Signal quality detection method and signal detection device

Abstract

The present invention discloses a signal quality detection method and a signal detection device thereof, which are used for an ear-hook type physiological measurement device. The signal quality detection method comprises: receiving a sensing signal by the ear-hook type physiological measurement device; filtering the sensing signal to generate a pre-processing signal; calculating a physiological index according to the pre-processing signal; and calculating a similarity between a red light AC component and an infrared light AC component in the pre-processing signal and a plurality of correlation coefficients of the red light AC component, and generating a trust index of the physiological index accordingly; wherein the trust index indicates one of a plurality of signal qualities.

Description

Technical Field

[0001] The present invention relates to a signal quality detection method and a signal detection device for an ear-hook type physiological measurement device, and in particular to a signal quality detection method and a signal detection device with higher accuracy that can instantly indicate a confidence index of a detected physiological index to assist a user in adjusting a wearing position. Background Art

[0002] Photoplethysmography (PPG) can non-invasively detect changes in blood volume, and then measure physiological indicators such as heart rate and blood oxygen concentration. Simply put, the sensing elements of the photoplethysmography method are mainly composed of light-emitting diodes and photodiodes. The light-emitting diodes emit light, and the photodiodes receive light signals passing through the blood vessels. Its alternating current (AC) component reflects the changes in blood caused by heart beats, while the direct current (DC) component reflects the unchanged part of light absorption by subcutaneous tissue, venous blood, etc. Among them, the optical signal measured by the photoplethysmography method is extremely susceptible to external noise interference. Low-frequency noise mostly comes from movement and breathing interference, which will cause amplitude drift in the DC component, and high-frequency noise mostly comes from ambient light sources, which will cause disturbances in the AC component.

[0003] Most of the current wearable devices for measuring blood oxygen concentration are finger-clip type blood oxygen pulse meters, which usually cover the object to be measured to isolate external light. Generally, blood oxygen measurement devices can be selected as clip-on or patch-type by replacing the probe. After comparing the measurement positions of fingers, earlobes, foreheads and toes, the current literature concludes that the probe on the earlobe has a relatively high accuracy. However, the existing ear-hook devices face two major challenges when measuring blood oxygen concentration. First, it is impossible to directly visually see whether there is a bad wearing situation. The user's ear cavity is prone to scattering light sources or receiving noise mixed with external light sources in the photodiode due to the lack of a fit device sensor. Secondly, the sensor uses a reflective optical module, which cannot completely isolate external noise. In this case, if the quality of the red light waveform signal used for blood oxygen concentration measurement is poor, it will greatly affect the accuracy of blood oxygen concentration calculation; in addition, physiological data will also fluctuate in value due to various factors, resulting in inaccurate records by users.

[0004] In view of this, the prior art needs to be improved. Summary of the invention

[0005] Therefore, the main purpose of the present invention is to provide a signal quality detection method and a signal detection device with higher accuracy that can instantly indicate the confidence index of the detected physiological index to assist the user in adjusting the wearing position.

[0006] The present invention discloses a signal quality detection method for an ear-hook type physiological measurement device. The signal quality detection method comprises: receiving a sensing signal by the ear-hook type physiological measurement device; filtering the sensing signal to generate a pre-processing signal; calculating a physiological index according to the pre-processing signal; and calculating a similarity between a red light AC component and an infrared light AC component in the pre-processing signal and a plurality of correlation coefficients of the red light AC component, and generating a trust index of the physiological index accordingly; wherein the trust index indicates one of a plurality of signal qualities.

[0007] The present invention further discloses a signal detection device for an ear-hook physiological measurement device, the signal detection device comprising: an input module for receiving a sensing signal from the ear-hook physiological measurement device; a pre-processing module for filtering the sensing signal to generate a pre-processing signal; a processor for executing a program; and a storage unit coupled to the processor for storing the program; wherein the program is used to instruct the processor to execute the following steps: calculating a physiological index according to the pre-processing signal; and calculating a similarity between a red light AC component and an infrared light AC component in the pre-processing signal and a plurality of correlation coefficients of the red light AC component, and generating a trust index of the physiological index accordingly; wherein the trust index indicates one of a plurality of signal qualities. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1 FIG. 1 is a schematic diagram of a signal detection device according to an embodiment of the present invention.

[0009] Figure 2 Schematic diagram of a similarity determination process according to a first embodiment of the present invention.

[0010] Figure 3 Schematic diagram of a red light cross-correlation judgment process according to an embodiment of the present invention.

[0011] Figure 4 It is a waveform diagram showing a high signal quality indicated by a trust index according to an embodiment of the present invention.

[0012] Figure 5 It is a waveform diagram showing a medium signal quality indicated by a trust index according to an embodiment of the present invention.

[0013] Figure 6 It is a waveform diagram showing that the trust index indicates low signal quality according to an embodiment of the present invention.

[0014] Figure 7 Schematic diagram of a trust index determination process according to an embodiment of the present invention.

[0015] Figure 8 FIG. 4 is a schematic diagram of a signal quality detection process according to an embodiment of the present invention.

[0016] The reference numerals are described as follows:

[0017] 10: Signal detection device

[0018] 100: Input module

[0019] 102: Pre-processing module

[0020] 104: Blood oxygen concentration calculation module

[0021] 106: Signal quality detection module

[0022] 108: Output module

[0023] 110: Bandpass filter

[0024] 112: Low pass filter

[0025] 20,30,70,80: Steps

[0026] 800~810:Process

[0027] SEN: Sense Signal

[0028] AC Red : Red light AC component

[0029] AC IR : Infrared light AC component

[0030] DC Red : Red light DC component

[0031] DC IR : Infrared light DC component

[0032] SPO2: blood oxygen concentration

[0033] RI: Trust Index

[0034] SAM Red : Red light sampling signal

[0035] SAM IR : Infrared light sampling signal

[0036] SQI RICORR : Similarity

[0037] SQI XCORR : Mutual correlation coefficient

[0038] TEM Red : Sample signal

[0039] T1 Red ,T2 Red : Red light test signal

[0040] RI t : Current Trust Index

[0041] RI t-2 ,RI t-4 : Previous Trust Index DETAILED DESCRIPTION

[0042] Generally speaking, photoplethysmography (PPG) is used to measure blood oxygen concentration using two different wavelength spectra, such as red light and infrared light, to determine the changes in the absorption of different components of human tissue. Specifically, the contraction and relaxation of the heart will cause pulsation and cause changes in blood vessel volume. The oxygenated and unoxygenated hemoglobin in the blood will also change accordingly, which will also affect the absorption of light. Therefore, the light sensor has different light intensities received and generates a tiny AC signal. The blood oxygen wave signals received by red light and infrared light can be divided into DC components. Red 、DC IR and AC component Red , AC IR , a ratio value R can be obtained according to formula (1), and then the blood oxygen concentration can be obtained through experimental analysis of the ratio value R:

[0043]

[0044] Since blood has a high absorption rate for red light, the difference in the intensity of the reflected light is small. In other words, the amplitude of the red light is small and the noise is relatively large. The AC component of the red light is DC. Red It is easily affected by interference and the waveform signal quality is poor, resulting in errors in calculating blood oxygen concentration.

[0045] Please refer to Figure 1 , Figure 1Schematic diagram of a signal detection device 10 according to an embodiment of the present invention. The signal detection device 10 is used in an ear-hook physiological measurement device (such as a Thor ear-hook device), which contacts the ear wall and includes a photoplethysmography module for sending and receiving green light, red light and infrared light to generate a sensing signal SEN. In short, the signal detection device 10 includes an input module 100, a pre-processing module 102, a blood oxygen concentration calculation module 104, a signal quality detection module 106 and an output module 108. The input module 100 can receive the sensing signal SEN from the ear-hook physiological measurement device (such as including a transmission module for wireless communication such as Bluetooth or WiFi or transmission through a transmission line), and the pre-processing module 102 can filter the sensing signal SEN to generate a pre-processing signal. For example, the pre-processing module 102 can include a bandpass filter 110 and a low-pass filter 112. The bandpass filter 110 performs bandpass filtering on the sensing signal SEN to generate a red light AC component AC in the pre-processing signal. Red and an infrared light AC component IR , and the low-pass filter 112 performs low-pass filtering on the sensing signal SEN to generate a red light DC component DC in the pre-processing signal Red and an infrared light DC component DC IR The blood oxygen concentration calculation module 104 can calculate a physiological index according to the pre-processed signal. For example, the blood oxygen concentration calculation module 104 can calculate a physiological index according to the red light AC component AC in the pre-processed signal. Red , infrared light AC component IR , Red light DC component Red And infrared light DC component IR , calculate a blood oxygen concentration SPO2 (or other physiological indicators such as heart rate, etc.). The signal quality detection module 106 calculates the red light AC component AC Red And infrared light AC component IR The similarity and red light AC component Red The multiple correlation coefficients are used to generate a trust index RI of the physiological indicator (such as blood oxygen concentration SPO2), and the trust index RI indicates one of the multiple signal qualities. The output module 108 outputs the physiological indicator (such as blood oxygen concentration SPO2) and the trust index RI (such as image output on the screen, flashing lights or sound output on the speaker, etc.).

[0046] In this case, the ear-hook physiological measurement device can be set to transmit the sensing signal SEN every 2 seconds, so that the signal detection device 10 can update and output the blood oxygen concentration SPO2 and the trust index RI every 2 seconds, achieving very real-time measurement. In addition, taking the minimum heart rate of 40 beats per minute (BPM), each heartbeat takes at least 1.5 seconds. The 2-second sampling window ensures at least one complete cycle of the pulse wave, and there will be no difference in the calculated values ​​due to sampling different sections. Then, each component in the filtered pre-processed signal will calculate the blood oxygen concentration SPO2 and the trust index RI of this section of the sensing signal SEN respectively. The user can adjust the wearing position by judging whether the signal quality indicated by the trust index RI is high signal quality, medium signal quality, or low signal quality, and judge the blood oxygen concentration according to the trust index RI. In this way, the present invention can instantly indicate the trust index RI of the detected physiological index, thereby assisting the user to adjust the wearing position with higher accuracy.

[0047] Specifically, after the pre-processing module 102 samples the sensing signal SEN and eliminates the ambient light, the bandpass filter 110 and the low-pass filter 112 (such as a Butterworth filter that can make the frequency response curve within the frequency band flat) can be used to extract the AC component AC. Red , AC IR and DC component Red 、DC IR It is worth noting that the cutoff frequency band of the bandpass filter 110 can be designed to be 0.5-5 Hz. For example, for the extreme heart rate values ​​of 40 and 250, the heart rates are 0.667 and 4.167 Hz respectively, both of which fall within the bandpass range. Compared with the high-pass filter, the bandpass filter 110 filters out the noise of the ambient light, making the waveform smoother, thereby being able to calculate a more accurate blood oxygen concentration.

[0048] Then, the blood oxygen concentration calculation module 104 can use the spectra of different wavelengths of red light and infrared light to estimate the saturation ratio of oxygen and deoxygen in the blood vessels, and calculate the saturation ratio of oxygen and deoxygen in the blood vessels through the red light AC component. Red , infrared light AC component IR , Red light DC component Red and an infrared light DC component DC IRThe ratio value R is obtained according to formula (1) and the blood oxygen concentration is obtained through experiments. In this case, the general calculation method of the ratio value R is to use the peak and trough detection method to calculate a blood perfusion index (PI) from the detected extreme value, that is, the ratio of the AC component to the DC component. The extreme value of each pulse wave can be successfully detected through a clean signal, but the waveform interfered by noise is prone to misjudgment or omission of the correct extreme point. Therefore, another calculation method is to perform the root mean squared (RMS) and then obtain the blood perfusion index from the ratio of the AC component to the DC component; compared with only calculating the peak and trough, this method uses each point in the signal to calculate the amplitude of the average square root. The present invention sets the calculation of the AC component and the DC component every 2 seconds to reduce the impact of the average amplitude. The calculation of the blood perfusion index by the peak and trough detection method or the root mean squared method is well known to ordinary technicians in the field and will not be repeated here for the sake of brevity. Thus, the embodiment of the present invention may adopt the mean square root method to calculate the ratio value R to effectively avoid motion artifacts and noise, but in other embodiments, the peak and valley detection method may be used, but is not limited thereto.

[0049] On the other hand, since clinical testing should avoid interference from external noise as much as possible, or light leakage caused by poor wearing that affects signal quality, but users usually find it difficult to judge whether the measurement is good or bad by themselves, and a bad waveform will lead to incorrect measurement values, the signal quality detection module 106 can generate a trust index RI to classify the signal quality of the sensing signal SEN into low, medium, and high, to help users easily judge whether the current sensing signal SEN is good. If the signal quality is displayed as low, it means that the estimated blood oxygen concentration SPO2 has a very low credibility, and the user should readjust the device wearing position; if the signal quality is displayed as medium, it means that the user can fine-tune the device wearing, and the data of the measured blood oxygen concentration SPO2 can be trusted; and if the signal quality is displayed as high, the blood oxygen concentration SPO2 is accurate data with high credibility. In other embodiments, the trust index RI can also indicate one of the other number of signal qualities, so that the user can adjust the device wearing accordingly.

[0050] Specifically, the signal quality detection module 106 uses two signal quality indicators (SQI) to achieve automatic classification of waveform signal quality using a combination of different weights, namely, red light AC component and Red Multiple cross-correlation coefficients SQI XCORR And the red light AC component Red And infrared light AC component IR SQI RICORR , multiple cross-correlation coefficients SQI of red light XCORRIt is used to judge whether the signal is stable and not interfered by noise. The similarity SQI between red light and infrared light RICORR Used to identify light leakage (the red light and infrared light emitters are located at different positions of the ear-hook physiological measurement device. If the peaks and troughs of the waveforms of the received red light and infrared light components do not correspond and are similar, it means that the ear-hook physiological measurement device is not fitted properly, causing light leakage).

[0051] Please refer to Figure 2 , Figure 2 FIG. 2 is a schematic diagram of a similarity determination process 20 according to an embodiment of the present invention. Figure 2 As shown, the signal quality detection module 106 captures the red light AC component AC within a sampling period. Red And infrared light AC component IR As a red light sampling signal SAM Red and an infrared light sampling signal SAM IR (If the sampling period is 2 seconds, the red light sampling signal SAM Red And infrared light sampling signal SAM IR The length of each is 2 seconds), and then the red light sampling signal SAM is calculated Red And infrared light sampling signal SAM IR SQI RICORR When the similarity SQI RICORR When the similarity SQI is less than a similarity threshold (such as 0.7), the confidence index RI indicates a signal with low signal quality. RICORR When it is greater than the similarity threshold (such as 0.7), continue to calculate the red light AC component Red Multiple cross-correlation coefficients SQI XCORR . Similarity SQI RICORR The calculation method of (such as correlation coefficient) is shown in formula (2) (where x and y can be the red light sampling signal SAM Red And infrared light sampling signal SAM IR ):

[0052]

[0053] Please refer to Figure 3 , Figure 3 FIG. 3 is a schematic diagram of a red light cross-correlation determination process 30 according to an embodiment of the present invention. Figure 2 Similarity SQI RICORR When the similarity threshold is greater than the similarity threshold, the red light cross-correlation judgment process 30 can be performed. Figure 3 As shown, the signal quality detection module 106 captures the red light AC component AC in the previous sampling period. Red As a sample signal TEM Red(i.e., the previous red light sampling signal), and at the same time, a red light sampling signal SAM captured in the current sampling period Red Divided into red light test signal T1 Red 、T2 Red (For example, the red light sampling signal SAM Red The red light test signal T1 is divided into two equal parts with a length of 1 second. Red 、T2 Red In other embodiments, other numbers and lengths of red light test signals may be used to generate red light test signals, respectively calculating the red light test signal T1 Red 、T2 Red With the sample signal TEM Red Two mutual correlation coefficients SQI XCORR Next, the signal quality detection module 106 detects the signal quality of the two signals according to the two correlation coefficients SQI XCORR and multiple cross-correlation thresholds to generate a confidence index RI indicating one of high, medium, and low signal quality. For example, if two cross-correlation coefficients SQI XCORR If one of the two cross-correlation coefficients SQI is less than a first cross-correlation threshold value (such as 0.5), a confidence index RI is generated to indicate low signal quality. XCORR If the two cross-correlation coefficients SQI XCORR If the signal quality is higher than a second cross-correlation threshold (eg, 0.7), a confidence index RI is generated indicating high signal quality.

[0054] For example, see Figure 4 , Figure 4 FIG. 1 is a waveform diagram showing that the reliability index RI indicates high signal quality according to an embodiment of the present invention. Figure 4 As shown, Figure 4 The upper and lower left half respectively show the infrared light sampling signal HSAM1 of a subject IR And red light sampling signal HSAM1 Red , their peaks and troughs are roughly aligned and have high similarity, and the red light sampling signal HSAM1 Red The signal quality is high because the signal is less affected by noise and has a regular periodic waveform (assuming the sample signal also has a regular periodic waveform). The upper and lower right half show the infrared light sampling signal HSAM2 of another subject. IR And red light sampling signal HSAM2 Red , also has similar characteristics but with high signal quality.

[0055] On the other hand, please refer to Figure 5 , Figure 5FIG. 1 is a waveform diagram of a medium signal quality indicated by a confidence index RI according to an embodiment of the present invention. Figure 5 As shown, Figure 5 The left half shows the infrared light sampling signal MSAM1 of a subject. IR And red light sampling signal MSAM1 Red , their peaks and troughs are roughly aligned and have high similarity, but the red light sampling signal MSAM1 Red Although the signal is greatly affected by noise, the waveform still has regular periodicity and has a medium cross-correlation coefficient (assuming that the sample signal also has a regular periodic waveform), so it is a medium signal quality. The upper and lower right half show the infrared light sampling signal MSAM2 of another subject. IR And red light sampling signal MSAM2 Red , also has similar characteristics and is of medium signal quality.

[0056] On the other hand, please refer to Figure 6 , Figure 6 FIG. 1 is a waveform diagram showing that the reliability index RI indicates low signal quality according to an embodiment of the present invention. Figure 6 As shown, Figure 6 The left half shows the infrared light sampling signal LSAM1 of a subject. IR And red light sampling signal LSAM1 Red , which is disturbed by noise or light leakage, resulting in the inability to align the peaks and valleys and having low similarity, thus having low signal quality. The upper and lower parts of the right half respectively show the infrared light sampling signal LSAM2 of another subject. IR And red light sampling signal LSAM2 Red , after translation, the peaks and troughs are roughly aligned and have a high enough similarity, but the red light sampling signal MSAM1 Red The waveform is greatly affected by noise, making it non-periodic and having a low cross-correlation coefficient (assuming that the sample signal is a waveform with regular periodicity), thus indicating low signal quality.

[0057] In addition, the cross-correlation coefficient SQI XCORR The calculation method of is as follows. First, a correlation coefficient ρ XY [n] represents a test signal X and a sample signal Y at time point n (the sample signal X and the sample signal Y can be the red light test signal T1 Red 、T2 Red One of them and the sample signal TEM Red ) is as follows:

[0058]

[0059] Among them, X NORM represents the normalized 1 second test signal, Y NROMis the normalized 2-second sample signal. In order to calculate the proportional value of the correlation, it is necessary to compare the maximum possible correlation coefficient, which is the autocorrelation coefficient max(ρ YY [n]), the maximum value occurs at the position where the time displacement is 0, and the cross-correlation coefficient compared with the 1-second test signal must be adjusted to half, so it can be simplified to formula (4):

[0060]

[0061] Cross-correlation coefficient SQI XCORR It can be defined as formula (5), which represents the correlation coefficient between the test signal and the sample signal compared with the autocorrelation coefficient of the sample signal. The closer the value is to 1, the closer they are to each other, and vice versa. The two 1-second test signals must have a certain degree of similarity to be considered as signals with good signal quality. If any of the 1-second test signals is below the standard, it will be considered as a signal with poor signal quality.

[0062]

[0063] In this case, the red light AC component Red Multiple cross-correlation coefficients SQI XCORR The pulse signal periodicity characteristic is adopted to convert the current 2-second red light sampling signal SAM Red Divide into 1 second each as the current red light test signal T1 Red 、T2 Red The red light sampling signal of the first 2 seconds is used as the sample signal TEM Red , each 1 second red light test signal T1 Red 、T2 Red With the sample signal TEM Red The advantages of calculating the mutual correlation coefficient are at least as follows:

[0064] 1. Reduce the false positive rate: two sampled test signals T1 Red 、T2 Red The mutual correlation coefficient SQI XCORR Both need to be higher than the threshold, reducing the occurrence of similar bad signals (two sampling signals T1 Red 、T2 Red Both are bad and have nothing to do with the sample signal TEM Red The likelihood of similarity is very low).

[0065] 2. Reduce the impact of filtering distortion: Test signal T1 Red 、T2 RedIt will be filtered first. Shorter signals are prone to slight distortion at the edge after filtering (such as not capturing the complete waveform, inconsistent effects when the signal sampling points are different, and greater impact on the calculation results when the edge is at a place where the slope changes greatly). Use 2 test signals T1 Red 、T2 Red The sample reference avoids the situation where one test signal is more easily distorted.

[0066] 3. Avoid misjudgment of surges: The high amplitude caused by surges will cause a high correlation coefficient (the surge signal is strong and the sample signal TEM Red The double sampling signal screening ensures the filtering of the surge signal (surges are sporadic and not continuous, which can avoid misjudging a single surge as high signal quality).

[0067] It is worth noting that the above embodiment mainly calculates the red light AC component AC Red And infrared light AC component IR Similarity and red light AC component AC Red The multiple cross-correlation coefficients are used to automatically detect the waveform quality and generate the trust index RI of the physiological index, so as to assist the user to adjust the wearing position to maintain the waveform with obvious periodicity and higher accuracy. Those with ordinary knowledge in this field can make modifications or changes accordingly, but are not limited to this. For example, the above embodiment is mainly described with blood oxygen concentration SPO2, but in other embodiments, the trust index RI of other physiological indicators such as heart rate can also be indicated. The above embodiment calculates the cross-correlation coefficients with two test signals and a sample signal and then generates the trust index RI, but in other embodiments, it can also be other numbers of test signals, and the trust index RI can also indicate one of the other numbers of signal qualities. In addition, the above embodiment is calculated with a sampling signal, a sample signal of 2 seconds and two 1-second test signals, but in other embodiments, it can also be other sampling periods. When the sampling period is extended, the sampling waveform matching is more rigorous, so it is more required that the wearer wears the watch well, and the measured blood oxygen is more accurate. For example, the sensing signal SEN is usually transmitted via Bluetooth. Due to the data length limitation, Bluetooth transmits a sampling signal of 2 seconds each time. The extended test sampling is 2n seconds. The signal lengths of different sampling periods are as follows:

[0068]

[0069] In this case, different designs can be made for different user scenarios to meet actual needs. For example, for home measurement and tracking, the accuracy of the present invention is slightly inferior to that of medical-grade equipment, but it can be tracked for a long time and dynamically, which is the advantage of smart wearable devices. In addition, home use does not require instant measurement, and the accuracy of calculation can be improved by sampling longer signals as listed in the above table. The common symptom of people infected with the new coronavirus is reduced blood oxygen content, and patients cannot even detect that their own blood oxygen is low. The weakness of medical instruments is that they can only measure once. For such patients with long-term chronic hypoxia, the present invention can perform blood oxygen detection around the clock to observe trend changes rather than absolute values. Instant long-term measurement is particularly important. On the other hand, for those who engage in specific sports such as mountaineering and free diving, what is needed is convenience and immediacy. By monitoring blood oxygen saturation concentration, the physical condition can be evaluated in real time, so as to adjust the current exercise intensity. This group needs real-time feedback to track the trend of blood oxygen concentration changes, which is suitable for real-time measurement with short sampling time as listed in the above table.

[0070] On the other hand, in the above embodiment, according to the current 2-second red light sampling signal SAM Red The current red light test signal T1 Red 、T2 Red Calculate multiple cross-correlation coefficients SQI XCORR The confidence index RI of the blood oxygen concentration SPO2 is directly output. However, in other embodiments, since the confidence index RI indicates the waveform signal quality score, the reflection is sensitive and immediate, the noise caused by a slight shake will cause the original signal quality index to drop from high to low, and the repeated flashing of the index will make the user at a loss. In addition, a stable periodic waveform requires a period of sampling as a sample signal TEM Red Reference, reference error sample signal TEM Red It should not be reflected in the user interface, so in addition to the current trust index, multiple previous trust indexes calculated previously can also be referenced to determine the trust index RI.

[0071] For example, an erroneous sampling signal will cause the current confidence index to indicate low quality in the current sampling period, and will also cause the next confidence index to indicate low quality when used as a sample signal in the next sampling period. Therefore, the signal quality indicated by the confidence index of the first two signal segments can be used as a reference, and a total of three signal qualities can be counted, and the highest one can be used as the current signal quality. For details, please refer to Figure 7 , Figure 7 FIG. 1 is a schematic diagram of a trust index determination process 70 according to an embodiment of the present invention. The signal quality detection module 106 can detect the red light sampling signal SAM in the current sampling period. Red And infrared light sampling signal SAM IR, generating a current confidence index RI of blood oxygen concentration SPO2 t , and then according to the current trust index RI t and the two previous confidence indices RI generated in the two previous sampling periods t-2 , RI t-4 , generate the trust index RI, such as the current trust index RI t The previous trust index RI t-2 , RI t-4 The highest signal quality among multiple signals is output as the confidence index RI. t The previous trust index RI t-2 , RI t-4 For the generation method of , reference may be made to the above-mentioned similarity judgment process 20 and the red light cross-correlation judgment process 30 , which will not be repeated here for the sake of brevity.

[0072] In this case, the trust index judgment process 70 can optimize the user experience, avoid the direct reaction of the noise of the small shaking causing the display of the waveform signal quality to switch too quickly, and can be post-processed and optimized, and the signal quality detection indicated by the trust index RI is more stringent (for example, the current signal is good but the signal quality is low because the referenced sample signal of the previous moment is noise), and can also refer to multiple time points for statistics to reduce errors and improve the reliability of waveform signal quality judgment. In this way, compared with the strict waveform screening, the present invention adopts a more relaxed mechanism, which is also in line with the user's wearing adjustment scenario.

[0073] It is worth noting that in the above embodiment, the current trust index RI t With two previous trust indices RI t-2 , RI t-4 , and outputs the highest signal quality as the trust index RI. However, in other embodiments, the number of previous trust indexes is not limited to two, and different weights can be set for the current trust index and multiple previous trust indexes to emphasize the most recent detection results while avoiding short-term minor shakes that cause the display of signal quality to switch too quickly. For example, the weights of the current trust index and the three previous trust indexes are set to 0.4, 0.2, 0.2, and 0.2 respectively. When short-term minor shakes only affect one sampling period, the trust index RI will not indicate low signal quality. When the wearer changes from good to poor wearing, it can be found in two sampling periods that the trust index RI indicates low signal quality, allowing the user to adjust the wearer to reflect the most recent detection results.

[0074] Therefore, the signal quality detection operation of the signal detection device 10 can be summarized as a signal quality detection process 80, such as Figure 8 As shown, it includes the following steps:

[0075] Step 800: Start.

[0076] Step 802: Receive a sensing signal SEN from the ear-hook type physiological measurement device.

[0077] Step 804: Filter the sensing signal SEN to generate a pre-processing signal.

[0078] Step 806: Calculate a physiological index according to the pre-processed signal.

[0079] Step 808: Calculate the red light AC component AC in the pre-processed signal Red And infrared light AC component IR Similarity and red light AC component AC Red A plurality of cross-correlation coefficients are obtained and a confidence index RI of the physiological indicator is generated accordingly, wherein the confidence index RI indicates one of a plurality of signal qualities.

[0080] Step 810: End.

[0081] The detailed operation of the signal quality detection process 80 may refer to the relevant contents of the signal detection device 10 , which will not be described again for the sake of brevity.

[0082] In addition, the signal detection device 10 may include a processing unit and a storage unit. The processing unit may be a microprocessor or an application-specific integrated circuit (ASIC). The storage unit may be any data storage device for storing a program code, and the processing unit reads and executes the program code to execute the functions of the blood oxygen concentration calculation module 104 and the signal quality detection module 106, thereby completing each step of the signal quality detection process 80. The storage unit may be a subscriber identity module (SIM), a read-only memory (ROM), a random-access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, etc., but is not limited thereto.

[0083] In summary, the present invention calculates the red light AC component AC Red And infrared light AC component IR Similarity and red light AC component AC RedThe multiple correlation coefficients are used to automatically detect the waveform quality and generate the physiological index RI to help users adjust the wearing position to maintain the waveform with obvious periodicity and higher accuracy.

[0084] The above description is only a preferred embodiment of the present invention. All equivalent changes and modifications made according to the claims of the present invention should fall within the scope of the present invention.

Claims

1. A signal quality detection method for an ear-hook physiological measurement device, comprising the following steps: Receiving a sensing signal from the ear-hook type physiological measurement device; filtering the sensing signal to generate a pre-processing signal; Calculate a physiological index according to the pre-processed signal; and Calculating a similarity between a red light AC component and an infrared light AC component in the pre-processed signal and a plurality of correlation coefficients of the red light AC component, and generating a trust index of the physiological indicator accordingly; in, The confidence index indicates one of a plurality of signal qualities, The step of calculating the similarity between the red light AC component and the infrared light AC component in the pre-processed signal and the multiple correlation coefficients of the red light AC component and generating the trust index of the physiological indicator accordingly comprises: The red light AC component captured in the previous sampling period is used as a sample signal; Dividing a red light sampling signal captured in a current sampling period into a plurality of red light test signals; The plurality of cross-correlation coefficients between the plurality of red light test signals and the sample signal are calculated.

2. The signal quality detection method according to claim 1, wherein: The step of filtering the sensing signal to generate the pre-processing signal comprises: The sensing signal is band-pass filtered to generate the red light AC component and the infrared light AC component in the pre-processing signal.

3. The signal quality detection method according to claim 1, wherein: The steps of calculating the similarity between the red light AC component and the infrared light AC component in the pre-processed signal and the multiple correlation coefficients of the red light AC component and generating the trust index of the physiological indicator accordingly include: Capturing the red light AC component and the infrared light AC component within a sampling period as a red light sampling signal and an infrared light sampling signal; Calculating the similarity between the red light sampling signal and the infrared light sampling signal; and When the similarity is less than a similarity threshold, the confidence index is generated to indicate a low signal quality among the plurality of signal qualities.

4. The signal quality detection method as claimed in claim 1 further comprises the following steps: When the similarity is greater than a similarity threshold, the plurality of correlation coefficients of the red light AC component are calculated.

5. The signal quality detection method according to claim 1, wherein: The step of calculating the similarity between the red light AC component and the infrared light AC component in the pre-processed signal and the multiple correlation coefficients of the red light AC component and generating the trust index of the physiological indicator accordingly further includes: The confidence index is generated according to the plurality of cross-correlation coefficients and a plurality of cross-correlation thresholds to indicate the one of the plurality of signal qualities.

6. The signal quality detection method according to claim 5, wherein: The step of generating the confidence index indicating the one of the plurality of signal qualities according to the plurality of cross-correlation coefficients and the plurality of cross-correlation threshold values ​​comprises: If one of the plurality of cross-correlation coefficients is less than a first cross-correlation threshold value, the confidence index is generated to indicate a low signal quality among the plurality of signal qualities; if the plurality of cross-correlation coefficients are between the first cross-correlation threshold value and a second cross-correlation threshold value, the confidence index is generated to indicate a medium signal quality among the plurality of signal qualities; if the plurality of cross-correlation coefficients are higher than the second cross-correlation threshold value, the confidence index is generated to indicate a high signal quality among the plurality of signal qualities.

7. The signal quality detection method according to claim 5, wherein: The steps of calculating the similarity between the red light AC component and the infrared light AC component in the pre-processed signal and the multiple correlation coefficients of the red light AC component and generating the trust index of the physiological indicator accordingly include: The trust index is generated according to a current trust index generated in the current sampling period and a plurality of previous trust indexes generated in a plurality of previous sampling periods.

8. The signal quality detection method according to claim 7, wherein: The step of generating the trust index according to the current trust index generated in the current sampling period and the multiple previous trust indexes generated in the multiple previous sampling periods comprises: The current confidence index and the plurality of previous confidence indexes indicating a highest one of the plurality of signal qualities are used as the confidence index.

9. The signal quality detection method according to claim 7, wherein: The step of generating the trust index according to the current trust index generated in the current sampling period and the multiple previous trust indexes generated in the multiple previous sampling periods comprises: The trust index is generated according to a plurality of weights corresponding to the current trust index and the plurality of previous trust indexes, the current trust index and the plurality of previous trust indexes.

10. A signal detection device for an ear-hook physiological measurement device, comprising: an input module, used for receiving a sensing signal from the ear-hook type physiological measurement device; a pre-processing module, used for filtering the sensing signal to generate a pre-processing signal; a processor for executing a program; and A storage unit, coupled to the processor, for storing the program; wherein the program is used to instruct the processor to perform the following steps: Calculate a physiological index according to the pre-processed signal; and Calculating a similarity between a red light AC component and an infrared light AC component in the pre-processed signal and a plurality of correlation coefficients of the red light AC component, and generating a trust index of the physiological indicator accordingly; in, The confidence index indicates one of a plurality of signal qualities, The step of calculating the similarity between the red light AC component and the infrared light AC component in the pre-processed signal and the multiple correlation coefficients of the red light AC component and generating the trust index of the physiological indicator accordingly comprises: The red light AC component captured in the previous sampling period is used as a sample signal; Dividing a red light sampling signal captured in a current sampling period into a plurality of red light test signals; The plurality of cross-correlation coefficients between the plurality of red light test signals and the sample signal are calculated.

11. The signal detection device according to claim 10, wherein: The pre-processing module further includes a band-pass filter for performing band-pass filtering on the sensing signal to generate the red light AC component and the infrared light AC component in the pre-processing signal.

12. The signal detection device according to claim 10, wherein: The steps of calculating the similarity between the red light AC component and the infrared light AC component in the pre-processed signal and the multiple correlation coefficients of the red light AC component and generating the trust index of the physiological indicator accordingly include: Capturing the red light AC component and the infrared light AC component within a sampling period as a red light sampling signal and an infrared light sampling signal; Calculating the similarity between the red light sampling signal and the infrared light sampling signal; and When the similarity is less than a similarity threshold, the confidence index is generated to indicate a low signal quality among the plurality of signal qualities.

13. The signal detection device according to claim 10, wherein: The program further instructs the processor to perform the following steps: When the similarity is greater than a similarity threshold, the plurality of correlation coefficients of the red light AC component are calculated.

14. The signal detection device according to claim 10, wherein: The step of calculating the similarity between the red light AC component and the infrared light AC component in the pre-processed signal and the multiple correlation coefficients of the red light AC component and generating the trust index of the physiological indicator accordingly further includes: The confidence index is generated according to the plurality of cross-correlation coefficients and a plurality of cross-correlation thresholds to indicate the one of the plurality of signal qualities.

15. The signal detection device according to claim 14, wherein: The step of generating the confidence index indicating the one of the plurality of signal qualities according to the plurality of cross-correlation coefficients and the plurality of cross-correlation threshold values ​​comprises: If one of the plurality of cross-correlation coefficients is less than a first cross-correlation threshold value, the confidence index is generated to indicate a low signal quality among the plurality of signal qualities; if the plurality of cross-correlation coefficients are between the first cross-correlation threshold value and a second cross-correlation threshold value, the confidence index is generated to indicate a medium signal quality among the plurality of signal qualities; if the plurality of cross-correlation coefficients are higher than the second cross-correlation threshold value, the confidence index is generated to indicate a high signal quality among the plurality of signal qualities.

16. The signal detection device according to claim 14, wherein: The steps of calculating the similarity between the red light AC component and the infrared light AC component in the pre-processed signal and the multiple correlation coefficients of the red light AC component and generating the trust index of the physiological indicator accordingly include: The trust index is generated according to a current trust index generated in the current sampling period and a plurality of previous trust indexes generated in a plurality of previous sampling periods.

17. The signal detection device according to claim 16, wherein: The step of generating the trust index according to the current trust index generated in the current sampling period and the multiple previous trust indexes generated in the multiple previous sampling periods comprises: The current confidence index and the plurality of previous confidence indexes indicating a highest one of the plurality of signal qualities are used as the confidence index.

18. The signal detection device according to claim 16, wherein: The step of generating the trust index according to the current trust index generated in the current sampling period and the multiple previous trust indexes generated in the multiple previous sampling periods comprises: The trust index is generated according to a plurality of weights corresponding to the current trust index and the plurality of previous trust indexes, the current trust index and the plurality of previous trust indexes.

Images