a single molecule, an early warning mechanism is further introduced to address potential unintended state change events. Through comprehensive analysis of time-series feature vectors and environmental condition data, the early warning model can identify the abnormal precursor features hidden in the signal before unintended molecular breakage or abnormal configurational changes actually occur, and proactively issue an early warning signal and mark relevant information when the probability of the event reaches the early warning threshold.

[0057] For example, to help understand the implementation flow of the single-molecule state monitoring method obtained in this embodiment combined with the above embodiment one, please refer to... Figure 4 , Figure 4 Here is a simplified flowchart of the single-molecule state monitoring method of this application, specifically: Before formal monitoring begins, sample preprocessing and monitoring system setup are performed, including preparing single-molecule samples to a suitable monitoring state and deploying and connecting the signal acquisition unit and intelligent processing unit. Following this, the AI ​​model enters an offline training phase. A training dataset containing positive and negative samples is constructed, and the preprocessing model, feature fusion processing model, and dynamic recognition model are iteratively trained until each model meets preset performance indicators. After model training, the real-time monitoring phase begins. The signal acquisition unit simultaneously acquires a first signal reflecting the single molecule's own state and a second signal reflecting its environmental conditions, which are then time-aligned to form time-varying data signals. Upon receiving these signals, the intelligent processing unit first performs denoising, spurious signal removal, and normalization processing by the preprocessing model to generate standardized data. Then, the feature fusion processing model extracts transient features and time-series correlation features, fusing them to generate a time-series feature vector. Finally, the dynamic recognition model classifies the single molecule to determine its current state type. Based on this, a comprehensive analysis of the time-series feature vector and environmental condition data is performed to determine whether an unexpected state change event will occur. When the probability of an event occurring reaches the warning threshold, an early warning signal is proactively issued and relevant information is marked. Simultaneously, the monitoring curve, identification results, and warning information are displayed and stored in real time. After the monitoring task is completed, all data is exported for organization and archiving.

[0058] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the single-molecule state monitoring method of this application. Any simple modifications based on this technical concept are within the protection scope of this application.

[0059] This application provides an electronic device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the single-molecule state monitoring method in Embodiment 1 above.

[0060] The following is for reference. Figure 5 The diagram illustrates a structural schematic of an electronic device suitable for implementing embodiments of this application. The electronic devices in these embodiments may include, but are not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), in-vehicle terminals (e.g., in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 5The electronic device shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of this application.

[0061] like Figure 5 As shown, the electronic device may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage device 1003 into a random access memory (RAM) 1004. The RAM 1004 also stores various programs and data required for the operation of the electronic device. The processing unit 1001, ROM 1002, and RAM 1004 are interconnected via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to the I / O interface 1006: input devices 1007 including, for example, a touchscreen, touchpad, keyboard, mouse, image sensor, microphone, accelerometer, gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; storage devices 1003 including, for example, magnetic tape, hard disk, etc.; and communication devices 1009. Communication device 1009 allows electronic devices to communicate wirelessly or wiredly with other devices to exchange data. While electronic devices with various systems are shown in the figures, it should be understood that implementation or possession of all the systems shown is not required. More or fewer systems may be implemented alternatively.

[0062] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from ROM 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0063] The electronic device provided in this application employs the single-molecule state monitoring method described in the above embodiments, which can solve the technical problem that traditional single-molecule monitoring technology is difficult to accurately identify molecular states. Compared with the prior art, the beneficial effects of the electronic device provided in this application are the same as those of the single-molecule state monitoring method provided in the above embodiments, and other technical features of this electronic device are the same as those disclosed in the method of the previous embodiment, and will not be repeated here.

[0064] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0065] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0066] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, the computer-readable program instructions being used to execute the single-molecule state monitoring method in the above embodiments.

[0067] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0068] The aforementioned computer-readable storage medium may be included in an electronic device or may exist independently without being assembled into an electronic device.

[0069] The aforementioned computer-readable storage medium carries one or more programs, which, when executed by an electronic device, cause the electronic device to: acquire the first signal and the second signal in real time during the state change of the single molecule through the signal acquisition unit, thereby obtaining a time-varying data signal; The intelligent processing unit preprocesses the time-varying data signal to generate standardized data. The standardized data is subjected to feature fusion processing to extract transient features from the first signal and temporal correlation features between the first signal and the second signal, thereby generating a temporal feature vector; Based on the dynamic recognition model, the temporal feature vector is classified to determine the current state type of the single molecule.

[0070] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a Local Area Network (LAN) or a Wide Area Network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0071] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0072] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0073] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the above-described single-molecule state monitoring method, thereby solving the technical problem that traditional single-molecule monitoring technologies struggle to accurately identify molecular states. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the single-molecule state monitoring method provided in the above embodiments, and will not be elaborated upon here.

[0074] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the single-molecule state monitoring method described above.

[0075] The computer program product provided in this application can solve the technical problem that traditional single-molecule monitoring technology is difficult to accurately identify molecular states. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as those of the single-molecule state monitoring method provided in the above embodiments, and will not be repeated here.

[0076] The above description is only a part of the embodiments of this application and does not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application and using the contents of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included in the patent protection scope of this application.

Claims

1. A method for monitoring the state of a single molecule, characterized in that, The method for monitoring the state of a single molecule is applied to a state monitoring system. The state monitoring system includes a signal acquisition unit for acquiring a first signal reflecting the state of the single molecule itself and a second signal reflecting the environmental conditions of the single molecule, and an intelligent processing unit electrically connected to the signal acquisition unit. The signal acquisition unit acquires the first signal and the second signal in real time during the state change of the single molecule to obtain time-varying data signals. The intelligent processing unit preprocesses the time-varying data signal to generate standardized data. By performing feature fusion processing on the standardized data, transient features in the first signal and temporal correlation features between the first signal and the second signal are extracted to generate a temporal feature vector. Based on the dynamic recognition model, the temporal feature vector is classified to determine the current state type of the single molecule; The step of acquiring the first signal and the second signal in real time during the state change of the single molecule through the signal acquisition unit to obtain time-varying data signals includes: The first signal is acquired at a preset sampling rate; At the same time as the acquisition of the first signal, the second signal is acquired synchronously; The first signal and the second signal are time-aligned respectively to obtain the time-varying data signal, so that the data at each time point in the time-varying data signal includes the state information of the single molecule and the corresponding environmental condition information. The step of generating a time-series feature vector by performing feature fusion processing on the standardized data to extract transient features from the first signal and temporal correlation features between the first signal and the second signal includes: The standardized data is input into the feature fusion processing model, which is composed of a convolutional neural network and a long short-term memory network; The first signal in the standardized data is extracted by the convolutional neural network to obtain transient features characterizing the peak value and waveform shape of the first signal. Through the Long Short-Term Memory Network, temporal correlation features are extracted from the first signal and the second signal in the standardized data to obtain continuous change features that characterize the rate of change and duration of change of the first signal, as well as the temporal correspondence between the change trend of the first signal and the change trend of the second signal. The transient features and the continuous change features are fused to generate the temporal feature vector.

2. The single-molecule state monitoring method as described in claim 1, characterized in that, The intelligent processing unit includes a preprocessing model, a feature fusion processing model, and a dynamic recognition model; Before the step of acquiring the first signal and the second signal in real time to obtain the time-varying data signal, the method further includes: Construct a training dataset, wherein the training dataset includes positive samples and negative samples, the positive samples include signal data of the single-molecule state change process and corresponding environmental condition data, and the negative samples include signal data of background noise; Based on the training dataset, the preprocessing model, the feature fusion processing model, and the dynamic recognition model are iteratively trained until each model meets the corresponding preset model training performance index.

3. The single-molecule state monitoring method as described in claim 2, characterized in that, The step of preprocessing the time-varying data signal through the intelligent processing unit to generate standardized data includes: The time-varying data signal is denoised using the preprocessing model in the intelligent processing unit. The time-varying data signal after noise reduction is subjected to pseudo signal removal processing to identify and remove pseudo signals, wherein the pseudo signals include at least one of fluorescent dye blinking, photobleaching and electrode interference. Normalization processing is performed on the time-varying data signal after pseudo-signal removal to adjust the signal components of different magnitudes in the first signal and the second signal to a unified numerical range, thereby obtaining the standardized data.

4. The single-molecule state monitoring method as described in claim 2, characterized in that, The step of classifying the temporal feature vector based on the dynamic recognition model to determine the current state type of the single molecule includes: The temporal feature vector is input into the dynamic recognition model; Self-attention weighting is applied to the dependencies between time steps in the time-series feature vector to generate the probability distribution of the single molecule in each preset state type at each time step. Based on the probability distribution, the preset state type with the highest probability is determined as the state type of the single molecule at the corresponding time step; Output the current state type of the single molecule, wherein the preset state type includes at least one of stable configuration state, configurational transition state, and molecular breakage state.

5. The single-molecule state monitoring method as described in claim 1, characterized in that, After the step of classifying the temporal feature vector based on the dynamic recognition model to determine the current state type of the single molecule, the method further includes: A comprehensive analysis is performed on the time-series feature vector and the environmental condition data corresponding to the second signal to determine whether the single molecule will undergo an unexpected state change event, wherein the unexpected state change event includes unexpected molecule breakage or abnormal configuration change. When the probability of the occurrence of the unexpected state change event reaches a preset warning threshold, a warning signal is generated, and the time point that triggers the warning, the corresponding environmental condition data, and the feature fragment of the time-series feature vector are marked.

6. An electronic device, characterized in that, The device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the single-molecule state monitoring method as described in any one of claims 1 to 5.

7. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, it implements the steps of the single-molecule state monitoring method as described in any one of claims 1 to 5.

8. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the steps of the single-molecule state monitoring method as described in any one of claims 1 to 5.

Images