Time series data processing device, time series data processing method, and computer program

Abstract

To provide a time series data processing device capable of data processing excluding arbitrariness.SOLUTION: Provided is a time series data processing device 10 including: a transformation unit 101 that transforms time series data into data of a frequency domain; a selection unit 102 that selects a predetermined frequency component from the data of the frequency domain; and an inverse transformation unit 103 that inversely transforms the frequency component selected by the selection unit 102 into data of a time domain. The selection unit 102 selects a frequency component with which a difference between the time series data and the data inversely transformed by the inverse transformation unit 103 is smaller than a predetermined threshold.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a time series data processing device, a time series data processing method, and a computer program. [Background technology]

[0002] In order to obtain accurate biometric information of a subject based on the subject's biometric data, it is necessary to remove noise superimposed on the biometric data. Therefore, a technology related to time-series data processing for removing noise from time-series data such as the subject's biometric data has been disclosed. Patent Document 1 discloses a technology that reduces the burden on the subject compared to when sleep polysomnography is used and detects heartbeat intervals with higher accuracy than conventional methods.

[0003] In detail, Patent Document 1 discloses a biological information acquisition device that calculates a first respiratory waveform based on a first interval and a vibration waveform, calculates a second respiratory waveform based on a second interval different in period from the first interval and the vibration waveform, calculates a first pulse waveform by subtracting the first respiratory waveform from the vibration waveform, and calculates a second pulse waveform by subtracting the second respiratory waveform from the vibration waveform, and calculates a heart rate variability index based on pulse waveforms with a small number of abnormal peak intervals from each pulse waveform.

[0004] In Patent Document 1, as preprocessing, a bandpass filter is used to remove components higher in frequency than normal body movement, breathing, and pulse wave periods, as well as direct current components, from the output of the microwave radar.In addition, Patent Document 1 uses a simple moving average method or a weighted moving average method when calculating the first respiratory waveform as preprocessing. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Japanese Patent Application Laid-Open No. 2017-104360 Summary of the Invention [Problem to be solved by the invention]

[0006] However, the technology disclosed in Patent Document 1 cannot determine whether necessary data is included in the frequency band components removed by the band-pass filter, or whether noise is included in the components that pass through the band-pass filter. Furthermore, the technology disclosed in Patent Document 1 cannot determine which frequency components have been manipulated using only a moving average. Furthermore, the technology disclosed in Patent Document 1 relies on the analyst's discretion for preprocessing, which is prone to arbitrariness, potentially leading to erroneous conclusions.

[0007] The present invention has been made in view of the above points, and aims to provide a time-series data processing device, a time-series data processing method, and a computer program that are capable of processing data without arbitrariness. [Means for solving the problem]

[0008] In order to achieve the above object, a time series data processing device according to one aspect of the present invention comprises a conversion unit that converts time series data into frequency domain data, a selection unit that selects predetermined frequency components from the frequency domain data, and an inverse conversion unit that inversely converts the frequency components selected by the selection unit into time domain data, wherein the selection unit selects frequency components such that the difference between the time series data and the data after inverse conversion by the inverse conversion unit is smaller than a predetermined threshold.

[0009] The selection unit may select frequency components that make the data after the inverse transformation by the inverse transformation unit self-consistent.

[0010] The selection unit may determine that the data after inverse transformation by the inverse transformation unit has become self-consistent when a difference between data after inverse transformation of a first frequency component and data after inverse transformation of a second frequency component including the first frequency component becomes smaller than a predetermined threshold.

[0011] The time-series data may be a biological signal or a signal based on a biological signal.

[0012] In order to achieve the above object, a time series data processing method according to another aspect of the present invention includes a processor that converts time series data into frequency domain data, selects predetermined frequency components from the frequency domain data, inversely converts the selected frequency components into time domain data, and selects frequency components such that the difference between the time series data and the data after the inverse conversion is smaller than a predetermined threshold.

[0013] In order to achieve the above object, a computer program according to another aspect of the present invention causes a computer to perform processes of transforming time series data into frequency domain data, selecting predetermined frequency components from the frequency domain data, inversely transforming the selected frequency components into time domain data, and selecting frequency components such that the difference between the time series data and the data after the inverse transformation is smaller than a predetermined threshold. [Effects of the Invention]

[0014] According to the present invention, it is possible to provide a time-series data processing device, a time-series data processing method, and a computer program that are capable of processing data without arbitrariness. [Brief explanation of the drawings]

[0015] [Figure 1] 1 is a diagram illustrating a schematic configuration of a time-series data processing system according to an embodiment. [Figure 2] FIG. 2 is a block diagram illustrating an example of a hardware configuration of a time-series data processing device. [Figure 3] FIG. 2 is a block diagram illustrating an example of a functional configuration of a time-series data processing device. [Figure 4] 1 is a flowchart showing the flow of time-series data processing by the time-series data processing device. [Figure 5] 1 is a graph showing an example of the convergence process of time-series data, for example, electroencephalogram data. [Figure 6]FIG. 2 is a diagram showing an example of pulse wave data, which is an example of time-series data. [Figure 7] FIG. 10 is a diagram showing an example of electroencephalogram data, which is an example of time-series data. [Figure 8] This is a graph in which the original pulse wave data and the time series data reconstructed based on frequency components whose errors are below a threshold are superimposed. [Figure 9] This is a graph that superimposes the original electroencephalogram data and time series data reconstructed based on frequency components whose errors are below a threshold. DETAILED DESCRIPTION OF THE INVENTION

[0016] An example of an embodiment of the present invention will be described below with reference to the drawings. The same reference numerals are used throughout the drawings to designate identical or equivalent components and parts. The dimensional proportions of the drawings are exaggerated for illustrative purposes and may differ from the actual proportions.

[0017] FIG. 1 is a diagram showing a schematic configuration of a time-series data processing system according to this embodiment.

[0018] The time series data processing system shown in FIG. 1 includes a biological signal measuring device 1 that measures a biological signal, which is an example of time series data, and a time series data processing device 10 that processes the biological signal measured by the biological signal measuring device 1.

[0019] The biological signals measured by the biological signal measuring device 1 include, for example, heart rate, electrocardiogram, magnetocardiogram, electroencephalogram, electroencephalogram, pulse wave, body temperature, blood pressure, electromyogram, skin potential, or signals processed from these, but the present invention is not limited to these examples. Examples of signals based on biological signals include signals that combine multiple of the above-mentioned biological signals, or numerical values ​​calculated based on the above-mentioned biological signals. The signal based on a biological signal may also be a signal that combines a biological signal with another signal.

[0020] The biosignal measured by the biosignal measurement device 1 may have noise superimposed on it. For example, if the biosignal is based on a pulse wave, it may include not only the pulse wave of the subject, but also vibrations or noise caused by the subject's body movement or breathing. When noise is superimposed on the biosignal, accurate processing cannot be performed. Therefore, the noise superimposed on the biosignal must be removed before processing the biosignal.

[0021] Therefore, the time-series data processing device 10 performs processing on the biosignal sent from the biosignal measuring device 1 to remove noise superimposed on the biosignal, and then performs processing related to the biosignal. When removing noise superimposed on the biosignal, the time-series data processing device 10 according to this embodiment converts the biosignal into frequency-domain data and selects predetermined frequency components from the frequency-domain data. The time-series data processing device 10 then inversely converts the selected frequency components into time-domain data and selects frequency components for which the difference between the biosignal and the inversely converted data is smaller than a predetermined threshold, thereby removing the noise superimposed on the biosignal sent from the biosignal measuring device 1.

[0022] The time-series data processing device 10 processes the biosignal after removing noise superimposed on the biosignal sent from the biosignal measuring device 1. The processing of the biosignal performed by the time-series data processing device 10 includes, for example, displaying a waveform based on the biosignal, and displaying heart rate, body temperature, blood pressure, and other bioinformation.

[0023] The time series data processing device 10 performs conversion to frequency domain data and inverse conversion in this way, and selects frequency components such that the difference between the biological signal and the data after inverse conversion is smaller than a predetermined threshold, thereby enabling data processing without arbitrariness.

[0024] The time-series data processing device 10 can be applied to devices in general that handle time-series data, such as servers on a cloud network, personal computers, smartphones, smart watches, smart bands, wearable devices, medical devices, diagnostic devices, etc. The time-series data handled by the time-series data processing device 10 is not limited to biological signals, and any type of data can be used as long as it is continuous over time.

[0025] FIG. 2 is a block diagram showing an example of the hardware configuration of the time-series data processing device 10. As shown in FIG.

[0026] 2, the time-series data processing device 10 includes a CPU (Central Processing Unit) 11, a ROM (Read Only Memory) 12, a RAM (Random Access Memory) 13, a storage 14, an input unit 15, a display unit 16, and a communication interface (I / F) 17. Each component is connected to each other via a bus 19 so as to be able to communicate with each other.

[0027] The CPU 11 is a central processing unit that executes various programs and controls each component. That is, the CPU 11 reads a program from the ROM 12 or the storage 14 and executes the program using the RAM 13 as a work area. The CPU 11 controls the above components and performs various arithmetic processing in accordance with the program recorded in the ROM 12 or the storage 14. In this embodiment, the ROM 12 or the storage 14 stores a time-series data processing program that removes noise superimposed on a biosignal sent from the biosignal measuring device 1.

[0028] The ROM 12 stores various programs and various data. The RAM 13 temporarily stores programs or data as a working area. The storage 14 is configured with a storage device such as an HDD (Hard Disk Drive), SSD (Solid State Drive), or flash memory, and stores various programs including the operating system and various data.

[0029] The input unit 15 includes a pointing device such as a mouse and a keyboard, and is used to perform various inputs.

[0030] The display unit 16 is, for example, a liquid crystal display, and displays various information. The display unit 16 may also function as the input unit 15 by adopting a touch panel system.

[0031] The communication interface 17 is an interface for communicating with other devices, and uses standards such as Ethernet (registered trademark), FDDI, and Wi-Fi (registered trademark).

[0032] When executing the above time series data processing program, the time series data processing device 10 uses the above hardware resources to realize various functions. The functional configuration realized by the time series data processing device 10 will be described.

[0033] Next, the functional configuration of the time-series data processing device 10 will be described.

[0034] FIG. 3 is a block diagram showing an example of the functional configuration of the time-series data processing device 10.

[0035] 3, the time series data processing device 10 has, as functional components, a conversion unit 101, a selection unit 102, an inverse conversion unit 103, and a data processing unit 104. Each functional component is realized by the CPU 11 reading and executing a time series data processing program stored in the ROM 12 or the storage 14.

[0036] The converter 101 converts the biological signal sent from the biological signal measuring device 1 into a frequency domain. Specifically, if the biological signal sent from the biological signal measuring device 1 is x(t) (t=[1:T]), the converter 101 converts the biological signal into a frequency domain by a Fourier transform F[x(t)].

[0037] The selection unit 102 selects a predetermined frequency component from the data in the frequency domain converted by the conversion unit 101. Specifically, the selection unit 102 selects the frequency components [m: T-m] of the Fourier transform F[x(t)] of the biological signal. The frequency components [m: T-m] selected by the selection unit 102 are denoted as F m , , , , ,

[0040] , , , , m ,

[0039] ,

[0041] , , , [x(t)]. Here, m is a value from 1 to T / 2. And the selection unit 102 selects the frequency component F m [x(t)] such that the difference between the data inverse-transformed by the inverse transformation unit 103 described later and the biological signal sent from the biological signal measurement device 1 is smaller than a predetermined threshold. The details of the selection method of the frequency component F m [x(t)] will be described later.

[0038] The inverse transformation unit 103 performs an inverse transformation of the data of the frequency component selected by the selection unit 102 into data in the time domain. Specifically, the inverse transformation unit 103 performs an inverse Fourier transform F m [x(t)] on the frequency component F -1 [F m [x(t)]]. When selecting the frequency component F m [x(t)], the selection unit 102 evaluates the error E(m) defined by the mean square error shown in Equation (1).

[0039] [Equation]

[0040] The selection unit 102 increases m from 1 to T / 2. When E(m) < C, it determines that the time series data has become self-collision-free, and adopts the time series data x m (t) that has been noise-removed. C is a threshold for determination. Being self-collision-free means that the value regarding the difference between the original time series data and the data after inverse transformation converges to less than a predetermined threshold C. That is, the selection unit 102 may select the frequency component so that the time series data reconstructed by the inverse transformation unit 103 becomes self-collision-free.

[0041] The data processing unit 104 performs data processing on the time series data x m (t) selected by the selection unit 102 such that E(m) < C. The data processing unit 104 performs processes such as display of waveforms based on biological signals, display of biological information such as heart rate, body temperature, blood pressure, and the like.

[0042] By having such a configuration, the time series data processing device 10 can perform conversion to and from frequency domain data, and select frequency components for which the difference between the biological signal and the data after inverse conversion is smaller than a predetermined threshold value, thereby enabling data processing that excludes arbitrariness. The time series data processing device 10 can guarantee the accuracy required for time series analysis by limiting the target frequency of the time series data and performing error evaluation in the time domain.

[0043] In this embodiment, Fourier transform is used for the conversion of data from the time domain to the frequency domain, but the present invention is not limited to such an example. The time series data processing device 10 may use any conversion method such as Laplace transform, wavelet transform, etc. for the conversion of data from the time domain to the frequency domain.

[0044] Next, the operation of the time series data processing device 10 will be described.

[0045] FIG. 4 is a flowchart showing the flow of time series data processing by the time series data processing device 10. The CPU 11 reads out the time series data processing program from the ROM 12 or the storage 14, expands it in the RAM 13, and executes it, thereby performing time series data processing. The time series data processing shown in FIG. 4 is preprocessing performed prior to data processing on the biological signal x(t) sent from the biological signal measurement device 1, and is processing for removing noise superimposed on the biological signal x(t).

[0046] First, in step S101, the CPU 11 converts the biological signal x(t) sent from the biological signal measurement device 1 into data in the frequency domain by Fourier transform F[x(t)].

[0047] Subsequently, in step S102, the CPU 11 selects a certain frequency component [m:T-m] from the data in the frequency domain. The CPU 11 first selects the frequency component [1:T-1] with m = 1. Let the frequency component selected by the CPU 11 in this step S102 be F1[x(t)].

[0048] Subsequently, in step S103, the CPU 11 performs an inverse Fourier transform on the frequency component F m [x(t)] to obtain F -1 [F m [x(t)]. When m = 1, the CPU 11 performs an inverse Fourier transform on the frequency component F1[x(t)] to obtain F -1 [F1[x(t)]].

[0049] Subsequently, in step S104, the CPU 11 obtains E(m) by using the above formula (1) and evaluates E(m). Specifically, the CPU 11 evaluates E(m) based on whether E(m) is less than a predetermined evaluation error C.

[0050] Subsequently, in step S105, the CPU 11 determines whether E(m) < C. In other words, the CPU 11 determines whether the time-series data reconstructed by the inverse Fourier transform is self-collision-free. As a result of the determination in step S105, if E(m) < C (step S105; Yes), the CPU 11 ends the series of processes. On the other hand, as a result of the determination in step S105, if E(m) < C is not satisfied (step S105; No), then in step S106, the CPU 11 increases m by 1 and returns to the frequency component selection process in step S102.

[0051] By having such processing, the time-series data processing device 10 can perform conversion to and inverse conversion from data in the frequency domain, and by selecting a frequency component for which the difference between the biological signal and the data after inverse conversion is smaller than a predetermined threshold value, data processing that eliminates arbitrariness becomes possible.

[0052] Next, the effects of the time-series data processing device 10 will be described.

[0053] FIG. 5 is a graph showing an example of the convergence process of time-series data, for example, electroencephalogram data. In the graph of FIG. 5, the vertical axis represents the root mean square (RMS), that is, E(m) obtained by Equation (1), and the horizontal axis represents the number of steps of the self-consistent field calculation, that is, the value of m. As shown in FIG. 5, it can be seen that as the value of m increases, the value of E(m) decreases.

[0054] FIG. 6 is a diagram showing an example of pulse wave data, which is an example of time-series data. The time-series data processing device 10 can change the selection range of the pulse wave data by changing the value of m. In the example of FIG. 6, it is assumed that m = 250 and E(m) < C. In this case, data in the range of approximately 166 Hz to 250 Hz and 750 to 834 Hz can be extracted as actual pulse wave data.

[0055] FIG. 7 is a diagram showing an example of electroencephalogram data, which is an example of time-series data. The time-series data processing device 10 can change the selection range of the electroencephalogram data by changing the value of m. In the example of FIG. 7, it is assumed that m = 250 and E(m) < C. In this case, data in the range of approximately 166 Hz to 250 Hz and 750 to 834 Hz can be extracted as actual electroencephalogram data.

[0056] FIG. 8 is a graph obtained by superimposing the original pulse wave data and the time-series data reconstructed based on the frequency components adopted at the time when m = 250 and E(m) < C. In FIG. 8, "original" is the original pulse wave data, and "smoothed" is the reconstructed time-series data. Thus, it can be seen that the time-series data reconstructed based on the frequency components when E(m) < C is approximated to the original pulse wave data, and the noise components superimposed on the original pulse wave data are removed.

[0057] FIG. 9 is a graph in which the original electroencephalogram data and the time-series data reconstructed based on the frequency components adopted at the time when E(m) < C and m = 250 are superimposed. In FIG. 9, "original" is the original electroencephalogram data, and "smoothed" is the reconstructed time-series data. Thus, it can be seen that the time-series data reconstructed based on the frequency components for which E(m) < C approximates the original electroencephalogram data and that the noise components superimposed on the original electroencephalogram data are removed.

[0058] From the above, the time-series data processing apparatus 10 according to the present embodiment can perform stable noise removal without depending on the discretion of an analyst in preprocessing such as time-series causal analysis by clearly handling a target frequency band in the frequency domain and directly evaluating an error in the time domain.

[0059] Note that in each of the above embodiments, the time-series data processing executed by the CPU by reading software (program) may be executed by various processors other than the CPU. Examples of the processor in this case include a PLD (Programmable Logic Device) whose circuit configuration can be changed after manufacture, such as an FPGA (Field-Programmable Gate Array), and a dedicated electric circuit such as a processor having a circuit configuration dedicated to executing a specific process, such as an ASIC (Application Specific Integrated Circuit). Further, the time-series data processing may be executed by one of these various processors, or may be executed by a combination of two or more processors of the same type or different types (for example, a plurality of FPGAs, a combination of a CPU and an FPGA, etc.). Further, the hardware structure of these various processors is, more specifically, an electric circuit combining circuit elements such as semiconductor elements.

[0060] In addition, in each of the above embodiments, the time-series data processing program is described as being pre-stored (installed) in a ROM or storage, but this is not limiting. The program may be provided in a form recorded on a non-transitory recording medium such as a CD-ROM (Compact Disk Read Only Memory), a DVD-ROM (Digital Versatile Disk Read Only Memory), or a USB (Universal Serial Bus) memory. The program may also be downloaded from an external device via a network.

[0061] Although the embodiments of the present invention have been described in detail above with reference to the accompanying drawings, the technical scope of the present invention is not limited to these examples. It is clear that a person skilled in the art of the present invention can conceive of various modifications or alterations within the scope of the technical idea described in the claims, and it is understood that these various modifications or alterations also fall within the technical scope of the present invention. [Explanation of symbols]

[0062] 1. Biosignal measurement device 10 Time series data processing device

Claims

1. a transformer for transforming time series data into frequency domain data; a selector that selects predetermined frequency components from the frequency domain data so that the frequency range narrows stepwise from the low frequency side and the high frequency side; an inverse transform unit that inversely transforms the frequency components selected by the selector into time domain data; Equipped with the selection unit selects a first frequency component when a difference between the time series data and data after inverse transformation by the inverse transformation unit converges to less than a predetermined threshold, that is, when a difference between data after inverse transformation of a first frequency component, which is a frequency component for which data after inverse transformation by the inverse transformation unit is self-consistent, and data after inverse transformation of a second frequency component including the first frequency component, becomes smaller than a predetermined threshold. Time series data processing device.

2. The time-series data processing device according to claim 1 , wherein the time-series data is a biological signal or a signal based on a biological signal.

3. The processor: Convert time series data into frequency domain data, Selecting predetermined frequency components from the frequency domain data so that the frequency range narrows stepwise from the low frequency side and the high frequency side; transforming the selected frequency components back into time domain data; When a difference between the time series data and the data after inverse transformation has converged to less than a predetermined threshold, that is, when a difference between data after inverse transformation of a first frequency component, which is a frequency component for which the data after inverse transformation by the inverse transformation is self-consistent, and data after inverse transformation of a second frequency component including the first frequency component, becomes smaller than a predetermined threshold, the first frequency component is selected. A time series data processing method that performs processing.

4. On the computer, Convert time series data into frequency domain data, selecting predetermined frequency components from the frequency domain data so that the frequency range narrows stepwise from the low frequency side and the high frequency side; transforming the selected frequency components back into time domain data; When a difference between the time series data and the data after inverse transformation has converged to less than a predetermined threshold, that is, when a difference between data after inverse transformation of a first frequency component, which is a frequency component for which the data after inverse transformation by the inverse transformation is self-consistent, and data after inverse transformation of a second frequency component including the first frequency component, becomes smaller than a predetermined threshold, the first frequency component is selected. A computer program that executes a process.

Images