Wearable device for detecting apnea index and index determination method thereof

Abstract

A wearable device includes at least one light source, a light sensor, and a processor. The processor generates a peak interval diagram according to one of the at least two light detection signals detected by the light sensor when the at least one light source is turned on, and generates a blood oxygen concentration diagram according to two of the at least two light detection signals detected by the light sensor when the at least one light source is turned on. The processor further determines a sleep apnea index score according to the peak value interval diagram, determines a blood oxygen unsaturation index score according to the blood oxygen concentration diagram, and fits out apnea syndrome indexes corresponding to the sleep apnea index score and the blood oxygen unsaturation index score.

Description

Technical Field

[0001] The present invention relates to an optical detection device, and more particularly to a wearable device suitable for optically detecting apnea indicators and an indicator determination method thereof. Background Art

[0002] The current standard assessment method for obstructive sleep apnea (OSA) is to perform polysomnography (PSG), which measures the user's respiratory airflow, brain wave graph, blood oxygen concentration, respiratory movement and electrocardiogram throughout the sleep process, and then is analyzed and interpreted by professionals. This standard method must be performed in institutions such as hospitals or sleep centers. Although users can currently operate a simplified version of the home PSG device to assess the risk of apnea, a nasal flow meter that measures respiratory airflow is still a necessary equipment. However, the nasal flow meter may make the user feel uncomfortable and affect the quality of sleep. Summary of the invention

[0003] In view of this, the present invention provides a wearable device applicable to detecting apnea index and an index determination method thereof. The user does not need to wear a nasal airway tube during measurement, so the sleep quality will not be affected.

[0004] The present invention provides a wearable device and an index determination method thereof that use light sources of at least two wavelengths to obtain a peak interval graph and a blood oxygen concentration graph, and determine a sleep apnea index (AHI) score and a blood oxygen desaturation index (ODI) score respectively based on the obtained peak interval graph and the blood oxygen concentration graph, and then use a predetermined machine learning model and parameters to derive an apnea syndrome index (OSA Index), thereby allowing users to easily record their own conditions for a long time.

[0005] The present invention provides a wearable device comprising a device body, at least one light source, a light sensor, and a processor. The device body has an inner surface for attaching to the user's skin. The at least one light source is disposed on the inner surface for emitting light of at least two wavelengths. The light sensor is disposed on the inner surface for outputting at least two PPG signals when emitting light relative to the at least one light source. The processor is used to generate a peak interval graph based on one of the at least two PPG signals, generate a blood oxygen saturation concentration graph based on two of the at least two PPG signals, determine a sleep apnea index (AHI) score based on the peak interval graph, determine a blood oxygen desaturation index (ODI) score based on the blood oxygen saturation concentration graph, and determine the OSA index based on the AHI score and the ODI score.

[0006] The present invention also provides a method for determining an AHI score, comprising the steps of: marking reasonable peak intervals to determine the peak interval graph; determining a steady segment in the peak interval graph; upsampling the peak interval data in the steady segment to generate upsampled interval data; calculating pairwise differences of the upsampled interval data to generate interval difference data; calculating variance data of the interval difference data; determining a first threshold based on the variance data and the peak interval data; determining a second threshold based on the first threshold and the number of data points in the variance data that are continuously smaller than the first threshold; and using the ratio of the count value in the variance data that is greater than the second threshold to the total time of the peak interval graph as the AHI score.

[0007] The present invention also provides a method for determining an ODI score, comprising the steps of: scanning the blood oxygen saturation concentration graph to determine a reasonable blood oxygen saturation concentration value; calculating a baseline blood oxygen saturation concentration according to a blood oxygen saturation concentration value within a predetermined time after the reasonable blood oxygen saturation concentration value; determining a low blood oxygen threshold according to the baseline blood oxygen saturation concentration; and taking the ratio of the count value of continuous data points below the low blood oxygen threshold after the reasonable blood oxygen saturation concentration value in the blood oxygen saturation concentration graph to the total time of the blood oxygen saturation concentration graph as the ODI score.

[0008] In order to make the above and other purposes, features and advantages of the present invention more obvious, the following will be described in detail with reference to the accompanying drawings. In addition, in the description of the present invention, the same components are represented by the same symbols, which are hereby described together. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1A is a schematic diagram of a wearable device for detecting an apnea index according to an embodiment of the present invention;

[0010] Figure 1B is a schematic diagram of the operation of a wearable device for detecting an apnea index according to an embodiment of the present invention;

[0011] Figure 2 is a block diagram of a wearable device for detecting an apnea index according to an embodiment of the present invention;

[0012] Figure 3 is a peak interval graph obtained by a wearable device suitable for detecting an apnea index according to an embodiment of the present invention;

[0013] Figure 4 is a blood oxygen concentration graph obtained by a wearable device suitable for detecting an apnea index according to an embodiment of the present invention;

[0014] Figure 5is a flow chart of a sleep apnea index (AHI) score obtained by a wearable device suitable for detecting apnea indicators according to an embodiment of the present invention;

[0015] Fig. 6A yes Figure 5 A schematic diagram of data points from steps S513 to S515;

[0016] Figure 6B yes Figure 5 A schematic diagram of data points of step S517;

[0017] Figure 7 is a flow chart of an oxygen desaturation index (ODI) score obtained by a wearable device suitable for detecting an apnea index according to an embodiment of the present invention;

[0018] Figure 8 is a schematic diagram of a wearable device suitable for detecting an apnea index according to an embodiment of the present invention determining an apnea (OSA) index based on an AHI score and an ODI score; and

[0019] Fig. 9 yes Figure 8 Schematic diagram of a conversion model for converting AHI and ODI scores into OSA indicators.

[0020] Description of Reference Numerals

[0021] 100 Wearable Devices

[0022] 11 Inner surface

[0023] 121, 122, 123 Light Source

[0024] 13 Light Sensor

[0025] 14 Processor

[0026] 141 Memory

[0027] 20 Display

[0028] 90 Skin DETAILED DESCRIPTION

[0029] The purpose of the present invention is to provide a wearable device that uses an all-optical method (i.e., without the need for a nasal airway tube) to obtain an obstructive sleep apnea (OSA) index. The wearable device includes a light sensor that detects a photoplethysmographic (PPG) signal associated with different light wavelengths. The PPG signal is used to generate a sleep apnea index (AHI) score and an oxygen desaturation index (ODI) score, and the AHI score and the ODI score are fitted to the OSA index through a predetermined model and parameters. The method of determining the degree of OSA using the OSA index is known, for example, referring to Figure 8 , so I will not go into details here.

[0030] Please refer to Figure 1A , Figure 1B and Figure 2 As shown, Figure 1A is a schematic diagram of a wearable device 100 suitable for detecting an apnea index according to an embodiment of the present invention; Figure 1B is a schematic diagram of the operation of a wearable device 100 suitable for detecting an apnea index according to an embodiment of the present invention; Figure 2 is a block diagram of a wearable device 100 suitable for detecting apnea indicators according to an embodiment of the present invention.

[0031] The wearable device 100 is selected from electronic devices or accessories that are in contact with the skin 90 when worn on the user's body, such as watches, bracelets, rings, earplugs, and eye masks, but is not limited thereto. The wearable device 100 includes a device body 100b and a detection chip 100c, such as Figure 1A shown.

[0032] The device body 100b has an inner surface 11 (refer to Figure 1A ) is used to be attached to the surface of the user's skin 90, wherein the position of the skin 90 relative to the body depends on the type of wearable device 100.

[0033] The detection chip 100c includes a first light source 121, a second light source 122, a third light source 123, a light sensor 13 and a processor 14, but the present invention is not limited thereto. In other embodiments, at least one of the first light source 121, the second light source 122, the third light source 123, the light sensor 13 and the processor 14 can be configured in different chips or modules.

[0034] The first light source 121, the second light source 122, and the third light source 123 are, for example, light emitting diodes or laser diodes, which are disposed on the inner surface 11 and are used to emit light of a first wavelength λ1, a second wavelength λ2, and a third wavelength λ3 toward the skin 90, respectively, wherein the first wavelength λ1, the second wavelength λ2, and the third wavelength λ3 are different from each other. For example, the light of the first wavelength λ1 is green light, the light of the second wavelength λ2 is red light, and the light of the third wavelength λ3 is infrared light, but the present invention is not limited thereto.

[0035] The light sensor 13 is, for example, a complementary metal oxide semiconductor (CMOS) image sensor or a single photon avalanche diode (SPAD) image sensor, but is not limited thereto. The light sensor 13 is disposed on the inner surface 11 and faces the skin 90, and is used to output a first photoplethysmographic (PPG) signal PPG1 when emitting light relative to the first light source 121, to output a second PPG signal PPG2 when emitting light relative to the second light source 122, and to output a third PPG signal PPG3 when emitting light relative to the third light source 123. In one embodiment, the first light source 121, the second light source 122, and the third light source 123 emit light in separate time periods. In another embodiment, in an embodiment in which the light sensor 13 is capable of distinguishing the light energy of different light sources (for example, a filter is configured or different light sources emit light of different characteristics), the first light source 121, the second light source 122, and the third light source 123 may also emit light at the same time.

[0036] It must be pointed out that although Figure 1A The first light source 121, the second light source 122, and the third light source 123 are shown to be arranged at four sides of the optical sensor 13, but the present invention is not limited thereto. As long as the first light source 121, the second light source 122, and the third light source 123 are arranged at positions that emit light toward the skin 90 and return the light to the optical sensor 13, the optical sensor 13 can generate PPG1, PPG2, and PPG3. In one embodiment, at least one of the first light source 121, the second light source 122, and the third light source 123 is arranged at a different side of the optical sensor 13. In another embodiment, at least one of the first light source 121, the second light source 122, and the third light source 123 has a different distance from the optical sensor 13.

[0037] The processor 14 is, for example, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a microprocessor unit (MCU) or a field programmable gate array (FPGA), and implements its functions using software, firmware and / or hardware, including Figure 2 The processor 14 is coupled to the optical sensor 13 to receive the first PPG signal PPG1, the second PPG signal PPG2 and the third PPG signal PPG3, and determines the OSA index according to the PPG signals PPG1, PPG2 and PPG3, as illustrated in FIG. Figure 4 and Figures 7 and 8 .

[0038] The wearable device 100 further includes a display 20 or an electronic device wirelessly coupled to the display 20. The display 20 is used to display the OSA index (e.g., display the OSA index value) or display the OSA category related to the OSA index, e.g. Figure 8 The displayed categories include normal category, mild OSA category, moderate OSA category and severe OSA category, so that the user can understand the assessed OSA risk through the display 20. The OSA index and OSA category can also be recorded in the memory 141 for access by an external device / system via wired or wireless communication.

[0039] For example, the processor 14 is used to generate a peak interval diagram according to the first PPG signal PPG1 (for example, refer to Figure 3 ), generate a blood oxygen saturation concentration map according to the second PPG signal PPG2 and the third PPG signal PPG3 (for example, refer to Figure 4 The processor 14 calculates the number of AHI events and ODI events in the total recording time (sometimes referred to as the total time in this description), so the processor 14 also receives the data point time ( Figure 2 The AHI and ODI scores are determined by the time (displayed as time).

[0040] In one embodiment, the peak interval diagram is formed by the RR interval (milliseconds) of the heart rhythm waveform at different times, wherein the meaning of the RR interval of the heart rhythm waveform is known, so it is not described here. In addition, the method of calculating the blood oxygen saturation concentration using the PPG signals related to two optical wavelengths is also known, so it is not described here. In the present invention, the optical wavelengths λ1, λ2 and λ3 refer to the main wavelengths of the optical spectrum, respectively. That is, the optical sensor 13 provides at least two PPG signals related to at least two optical wavelength ranges to the processor 14 for calculating the peak interval and SPO2 respectively. When calculating SPO2, the two optical wavelength ranges used may partially overlap or not overlap. The same optical wavelength range can be used to calculate the peak interval diagram and the SPO2 diagram respectively.

[0041] In the present invention, the AHI score is, for example, the number of sleep apnea occurring during the total recording time of the peak interval graph (for example, determined by the user manually controlling the start of recording before going to bed and manually controlling the end of recording when getting up, or determined by the wearable device 100 having automatic judgment or presetting the start and end of recording). For example, the processor 14 is used to count the number of times the variance of the peak interval difference in the total recording time exceeds a threshold interval (explained below by way of example) as the number of sleep apnea occurring, wherein the threshold interval is determined according to the peak interval graph determined each time, for example, the peak interval graph and the threshold interval measured each night are different.

[0042] In the present invention, the ODI score is the number of times blood oxygen desaturation occurs during the total recording time of the blood oxygen saturation concentration graph (as described above). For example, the processor 14 is used to count the number of consecutive data points in which the blood oxygen saturation concentration is less than the low blood oxygen threshold during the total recording time as the number of times blood oxygen desaturation occurs, wherein the low blood oxygen threshold is determined according to the blood oxygen saturation concentration graph determined each time, for example, the blood oxygen saturation concentration graph and the low blood oxygen threshold measured every night are different.

[0043] Finally, processor 14 uses a pre-determined machine learning model (e.g. Figure 8 The AHI score and the ODI score are converted / fitted to OSA indicators in S810 shown. The machine learning model is established based on, for example, partial least squares (PLS), principal component (PC), multiple linear regression (MLR), artificial neural network (ANN), K nearest neighbor (KNN), decision tree (DT), support vector machine (SVM), etc., but is not limited thereto. The parameters of the machine learning model can be established based on partial least squares, principal component, multiple linear regression, artificial neural network, K nearest neighbor, decision tree, support vector machine, etc., and can be updated later using edge training. For example Fig. 9 A method of converting / fitting the AHI score and the ODI score to OSA indices using multiple linear regression is shown. Fig. 9 It can be seen that once a set of AHI scores and ODI scores are determined, an OSA index can be determined.

[0044] Please also refer to Figure 2 and Figure 5 As shown, a method for determining the AHI score is described below by way of example.

[0045] The processor 14 calculates the AHI score according to the peak interval graph and the corresponding recording time.

[0046] Step S510: The processor 14 marks the reasonable peak interval in PPG1 to determine the peak interval map. In one embodiment, when the peak interval is the RR interval of the heart rhythm waveform, the reasonable peak interval includes the peak interval between 400 milliseconds and 1800 milliseconds.

[0047] Steps S511 to S512: Then, the processor 14 determines a stable segment in the peak interval graph. When processing the entire peak interval graph, multiple stable segments can be determined. In one embodiment, the stable segment is a segment in which the difference between adjacent peak interval data is less than 2 times the standard deviation of the peak interval data in the reasonable peak interval. Simply put, the stable segment means that the change in peak interval data is within a predetermined range. The stable segment is to eliminate the interference of user movements during sleep (such as turning over). When it is determined that the difference between adjacent peak interval data is greater than a predetermined range (such as 2 times the standard deviation), the processor 14 generates a stable segment for subsequent steps. That is, the following steps S513 to S523 are performed in sequence for each stable segment.

[0048] Step S513: In order to improve the calculation accuracy, the processor 14 up-samples the peak interval data in the stable segment to generate up-sampled interval data, which can reduce the sampling frequency of the optical sensor 13 when acquiring the PPG signal. In one embodiment, the up-sampling multiple is, for example, 10 to 40 times the peak interval data. Fig. 6A Upsampled interval data 610 is displayed.

[0049] Step S514: The processor 14 then calculates the pairwise differences of the up-sampled interval data to generate interval difference data, for example Fig. 6A The interval difference data is displayed 620. The inventive concept of the present invention is to use the difference (ie, time change) of the up-sampled interval data to determine whether sleep apnea occurs.

[0050] Step S516: The processor 14 calculates the variance data of the interval difference data, such as calculating the standard deviation square of the interval difference data, but not limited thereto. In order to eliminate noise and improve accuracy, the processor 14 may first perform a bandpass filter on the interval difference data 620 (step S515) and then calculate the variance data. Fig. 6A The bandpass filtered difference data 630 is displayed.

[0051] Step S517: The processor 14 determines a first threshold TH1 based on the variance data (or the bandpass filtered variance data 630) and the peak interval data. In one embodiment, the first threshold TH1 is determined based on the average Avg1 of the variance data (or the bandpass filtered variance data 630) and the average Avg2 of the peak interval data of the stable segment, for example (Avg1 / Avg2)×N, where N is a preset positive number, which can be obtained from clinical data of PPG with apnea and hypopneas (for example, between 500-1000). In order to eliminate noise to improve accuracy, the processor 14 may first perform a low-pass filter on the variance data (or the bandpass filtered variance data 630) (step S517) and then calculate the first threshold TH1. Figure 6B Displays low-pass filtered variance data.

[0052] Steps S519 to S521: The processor 14 determines the second threshold TH2 according to the first threshold TH1 and the number of data points in the variance data (or low-pass filtered variance data) that are continuously smaller than the first threshold TH1. For example, assuming that the number of continuous data points that are smaller than the first threshold TH1 is M, then TH2 = TH1 × M. That is, the second threshold TH2 is positively correlated with the number of data points that are continuously smaller than the first threshold TH1. The larger the value of M, the higher the value of the second threshold TH2.

[0053] Steps S522 to S524: The processor 14 uses the ratio of the count value greater than the second threshold TH2 in the variance data (or low-pass filtered variance data) to the total time of the peak interval graph as the AHI score. Figure 6B If a stable segment has two sleep apnea events AHI1 and AHI2, the processor 14 will add 2 sleep apnea events to the stable segment. Figure 5 The AHI event value of step S523 in the above method is incremented by 1 for each sleep apnea event. That is, the AHI score is the total number of AHI events divided by the total time of the peak interval graph.

[0054] The processor 14 updates the cumulative number of AHI events obtained in step S523 for each stable segment (if an OSA event occurs) as the total number of events. The total time depends on the total sleep time of the user for one night (assuming it is measured at night). The wearable device 100 of the present invention can start recording when the user starts to sleep, and the processor 14 can use the total recorded time as the total time, or deduct the predetermined time (e.g., 5-10 minutes) of the first segment (just falling asleep) and the last segment (before waking up) from the total recorded time to use as the total time.

[0055] Please also refer to Figure 2 and Figure 7As shown, a method for determining an ODI score is described below with an example.

[0056] The processor 14 calculates the ODI score according to the blood oxygen saturation concentration graph and the corresponding recording time.

[0057] Steps S710 to S711: The processor 14 scans the blood oxygen saturation concentration graph to determine a reasonable blood oxygen saturation concentration value. In one embodiment, the reasonable blood oxygen saturation concentration value is set to the blood oxygen saturation concentration value of the blood oxygen saturation concentration graph within a predetermined time (e.g., one minute to two minutes) where the standard deviation of the blood oxygen saturation concentration is less than 2, but is not limited to less than 2. The processor 14 enters the following steps only after confirming that a reasonable blood oxygen saturation concentration value appears. In one embodiment, the processor 14 only performs step S711 once in a blood oxygen saturation concentration graph (e.g., relative to one sleep).

[0058] Step S712: After a reasonable blood oxygen saturation concentration value appears, the processor 14 then calculates a baseline blood oxygen saturation concentration based on the blood oxygen saturation concentration values ​​within a predetermined time (e.g., one to two minutes) (all of which are considered reasonable values ​​at this time). In one embodiment, the baseline blood oxygen saturation concentration is the average of the blood oxygen saturation concentration values ​​within the predetermined time, but is not limited thereto. In the present invention, the baseline blood oxygen saturation concentration may be different according to each recorded blood oxygen saturation concentration graph (e.g., relative to each sleep).

[0059] Step S713: After the baseline blood oxygen saturation concentration is determined, the processor 14 determines a hypoxemia threshold value according to the baseline blood oxygen saturation concentration. In one embodiment, the hypoxemia threshold value is set to 2% to 5% of the baseline blood oxygen saturation concentration, but is not limited thereto.

[0060] Steps S714 to S716: In the present invention, during the period of continuously scanning the blood oxygen saturation concentration value after the reasonable blood oxygen saturation concentration value, when the blood oxygen saturation concentration value of the blood oxygen saturation concentration graph has a predetermined number of consecutive data points below the hypoxemia threshold, the processor 14 adds 1 to the blood oxygen desaturation index (ODI) event (step S716). In one embodiment, the predetermined number of consecutive data points is at least greater than or equal to 4 data points, but is not limited to 4. In other words, when only less than 4 data points are below the hypoxemia threshold, the ODI event is not added by 1.

[0061] Step S717: Finally, the processor 14 uses the ratio of the count value of the continuous data points below the hypoxemia threshold in the blood oxygen saturation concentration values ​​(all considered as reasonable values ​​at this time) of the blood oxygen saturation concentration graph to the total time of the blood oxygen saturation concentration graph as the ODI score. That is, the ODI score is the total number of ODI events divided by the total time of the blood oxygen saturation concentration graph.

[0062] Although Figure 7 The upsampled SpO2 data is not shown, but the present invention is not limited thereto. The processor 14 may also upsample the SpO2 data before step S714.

[0063] In one embodiment, the method for determining the AHI score and the ODI score of the present invention can be processed by the processor 14 after the user wakes up. For example, all PPG signals (such as PPG1, PPG2, and PPG3) measured by the light sensor 13 during sleep are first stored in the memory 141. In another embodiment, the method for determining the AHI score and the ODI score of the present invention can be processed by the processor 14 once every period of time (such as one to two hours) when the user is sleeping, and the updated count value (such as steps S523 and S716) is recorded in the memory 141, that is, the accumulated AHI event value and ODI event value are recorded. The processor 14 calculates the ratio of the final (end of sleep time) AHI event value and ODI event value to the total time to obtain the AHI score and the ODI score.

[0064] It must be noted that, although the above embodiment is described by taking three light sources as an example, the present invention is not limited thereto. In other embodiments, the peak interval graph can be obtained using one of the two light sources (e.g., red light) used to detect the blood oxygen concentration graph. In another embodiment, at least two wavelengths of light are generated by one light source, for example, by adjusting the driving parameters of the one light source to change the light emission wavelength.

[0065] It must be noted that the numerical values ​​in the above embodiments, such as the number of data points, time, ratio, percentage, etc., are only for illustration and are not intended to limit the present invention.

[0066] In summary, the existing standard equipment for determining apnea indicators includes a nasal airway tube for users to wear while sleeping, which affects the quality of sleep. Therefore, the present invention also provides an all-optical wearable device suitable for detecting apnea indicators (refer to Figure 2 ) and its index detection method (refer to Figure 5 and Figure 7-8 ), which uses PPG signals related to different light wavelengths to determine the sleep apnea index score and the blood oxygen desaturation index score, that is, the corresponding OSA index can be determined according to the pre-established model and parameter data. The apnea index detection results of the present invention have a certain degree of accuracy (diagnostic thresholds RI>5 and RI>15) through verification with the detection results of existing standard equipment, which is sufficient for users to use as a reference for long-term observation.

[0067] Although the present invention has been disclosed through the above examples, they are not intended to limit the present invention. Any person skilled in the art with ordinary knowledge in the art to which the present invention belongs can make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the scope defined by the appended claims.

Claims

1. A wearable device for obtaining an OSA index of a user, the wearable device comprising: a device body having an inner surface for attaching to the skin of the user; At least one light source, the at least one light source is disposed on the inner surface and is used to emit light of at least two wavelengths; a light sensor disposed on the inner surface and configured to output at least two photoplethysmographic (PPG) signals when emitting light relative to the at least one light source; and processor, which is used to generating a peak interval graph according to one of the at least two PPG signals, generating a blood oxygen saturation concentration map according to two of the at least two PPG signals, determining a sleep apnea index (AHI) score based on the peak interval graph, determining an oxygen desaturation index (ODI) score according to the blood oxygen saturation concentration graph, and The OSA index is determined according to the AHI score and the ODI score.

2. The wearable device according to claim 1, wherein: One of the at least two wavelengths of light is red light, and Another one of the at least two wavelengths of light is infrared light.

3. The wearable device according to claim 1, wherein: The wearable device further includes a display or the wearable device is wirelessly coupled to an electronic device having a display, and The display is used to display the OSA index or display the OSA category related to the OSA index.

4. The wearable device according to claim 1, wherein: The AHI score is the number of sleep apnea episodes that occurred during the total recording time of the peak interval graph.

5. The wearable device according to claim 4, wherein: The processor is further configured to: The number of times that the variance of the peak interval difference in the total recording time exceeds the threshold interval is counted as the number of times sleep apnea occurs.

6. The wearable device according to claim 5, wherein: The threshold interval is determined according to the peak interval map determined each time.

7. The wearable device according to claim 1, wherein: The ODI score is the number of times blood oxygen desaturation occurs during the total recording time of the blood oxygen saturation concentration graph.

8. The wearable device according to claim 7, wherein: The processor is further configured to: The number of consecutive data points in which the blood oxygen saturation concentration is less than the low blood oxygen threshold value during the total recording time is counted as the number of times blood oxygen desaturation occurs.

9. The wearable device according to claim 8, wherein: The hypoxemia threshold is determined according to the blood oxygen saturation concentration graph determined each time.

10. The wearable device according to claim 1, wherein: The processor is further configured to: The AHI score and the ODI score are converted into the OSA index using a predetermined machine learning model.

11. A method for determining the AHI score of claim 1, the method comprising: Marking reasonable peak intervals to determine the peak interval diagram; determining a stable segment in the peak interval graph; upsampling the peak interval data in the stable segment to generate upsampled interval data; Calculating the pairwise differences of the upsampled interval data to generate interval difference data; Calculate the variance data of the interval difference data; Determine a first threshold value according to the variance data and the peak interval data; Determine a second threshold value according to the first threshold value and the number of data points in the variance data that are continuously smaller than the first threshold value; as well as The ratio of the count value greater than the second threshold in the variance data to the total time of the peak interval graph is used as the AHI score.

12. The determination method according to claim 11, wherein: The reasonable peak interval includes a peak interval between 400 milliseconds and 1800 milliseconds.

13. The determination method according to claim 11, wherein: The stable section is set as a difference between adjacent peak interval data that is less than 2 times the standard deviation of the peak interval data in the reasonable peak interval.

14. The determination method according to claim 11, wherein: The first threshold is determined according to an average of the variance data and an average of the peak interval data in the stable segment.

15. The determination method according to claim 11, wherein: The second threshold is positively correlated with the number of the data points that are continuously smaller than the first threshold.

16. The determination method according to claim 11, further comprising: performing bandpass filtering on the interval difference data; and The variance data is low-pass filtered.

17. A method for determining the ODI score of claim 1, the method comprising: Scanning the blood oxygen saturation concentration graph to determine a reasonable blood oxygen saturation concentration value; Calculating a baseline blood oxygen saturation concentration according to a blood oxygen saturation concentration value within a predetermined time after the reasonable blood oxygen saturation concentration value; Determining a hypoxemia threshold according to the reference blood oxygen saturation concentration; and The ODI score is the ratio of the count value of the continuous data points below the low blood oxygen threshold value after the reasonable blood oxygen saturation concentration value in the blood oxygen saturation concentration graph to the total time of the blood oxygen saturation concentration graph.

18. The determination method according to claim 17, wherein: The reasonable blood oxygen saturation concentration value is the blood oxygen saturation concentration value of the blood oxygen saturation concentration graph in which the standard deviation of the blood oxygen saturation concentration within a predetermined time is less than 2.

19. The determination method according to claim 17, wherein: The reference blood oxygen saturation concentration is the average of the blood oxygen saturation concentration values ​​within the predetermined time, and The hypoxemia threshold is set to 2% to 5% of the baseline blood oxygen saturation concentration.

20. The determination method according to claim 17, wherein: The continuous data points are at least greater than or equal to 4 data points.

Images