Multi-source physiological information acquisition method and system
By constructing a dynamic programming matrix D to align multi-source physiological data in timing, the problem of inconsistent timing of sensor data is solved, and the accuracy and credibility of data are improved.
Patent Information
- Application Number
- CN202510455357.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-11
- Publication Date
- 2025-06-24
AI Technical Summary
During the multi-source physiological information acquisition process, there are time differences in the data collected by each sensor, resulting in inconsistent timing of the collected physiological information and lack of rigor.
By constructing a dynamic programming matrix D, physiological data are input into the dynamic programming matrix D, and the best matching paths that characterize physiological data are arranged in chronological order are obtained, and the physiological data aligned after timing is obtained based on the chronological order of the best matching path.
It solves the problem of inconsistency in the timing of physiological information caused by the time difference in data collection, improves the accuracy and credibility of the data, and ensures the stability and reliability of the obtained physiological data.
Smart Images
Figure CN120197111A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of physiological information acquisition, and particularly relates to a multi-source physiological information acquisition method and system. Background Art
[0002] In today's medical and health field, increasing attention is paid to the collection and analysis of physiological information. By collecting and analyzing physiological signals, doctors can accurately evaluate the health status of patients, predict disease risks, and accordingly develop personalized treatment plans, which is crucial for the early detection, prevention, and treatment of diseases.
[0003] Currently, sensors are used to obtain multi-source physiological information. The process includes: configuring corresponding sensors according to the requirements of experiments or monitoring; after the connection between the sensors and the data acquisition device is completed, the system starts to collect various physiological signals; the system performs conditioning processes such as amplification and filtering on the collected signals, and the conditioned analog signals are converted into digital signals, which are then stored in the storage device of the system; the storage device further processes the digital signals, such as feature extraction and data analysis, to obtain physiological information.
[0004] The defects of the above-mentioned existing technologies are as follows: during the acquisition process of multi-source physiological information, there are time differences in the data collected by each sensor, resulting in inconsistent time sequences of the collected physiological information and lack of rigor in the data. Summary of the Invention
[0005] Based on this, it is necessary to provide a multi-source physiological information acquisition method and system for the above-mentioned technical problems.
[0006] An embodiment of the present invention provides a multi-source physiological information acquisition method, including:
[0007] Obtaining a plurality of physiological data with time sequence differences collected by sensors corresponding to different parts of the human body;
[0008] Constructing a dynamic programming matrix D;
[0009] Inputting the physiological data into the dynamic programming matrix D to obtain an optimal matching path representing the chronological arrangement of the physiological data; obtaining the physiological data after time sequence alignment based on the chronological order of the optimal matching path, and analyzing the physiological data after time sequence alignment to obtain physiological information representing the health status of the human body;
[0010] Wherein, the constructing of the dynamic programming matrix D includes:
[0011] Initialize the dynamic programming matrix to obtain a multidimensional dynamic programming matrix D0 containing multiple dimensions N1×N2×…×N7; initialize the boundary conditions of the multidimensional dynamic programming matrix D0 so that the data points in the first dimension and the first data point in each dimension meet the boundary conditions, and gradually fill in the elements in the multidimensional dynamic programming matrix D0 to obtain the dynamic programming matrix D.
[0012] In addition, the physiological data includes: electroencephalogram data, electrocardiogram data, electromyography data, blood pressure data, blood oxygen data, body temperature data and respiratory data.
[0013] In addition, before the physiological data is input into the dynamic programming matrix D, the physiological data is subjected to noise filtering and anti-interference processing, specifically including:
[0014] For EEG data, the bandpass filtering method was used to retain the data of 0.1-45Hz, and independent component analysis was performed to remove the electrooculogram, electromyography and electrocardiogram components contained in the EEG signal;
[0015] For ECG data and EMG data, wavelet transform is used to decompose physiological signals into sub-signals of different frequency ranges, and filtering is performed according to the energy of the sub-signals;
[0016] The blood pressure data, blood oxygen data, body temperature data and respiratory data are denoised through digital filtering.
[0017] In addition, the use of wavelet transform to decompose the physiological signal into sub-signals in different frequency ranges specifically includes:
[0018] The ECG data and EMG data x(t) are matched with the Daubechies wavelet basis function to obtain the wavelet coefficients of each scale, and the formula is:
[0019]
[0020] For each scale wavelet coefficient W j,k , if its absolute value is less than the set threshold T, it is set to zero, otherwise the original value is retained. The formula is:
[0021]
[0022] The wavelet coefficients after anti-interference processing are inversely transformed to obtain the reconstructed signal; the reconstructed signal is synthesized into the original signal through inverse wavelet transform, and the formula is:
[0023]
[0024] Among them, W j,k represents the wavelet coefficient of the jth scale and the kth position, ψ j,k(t) represents the Daubechies wavelet basis function, and x(t) represents the physiological signal. represents the reconstructed signal.
[0025] In addition, initializing the boundary conditions of the multi-dimensional dynamic programming matrix D0 specifically includes:
[0026] D0(i1,i2,...,i7) represents the distance between the first i1 elements of sequence X1, the first i2 elements of sequence X2,..., and the first i7 elements of sequence X7;
[0027] Initialize the boundary conditions of the multi-dimensional dynamic programming matrix D0 so that the first dimension and the first element of each dimension satisfy the boundary conditions, usually initialized to infinity or zero:
[0028] D0(0,i2,...,i7) = ∞, D0(i1,0,...,i7) = ∞,...
[0029] where D(i1, i2,..., i7) represents the distance between the first i1 elements of sequence X1, the first i2 elements of sequence X2,..., and the first i7 elements of sequence X7.
[0030] In addition, the formula for gradually filling the elements in the multi-dimensional dynamic programming matrix D0 is:
[0031] For each element D(i1,i2,…,i7) in the sequence,
[0032]
[0033] where, represents the distance between the i1-th element of sequence X1, the i2-th element of sequence X2,..., and the i7-th element of sequence X7, using the Euclidean distance.
[0034] In addition, a multi-source physiological information acquisition system includes:
[0035] A multi-source physiological information acquisition module 1 for acquiring multiple physiological data with time series differences collected by sensors corresponding to different parts of the human body;
[0036] A data integration module 3 for constructing a dynamic programming matrix;
[0037] A data analysis and interaction module 4 for inputting physiological data into the dynamic programming matrix D to obtain the optimal matching path representing the chronological arrangement of physiological data; obtaining the chronologically aligned physiological data based on the chronological order of the optimal matching path, analyzing the chronologically aligned physiological data, and obtaining physiological information representing the human health status.
[0038] Additionally, it further includes a data anti-interference module 2 for performing noise filtering and anti-interference processing on physiological data.
[0039] The above-mentioned method and system for obtaining multi-source physiological information provided by the embodiments of the present invention have the following beneficial effects compared with the prior art:
[0040] In the prior art, when multiple sensors are used for synchronous acquisition of multi-source physiological information, there are time differences in the data collected by each sensor, resulting in inconsistent time sequences of the collected physiological information and lack of rigor in the data.
[0041] However, in the present invention, by constructing a dynamic programming matrix D and inputting physiological data into the dynamic programming matrix D, the time sequences of the collected physiological information are uniformly sorted out, solving the problem that the inconsistent time sequences of physiological information are caused by the time differences in data collection, obtaining the best matching path representing the chronological correspondence of physiological data, and obtaining the physiological data after time sequence alignment based on the chronological order of the best matching path, improving the accuracy and credibility of the data, and ensuring the stability and reliability of the obtained physiological data. Description of the Drawings
[0042] Figure 1 It is a block diagram of the composition of a multi-source physiological information acquisition system provided in an embodiment. Detailed Embodiments
[0043] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.
[0044] In one embodiment, a method for obtaining multi-source physiological information is provided, and the method includes:
[0045] 1. Obtain multiple physiological data with time sequence differences collected by sensors corresponding to different parts of the human body.
[0046] 2. Construct a dynamic programming matrix D, and the construction of the dynamic programming matrix D includes: initializing the dynamic programming matrix to obtain a multi-dimensional dynamic programming matrix D0 including multiple dimensions N1×N2×…×N7; initializing the boundary conditions of the multi-dimensional dynamic programming matrix D0 so that the data points in the first dimension and the first data points in each dimension meet the boundary conditions, and gradually filling the elements in the multi-dimensional dynamic programming matrix D0 to obtain the dynamic programming matrix D.
[0047] 3. Input the physiological data into the dynamic programming matrix D to obtain the optimal matching path representing the chronological arrangement of the physiological data; based on the chronological order of the optimal matching path, obtain the physiologically data after time alignment, and analyze the physiologically data after time alignment to obtain the physiological information representing the human health status.
[0048] In one embodiment, a multi-source physiological information data acquisition system is provided, as Figure 1 shown, including a multi-source physiological information acquisition module 1, a data anti-interference module 2, a data integration module 3, and a data analysis and interaction module 4.
[0049] The multi-source physiological information acquisition module 1 is used to acquire a plurality of physiologically data with time series differences collected by sensors corresponding to different parts of the human body.
[0050] The data anti-interference module 2 is used to perform noise filtering and anti-interference processing on the physiological data.
[0051] The data integration module 3 is used to construct a dynamic programming matrix.
[0052] The data analysis and interaction module 4 is used to input the physiological data into the dynamic programming matrix D to obtain the optimal matching path representing the chronological arrangement of the physiological data; based on the chronological order of the optimal matching path, obtain the physiologically data after time alignment, and analyze the physiologically data after time alignment to obtain the physiological information representing the human health status.
[0053] When performing multi-source physiological information data acquisition, the multi-source physiological information acquisition module 1 is responsible for collecting physiological data from different sensors, including electroencephalogram data, electrocardiogram data, electromyogram data, blood pressure data, blood oxygen data, body temperature data, and respiration data. The acquisition module 1 transmits the collected data to the data anti-interference module 2.
[0054] The data anti-interference module 2 receives the data from the multi-source physiological information acquisition module 1 and performs noise filtering and anti-interference processing to ensure the quality and stability of the collected data. The processed data is transmitted to the data integration module 3.
[0055] The data integration module 3 receives the data from the data anti-interference module 2 and is responsible for integrating the data collected by different sensors and ensuring that they have the same time reference. In this way, even if the data from different sensors have different sampling rates or time series characteristics, the system can uniformly process and analyze the data.
[0056] The data analysis and interaction module 4 is the core part of the system, responsible for real-time analysis and processing of the collected physiological data, and providing a user interface for interaction and feedback. This module has functions such as a data visualization interface, allowing users to view the collected physiological data and analysis results in real time, and perform corresponding operations and decisions.
[0057] The specific implementation process is as follows:
[0058] Step 1: The multi-source physiological data acquisition module 1 is responsible for collecting physiological data from different sensors, including EEG, ECG, EMG, blood pressure, blood oxygen, body temperature, respiration, etc. These sensors may be distributed in different parts of the human body or use different acquisition technologies. For example, EEG sensors are used to record brain electrical activity, ECG sensors are used to record heart electrical activity, and EMG sensors are used to record muscle contraction. The acquisition module 1 transmits the data collected by these sensors to the data anti-interference module 2.
[0059] Step 2: The data anti-interference module 2 receives the data from the multi-source physiological information acquisition module 1 and performs noise filtering and anti-interference processing.
[0060] For EEG signals, bandpass filtering is used to retain data from 0.1 to 45 Hz, and independent component analysis is used to remove the electrooculographic, electromyographic and electrocardiographic components contained in the EEG signals.
[0061] For electrocardiogram and electromyography signals, the present invention uses the wavelet transform method to decompose physiological signals into sub-signals in different frequency ranges, and performs filtering processing according to their energy size to remove the influence of environmental interference, equipment noise and other factors on the data, thereby ensuring the quality and stability of the collected data.
[0062] The ECG and EMG signals x(t) are decomposed into wavelet coefficients of different scales. Daubechies wavelet transform is usually performed in an iterative manner to decompose the signal into multiple levels of scales. During the decomposition process, the ECG and EMG signals are matched with the Daubechies wavelet basis functions through convolution operations to obtain the wavelet coefficients of each scale.
[0063]
[0064] Among them, W j,k represents the wavelet coefficient at the jth scale and the kth position, ψ j,k (t) represents the Daubechies wavelet basis function.
[0065] After obtaining the wavelet coefficients, anti-interference processing is performed by setting a threshold. For each scale of wavelet coefficients W j,k, if the absolute value is less than the set threshold T, it is set to zero; if the absolute value is greater than the set threshold T. This step can be achieved by methods such as soft threshold or hard threshold.
[0066]
[0067] Finally, perform the inverse transform on the wavelet coefficients after anti-interference processing to obtain the reconstructed signal.
[0068] Through the inverse wavelet transform, synthesize the original signal from the reconstructed signal.
[0069]
[0070] After the Daubechies wavelet transform and anti-interference processing, electrocardiogram and electromyogram signals with better anti-interference performance can be obtained, retaining the main features of the signals while effectively removing the interference components.
[0071] For physiological information such as blood pressure, blood oxygen, respiration, and body temperature, use digital filtering methods to reduce noise and remove external interference.
[0072] The data anti-interference module 2 will transmit the processed physiological data to the data integration module 3.
[0073] Step 3: The data integration module 3 receives the data from the data anti-interference module 2 and is responsible for integrating the data collected by different sensors and ensuring that they have the same time reference. The data integration module 3 performs data alignment and time synchronization processing to ensure that the data from different sensors is consistent in time, so as to ensure that the data from different sensors has different sampling rates or timing characteristics, and the system can also perform unified processing and analysis on the data.
[0074] Align the 7 time series X1, X2, …, X7 of electroencephalogram, electrocardiogram, electromyogram, blood pressure, blood oxygen, body temperature, and respiration data. The lengths of each sequence are N1, N2, …, N7 respectively. Perform dynamic time warping (DTW) alignment on these 7 time series, specifically including:
[0075] First, initialize the dynamic programming matrix to obtain a multi-dimensional dynamic programming matrix D0 of N1×N2×…×N7.
[0076] Initialize the boundary conditions of the multi-dimensional dynamic programming matrix D0 so that the data points in the first dimension and the first data point in each dimension satisfy the boundary conditions, usually initialized to infinity or zero.
[0077] D0(0, i2,..., i7) = ∞, D0(i1, 0,..., i7) = ∞,...
[0078] Among them, D0(i1, i2,..., i7) represents the distance between the first i1 elements of sequence X1, the first i2 elements of sequence X2,..., and the first i7 elements of sequence X7.
[0079] Gradually fill the elements in the multi-dimensional dynamic programming matrix D0 until the best matching path is found, and obtain the dynamic programming matrix D. Specifically, for each element D(i1, i2,…, i7) in the sequence, calculate its value as:
[0080]
[0081] Among them, represents the distance between the i1-th element of sequence X1, the i2-th element of sequence X2,..., and the i7-th element of sequence X7, and the Euclidean distance is adopted.
[0082] According to the dynamic programming matrix D, starting from the point where the last sequence ends in the time dimension, backtrack according to the principle of the minimum path to obtain the best matching path representing the chronological correspondence of the physiological data. The specific process is as follows:
[0083] (1) Start from the point where the last sequence in the dynamic programming matrix ends in the time dimension.
[0084] (2) According to the principle of the minimum path, the principle of the minimum path means selecting the path with the minimum cumulative cost in the dynamic programming matrix, and selecting the adjacent point with the minimum cumulative cost as the next backtracking point.
[0085] (3) Repeat step (2) until backtracking to the starting point.
[0086] (4) According to the points selected during the backtracking process, obtain the best matching path.
[0087] Obtain the physiologic data after time series alignment based on the chronological order of the best matching path.
[0088] Table 17 aligned physiologic data (the first 0.1 seconds)
[0089]
[0090]
[0091] Step 4: The data analysis and interaction module 4 is the core part of the system, responsible for performing real-time analysis and processing on the collected physiological data, and providing a user interface for interaction and feedback.
[0092] In this step, the physiological data after time series alignment is analyzed using the deep learning algorithm LSTM to obtain physiological information characterizing the human health condition. At the same time, the system may also provide a user interface that enables the user to view the collected physiological data and analysis results in real time and perform corresponding operations and decisions.
[0093] The above-described embodiments merely represent several implementation manners of the present invention. The description thereof is relatively specific and detailed, but it should not be construed as a limitation on the scope of the invention patent. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can still be made, and these all fall within the protection scope of the present invention. Therefore, the protection scope of the present invention patent shall be subject to the appended claims.
Claims
1. A method for acquiring multi-source physiological information, characterized in that: include: Acquire multiple physiological data with time series differences collected by corresponding sensors at different parts of the human body; Construct dynamic programming matrix D; Input the physiological data into the dynamic programming matrix D to obtain the best matching path representing the corresponding arrangement of the physiological data in time sequence; Acquire the physiological data after time sequence alignment based on the time sequence of the best matching path, analyze the physiological data after time sequence alignment, and obtain the physiological information representing the health status of the human body; Wherein, constructing the dynamic programming matrix D includes: Initialize the dynamic programming matrix to obtain a multidimensional dynamic programming matrix D0 containing multiple dimensions N1×N2×…×N7; initialize the boundary conditions of the multidimensional dynamic programming matrix D0 so that the data points in the first dimension and the first data point in each dimension meet the boundary conditions, and gradually fill in the elements in the multidimensional dynamic programming matrix D0 to obtain the dynamic programming matrix D.
2. A multi-source physiological information acquisition method as claimed in claim 1, characterized in that: The physiological data includes: electroencephalogram data, electrocardiogram data, electromyography data, blood pressure data, blood oxygen data, body temperature data and respiratory data.
3. A multi-source physiological information acquisition method as claimed in claim 2, characterized in that: Before the physiological data is input into the dynamic programming matrix D, the physiological data is subjected to noise filtering and anti-interference processing, including: For EEG data, the bandpass filtering method was used to retain the data of 0.1-45Hz, and independent component analysis was performed to remove the electrooculogram, electromyography and electrocardiogram components contained in the EEG signal; For ECG data and EMG data, wavelet transform is used to decompose physiological signals into sub-signals of different frequency ranges, and filtering is performed according to the energy of the sub-signals; The blood pressure data, blood oxygen data, body temperature data and respiratory data are denoised through digital filtering.
4. A multi-source physiological information acquisition method as claimed in claim 3, characterized in that: The method of using wavelet transform to decompose the physiological signal into sub-signals in different frequency ranges specifically includes: The ECG data and EMG data x(t) are matched with the Daubechies wavelet basis function to obtain the wavelet coefficients of each scale, and the formula is: For each scale wavelet coefficient W j,k , if the absolute value is less than the set threshold T, it is set to zero; if the absolute value is greater than the set threshold T, the original value is retained. The formula is: The wavelet coefficients after anti-interference processing are inversely transformed to obtain the reconstructed signal; the reconstructed signal is synthesized into the original signal through inverse wavelet transform, and the formula is: Among them, W j,k represents the wavelet coefficient of the jth scale and the kth position, ψ j,k (t) represents the Daubechies wavelet basis function, x(t) represents the physiological signal, represents the reconstructed signal.
5. The multi-source physiological information acquisition method according to claim 1, characterized in that: The boundary conditions of the multidimensional dynamic programming matrix D0 are initialized, which specifically includes: D0(i1,i2,...,i7) represents the distance between the first i1 elements of sequence X1, the first i2 elements of sequence X2, ... the first i7 elements of sequence X7; Initialize the boundary conditions of the multidimensional dynamic programming matrix D0 so that the first dimension and the first element of each dimension meet the boundary conditions, usually initialized to infinity or zero: D0(0,i2,...,i7)=∞,D0(i1,0,...,i7)=∞,... Wherein, D(i1, i2, ..., i7) represents the distance between the first i1 elements of sequence X1, the first i2 elements of sequence X2, ... the first i7 elements of sequence X7.
6. The multi-source physiological information acquisition method according to claim 5, characterized in that: The formula for gradually filling the elements in the multidimensional dynamic programming matrix D0 is: For each element D(i1,i2,…,i7) in the sequence, D(i1,i2-1,…,i7),…,D(i1,i2,…,i7-1)} in, Represents the distance between the i1th element of sequence X1, the i2th element of sequence X2, ..., the i7th element of sequence X7, using Euclidean distance.
7. A multi-source physiological information acquisition system, characterized in that: include: The multi-source physiological information acquisition module 1 is used to acquire multiple physiological data with time sequence differences collected by corresponding sensors at different parts of the human body; Data integration module 3, used to construct a dynamic programming matrix; The data analysis and interaction module 4 is used to input the physiological data into the dynamic programming matrix D to obtain the best matching path representing the corresponding arrangement of the physiological data in time sequence; The physiological data after time alignment is obtained based on the time sequence of the best matching path, and the physiological data after time alignment is analyzed to obtain physiological information representing the health status of the human body.
8. A multi-source physiological information acquisition system as claimed in claim 7, characterized in that: It also includes a data anti-interference module 2 for performing noise filtering and anti-interference processing on physiological data.