Multi-spectral correction method for wearable blood oxygen

Abstract

The invention provides a multispectral correction method for wearable blood oxygen, which comprises the following steps of: 1, setting the same distance between three-color light and a light sensor, and alternately turning on the three-color light; 2, the optical sensor obtains three paths of synchronizing signal waveforms, and three paths of signals are extracted respectively; step 3, taking the green light waveform as a reference, calibrating wave crest and wave trough positions, and obtaining time axis coordinates of the wave crest and the wave trough; step 4, finding corresponding red light and infrared light waveform periods according to the time axis projection coordinates; 5, finding out a plurality of alternative wave crest and wave trough amplitudes in the red light and infrared light periods, forming a group by two wave troughs and one wave crest, fitting a waveform function of each pulse wave period according to the green light waveform, scaling down the green light waveform function in equal proportion, comparing the green light waveform function with the actual red light and infrared light waveforms, calculating an error, and determining the pulse wave period according to the error. Selecting a group with the minimum error as final peak and trough feature points; and step 6, calculating a blood oxygen value by using the group of wave crest and wave trough feature points. The wearable blood oxygen detection device has the beneficial effects that the precision of wearable blood oxygen detection is improved, and accurate blood oxygen measurement can be realized at multiple parts, so that all-weather blood oxygen data monitoring is realized.

Description

Technical Field

[0001] The invention relates to the field of medical care information technology, in particular to a multi-spectral correction method for wearable blood oxygen. Background Art

[0002] In order to measure accurate blood oxygen data, existing wearable blood oxygen detection technologies are mostly finger-clip detection, which is not comfortable to wear for a long time and difficult to achieve all-weather blood oxygen monitoring. Because the skin in areas such as the wrist is thicker, the response of red light and infrared light to hemoglobin becomes poor, and the accuracy of blood oxygen detection equipment is not enough. Summary of the invention

[0003] The present invention proposes a multi-spectral correction algorithm for wearable blood oxygen, which increases the accuracy of blood oxygen based on the existing spectral analysis to obtain blood oxygen. This is achieved through the following technical solutions.

[0004] A multi-spectral correction method for wearable blood oxygen, comprising: Step 1. Set the three-color lights to be at the same distance from the light sensor and turn them on alternately at a frequency of 100 Hz. Step 2: The optical sensor obtains three-way synchronous signal waveforms and extracts three-way signals respectively; Step 3: Using the green light waveform as a reference, calibrate the peak and trough positions, and obtain the time axis coordinates of the peak and trough; Step 4: Find the corresponding red light and infrared light waveform periods according to the time axis projection coordinates; Step 5, find multiple candidate peak and trough amplitudes in the red light and infrared light cycles, two troughs and one peak as a group, fit the waveform function of each pulse wave cycle according to the green light waveform, proportionally reduce the green light waveform function, compare it with the actual red light and infrared light waveforms, calculate the error, and select the group with the smallest error as the final peak and trough feature points; Step 6: Calculate the blood oxygen value using the group of peak and trough feature points.

[0005] Furthermore, projecting the time axis coordinates of the peaks and troughs onto the red and infrared waveforms can solve the problem of misjudgment of the characteristic point markings of the peaks and troughs of red and infrared light.

[0006] The beneficial effect of the present invention is that the accuracy of wearable blood oxygen detection is improved, so that it can achieve accurate blood oxygen measurement in multiple parts, thereby realizing all-weather blood oxygen data monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] Figure 1 It is a flow chart of the multi-spectrum correction method of an embodiment of the present invention.

[0008] Figure 2 It is a schematic diagram of the peaks and troughs of the pulse waves of different colors of the present invention. DETAILED DESCRIPTION

[0009] The embodiments of the present invention are described in detail below with reference to the accompanying drawings, but the present invention can be implemented in many different ways as defined and covered by the claims.

[0010] In an embodiment of the present invention, a wearable watch or a ring is configured, and a photoelectric sensor is configured at the bottom of the wearable watch or the inner side of the ring, and the photoelectric sensor has a three-color light source.

[0011] A blood oxygen monitoring hardware with a three-color light source is configured. The three-color light source is at the same point and maintains the same distance from the light sensor, which is 3-7mm.

[0012] like Figure 1 As shown, a multi-spectral correction method for wearable blood oxygen comprises: S1. Set the three-color lights to be at the same distance from the light sensor, and turn them on alternately at a frequency of 100 Hz. The green light, red light, and infrared light are turned on alternately with the same current, and the switching frequency is 100 Hz. Three-way synchronous original signals can be obtained. S2, the optical sensor obtains three-way synchronous signal waveforms and extracts three-way signals respectively; S3, using the green light waveform as a reference, calibrating the peak and trough positions, and obtaining the time axis coordinates of the peak and trough; S4, finding the corresponding red light and infrared light waveform periods according to the time axis projection coordinates; S5. Find multiple candidate peak and trough amplitudes in the red light and infrared light cycles, with two troughs and one peak as a group, fit the waveform function of each pulse wave cycle according to the green light waveform, proportionally reduce the green light waveform function, compare it with the actual red light and infrared light waveforms, calculate the error, and select the group with the smallest error as the final peak and trough feature points; S6. Calculate the blood oxygen value using the group of peak and trough feature points.

[0013] like Figure 2 As shown, in step 2-4, the green light signal is extracted, the peak and trough points are found, and the horizontal and vertical coordinates of the peak and trough points (x1, y1) (x2, y2)...(xn, yn) are recorded; each pulse wave cycle waveform is simplified into a triangle, where P1 (x1, y1) and P3 (x3, y3) are troughs, and P2 (x2, y2) is the peak, and a triangular function can be fitted.

[0014] The method for finding troughs and peaks is: (1) First locate the horizontal coordinate position x1 of the first trough, and use the autocorrelation method to calculate the period n1 of the waveform in the first 3 seconds, and set the initial jump step length to x1=n1; (2) Perform minimum value detection on the data points whose horizontal coordinates are within the range of μ / 3 from x1+μ1. The minimum value is the second trough x2, and the jump step size is updated to μ2=x2-x1; (3) Perform minimum value detection on the data points whose horizontal coordinates are within the range of μ / 3 from x2+μ2. The minimum value is the third trough x3. At the same time, update the jump step size to: μ3=x3-x2, and so on.

[0015] Project the peak and trough horizontal coordinates onto the coordinate axes of the red and infrared light waveforms, find the y coordinate values ​​of the five coordinates near x1, x2, and x3, and generate five pairs of coordinate values. Take red light x1 as an example: the five coordinate values ​​are: (x1-2, yr11) (x1-1, yr12) (x1, yr13) (x1+1, yr14) (x1+2, yr15).

[0016] Use the same method to find x2 and x3. The 5 coordinates of red light x2 are: (x2-2, yr21) (x2-1, yr22) (x2, yr23) (x2+1, yr24) (x2+2, yr25); the 5 coordinates of red light x3 are: (x3-2, yr31) (x3-1, yr32) (x3, yr33) (x3+1, yr34) (x3+2, yr35) The five pairs of peak and trough coordinates are fitted into triangles, and the variance is used to determine the difference between the triangle and the actual original signal. The group with the smallest difference is selected as the final peak and trough points.

[0017] Repeat the above steps on the infrared waveform.

[0018] The blood oxygen value is calculated using the peak and trough points according to the general blood oxygen formula. That is, the formula for calculating blood oxygen saturation is: SpO2 = bound hemoglobin ÷ (bound hemoglobin + free hemoglobin) × 100%.

[0019] Since there is a certain ratio between oxygenated hemoglobin HbO2 and hemoglobin Hb in the blood, the absorption characteristics of oxygenated hemoglobin HbO2 and hemoglobin Hb for light with a wavelength of 600~1000nm are that the absorption coefficient of Hb is higher between 600~800nm, and the absorption coefficient of HbO2 is higher between 800~1000nm. Therefore, the PPG signals of HbO2 and Hb can be detected respectively using red light (600~800nm) and light close to IR (800~1000nm), and then the corresponding ratio is calculated through program processing, so that the blood oxygen value is obtained.

[0020] Although the blood oxygen value can be easily calculated by red light, the wrist skin responds poorly to red light and infrared light to hemoglobin, and the PPG signal-to-noise ratio measured by green light is high, and the requirements for detection and signal conditioning are relatively low. The method of the embodiment of the present invention makes the output value of blood oxygen more accurate.

[0021] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A multi-spectral correction method for wearable blood oxygen, characterized in that: include: Step 1. Set the three-color lights to be at the same distance from the light sensor and turn them on alternately; Step 2: The optical sensor obtains three-way synchronous signal waveforms and extracts three-way signals respectively; Step 3: Using the green light waveform as a reference, calibrate the peak and trough positions, and obtain the time axis coordinates of the peak and trough; Step 4: Find the corresponding red light and infrared light waveform periods according to the time axis projection coordinates; Step 5, find multiple candidate peak and trough amplitudes in the red light and infrared light cycles, two troughs and one peak as a group, fit the waveform function of each pulse wave cycle according to the green light waveform, proportionally reduce the green light waveform function, compare it with the actual red light and infrared light waveforms, calculate the error, and select the group with the smallest error as the final peak and trough feature points; Step 6: Calculate the blood oxygen value using the group of peak and trough feature points.

2. The multi-spectral correction algorithm according to claim 1, characterized in that: The time axis coordinates of the peaks and troughs are projected onto the red and infrared waveforms to solve the problem of misjudgment of the characteristic point markings of the peaks and troughs of the red and infrared light.

3. The low-power consumption snoring automatic recording method according to claim 1, characterized in that: The three-color light sources are at the same point and maintain the same distance from the light sensor, which is 3-7mm.

Images