Method and apparatus for monitoring user physiology using photoplethysmography signals

Abstract

This disclosure relates generally to a system and a computerized method (200) for monitoring a user's physiology using photoplethysmography (PPG) signals (310). The method (200) includes the steps of: receiving a set of PPG signals (310), each PPG signal being measured from a different location on the user; identifying a plurality of 2D pulse wavelets (312) from each PPG signal (310); 3D aligning each of the 2D pulse wavelets (312); and constructing a set of 3D morphological representations (400) from the aligned pulse wavelets (312), on which a set of the user's physiological parameters (410) can be measured.

Description

Technical Field

[0001] Cross - reference to Related Applications This disclosure claims the benefit of Singapore Patent Application No. 10202300873U, filed on March 30, 2023, the entire content of which is incorporated herein by reference.

[0002] This disclosure generally relates to methods and apparatuses for monitoring user physiology using photoplethysmography signals. More particularly, this disclosure describes various embodiments of methods and apparatuses for monitoring user physiological parameters, such as blood pressure, using photoplethysmography signals measured from a user.

Background Art

[0003] Many monitoring systems use various mechanical, electrical, and optical components to measure the physiological state of a user, such as a heart monitoring system for measuring cardiovascular parameters. One example is an ambulatory blood pressure monitor incorporating an inflatable cuff that wraps around a user's limb. Another example is an electrocardiogram, such as a Holter monitor, which requires bothersome and obstructive patch electrodes. The application of such electrodes and cuffs can be very bothersome and impractical during long - term use or during strenuous physical activity. In addition, the inflation of the cuff causes significant constriction of the user's limb, making the cuff inappropriate for daytime or, more critically, nighttime continuous wear.

[0004] Blood pressure can be measured indirectly using auscultation or oscillometric methods. Auscultation is based on detecting Korotkoff sounds emitted from an acoustic transducer signal, while oscillometric methods are based on detecting changes in cuff volume. Both auscultation and oscillometric methods measure blood pressure through arterial occlusion, making the measurement protocols unsuitable for rapid measurement. While the upper arm cuff method is the clinically recognized standard for blood pressure monitoring, it is not suitable for certain situations where continuous blood pressure monitoring is desirable, such as routine orthostatic hypotension. For example, cuff-based blood pressure monitors cannot be used to monitor sleep patterns without disturbing the patient, as the expansion of the upper arm or wrist cuff pressure, combined with an increase in systemic blood pressure, can cause disturbances.

[0005] Orthostatic hypotension, or postural hypotension, is an abnormally rapid drop in blood pressure from a sitting or lying position, causing reduced blood flow to the brain, as illustrated in Figure 1A. This can lead to symptoms such as dizziness or lightheadedness, unsteadiness, visual disturbances, and sometimes fainting. The risk of adverse outcomes is closely linked to the risk of falls and can affect an individual's quality of life; therefore, the risk increases even if symptoms are absent.

[0006] There are four main subtypes of orthostatic hypotension: initial orthostatic hypotension, delayed blood pressure recovery, classic orthostatic hypotension, and delayed orthostatic hypotension. Clinical symptoms vary among subtypes, ranging from cognitive impairment with unconscious hypotension or falls of unknown cause to classic syncope and syncope.

[0007] Orthostatic hypotension is diagnosed when blood pressure drops by 20 mmHg or more systolic and 10 mmHg or more diastolic within 3 minutes of standing up after lying supine or at a 60° angle on a tilt table for 5 minutes. The sudden drop in blood pressure may be due to autonomic reflex dysfunction, insufficient dosage, or adverse reactions to medication. Symptoms are usually associated with reduced blood flow to the brain, but many patients may be asymptomatic. This disease process is characterized by frequent falls, high morbidity and mortality rates, and numerous hospitalizations.

[0008] Diagnosing orthostatic hypotension is difficult, and current practice requires a detailed medical history, examinations, and both supine and upright blood pressure measurements. In-clinic evaluation requires the patient to undergo blood pressure measurements in three or four scenarios: 5 minutes after lying down, 1 minute after standing, 3 minutes after standing, and tilt table measurements may be required for these measurements.

[0009] These measurements are time-consuming, labor-intensive, and require a physician to perform. Furthermore, orthostatic hypotension can be asymptomatic and may not present any signs in a clinic setting, making measurements unreliable. Therefore, a physician may ask the patient to wear a 24-hour ambulatory blood pressure monitor, as illustrated in Figure 1B. However, ambulatory blood pressure monitors, consisting of a blood pressure machine 100 and cuff 110, are obstructive, uncomfortable, and do not measure changes in the patient's posture.

[0010] Conventional monitoring devices are used in clinical settings where the user is typically stationary, such as when a patient under medical observation is sitting or lying down. However, these devices are not suitable for individuals who frequently need access to blood pressure or cardiac data while performing a series of daily routines, such as during office work, driving, or sports activities. Conventional devices designed for the clinical assessment of such physiological parameters are ineffective or impractical for continuous monitoring during daily activities or sleep. [Overview of the Initiative] [Problems that the invention aims to solve]

[0011] Therefore, in order to address or mitigate at least one of the aforementioned problems and / or disadvantages, it is necessary to provide improved methods and apparatus for monitoring the physiological parameters of users. [Means for solving the problem]

[0012] According to a first aspect of this disclosure, there are systems and computerized methods for monitoring a user's physiology using PPG signals. The method is A step of receiving a set of PPG signals measured from a user, wherein each PPG signal is measured from a different location on the user. A step of identifying multiple pulse waves from each PPG signal, wherein each pulse wavelet is defined 2D by an amplitude axis and a first time axis, For each PPG signal, the steps include: aligning each 2D pulse wavelet in 3D along a second time axis; Steps to construct a set of 3D morphological representations from aligned pulse wavelets and Includes, Based on the constructed 3D morphological representation, a set of the user's physiological parameters can be measured.

[0013] According to a second aspect of this disclosure, there is a measuring device for monitoring a user's hemodynamic state using PPG signals. This measuring device is Multiple PPG sensors for measuring multiple PPG signals from different locations on the user, An inertial measurement unit for measuring motion data from the user, It is a processor, The first hemodynamic profile of the user is calculated from the PPG signal and the first machine learning model. The system calculates the user's second hemodynamic profile from the first hemodynamic profile, motion data, and a second machine learning model. A processor and It is equipped with.

[0014] According to a third aspect of this disclosure, there are systems and computerized methods for monitoring a user's hemodynamic state using PPG signals. The method is Receiving a plurality of PPG signals measured from a user, wherein each PPG signal is measured from a different position on the user; Receiving motion data measured from a user using an inertial measurement unit; Generating a first hemodynamic profile of the user from the PPG signals and a first machine learning model; Generating a second hemodynamic profile of the user from the first hemodynamic profile, the motion data, and a second machine learning model; and including.

[0015] A method and apparatus for monitoring user physiology using a photoplethysmography signal according to the present disclosure are disclosed herein for that purpose. Various features, aspects, and advantages of the present disclosure will become more apparent from the following detailed description of embodiments of the present disclosure and from the accompanying drawings, which are presented by way of non-limiting examples only.

Brief Description of the Drawings

[0016] [Figure 1A] A diagram of an existing method for measuring orthostatic hypotension. [Figure 1B] A diagram of an existing method for measuring orthostatic hypotension. [Figure 2A] A diagram of a method for monitoring a user's physiology using a PPG signal according to an embodiment of the present disclosure. [Figure 2B] A diagram of a method for monitoring a user's physiology using a PPG signal according to an embodiment of the present disclosure. [Figure 3A] A diagram of a PPG sensor disposed on a user and PPG signals measured by the PPG sensor. [Figure 3B] A diagram of a PPG sensor disposed on a user and PPG signals measured by the PPG sensor. [Figure 4A] A diagram for identifying a pulse wavelet of a PPG signal. [Figure 4B] A diagram for identifying a pulse wavelet of a PPG signal. [Figure 4C] It is a diagram for identifying the pulse wavelet of the PPG signal. [Figure 4D] It is a diagram for identifying the pulse wavelet of the PPG signal. [Figure 4E] It is a diagram for identifying the pulse wavelet of the PPG signal. [Figure 4F] It is a diagram for identifying the pulse wavelet of the PPG signal. [Figure 5A] It is a diagram for aligning the pulse wavelet of the PPG signal. [Figure 5B] It is a diagram for aligning the pulse wavelet of the PPG signal. [Figure 6A] It is a diagram of a 3D morphological representation constructed from the aligned pulse wavelets. [Figure 6B] It is a diagram of a 3D morphological representation constructed from the aligned pulse wavelets. [Figure 6C] It is a diagram of a 3D morphological representation constructed from the aligned pulse wavelets. [Figure 6D] It is a diagram of a 3D morphological representation constructed from the aligned pulse wavelets. [Figure 7A] It is a diagram of a grid matrix for aligning the pulse wavelet. [Figure 7B] It is a diagram of a grid matrix for aligning the pulse wavelet. [Figure 7C] It is a diagram of a grid matrix for aligning the pulse wavelet. [Figure 7D] It is a diagram of a grid matrix for aligning the pulse wavelet. [Figure 8] It is a diagram of another grid matrix for aligning the pulse wavelet. [Figure 9A] It is a diagram of another grid matrix for aligning the pulse wavelet. [Figure 9B]This is a diagram of another grid matrix for aligning pulse wavelets. [Figure 9C] This is a diagram of another grid matrix for aligning pulse wavelets. [Figure 9D] This is a diagram of another grid matrix for aligning pulse wavelets. [Figure 9E] This is a diagram of another grid matrix for aligning pulse wavelets. [Figure 9F] This is a diagram of another grid matrix for aligning pulse wavelets. [Figure 9G] This is a diagram of another grid matrix for aligning pulse wavelets. [Figure 9H] This is a diagram of another grid matrix for aligning pulse wavelets. [Figure 9I] This is a diagram of another grid matrix for aligning pulse wavelets. [Figure 9J] This is a diagram of another grid matrix for aligning pulse wavelets. [Figure 10A] This is a diagram of a 3D morphological representation constructed from a grid matrix and a different number of PPG signals. [Figure 10B] This is a diagram of a 3D morphological representation constructed from a grid matrix and a different number of PPG signals. [Figure 10C] This is a diagram of a 3D morphological representation constructed from a grid matrix and a different number of PPG signals. [Figure 10D] This is a diagram of a 3D morphological representation constructed from a grid matrix and a different number of PPG signals. [Figure 11A] This is a diagram showing 3D form representations for different users. [Figure 11B] This is a diagram showing 3D form representations for different users. [Figure 11C] This is a diagram showing 3D form representations for different users. [Figure 12A]This is a diagram of two PPG signals and their pulse wavelets. [Figure 12B] This is a diagram of two PPG signals and their pulse wavelets. [Figure 13A] This diagram shows how to calculate pulse propagation time from two PPG signals. [Figure 13B] This diagram shows how to calculate pulse propagation time from two PPG signals. [Figure 13C] This diagram shows how to calculate pulse propagation time from two PPG signals. [Figure 13D] This diagram shows how to calculate pulse propagation time from two PPG signals. [Figure 13E] This diagram shows how to calculate pulse propagation time from two PPG signals. [Figure 14A] This diagram shows how to estimate blood glucose levels from a 3D morphological representation. [Figure 14B] This diagram shows how to estimate blood glucose levels from a 3D morphological representation. [Figure 14C] This diagram shows how to estimate blood glucose levels from a 3D morphological representation. [Figure 15A] This figure shows a system and method for monitoring a user's hemodynamic state using PPG signals, according to an embodiment of the present disclosure. [Figure 15B] This figure shows a system and method for monitoring a user's hemodynamic state using PPG signals, according to an embodiment of the present disclosure. [Figure 16A] This is a diagram of a measuring device for measuring PPG signals and motion data to monitor a user's hemodynamic state, according to an embodiment of the present disclosure. [Figure 16B] This is a diagram of a measuring device for measuring PPG signals and motion data to monitor a user's hemodynamic state, according to an embodiment of the present disclosure. [Figure 17A] This figure shows the estimated and actual blood pressure results for a group of subjects. [Figure 17B] This figure shows the estimated and actual blood pressure results for a group of subjects. [Figure 17C] This figure shows the estimated and actual blood pressure results for a group of subjects. [Figure 18] This is a diagram of a system for monitoring a user's physiological state using PPG signals, motion data, and a 3D morphological representation, according to an embodiment of the present disclosure. [Modes for carrying out the invention]

[0017] For the purposes of brevity and clarity, the description of embodiments of this disclosure, as shown in the drawings, relates to methods and apparatus for monitoring user physiology using photoplethysmography signals. While embodiments provided herein will be described in conjunction with the embodiments described herein, it will be understood that they are not intended to limit the disclosure to these embodiments. Conversely, this disclosure is intended to encompass alternatives, variations, and equivalents of the embodiments described herein, which fall within the scope of the disclosure as defined by the appended claims. Furthermore, specific details are provided in the following detailed description to provide a complete understanding of the disclosure. However, it will be recognized by persons with ordinary art in the art that the disclosure may be implemented without specific details and / or with numerous details arising from combinations of embodiments of a particular set of embodiments. In some cases, well-known systems, methods, procedures, and components are not described in detail so as not to unnecessarily obscure the embodiments of this disclosure.

[0018] In embodiments of this disclosure, the depiction of a given element in a particular figure, or the consideration or use of a particular element number, or a reference to it in the corresponding descriptive material, may encompass the same, equivalent, or similar element or element number identified in another figure or descriptive material associated therewith.

[0019] References to “one embodiment / example,” “another embodiment / example,” “some embodiments / examples,” “some other embodiments / examples,” etc., indicate that the embodiments / examples described in that way may include certain features, structures, characteristics, properties, elements, or limitations, but not all embodiments / examples necessarily include those particular features, structures, characteristics, properties, elements, or limitations. Furthermore, repeated use of the phrase “in one embodiment / example” or “in another embodiment / example” does not necessarily refer to the same embodiment / example.

[0020] The terms "equipped with," "includes," "possess," etc., do not preclude the existence of other features / elements / steps other than those listed in one embodiment. The enumeration of certain features / elements / steps in different embodiments does not indicate that combinations of these features / elements / steps cannot be used in one embodiment.

[0021] Where used herein, the terms “a (one)” and “an (one)” are defined as one or more. The use of “ / ” in figures or related text is understood to mean “and / or” unless otherwise indicated. The term “set” is defined, according to a known mathematical definition, as a non-empty finite organization of elements that mathematically exhibit a cardinality of at least 1 (for example, a set as defined herein may correspond to a unit, a unary or single-element set, or a multi-element set). Enumerations of particular numbers or ranges of values ​​herein are understood to include or be enumerations of approximate numbers or ranges of values. The terms “first,” “second,” etc., are used merely as labels or identifiers and are not intended to impose numerical requirements on their related terms.

[0022] In a representative or illustrative embodiment of the present disclosure, as illustrated in Figures 2A and 2B, there is a computerized implementation or method 200 for monitoring a user's physiology using photoplethysmography (PPG) signals 310. The method 200 includes the step 210 of receiving a collection of PPG signals 310 measured from a user, each PPG signal 310 being measured from a different location on the user. The PPG signals 310 are measured from a collection of PPG sensors 300 positioned at one or more locations on the user's body. For example, the PPG sensors 300 are configured to measure the PPG signals 310 at a frequency of at least 300 or 400 Hz.

[0023] In some embodiments, the set of PPG signals 310 includes only one PPG signal 310 measured from a single location on the user's body. In other embodiments, the set of PPG signals 310 includes multiple PPG signals 310. Preferably, an array of PPG sensors 300 are placed at multiple locations on the user to measure a large number of PPG signals 310.

[0024] The PPG sensor 300 optically measures changes in arterial blood volume and is synchronized with the cardiac cycle. Each PPG sensor 300 has a light-emitting diode (LED) and a photodiode. The LED emits light toward the user's skin at its respective position, and the change in blood volume at that position is measured from the amount of light transmitted or reflected by the photodiode. For example, a PPG sensor 300 placed on the user's forehead measures the light reflection from the forehead, while a PPG sensor 300 placed on the user's fingertip measures the light transmission and absorption by the fingertip.

[0025] The PPG sensors can be positioned at various locations on the user. For example, as shown in Figure 3A, PPG sensor 300a is placed near the temporal artery on the forehead, PPG sensor 300b is placed on the earlobe, PPG sensor 300c is placed near the axillary artery on the clavicle, PPG sensor 300d is placed near the brachial artery on the inside of the elbow, PPG sensor 300e is placed near the radial artery at the proximal wrist, PPG sensor 300f is placed near the radial artery at the distal wrist, PPG sensor 300g is placed near the palmar digital artery at the fingertips, and PPG sensor 300h is placed on the toes. Figure 3B illustrates the respective PPG signals 310 measured from the forehead PPG sensor 300a, earlobe PPG sensor 300b, clavicle PPG sensor 300c, proximal wrist PPG sensor 300e, fingertip PPG sensor 300g, and toe PPG sensor 300h.

[0026] Method 200 includes step 220 of identifying a plurality of pulse wavelets 312 from each PPG signal 310. Each pulse wavelet 312 represents a single pulse in the cardiac cycle and exhibits a wave-like oscillation. Each pulse wavelet 312 is defined in two dimensions (2D) by an amplitude axis and a first time axis. In particular, the amplitude axis is the vertical axis and the first time axis is the horizontal axis.

[0027] Various methods can be used to identify pulse wavelets 312. For example, good quality pulse wavelets 312 can be identified from the PPG signal 310 while low quality pulse wavelets 312 are discarded. Figure 4A illustrates a periodic windowing algorithm used to identify good quality pulse wavelets 312. This algorithm detects periodicity in time series analysis and requires four steps to recognize repeating patterns at regular intervals. In the first transformation step, the PPG signal 310 is transformed from the time domain to the frequency domain using a Discrete Fourier Transform (DFT), thereby recognizing periodicity and facilitating the removal of high-frequency regions that do not typically occur biologically, i.e., noise from the PPG signal 310.

[0028] In the second step, important morphological features 314 of the PPG signal 310 are detected in order of priority of occurrence. Peaks corresponding to the frequencies of periodic components are detected in both the time and frequency domains. Important morphological features 314, including peaks and troughs of the PPG signal 310, are identified and subjected to a set of conditions that classify them as one of the following: systolic peak 314a, trough 314b, overlapping notch 314c, or diastolic peak 314d (see Figure 13B). A morphological structure scale based on periodic features may be extracted from the sequence of morphological features 314. For more meaningful structural clustering and visualization of the sequence of morphological features 314, a periodic distance scale may be used, whether they are periodic or not. Once peaks 314a and trough 314b are identified, the PPG signal 310 can be split or windowed into individual pulse wavelets 312 with their respective prominent peaks 314a and trough 314b.

[0029] As illustrated in Figure 4B, the order of occurrence of morphological features 314 in each pulse wavelet 312 is trough 314b, systolic peak 314a, overlapping notch 314c, diastolic peak 314d, and trough 314b. Initial windowing is performed to show that a single pulse wavelet 312 consists of a single systolic peak 314a located between two troughs 314b. The overlapping notch 314c is identified by limiting the second derived pulse and frequency range. The diastolic peak 314d is determined by limiting the subsequent gradient change of the first derived pulse, corresponding to the point where the first derived pulse intersects the x-axis or the first time axis. Each pulse wavelet 312 should have exactly one systolic peak 314a, one overlapping notch 314c, and one diastolic peak 314d. The detection of both overlapping notches 314c and diastolic peaks 314d is acceptable, but the absence of either is unacceptable. Thus, pulse wavelets 312 are classified as good or poor quality based on the presence of these morphological features 314 that indicate the pulse wavelet 312 is morphologically sound.

[0030] Alternatively or additionally, pulse wavelets 312 may be classified as good or bad quality based on a time interval-based segmentation method. The PPG signal 310 is segmented based on fixed time intervals. The interval would be classified as 'good' if potential pulse wavelets 312 can be found within the interval, and otherwise as 'noisy' or 'poor'. For example, pulse wavelets 312 can never be shorter than 50 ms or longer than 3 seconds, since these would indicate heart rates faster than 300 BPM (beats per minute) or slower than 20 BPM, respectively, both of which are impossible for humans.

[0031] In the third step, an autocorrelation function is calculated to detect arbitrary self-similarity in time series at different delays that may exhibit periodicity. Thus, the autocorrelation is the inverse Fourier transform of the periodogram, meaning that the autocorrelation function can be considered as the dual of the periodogram from the time domain to the frequency domain. Next, based on predetermined selection criteria, each individual pulse wavelet 312 will be classified as good or poor quality with respect to their similarity to a previously occurring pulse wavelet 312. For example, as illustrated in Figure 4C, if the difference between the reconstruction of the input pulse wavelet 312' and the preceding pulse wavelet 312'' is small, the input pulse wavelet 312' will be classified as good quality. Conversely, as illustrated in Figure 4D, if the difference between the reconstruction of the input pulse wavelet 312' and the preceding pulse wavelet 312'' is large, the input pulse wavelet 312' will be classified as poor quality.

[0032] In the fourth step, the detected periodicities are clustered and filtered to remove false positives and identify dominant periodicities. The resulting good-quality pulse wavelets 312 are then spliced ​​together to construct a good-quality PPG signal 310', as shown in Figure 4A. The good-quality PPG signal 310' can be used to interpret the recognized periodicities in the context of the original time series of the original PPG signal 310.

[0033] In some embodiments, as illustrated in Figures 4E and 4F, every other pulse wavelet 312 may be identified from the good quality pulse wavelet 312 to construct the good quality PPG signal 310'. For example, an even number of good quality pulse wavelets 312 may be identified and spliced ​​together to construct the PPG signal 310'.

[0034] Method 200 includes step 230 of aligning a 2D pulse wavelet 312 along a second time axis for each PPG signal 310. In particular, the second time axis is perpendicular to the amplitude axis and the first time axis. Aligning the pulse wavelets 312 may be done using a process known as image alignment or image registration, which requires transforming image data from each pulse wavelet 312 into a common coordinate system. For example, a feature-based method may be used to image align or register the pulse wavelets 312. The feature-based method may establish correspondences between sets of morphological features 314 of the pulse wavelets 312, determine geometric transformations, and thereby establish feature-to-feature correspondences between the pulse wavelets 312. For example, the morphological features 314 may include one or more of a systolic peak 314a, a pulse trough 314b, a overlapping notch 314c, and an diastolic peak 314d. Figures 5A and 5B illustrate an example of aligning pulse wavelets 312 based on their morphological features 314, including at least a systolic peak 314a.

[0035] Method 200 includes step 240 of constructing a set of three-dimensional (3D) morphological representations 400 from aligned pulse wavelets 312. Figures 6A to 6D show illustrative 3D morphological representations 400 constructed from aligned pulse wavelets 312 derived from PPG signals 310 measured over periods of 30 seconds, 3 minutes, 30 minutes, and 2 hours, respectively. In particular, the 3D morphological representation 400 is defined by a vertical amplitude axis and two horizontal time axes. The 3D morphological representation 400 may also be referred to as a photoplethysmorphogram.

[0036] In some embodiments, as illustrated in Figures 7A to 7D, aligning pulse wavelets 312 may involve positioning each pulse wavelet 312 across a grid matrix 500 having an array of cells 510. From each pulse wavelet 312, a set of morphological features 314 is identified, such as one or more systolic peaks, pulse troughs, overlapping notches, and diastolic peaks. As shown in Figure 7A, when the morphological features 314 are aligned with their respective cells 510, they are added to the cells 510 of the grid matrix 500. Subsequent pulse wavelets 312 are successively added to their respective cells 510 as their respective morphological features 314 are aligned with their respective cells 510, as shown in Figures 7B to 7D. The resulting grid matrix 500 is the sum of the morphological features 314 from the thus aligned pulse wavelets 312.

[0037] In the examples illustrated in Figures 7A to 7D, the cells 510 of the grid matrix 500 are arranged in a 5x5 grid, and four pulse wavelets 312 are summed in the grid matrix 500. In another example illustrated in Figure 8, six pulse wavelets 312 are summed in a second grid matrix 500 having cells 510 arranged in a 6x6 grid. In yet another example illustrated in Figures 9A to 9J, ten pulse wavelets 312 are summed in a third grid matrix 500 having cells 510 arranged in a 10x10 grid. The third grid matrix 500 can be used to construct a finer, higher-resolution 3D morphological representation 400 than the first and second grid matrices 500.

[0038] It will be recognized that the grid matrix 500 can have any number of cells 510 arranged in any number of rows and columns. It will also be recognized that any number of pulse wavelets 312 can be added to the grid matrix. For example, Figures 10A to 10D show an illustrative grid matrix 500 and 3D morphological representations 400 constructed from pulse wavelets 312 extracted from 5, 10, 20, and 30 PPG signals 310, respectively.

[0039] As described above, a set of 3D morphological representations 400 is constructed from the aligned pulse wavelets 312. The set of 3D morphological representations 400 represents a concatenation of PPG signals 310 measured from one or more locations on the user's body. The set of 3D morphological representations 400 is unique to the user, and a set of the user's physiological parameters 410 can be measured based on each set of 3D morphological representations 400. For example, the physiological parameters may include blood pressure, blood glucose, arterial distensibility, vascular perfusion, heart rate variability, respiratory rate, and / or stress parameters.

[0040] Figures 11A–11C show some illustrative 3D morphological representations 400 specific to different users. Each 3D morphological representation 400 may be compared to other 3D morphological representations 400 to measure physiological parameters 410 and / or diagnose a medical condition in the user. For example, a user's 3D morphological representation 400 may be compared to previous 3D morphological representations 400 for the same user, a group of 3D morphological representations 400 for users with similar health profiles, and / or 3D morphological representations 400 for a larger population of people.

[0041] As shown in Figure 11A, the first 3D morphological representation 400A was constructed using a PPG signal 310 measured over a 3-minute period for a 28-year-old male user. The first 3D morphological representation 400A shows that the user has a blood pressure of 111 / 76 mmHg and a blood glucose of 74 mg / dL and no significant medical conditions.

[0042] As shown in Figure 11B, a second 3D morphological representation 400B was constructed using a PPG signal 310 measured over a 3-minute period for a 34-year-old female user. The second 3D morphological representation 400B indicates that the user had a blood pressure of 119 / 68 mmHg and a blood glucose level of 66 mg / dL, and no significant medical conditions.

[0043] As illustrated in Figure 11C, a third 3D morphological representation 400C was constructed using a PPG signal 310 measured over a 3-minute period for a 46-year-old male user. The third 3D morphological representation 400C shows that the user had a blood pressure of 114 / 74 mmHg and a blood glucose level of 97 mg / dL, and no significant medical conditions.

[0044] In some embodiments, constructing a set of 3D morphological representations 400 may involve constructing each 3D morphological representation 400 for each PPG signal 310 from the respective pulse wavelets 312 identified from each PPG signal 310, such that each 3D morphological representation 400 corresponds to one PPG signal 310. If a large number of PPG signals 310 are measured by the user, the same number of 3D morphological representations 400 will be constructed for the user. For example, as shown in Figure 12A, two PPG signals 310A and 310B are measured by the user at different locations. Pulse wavelets 312A and 312B are identified from each PPG signal 310A and 310B. The pulse wavelets 312A and 312B from each PPG signal 310A and 310B are aligned, thereby constructing the respective 3D morphological representations 400 corresponding to each PPG signal 310A and 310B.

[0045] In some embodiments, constructing a set of 3D morphological representations 400 may include constructing a single 3D morphological representation 400 from all pulse wavelets 312 identified from all PPG signals 310. If a large number of PPG signals 310 are measured by the user, only one 3D morphological representation 400 will be constructed for the user. For example, as shown in Figure 12B, pulse wavelets 312A and 312B are extracted from each PPG signal 310A and 310B. Figure 12B illustrates two sets of illustrative pulse wavelets 312A and 312B from two time windows—shown as 312A' and 312B' in the first time window, and 312A'' and 312B'' in the second time window. The pulse wavelets 312A and 312B from all PPG signals 310A and 310B are collectively aligned to construct a single 3D morphological representation 400 for the user.

[0046] In some embodiments, as illustrated in Figure 13A, there are a first PPG sensor 300A and a second PPG sensor 300B positioned proximal and distal on the user to measure changes in blood volume along an artery. Arrow A1 indicates the direction of blood flow along the artery, and arrow A2 indicates the hemodynamic force acting on the arterial wall. The PPG signal 310 includes a first PPG signal 310A measured from the proximal position by the first PPG sensor 300A and a second PPG signal 310B measured from the distal position by the second PPG sensor 300B. The first PPG sensor 300A and the second PPG sensor 300B may be separated by a distance of at least 25 mm.

[0047] Various morphological features 314 of the pulse wavelet 312 can be determined from the PPG signal 310, some of which are listed below with reference to Figure 13B. 314a - Systolic peak 314b- Trough 314c- Duplicate notch 314d - Diastolic peak 314e- Time from systolic peak to diastolic peak 314f and 314g - Range for determining the augmentation index 314h - Contraction time 314i - Extended Time 314j - Pulse propagation time 314k - Time from trough to diastolic peak 314l- Ascending limb 314m - Time from overlapping notch to diastolic peak 314n- Descending leg 314o- Time from diastolic peak to trough 314p - pulse width

[0048] Method 200 may include the step of calculating the pulse propagation time between a first PPG signal 310A and a second PPG signal 310B. As illustrated by morphological feature 314j in Figure 13B, the pulse propagation time may be defined as the time required for the arterial blood pulse to change between the two arterial measurement sites within the same cardiac cycle of a single systolic-diastolic event. For example, as illustrated in Figures 13B and 13C, the pulse propagation time is defined as the period between the systolic peaks of the first PPG signal 310A and the second PPG signal 310B. Figure 13D illustrates various pulse propagation times between the corresponding pulse wavelets 312A and 312B from the first and second PPG signals 310A and 310B.

[0049] Multiple PPG sensors 300 may be placed on the user, and pulse propagation time may be calculated for any pair of PPG sensors 300 that are spatially separated from each other. For example, referring to Figure 3A, Figure 13E illustrates four PPG sensors 300 placed on the forearm—elbow PPG sensor 300d, proximal wrist PPG sensor 300e, distal wrist PPG sensor 300f, and fingertip PPG sensor 300g. Figure 13E also illustrates pulse propagation time for various pairs of the four PPG sensors 300d, 300e, 300f, and 300g.

[0050] Referring to the arrangement of the PPG sensors 300 shown in Figure 3A, the pulse propagation time may be calculated for each of the six pairs of PPG sensors 300. 1. Fingertip PPG sensor 300g and toe PPG sensor 300h 2. Forehead PPG sensor 300a and fingertip PPG sensor 300g 3. Forehead PPG sensor 300a and toe PPG sensor 300h 4. Forehead PPG sensor 300a and proximal wrist PPG sensor 300e 5. Proximal wrist PPG sensor 300e and fingertip PPG sensor 300g 6. Proximal wrist PPG sensor 300e and toe PPG sensor 300h

[0051] The experimental results revealed that pairs of fingertip PPG sensors 300g and toe PPG sensors 300h had a pulse propagation time of -50ms. This means that the pulse at the fingertip arrived 50ms before the pulse at the toe, which is logical since the fingertips are closer to the heart. Pairs of forehead PPG sensors 300a and fingertip PPG sensors 300g also had a pulse propagation time of -50ms. This means that the pulse at the forehead arrived 50ms before the pulse at the fingertips. Pairs of forehead PPG sensors 300a and toe PPG sensors 300h had a pulse propagation time of -100ms. Pairs of forehead PPG sensors 300a and proximal wrist PPG sensors 300e had a pulse propagation time of -30ms. Pairs of proximal wrist PPG sensors 300e and fingertip PPG sensors 300g had a pulse propagation time of -10ms, which is expected due to the short distance between the wrist and fingertips. The pair of proximal wrist PPG sensor 300e and toe PPG sensor 300h had a pulse propagation time of -100 ms.

[0052] The set of physiological parameters 410 can be additionally measured based on pulse propagation time. For example, the physiological parameters 410 may be measured based on the pulse propagation time and spatial distance between the first and second PPG sensors 300A and 300B. For example, pulse propagation time may be supplemented by a 3D morphological representation 400 to measure the physiological parameters 410 for the user.

[0053] A computerized method 200 for monitoring a user's physiology using one or more PPG sensors 300 may be performed on a system having a processor, where various steps of the computerized method 200 are performed in response to non-temporary instructions operated or executed by the processor. Non-temporary instructions are stored in memory and may be referred to as computer-readable storage media and / or non-temporary computer-readable media. Non-temporary computer-readable media include all computer-readable media, with the sole exception being the temporary propagation signal itself.

[0054] A system for performing Method 200 may include PPG sensors 300, such as an array of PPG sensors 300 as illustrated in Figure 3A. The PPG sensors 300 are isolated from the processor and can communicate with the processor, such as to communicate PPG signals 310 to the processor for analysis. Alternatively, PPG sensors 300, such as a first PPG sensor 300A and a second PPG sensor 300B that are spatially separated from each other, may be integrated into a wearable device for the user, such as a portable watch. The wearable device may include a processor configured to perform Method 200. Alternatively, the processor is separate from the wearable device, and the wearable device can communicate with the processor. For example, a separate computer device or server may include the processor, and the wearable device communicates PPG signals 310 to the computer device or server.

[0055] The system and method 200 provides non-invasive continuous monitoring of a user's physiological state, including hemodynamic and cardiac metabolic status, through processing of a PPG signal 310 measured by a PPG sensor 300, without the use of an occlusive cuff. The PPG signal 310 is an easily measurable parameter, and the PPG sensor 300 is relatively inexpensive and commonly used for pulse oximetry. Method 200 processes pulse wavelets 312 of the PPG signal 310 to construct one or more 3D morphological representations 400 that are unique to the user. The entire time series of the PPG signal 310 can be, for example, about 3 minutes, and can therefore be compressed into a 3D morphological representation 400 that can be easily visualized for monitoring the user's physiology.

[0056] In detail, the morphology of the 3D morphological representation 400 can be analyzed to measure various physiological parameters 410 and diagnose various medical conditions. For example, the user's blood pressure and blood glucose may be measured from the 3D morphological representation 400 without the need to attach a cuff or draw blood. The 3D morphological representation 400 improves diagnostic capabilities and enables notification of clinical efficacy and tolerance through a comprehensive assessment of various physiological abnormalities, including but not limited to blood pressure fluctuations (e.g., orthostatic hypotension and vasovagal syncope in the elderly), blood glucose fluctuations, gastrointestinal perfusion, peripheral vascular disease in diabetes, and arterial elasticity.

[0057] For example, Figure 14A illustrates the difference between pulse wavelets 312 for different blood glucose levels 420. For example, Figure 14B illustrates different pulse wavelets 312 used to estimate blood glucose levels 420 at various time points by constructing their respective 3D morphological representations 400 at these time points. In particular, the morphology of pulse wavelets 312 and 3D morphological representations 400 changes depending on an individual's blood glucose level. An individual's blood glucose level changes throughout the day due to the multiple dynamics effects of blood on the vascular walls. Figure 14C illustrates various pulse wavelets 312 and corresponding blood glucose levels at time points 0, 30 minutes, 90 minutes, and 120 minutes.

[0058] Individual users can monitor their own unique 3D morphological representation 400 over long periods to manage their personal health. For example, the 3D morphological representation 400 can be used to monitor the autonomic innervation of the heart for diagnostic and long-term monitoring of cardiac metabolic disease, enabling improved therapeutic approaches to cardiac metabolic disease in individuals. For individuals with chronic diseases, the 3D morphological representation 400 can provide periodic measurements of cardiac metabolic indicators, thereby using time-dependent changes to monitor peripheral vascular disease in diabetic patients and orthostatic hypotension in the elderly.

[0059] The use of these 3D morphological representations 400 for diagnosis opens up opportunities for the development of advanced telemonitoring techniques that avoid the need for electrocardiogram input and mechanical occlusion. The 3D morphological representations 400 are compressed from redundant time-series data of PPG signals 310, minimizing data loss and reducing memory requirements and hardware complexity. These 3D morphological representations 400 can help improve the monitoring and treatment of medical conditions such as hypertension, diabetes, and peripheral vascular disease.

[0060] In some embodiments, two or more spatially separated PPG sensors 300 are used to measure a number of PPG signals 310 that are later processed to construct one or more 3D morphological representations 400. For example, two temporally and morphologically distinct PPG signals 310 can provide insights into hemoperfusion and pulse dynamics that are not possible with conventional electrocardiograms or blood pressure monitors. Hemoperfusion is the local fluid flow through the capillary network and is a critical determinant of organ health.

[0061] In a representative or illustrative embodiment of this disclosure, as illustrated in Figures 15A and 15B, there is a computerized implementation or method 600 for monitoring a user's hemodynamic state using a PPG signal 310. For example, method 600 can be used to monitor blood pressure and hemodynamic states such as orthostatic hypotension.

[0062] Method 600 includes step 610 of receiving a plurality of PPG signals 310 measured from a user, each PPG signal 310 measured from a different location on the user. Preferably, the PPG signals 310 include a first PPG signal 310A measured from a proximal location on the user and a second PPG signal 310B measured from a distal location on the user. Method 600 includes step 620 of receiving motion data measured from the user using an inertial measurement unit (IMU) 320. Method 600 includes step 630 of generating a first hemodynamic profile of the user from the PPG signals 310 and a first machine learning model 710. For example, the first hemodynamic profile may include blood pressure and / or blood glucose data. Method 600 includes step 640 of generating a second hemodynamic profile of the user from the first hemodynamic profile, motion data, and a second machine learning model 720. For example, the second hemodynamic profile may include blood pressure and / or blood glucose data. Preferably, the second hemodynamic profile includes orthostatic blood pressure data, and more preferably orthostatic hypotension data.

[0063] The computerized method 600 can be performed on a system 700 having a processor, and the various steps of the computerized method 600 are performed in response to non-temporary instructions operated or executed by the processor. Non-temporary instructions are stored in memory and may be referred to as computer-readable storage media and / or non-temporary computer-readable media. Non-temporary computer-readable media include all computer-readable media, with the sole exception being the temporary propagation signal itself.

[0064] A system 700 for performing method 600 may include a PPG sensor 300 and an IMU 320 that are separated from and able to communicate with the processor. Alternatively, the PPG sensors 300, such as a first PPG sensor 300A and a second PPG sensor 300B, spatially separated from each other, as well as the IMU 320, may be integrated into a measuring device 800, as shown in Figure 16A. The measuring device 800 may include a processor configured to perform method 600, in detail to generate first and second hemodynamic profiles using machine learning models 710, 720. Alternatively, the processor may be separate from the measuring device 800, and the measuring device 800 can communicate with the processor. For example, a separate computer or server may include the processor, and the measuring device 800 communicates PPG signals 310 and IMU motion data to the computer or server.

[0065] The measuring device 800 may be in the form of a wearable patch that can be worn discreetly under clothing by the user, similar to a holter patch electrode. For example, the measuring device 800 can be placed over the deltoid muscle area of ​​the user's chest to measure the PPG signal 310 from the axillary artery.

[0066] The IMU320 is configured to measure motion data such as posture localization and velocity. The IMU320 may include a 3-axis accelerometer for measuring acceleration, a 3-axis gyroscope for measuring angular velocity, and a 3-axis magnetometer for measuring magnetic localization. The IMU320 thus has 9 degrees of freedom and can measure motion data at frequencies up to 20 Hz. For example, the motion data can be used to determine the user's posture, such as standing, sitting, or lying down, as illustrated in Figure 16B.

[0067] As described above, Method 600 uses two machine learning models 710, 720 to process PPG signals 310 from the PPG sensor 300 and motion data from the IMU 320 to generate a hemodynamic profile. Referring to Figure 15B as an example, the first machine learning model 710 includes an artificial neural network (ANN) 712, and the second machine learning model 720 includes a long-short-term memory (LSTM) network 722. The PPG signals 310 are input to the ANN 712 to generate a first hemodynamic profile, which may include a classification of the user's blood pressure. For example, the first machine learning model 710 includes a softmax layer for classifying blood pressure. The motion data and blood pressure classification are then concatenated in the LSTM network 722. The LSTM network 722 includes a series of regression layers to generate a second hemodynamic profile, which may include a prediction of the user's orthostatic blood pressure, such as orthostatic hypotension.

[0068] The machine learning models 710 and 720 combine the ANN712 and LSTM network 722 into a hybrid machine learning model that recognizes different body poses and takes into account physiological data from the PPG signal 310, thereby adjusting the predictive output accordingly. Furthermore, classification and regression tasks can be learned in parallel, and the hybrid machine learning model can learn feature representations from the intermediate layers and is expected to optimize its internal parameters according to gradient descent convergence to the minimum error of the predictive output. The hybrid machine learning model has the advantage of improving learning effectiveness, efficiency, and predictive accuracy.

[0069] In Method 600 and System 700, the PPG signal 310 and IMU motion data are combined to monitor the user's hemodynamic state, such as to predict orthostatic hypotension by postural awareness from the motion data. The measuring device 800 is designed to continuously measure the PPG signal 310 and motion data without an occlusive cuff. The measuring device 800 does not affect normal lifestyles and leads to improved awareness and acceptance of the measuring device 800 by various users. For example, a user can wear the measuring device 800 on their chest to continuously measure data and monitor their health, such as for the diagnosis, management, and treatment of orthostatic hypotension.

[0070] In some embodiments, method 600 may include the step of calculating the pulse propagation time between a first PPG signal 310A and a second PPG signal 310B. Hemodynamic parameters may be calculated from the pulse propagation time. For example, a first hemodynamic profile, such as one containing blood pressure data, may be calculated from the pulse propagation time.

[0071] In some embodiments, the first machine learning model 710 includes a support vector machine (SVM) to generate a first hemodynamic profile. Specifically, a set of morphological features 314 of PPG signals 310A, 310B is extracted and sent to the SVM along with the calculated pulse propagation time. The SVM then predicts blood pressure (systolic and diastolic) based on the morphological features 314 of PPG signals 310A, 310B and the pulse propagation time.

[0072] An experiment was conducted to evaluate the performance of SVM in predicting blood pressure. Blood pressure estimation was performed on a group of 78 participants and compared to actual blood pressure measurements. The 78 participants ranged in age from 21 to 79 years, with a mean age of 36.7 years. Their systolic and diastolic blood pressure were estimated using SVM and compared to actual blood pressure measured using a blood pressure monitor. Figure 17A illustrates the results for estimated and actual systolic blood pressure, and Figure 17B illustrates the results for estimated and actual diastolic blood pressure. Figure 17C illustrates the error or deviation between estimated and actual systolic blood pressure. It was found that the estimated blood pressure was mostly within 5 mmHg of the actual blood pressure, which is within the international standard for blood pressure estimation established by the American Association for the Development of Medical Devices (AAMI). The results demonstrate that SVM can predict blood pressure with reasonable accuracy using the PPG signal 310.

[0073] In some embodiments, as illustrated in Figure 18, PPG signals 310A, 310B from two or more PPG sensors and motion data from an IMU 320 may be used together with a 3D morphological representation 400 constructed from the PPG signals 310 to monitor the user's physiological parameters 410.

[0074] In the detailed description above, embodiments of the present disclosure relating to a method and apparatus for monitoring user physiology using photoplethysmography signals are described with reference to the provided figures. The descriptions of various embodiments herein are intended to illustrate non-limiting examples of the present disclosure, rather than relating to or limiting to any specific or particular expression of the present disclosure. The present disclosure helps to address at least one of the above-mentioned problems and issues related to the prior art. Although only some embodiments of the present disclosure are disclosed herein, it will be apparent to those skilled in the art that various changes and / or modifications can be made to the disclosed embodiments without departing from the scope of the present disclosure in view of the present disclosure. Accordingly, the following claims are not limited to the embodiments described herein, except as the scope of the present disclosure. [Explanation of Symbols]

[0075] 100 Blood pressure monitor 110 Cuff 200 ways 300 PPG sensor 300A First PPG Sensor 300B Second PPG Sensor 300a PPG sensor, forehead PPG sensor 300b PPG sensor, earlobe PPG sensor 300c PPG sensor, clavicle PPG sensor 300d PPG sensor, elbow PPG sensor 300e PPG sensor, proximal wrist PPG sensor 300f PPG sensor, distal wrist PPG sensor 300g PPG sensor, fingertip PPG sensor 300h PPG sensor, toe PPG sensor 310 PPG signal 310A PPG signal 310B PPG signal 310' High-quality PPG signal 312 Pulse wavelets 312' Input pulse wavelet 312'' Leading pulse wavelet 312A Pulse Wavelet 312B Pulse Wavelet 314 Morphological Characteristics 314a Systolic peak 314b trough 314c Duplicate notch 314d Diastolic peak 320 Inertial Measurement Unit (IMU) 400 Three-dimensional (3D) morphological representation 400A First 3D form representation 400B Second 3D form representation 400C Third 3D Form Representation 410 Physiological parameters 420 blood sugar level 500 Grid Matrix 510 cells 700 System 710 The first machine learning model 712 Artificial Neural Networks (ANNs) 720 Second Machine Learning Model 722 Long-Term Memory (LSTM) Network 800 measuring devices

Claims

1. A computerized method for monitoring a user's physiology using photoplethysmography (PPG) signals, A step of receiving a set of PPG signals measured from the user, wherein each PPG signal is measured from a different location on the user. A step of identifying a plurality of pulse wavelets from each PPG signal, wherein each pulse wavelet is defined in 2D by an amplitude axis and a first time axis. For each PPG signal, the steps include aligning each of the 2D pulse wavelets in 3D along a second time axis, The steps include constructing a set of 3D morphological representations from the aforementioned aligned pulse wavelets. Includes, A computerized method by which the set of physiological parameters of the user is measurable based on the constructed 3D morphological representation.

2. The method according to claim 1, wherein the step of constructing the set of 3D morphological representations includes, for each PPG signal, the step of constructing each 3D morphological representation from each pulse wavelet identified from each PPG signal, such that each 3D morphological representation corresponds to one PPG signal.

3. The method according to claim 1, wherein the step of constructing the set of 3D morphological representations includes the step of constructing a single 3D model from all the pulse wavelets identified from all the PPG signals.

4. The method according to any one of claims 1 to 3, wherein the PPG signal includes a first PPG signal measured from a proximal position on the user and a second PPG signal measured from a distal position on the user.

5. The method according to claim 4, further comprising the step of calculating the pulse propagation time between the first PPG signal and the second PPG signal, wherein the physiological parameters are additionally measurable based on the pulse propagation time.

6. The method according to any one of claims 1 to 5, wherein the physiological parameters include blood pressure, blood glucose, and / or arterial distensibility.

7. A system for monitoring a user's physiological processes using PPG signals, A processor configured to perform the computerized method according to any one of claims 1 to 6. A system equipped with these features.

8. A non-temporary computer-readable storage medium storing computer-readable instructions, wherein, when executed, the computer-readable instructions cause a processor to perform the computerized method described in any one of claims 1 to 6.

9. A measuring device for monitoring a user's hemodynamic state using photoplethysmography (PPG) signals, Multiple PPG sensors for measuring multiple PPG signals from different locations on the user, An inertial measurement unit for measuring motion data from the user, It is a processor, The first hemodynamic profile of the user is calculated from the PPG signal and the first machine learning model. The second hemodynamic profile of the user is calculated from the first hemodynamic profile, the motion data, and the second machine learning model. A processor and A measuring device equipped with the following features.

10. The measuring device according to claim 9, wherein the first hemodynamic profile includes blood pressure and / or blood glucose data, and the second hemodynamic profile includes orthostatic blood pressure data.

11. The aforementioned multiple PPG sensors A first PPG sensor for measuring a first PPG signal from a proximal position on the user, A second PPG sensor for measuring a second PPG signal from a distal position on the user, The measuring apparatus according to claim 9 or 10, including the following:

12. The measuring device according to claim 11, wherein the computer processor is configured to calculate the pulse propagation time between the first PPG signal and the second PPG signal.

13. The method according to claim 12, wherein the first machine learning model comprises a support vector machine for generating the first hemodynamic profile, which includes blood pressure data from the PPG signal and the pulse propagation time.

14. A computerized method for monitoring a user's hemodynamic status using photoplethysmography (PPG) signals, A step of receiving a plurality of PPG signals measured from the user, wherein each PPG signal is measured from a different location on the user. The steps include receiving motion data measured from the user using an inertial measurement unit, The steps include generating a first hemodynamic profile of the user from the PPG signal and a first machine learning model, The steps include generating a second hemodynamic profile of the user from the first hemodynamic profile, the motion data, and the second machine learning model, and Computerized methods, including those mentioned above.

15. The method according to claim 14, wherein the plurality of PPG signals include a first PPG signal measured from a proximal position on the user and a second PPG signal measured from a distal position on the user.

16. The method according to claim 15, further comprising the step of calculating the pulse propagation time between the first PPG signal and the second PPG signal.

17. The method according to claim 16, wherein the step of generating the first hemodynamic profile includes the step of predicting blood pressure from the PPG signal and the pulse propagation time using a support vector machine of the first machine learning model.

18. The method according to any one of claims 14 to 17, wherein the first machine learning model comprises an artificial neural network, and the second machine learning model comprises a long short-term memory (LSTM) network.

19. A system for monitoring a user's hemodynamic status using PPG signals, A processor configured to perform the computerized method according to any one of claims 14 to 18. A system equipped with these features.

20. A non-temporary computer-readable storage medium storing computer-readable instructions, wherein, when executed, the computer-readable instructions cause a processor to perform the computerized method described in any one of claims 14 to 18.

Images