Heart rate estimation method based on optical measurement, electronic equipment and storage medium

Abstract

The invention discloses a heart rate estimation method based on optical measurement, which comprises the following steps: a confidence level setting step: setting more than three confidence levels; a heart rate jump threshold setting step: setting a heart rate jump threshold for each confidence level; an initial state setting step: randomly selecting one confidence level as an initial confidence level; a confidence level evaluation step of evaluating the current confidence level according to a preset index; and a heart rate value estimation step: correcting the candidate heart rate value by using the heart rate jump threshold value of the current confidence level to obtain a current heart rate value Ncur. According to the method, the strategy of dynamically adjusting the heart rate jump threshold is adopted in the link of tracking the heart rate according to the threshold, and the appropriate heart rate jump threshold is selected according to the index of the real-time signal, so that the accuracy and continuity of heart rate value estimation are ensured, the problems that a spectrum peak is easily tracked by mistake and a physiologically unreasonable heart rate value is generated are solved, and the user experience is improved. The invention further provides a parameter setting method for controlling the current confidence level evaluation condition, and the portability of the program is improved.

Description

Technical Field

[0001] This invention belongs to the field of physiological signal processing technology in electronic devices, and specifically relates to a heart rate estimation method based on optical measurement for low-power embedded devices. Background Technology

[0002] With the increasing awareness of public health management, activity monitors utilizing optical measurements have become a key type of equipment in the fields of sports health monitoring and medical auxiliary monitoring due to their core advantage of non-invasive monitoring. These devices can collect human physiological signals through built-in specific optical sensors, thereby achieving real-time monitoring of core health parameters. Among them, photoplethysmograph (PPG) sensors are the most widely used core physiological signal acquisition modules.

[0003] PPG sensors, due to their low cost, non-invasiveness, and simple structure, and their natural compatibility with optical measurement principles, are widely integrated into various wearable activity monitors based on optical measurement. Typical products include mainstream wrist-worn devices such as smartwatches and smart bracelets. Leveraging the signal acquisition capabilities of PPG sensors, these optical measurement activity monitors can accurately monitor key physiological parameters such as heart rate and pulse oxygen saturation (SpO2). They can provide real-time exercise intensity feedback for athletes and assist healthcare professionals in assessing patients' cardiovascular health, possessing significant clinical application value and broad commercial prospects.

[0004] As a core technology supporting the monitoring of physiological parameters using optical measurement activity monitors, the core principle of PPG technology is to indirectly measure physiological signals based on the interaction between optical signals and biological tissues. The specific process is as follows: The PPG sensor module of the optical measurement activity monitor emits near-infrared or visible light of a specific wavelength. After the light shines directly on the skin surface and penetrates the epidermis, it interacts with biological tissues such as subcutaneous tissue and blood in blood vessels. Some of the light is absorbed and some is scattered. Because the periodic beating of the heart drives the blood volume of subcutaneous microvessels to fluctuate synchronously and periodically, and the amount of light absorbed by the blood at a specific wavelength changes with the blood volume, the intensity of light reflected or transmitted by the skin will fluctuate periodically. After the PPG sensor detects the change signal of this light intensity in real time (i.e., the PPG signal), the physiological information directly related to cardiovascular activity can be deduced in reverse, providing basic data support for subsequent health parameter calculations.

[0005] Heart rate monitoring is one of the core functions of various health monitoring devices. Its measurement results are usually represented by heart rate per minute (BPM) and have a wide range of applications: for athletes, heart rate monitoring data can provide real-time feedback on exercise intensity, providing a basis for the formulation and adjustment of scientific training plans; for healthcare professionals, heart rate monitoring is a key means of assessing patients' cardiovascular health status and assisting in disease diagnosis and rehabilitation monitoring.

[0006] Among them, the wrist-worn optical measurement activity monitor is more suitable for daily health monitoring and sports scenarios due to its advantages of small size, convenient wearing and high integration. However, it should be noted that, unlike static monitoring scenarios, the PPG signal collected by its built-in PPG sensor is easily affected by a variety of interference factors in dynamic sports environments, resulting in serious deterioration of signal quality and consequently poor accuracy of heart rate measurement. Summary of the Invention

[0007] During exercise, the acquired PPG signal is a mixture of the target physiological signal (heart rate-related blood volume fluctuation signal) and various interference signals. The interference signals mainly come from the following aspects: First, motion artifacts, which are the most important source of interference. During exercise, the swinging of limbs and the contraction and relaxation of muscles will cause periodic or non-periodic displacement of skin tissue, resulting in changes in the relative position between the sensor and the skin and changes in the light propagation path. This introduces strong interference components related to the motion frequency into the acquired signal. The amplitude of this component is often much larger than the amplitude of the target heart rate signal, which can easily mask the target signal. Second, environmental and equipment noise, including ambient light interference, thermal noise of the sensor circuit itself, and power frequency interference of the equipment power supply module, etc. This type of noise will be superimposed on the PPG signal, further reducing the signal-to-noise ratio. Third, wearing and skin condition interference. During exercise, sweating and changes in the tightness of the device can lead to poor contact between the sensor and the skin, causing signal baseline drift, signal disconnection, or amplitude abrupt changes. At the same time, individual characteristics such as skin texture and skin color differences will also exacerbate the instability of the signal.

[0008] To address this, the industry has applied various signal processing algorithms to the PPG heart rate calculation process to improve monitoring performance. In the noise suppression stage, traditional signal processing methods such as Active Noise Cancellation (ANC) and wavelet thresholding have been used for PPG signal preprocessing to initially filter out some interference components. In the accurate heart rate estimation stage, machine learning and deep learning models, due to their powerful feature learning and nonlinear fitting capabilities, can better adapt to signal characteristics under complex interference environments, exhibiting superior heart rate estimation performance. Typical examples include Deep Temporal Neural Networks (DTNNs) and Recurrent Neural Networks (RNNs), which have been widely researched and applied.

[0009] However, current complex algorithms (especially neural network models) with high-precision heart rate estimation capabilities typically come with high computational resource requirements and power consumption, specifically large model parameter sizes, high computational complexity, and large memory footprints. Portable wearable devices mostly use low-power embedded processors, whose memory capacity, data processing capabilities, and power consumption are all subject to strict constraints. This poses a significant challenge to achieving real-time and efficient heart rate calculation on the device side, limiting the large-scale application of high-precision PPG motion heart rate monitoring technology in low-power wearable devices.

[0010] To address this, the industry has further developed a heart rate estimation method suitable for low-power devices. For example... Figure 1 As shown, taking PPG signal processing as an example, this method consists of seven steps: data acquisition, data preprocessing, signal denoising, spectrum calculation, peak selection, heart rate tracking based on threshold, and output of heart rate value. The details of each step are as follows:

[0011] 1. Data Collection

[0012] This step involves acquiring raw PPG signals reflecting fluctuations in microvascular blood volume to provide fundamental data for subsequent heart rate calculations. Appropriate sampling parameters must be set during the acquisition process. This step will obtain raw time-series PPG signals (containing a mixture of periodic physiological signal components related to actual heart rate, motion artifacts, ambient light interference, etc.).

[0013] 2. Data Preprocessing

[0014] The original PPG signal was initially purified to remove obvious abnormal interference and standardize the data format, laying the foundation for subsequent signal denoising and spectrum analysis.

[0015] 3. Signal denoising

[0016] The PPG signal obtained after data preprocessing is targeted to suppress multi-source interference (motion artifacts, ambient light noise, circuit noise, etc.), improve the signal-to-noise ratio (SNR), and retain the periodic physiological signal components related to the actual heart rate, resulting in a purer PPG signal with significantly suppressed interference components and a more prominent actual heart rate signal.

[0017] 4. Spectrum Calculation

[0018] The relatively clean time-domain PPG signal obtained after signal denoising is converted to the frequency domain, and the periodic fluctuations corresponding to the heart rate are converted into characteristic peaks in the frequency domain, realizing the frequency domain extraction of heart rate information and obtaining the frequency domain spectrum (including the frequency-amplitude distribution of heart rate characteristic peaks) corresponding to each frame of signal.

[0019] 5. Select peak value

[0020] Identify characteristic peaks corresponding to heart rate from the frequency domain spectrum, eliminate harmonic interference and residual noise peaks, and determine the candidate heart rate value for each frame of signal (unit: Hz, which can be converted to BPM).

[0021] 6. Heart rate tracking based on thresholds

[0022] Based on the preset heart rate jump threshold, the results of spectral analysis and peak selection are further corrected to ensure that they do not exceed the maximum physiological range of heart rate change in the previous second, thereby ensuring the rationality of the heart rate results.

[0023] 7. Output heart rate value

[0024] Based on all the aforementioned steps, the final output is a heart rate value for the user's reference.

[0025] Throughout the process, data acquisition is fundamental, preprocessing and noise reduction are the core purification steps, spectrum calculation realizes the feature transformation from the time domain to the frequency domain, peak selection locates heart rate information, and heart rate tracking based on thresholds transforms heart rate data into monitoring results with practical application value, adapting to the needs of multiple scenarios such as sports and health, and medical assistance. The output of each step serves as the input for the next step, ensuring the coherence and rigor of the technical process.

[0026] Of course, there is still room for improvement in this process. Current technologies generally use a fixed threshold strategy for heart rate tracking. When the PPG signal experiences a significant decrease in spectral quality due to strenuous exercise, using a single static threshold can easily lead to problems such as mistracking of spectral peaks, excessively small jumps resulting in slow tracking, or excessively large jumps exceeding physiological limits.

[0027] The purpose of this invention is to overcome the problem in existing technologies that, due to the use of a fixed threshold, are prone to mistracking spectral peaks and generating physiologically unreasonable heart rate values ​​when the PPG signal quality changes. To address this, a dynamic adjustment strategy for the heart rate jump threshold is proposed to ensure the accuracy and continuity of heart rate value estimation.

[0028] Specifically, this invention proposes a heart rate estimation method based on optical measurements, including:

[0029] The confidence level setting step involves setting three or more confidence levels, including a confidence state, a verification state, and a transient state. The confidence state has the highest confidence level, the transient state has the lowest confidence level, and the verification state has a confidence level between the confidence state and the transient state.

[0030] The heart rate jump threshold setting steps are as follows: set a heart rate jump threshold for each confidence level; the heart rate jump threshold for the confidence state is the confidence state threshold θ1, the heart rate jump threshold for the verification state is the verification state threshold θ2, and the heart rate jump threshold for the transient loss state is the transient loss state threshold θ3; θ3 < θ2 < θ1.

[0031] The initial state setting step involves selecting any confidence level as the initial confidence level.

[0032] The confidence level assessment step involves determining the current confidence level based on preset indicators; this current confidence level is one of the confidence levels.

[0033] The heart rate estimation steps involve estimating the current heart rate value N using the following formula. cur :

[0034]

[0035] in, It is a candidate value for heart rate; N prev It is the previous heart rate value, i.e., the previous estimated heart rate value; θ is the heart rate jump threshold for the current confidence level; N cur It is the result after heart rate jump threshold correction, and is used as the current heart rate value.

[0036] The beneficial effects of this invention include the following:

[0037] 1. By setting multiple confidence levels with different confidence levels and assigning heart rate jump thresholds to each, and then setting different current confidence levels, the heart rate jump threshold parameter used in the heart rate value estimation step is dynamically switched between different heart rate jump thresholds. That is, a strategy of dynamically adjusting the heart rate jump threshold is adopted in the heart rate tracking step based on the threshold. The appropriate heart rate jump threshold is selected according to the indicators of the real-time signal to ensure the accuracy and continuity of heart rate value estimation and avoid the problems of easily mistracking spectral peaks and generating physiologically unreasonable heart rate values.

[0038] 2. Prioritize and assign weights to each spectral quality parameter based on its importance, and combine this with the corresponding spectral quality thresholds to determine the current confidence level. This incorporates spectral quality factors into the heart rate estimation.

[0039] 3. It incorporates a heart rate continuity index, which can improve the continuity of heart rate value estimation, especially reducing the occurrence of physiologically unreasonable data;

[0040] 4. It incorporates exercise intensity compliance indicators, providing another perspective for calculating candidate heart rate values. The physiological rationality;

[0041] 5. Based on the user's selected exercise type and intensity, a heart rate value is estimated, and an adjustment factor N is assigned to the current heart rate value. cur Greater flexibility, making it more adaptable to the motion characteristics of real-world scenarios;

[0042] 6. It enables heart rate estimation data to better reflect the user's individual characteristics, laying the foundation for further accumulation of more data with similar characteristics;

[0043] 7. A robust parameter setting method is provided to control the assessment criteria for the current confidence level. This allows for both strict parameter setting to raise the assessment standard and lowering it by lowering the parameter standard. This is applicable to various application scenarios, improving the ease of use and portability of the program.

[0044] 8. Adding this questioning state allows for further subdivision of the confidence level, enriching the confidence level and avoiding the one-size-fits-all treatment of data below the verification state as transient out-of-state. This accumulates more usable data for states with poor spectral quality, which is beneficial for further improving the level of heart rate estimation.

[0045] The following will further explain the concept, specific structure and technical effects of the present invention with reference to the accompanying drawings, so as to fully understand the purpose, features and effects of the present invention. Attached Figure Description

[0046] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0047] Figure 1 This is a flowchart illustrating existing heart rate estimation methods.

[0048] Figure 2This is a schematic diagram illustrating the confidence level (level 3) variation of the heart rate estimation method based on optical measurement according to the present invention.

[0049] Figure 3 This is a schematic diagram illustrating the confidence level (level 4) variation of the heart rate estimation method based on optical measurement according to the present invention. Detailed Implementation

[0050] To facilitate understanding by those skilled in the art, some terms appearing in this document are explained and clarified.

[0051] In this document, the singular forms “an,” “a,” and “the” include their plural forms, unless the context otherwise requires.

[0052] In this document, unless otherwise stated, the terms “comprising,” “including,” “having,” or “containing” mean that the listed technical features are included, but do not exclude the inclusion of other technical features.

[0053] In this article, ordinal numbers such as "first" and "second" added before a component are used to distinguish these components, not to sort them according to the order of the ordinal numbers.

[0054] First aspect: Heart rate estimation methods based on optical measurements

[0055] Example 1:

[0056] A heart rate estimation method based on optical measurements includes:

[0057] The confidence level setting step involves setting three or more confidence levels, including a confidence state, a verification state, and a transient state. The confidence state has the highest confidence level, the transient state has the lowest confidence level, and the verification state has a confidence level between the confidence state and the transient state.

[0058] The heart rate jump threshold setting steps are as follows: set a heart rate jump threshold for each confidence level; the heart rate jump threshold for the confidence state is the confidence state threshold θ1, the heart rate jump threshold for the verification state is the verification state threshold θ2, and the heart rate jump threshold for the transient loss state is the transient loss state threshold θ3; θ3 < θ2 < θ1.

[0059] The initial state setting step involves selecting any confidence level as the initial confidence level.

[0060] The confidence level assessment step involves determining the current confidence level based on preset indicators; this current confidence level is one of the confidence levels.

[0061] The heart rate estimation steps involve estimating the current heart rate value N using the following formula. cur :

[0062]

[0063] in, It is a candidate value for heart rate; N prev It is the previous heart rate value, i.e., the previous estimated heart rate value; θ is the heart rate jump threshold for the current confidence level; N cur It is the result after heart rate jump threshold correction, and is used as the current heart rate value.

[0064] This embodiment presents an improved solution for the heart rate tracking process based on thresholds in existing technologies. After determining candidate heart rate values ​​in the peak selection stage, the improved solution presented in this embodiment is implemented. In existing technologies, after selecting candidate heart rate values, the terminal device typically uses a fixed threshold strategy for heart rate tracking. When the PPG signal experiences a significant decrease in spectral quality due to strenuous exercise, using a single static threshold can easily lead to problems such as mistracking of spectral peaks, excessively small jumps resulting in slow tracking, or excessively large jumps exceeding physiological rationality. Therefore, a dynamic adjustment strategy for the heart rate jump threshold is proposed to ensure the accuracy and continuity of heart rate estimation.

[0065] The confidence level setting step, the heart rate jump threshold setting step, and the initial state setting step are the program initialization processes, which run when the device is powered on or when switching to a mode requiring heart rate estimation. The confidence level setting step is used to set the confidence level. The confidence level can be set to multiple levels, from high to low, depending on the confidence level. This embodiment provides an example with three levels, but obviously, more levels can be set. (The provided example is...) Figure 2Of the three levels shown, the confidence level represents the highest level of confidence, indicating reliable data that can be used directly; the transient loss level represents the lowest level of confidence, indicating unreliable data that cannot be used; and the validation level has a confidence level between the two, indicating a relatively high probability of data reliability. Regarding the validation level, it's important to note that "relatively high probability of reliability" is relative to the confidence and transient loss levels. Its reliability probability falls between the two, including situations where the validation level's confidence level is close to the confidence level, indicating a high probability of reliability and allowing for relatively direct use, and situations where the reliability probability is slightly higher than the transient loss level, indicating less reliability but usable temporarily after correction. The specific choice must be made based on the usage scenario, data collection environment, data requirements, and adoption conditions. It should also be noted that the standards for data reliability and usability vary with technological advancements and changing needs. This example only provides a data processing method based on a custom confidence level classification tailored to the specific usage scenario. The heart rate jump threshold setting steps involve setting the heart rate jump threshold for each confidence level, based on the usage scenario and the spectral quality of the currently acquired signal. For lower confidence levels (verification state, transient loss state, etc.), the heart rate jump threshold setting requires more restrictions, and the set value should be lower than the value in the confidence state. The heart rate jump threshold setting varies depending on the exercise state. For more intense exercise, the heart rate jump threshold value in the confidence state is higher than that in static activity. The heart rate jump thresholds for each confidence level are mostly set uniformly with the product in the initial state. As user historical data accumulates, personalized settings can be provided. In the initial state setting steps, any confidence level can be selected as the initial confidence level. This setting only affects the first few cycles; subsequent acquisition cycles will automatically adjust to the appropriate confidence level. It is recommended to set the initial state to the confidence state, as the signal quality is likely to be relatively good at this time, as the user may have just put on the device or begun exercising. It should also be noted that the confidence level setting step, the heart rate jump threshold setting step, and the initial state setting step do not need to be executed in exact order. They only need to be completed when powering on or switching to a mode requiring heart rate estimation. This distinction between steps is for descriptive purposes, not for the method itself.

[0066] The confidence level assessment process involves determining the current confidence level based on preset indicators. Specifically, it selects a matching confidence level from the existing settings as the current (i.e., from the start of this setting until the next setting). "Effective" here means that the heart rate jump threshold of this confidence level will be adopted as a parameter in heart rate estimation.

[0067] The heart rate estimation step is characterized by a dynamically adjusted heart rate jump threshold used in the formula, varying according to the current confidence level given in the confidence level assessment step. Current heart rate value N cur This is an estimated value corrected for the heart rate jump threshold. The heart rate jump threshold referred to here is the threshold for the current confidence level. Because the current confidence level is variable, the heart rate jump threshold changes when different confidence levels are set. Therefore, the selection of the heart rate jump threshold is dynamic, hence the term "dynamic heart rate jump threshold." When the heart rate candidate value minus the previous heart rate value (i.e., the current heart rate value obtained in the previous running cycle) exceeds the heart rate jump threshold allowed by the current confidence level, the current heart rate value N is reset. cur The estimated value is the sum of the previous heart rate value and the heart rate jump threshold corresponding to the current confidence level. Conversely, if the difference between the previous heart rate value (i.e., the current heart rate value obtained in the previous cycle) and the candidate heart rate value exceeds the heart rate jump threshold allowed by the current confidence level, then the current heart rate value N is adjusted. cur The estimated value is the difference between the previous heart rate value and the heart rate jump threshold corresponding to the current confidence level. In other cases, where the absolute value of the difference between the previous heart rate value (i.e., the current heart rate value obtained in the previous cycle) and the candidate heart rate value is less than the heart rate jump threshold allowed by the current confidence level, the candidate heart rate value is directly used as the current heart rate value N. cur It's important to note the equals sign in the inequality formula here. Both critical points of the inequality are included in the corresponding calculations, which is for stylistic purposes rather than the method itself. If one or both critical points are categorized as "other," the candidate heart rate value is directly used as the current heart rate value N. cur This also falls within the scope of the present invention.

[0068] In this embodiment, multiple confidence levels with different confidence levels are set, and heart rate jump thresholds are assigned to each level. Then, by setting different current confidence levels, the heart rate jump threshold parameter used in the heart rate value estimation step is dynamically switched between different heart rate jump thresholds. That is, a strategy of dynamically adjusting the heart rate jump threshold is adopted in the heart rate tracking step based on the threshold. The appropriate heart rate jump threshold is selected according to the indicators of the real-time signal to ensure the accuracy and continuity of heart rate value estimation and avoid the problems of easily mistracking spectral peaks and generating physiologically unreasonable heart rate values.

[0069] Example 2

[0070] A heart rate estimation method based on optical measurements:

[0071] In this confidence level assessment step, the preset indicators include spectral quality indicators for evaluating spectral quality; the spectral quality indicators include one or more spectral quality parameters; each spectral quality parameter has one or more spectral quality thresholds; based on the priority of the spectral quality parameters and in combination with the corresponding spectral quality thresholds, the current confidence level is assessed.

[0072] In this embodiment, various parameters, including but not limited to spectral entropy, spectral kurtosis, peak factor, or fusion indices based on peak factors within local frequency bands, can be used as the spectral quality parameter. A threshold can be set for each parameter used as the spectral quality parameter, and a priority order and weighting can be assigned according to the priority of each spectral quality parameter, i.e., according to its importance. Combined with the corresponding spectral quality threshold, the current confidence level is assessed, thus incorporating spectral quality factors into the heart rate value estimation.

[0073] Example 3

[0074] A heart rate estimation method based on optical measurements:

[0075] In this confidence level assessment step, the preset indicators also include a heart rate continuity indicator, which is used to evaluate the continuity between the candidate heart rate value and the previous heart rate value; combined with one or more preset continuity thresholds, the current confidence level is assessed.

[0076] In this embodiment, candidate heart rate values ​​are used. Compared to the previous heart rate value N prev The variation between the two measures whether heart rate changes remain continuous. Specifically, the magnitude of the change can be measured by the difference between the two values, the ratio between the two values, or a combination of both. For example, it can be required that... Here, α is the threshold for the difference between adjacent estimated heart rates, measured in bpm. It can be set independently or selected from the pre-set heart rate jump thresholds for each confidence level (θ1 is preferred). This value can also be dynamically changed, such as always using the heart rate jump threshold corresponding to the current confidence level. Measuring the magnitude of the change by the ratio of the two values ​​also requires setting a corresponding threshold. This threshold can be static or dynamic, changing with the previous heart rate value. For example, a 10% change, from 100 to 110, is different from an increase from 150 to 165. The heart rate continuity index needs to consider the physiological characteristics of the human body, setting a reasonable threshold range, or incorporating other indicators as references, such as maximum oxygen uptake, resting heart rate, current exercise pace (measured by the same product, read from other devices, or entered by the user). The beneficial effect is that incorporating the heart rate continuity index can improve the continuity of heart rate estimation, especially reducing the occurrence of physiologically unreasonable data.

[0077] Example 4

[0078] A heart rate estimation method based on optical measurements:

[0079] In the confidence level assessment step, the preset index also includes an exercise intensity compliance index, which is used to determine whether the heart rate candidate value is within a reasonable range of variation. The reasonable range of variation is calculated based on the heart rate value estimated in the previous confidence state combined with the exercise intensity value. When the heart rate candidate value exceeds the reasonable range of variation, the temporary loss threshold θ3 is reset to θ3'; θ3' > θ3.

[0080] In this embodiment, the preset index also includes the exercise intensity compliance index, which calculates the candidate heart rate value from another perspective. The physiological rationality of this is considered. Current heart rate monitoring electronic devices generally incorporate accelerometers, which can classify different exercise intensity levels based on the acceleration parameters provided. When a certain time elapses since the last confidence state, such as 20 seconds, the current reasonable heart rate range is extrapolated based on the trend of exercise intensity changes. If the estimated current heart rate value is outside this range, the threshold for temporary loss of confidence can be reset to be larger, ensuring that the final output current heart rate value N is accurate. cur The heart rate should increase or decrease rapidly until it enters a reasonable range. If the current heart rate value is within this range, the parameter settings are reasonable and can be continued for slow tracking.

[0081] Example 5

[0082] A heart rate estimation method based on optical measurements:

[0083] The heart rate estimation step also includes adjusting the current heart rate value N based on the user's selected exercise type and intensity. cur .

[0084] In this embodiment, the user's selected exercise type and exercise intensity value are combined to assign an adjustment value N to the current heart rate. cur Greater flexibility allows it to better adapt to the characteristics of movement in real-world scenarios. Heart rate fluctuations vary significantly across different types of exercise. For example, heart rate is generally stable during yoga, while exercises like interval sprinting often involve sudden increases in speed followed by abrupt rest. In such cases, the heart rate threshold can be adjusted based on the characteristics of the exercise itself and the user's current exercise intensity. This can be done either by maintaining the heart rate threshold or by adjusting the current heart rate value N while simultaneously changing the heart rate threshold. cur Make timely adjustments and corrections.

[0085] Example 6

[0086] A heart rate estimation method based on optical measurements:

[0087] In the heart rate estimation step, for users with historical exercise data, the current heart rate value N is adjusted by combining the user's personal heart rate characteristic curve downloaded from the cloud platform. cur .

[0088] In this embodiment, the use of historical exercise data fully considers the personalized characteristics of users. When collecting user data, authorization should be requested from the user to store this data in the cloud, analyze it, and use it to provide better services. After obtaining the user's authorization, the user's historical exercise data can be analyzed through the cloud platform to establish a heart rate change model that includes factors such as the user's exercise habits and physical condition, i.e., the user's personal heart rate characteristic curve. For example, by repeatedly collecting the user's steady-state heart rate at different paces, the user's maximum oxygen uptake can be inferred. The user's maximum oxygen uptake can be used to directly infer the user's heart rate level in various intensity ranges (such as brisk walking, marathon pace, lactate threshold running), thereby correcting the heart rate model. For situations where the user's signal quality is poor or they enter a transient loss state for an extended period, the model is used to adjust the current heart rate value N. cur This can compensate for the problem of inaccurate estimation caused by temporary loss of signal and limited heart rate change rate (due to the low heart rate change threshold θ3). This allows the data to better reflect the user's individual characteristics, laying the foundation for further accumulation of more data with similar characteristics.

[0089] Example 7

[0090] A heart rate estimation method based on optical measurements, further comprising the following steps in the confidence level assessment:

[0091] In this confidence level assessment step, the current confidence level is assessed in conjunction with time parameters; these time parameters include the time of temporary loss N1, the verification time N2, and the confidence time N3.

[0092] Indicator A is used as the criterion for judging whether the spectral quality index meets the confidence state condition; Indicator B is used as the criterion for judging whether the heart rate continuity index meets the confidence state condition.

[0093] The condition for this temporary loss of confidence is set at a higher confidence level as follows: N1 consecutive seconds of non-compliance with index A.

[0094] The condition for setting from a lower confidence level to the verification state is that index A is met for N2 consecutive seconds;

[0095] The condition for changing from the verification state to the confidence state is that both indicator A and indicator B are met simultaneously for N3 consecutive seconds.

[0096] The condition for setting a confidence level from a confidence state to a lower confidence level than the verification state is that it cannot simultaneously meet both indicator A and indicator B.

[0097] This embodiment provides a method for further fine-tuning the technical solution of the present invention. The core of this method lies in controlling the current confidence level assessment conditions, which can be adjusted as needed. Each assessment is equivalent to an evaluation; a positive assessment results in a promotion from a lower confidence level to a higher one, while a negative assessment results in a decrease from a higher confidence level to a lower one. It is important to note that confidence levels are relative. Sorted from highest to lowest confidence level, the three confidence levels given in the example are, in order, the confidence state, the verification state, and the transient loss state. The transient loss state is the lowest level of confidence; compared to it, all other confidence levels are high confidence levels, and any subsequent increases in confidence level will also be higher than the transient loss state. The confidence state is the highest level of confidence; compared to it, all other confidence levels are low confidence levels. The low confidence levels relative to the verification state include the transient loss state and any subsequent increases in confidence level. That is, the confidence state is the highest confidence level, the verification state is the second highest confidence level, and the transient loss state is the lowest confidence level. To move from a low confidence level to a high confidence level, positive feedback is needed; to remain in the current confidence state, positive feedback is needed continuously; receiving negative feedback will lower the confidence level; and continuously receiving negative feedback will keep the confidence level in the transient loss state. A more cautious approach than simply raising or lowering the confidence level after receiving a single positive or negative feedback is to wait N seconds before changing the confidence level. Therefore, it should be understood that this scheme can also be designed so that upgrading or downgrading between two adjacent confidence levels requires M1 seconds and M2 seconds respectively; the M1 seconds and M2 seconds between each pair of adjacent confidence levels can be set individually or uniformly. This embodiment presents a preferred and more effective scheme, such as... Figure 2 As shown. The condition for downgrading from other confidence levels to a transiently lost state is that the state fails to meet indicator A for N1 consecutive seconds. Even if the state was originally at that confidence level, it is directly downgraded to the transiently lost state, resulting in a very rapid and strict downgrade. However, the promotion of confidence levels is more stringent. Before being promoted to a confidence state, it must first be promoted to a verification state. The condition for setting from a low confidence level to a verification state is that the state meets indicator A for N2 consecutive seconds. The condition for promoting from a verification state to a confidence state is that the state simultaneously meets both indicator A and indicator B for N3 consecutive seconds. In addition to the time parameter, the promotion condition also includes the requirement to simultaneously meet indicator B. Selectively increasing the types of indicators that must be met ensures that the data in the confidence state is relatively reliable in terms of both spectral quality and physiological rationality, without imposing excessively high standards when the signal quality is poor and the confidence level is low. The criteria for downgrading from a confidence state are also quite strict. If indicator A and indicator B cannot be met simultaneously, the confidence level is immediately reduced, and it is reduced to a confidence level below the verification state. In this embodiment, a confidence level lower than the verification state is a transiently lost state. However, if the confidence level is increased, the number of confidence levels lower than the verification state will not be limited to transiently lost states. Overall, as... Figure 2 As shown, this embodiment provides a relatively rigorous parameter setting method to control the assessment conditions of the current confidence level. It can both improve the assessment standard by strictly setting the parameters and lower the assessment standard by lowering the parameter standard. It is applicable to a variety of different application scenarios and improves the convenience and portability of the program.

[0098] Example 8

[0099] A heart rate estimation method based on optical measurements:

[0100] The confidence level also includes the questioning state; the confidence level of the questioning state is between the verified state and the transient loss state; the heart rate jump threshold of the questioning state is the questioning state threshold θ4; θ3 < θ4 < θ2; for the transient loss state, the confidence level C < k3;

[0101] The condition for changing from a verification state to a questioning state is that it cannot meet indicator A;

[0102] When both indicator A and indicator B cannot be met simultaneously, the lower confidence level that the confidence state drops from the verified state is the questioned state.

[0103] In this embodiment, the questioning state is added based on the aforementioned embodiments, such as... Figure 3As shown, the confidence level of the questioning state lies between the validating state and the transient loss state, making the heart rate estimation highly unreliable. Adding this questioning state allows for further subdivision of the confidence level, enriching the confidence tiers and avoiding a blanket treatment of all data below the validating state as transient loss states. This accumulates more usable data for states with poor spectral quality, which is beneficial for further improving the heart rate estimation level. It should be noted that this embodiment only illustrates the case of adding up to four confidence levels; this is not a limitation on the number of confidence levels. The technical solution of this invention fully supports setting more confidence levels when necessary.

[0104] Second aspect: Electronic devices

[0105] Example 9

[0106] An electronic device includes a memory and a processor; the memory is used to store a computer program; the processor is used to implement any of the heart rate estimation methods shown in the first aspect of the invention when the computer program is executed.

[0107] Third aspect: Computer-readable storage media

[0108] A computer-readable storage medium storing a computer program that, when executed by a processor, implements any of the heart rate estimation methods shown in the first aspect of the invention.

[0109] Comprehensive Example

[0110] The sampling frequency was set to 25Hz. After a 1024-point Fourier transform, the frequency domain resolution was 0.024Hz. The analyzed frequency band was from 0.48Hz to 3.44Hz, containing a total of 120 frequency points. Let the spectrum of this frequency band be X, where X(i) represents the i-th frequency point (the value of i ranges from 1 to 120).

[0111] Step 1:

[0112] Set the confidence threshold θ1 = 5 bpm, the verification threshold θ2 = 4 bpm, the challenge threshold θ4 = 3 bpm, and the temporary loss threshold θ3 = 2 bpm. Set the initial state to the confidence state.

[0113] Step 2:

[0114] For index A, the peak factor within a local frequency band near the peak value is used as the spectral quality satisfaction condition. The peak factor CF within the local frequency band is calculated as follows:

[0115]

[0116] p represents the selected peak point. τ = 15, k = 2.5, with CF > 15 as the condition. RMS indicates root mean square calculation.

[0117] For indicator B, set the threshold α for the difference between adjacent estimated heart rates to 5 bpm.

[0118] For index C, a 5-level exercise intensity k is set according to the triaxial acceleration plan. The value of k ranges from 1 to 5. When the distance from the last confidence state exceeds 20 seconds, the current reasonable heart rate range is extrapolated based on the trend of the change in exercise intensity. If the estimated heart rate is not within this range, a threshold setting of θ3 = 5 bpm is used for fast tracking; otherwise, θ3 = 3 bpm is set for slow tracking.

[0119] The tracking method in transient loss state can avoid losing tracking flexibility in transient loss state. Simultaneously, other extrapolation methods for transient loss state are enabled: in cases of poor spectral quality and near-failure, heart rate trend estimation is performed based on the user's selected exercise type (yoga, rope skipping, etc.) and exercise intensity k according to the motion model. When historical user exercise data is available, estimation is also performed using the user's individual heart rate level.

[0120] Step 3:

[0121] Let N1 = 5s, meaning that the signal quality deteriorates for 5 consecutive seconds, and then proceed to a state transition from a higher confidence level to a temporary loss of state.

[0122] Let N2 = 3s, meaning that the signal quality is good for 3 consecutive seconds, and then proceed with the state transition from the lower confidence level to the verification state.

[0123] Let N3 = 4s, that is, the condition is met for 4 consecutive seconds, and then the state transition from the verification state to the confidence state is carried out.

[0124] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Modifications and improvements to the present invention are possible without exceeding the concept and scope defined by the claims. Therefore, the content of the embodiments in this specification should not be construed as a limitation of the present invention.

Claims

1. A heart rate estimation method based on optical measurements, characterized in that, include: The confidence level setting step involves setting three or more confidence levels; the confidence levels include a confidence state, a verification state, and a transient loss state; in the confidence state, the confidence level C ≥ k1, the heart rate estimation is reliable; in the verification state, k2 ≤ confidence level C < k1, the heart rate estimation is relatively reliable; in the transient loss state, the confidence level C < k2, the heart rate estimation is unreliable. The heart rate jump threshold setting steps involve setting a heart rate jump threshold for each confidence level; the heart rate jump threshold for the confidence state is the confidence state threshold θ1, the heart rate jump threshold for the validation state is the validation state threshold θ2, and the heart rate jump threshold for the transient loss state is the transient loss state threshold θ3; θ3 < θ2 < θ1 In the initial state setting step, one of the confidence levels is selected as the initial confidence level; The confidence level assessment step involves assessing the current confidence level based on preset indicators; the current confidence level is one of the aforementioned confidence levels. a heart rate value estimation step, estimating a current heart rate value N according to the following formula cur : in, N is a candidate heart rate value determined by selecting peak values. prev This is the previous heart rate value, θ is the heart rate jump threshold for the current confidence level, and N is the previous heart rate value. cur It is the result after heart rate jump threshold correction, and is used as the current heart rate value.

2. The heart rate estimation method according to claim 1, characterized in that: In the confidence level assessment step, the preset indicators include spectral quality indicators for evaluating spectral quality; the spectral quality indicators include one or more spectral quality parameters; each spectral quality parameter has one or more spectral quality thresholds; the current confidence level is assessed based on the priority of each spectral quality parameter and in combination with the corresponding spectral quality thresholds.

3. The heart rate estimation method according to claim 2, characterized in that: In the confidence level assessment step, the preset indicators also include a heart rate continuity indicator, which is used to evaluate the continuity between the candidate heart rate value and the previous heart rate value; and the current confidence level is assessed by combining one or more preset continuity thresholds.

4. The heart rate estimation method according to claim 3, characterized in that: In the confidence level assessment step, the preset index also includes an exercise intensity compliance index, which is used to determine whether the candidate heart rate value is within a reasonable range of variation. The reasonable range of variation is calculated based on the heart rate value estimated in the previous confidence state combined with the exercise intensity value. When the candidate heart rate value exceeds the reasonable range of variation, the temporary loss threshold θ3 is reset to θ3'; θ3' > θ3.

5. The heart rate estimation method according to claim 4, characterized in that: In the heart rate value estimating step, further comprising adjusting the current heart rate value N in combination with a user-selected exercise type and the exercise intensity value cur .

6. The heart rate estimation method according to claim 5, characterized in that: In the heart rate value estimation step, for users with historical motion data, the current heart rate value N is adjusted in combination with the user's personal heart rate characteristic curve downloaded from the cloud platform cur .

7. The heart rate estimation method according to claim 3, characterized in that, The confidence level assessment step also includes: In the confidence level assessment step, the current confidence level is assessed in conjunction with time parameters; the time parameters include the time of temporary loss N1, the verification time N2, and the confidence time N3. Indicator A is used as the criterion for judging whether the spectral quality index meets the confidence state condition; Indicator B is used as the criterion for judging whether the heart rate continuity index meets the confidence state condition. The condition for setting the transient loss state from a higher confidence level is that the index A is not met for N1 consecutive seconds; The condition for setting the verification state from a lower confidence level is that the index A is met for N2 consecutive seconds; The condition for setting the state from the verification state to the confidence state is that both indicator A and indicator B are met simultaneously for N3 consecutive seconds. The condition for setting the confidence state to a lower confidence level than the verification state is that it cannot simultaneously meet both indicator A and indicator B.

8. The heart rate estimation method according to claim 7, characterized in that: The confidence level also includes a questioning state; the confidence level of the questioning state is between the verification state and the transient loss state; the heart rate jump threshold of the questioning state is the questioning state threshold θ4; θ3 < θ4 < θ2; The condition for changing from the verification state to the challenged state is that it cannot meet indicator A; When both Indicator A and Indicator B cannot be met simultaneously, the lower confidence level that the confidence state drops from is the questioning state.

9. An electronic device, characterized in that, It includes a memory and a processor; the memory is used to store a computer program; the processor is used to implement the heart rate estimation method as described in any one of claims 1-8 when the computer program is executed.

10. A computer-readable storage medium, characterized in that, The storage medium stores a computer program that, when executed by a processor, implements the heart rate estimation method as described in any one of claims 1-8.

Images