Method for deriving metrics of PPG variability

Abstract

A method for deriving a measure of PPG variability indicative of fluid reactivity. The PPG signal is processed with an adaptive resampling algorithm that resamples the signal with a sampling density as a function of the instantaneous pulse rate. When the PPG signal is subsequently transformed using a discrete frequency domain transform, such as FFT, the resulting spectrum reflects the variability of the sample, where the FM contribution is greatly cancelled. This provides variability information that more reflects AM contributions, which are dominant contributions associated with respiration.

Description

Technical Field

[0001] The present invention relates to a method for deriving a measure of PPG variability. Background Art

[0002] Maintaining adequate oxygen delivery through the body is a core part of critical care medicine. Fluid administration is often used for hemodynamic management to improve cardiac output (CO). The more the myocardium is stretched during filling, the greater the amount of blood pumped into the aorta. The patient's fluid responsiveness depends on the ability of the heart to adapt to the increased volume. This is important in Figure 1 is shown in the figure, Figure 1 A so-called Frank-Starling curve is shown which relates stroke volume (y-axis) to cardiac preload (x-axis).

[0003] If the heart is on the steep part of this curve (e.g. Figure 1 If the heart is operated at the plateau region of the curve (e.g., point A, low preload), an increase in preload (via volume expansion) will result in a significant increase in stroke volume (SV). If the heart is operated at the plateau region of the curve (e.g., point B, higher preload), the fluid responsiveness will be low.

[0004] Fluid responsiveness can be assessed using SV or CO measurements before and after fluid administration. This requires injection of a fluid bolus with the risk of volume overload.

[0005] Cardiopulmonary interactions have been shown to be useful in predicting fluid responsiveness. Differences in intrathoracic pressure during the respiratory cycle result in changes in SV during breathing. SV changes (SVV) are Figure 1 The point A of the curve (SVV A ) than point B (SVV B SV and arterial blood pressure (ABP) show a strong correlation, and there is also respiratory modulation of pulse pressure (PP).

[0006] Photoplethysmography (PPG) can also be used to assess this variability. Figure 2 The relationship between ABP and respiratory cycle is shown graphically. Figure 2 The modulation of arterial blood pressure during the respiratory cycle is shown, and the maximum PP value PP is shown. MAX Consistent with maximum breathing and minimum PP value PP MIN Consistent with minimal respiration. Pulse pressure is the amplitude of the pulse wave.

[0007] For mechanically ventilated patients, measurements of SVV, pulse pressure variation (PPV), and / or variability of the PPG signal during breathing ('PPG variability') can be used to predict fluid responsiveness. By way of example, PPG variability indices include pulse oximetry plethysmography (POP) variability and / or Masimo's proprietary index PVI (Pleth Variability Index). In several clinical studies, thresholds for fluid responders and non-responders have been obtained for these two indices. For spontaneously breathing patients, this variability approach is less effective because the respiratory modulation effect is less pronounced and varies from breath to breath during breathing. Preload redistribution maneuvers such as the passive leg raise test are also used in clinical practice to predict fluid responsiveness, but are considered to have lower operational performance.

[0008] In the prior art, SVV provides the strongest predictor for fluid responsiveness, followed by PPV, and then PPG variability. However, SVV measurement requires an invasive central line and is therefore generally only feasible for patients in the most critical conditions. For PPV measurement, a radial arterial line is adequate but still invasive. Non-invasive pulse pressure measurement is also possible but generally has reduced signal quality. PPG variability as a proxy signal for fluid responsiveness has the advantages of being a completely non-invasive measurement and of being measured by widely available and inexpensive pulse oximeters. However, currently, PPG variability provides the least accurate predictive signal for fluid responsiveness.

[0009] It would be beneficial to provide an improved measure of PPG variability so that it could be used as a more reliable predictive measure of fluid responsiveness.

[0010] US2016 / 287090A1 discloses a plethysmographic respiratory processor that responds to respiratory effects appearing on a blood volume waveform and a corresponding detected intensity waveform measured at a blood-perfused peripheral tissue site with an optical sensor to provide a measurement of respiratory rate.

[0011] US 2014 / 094664 A1 discloses a method for determining a cardiopulmonary interaction factor of a subject. Summary of the invention

[0012] The invention is defined by the claims.

[0013] According to an example according to one aspect of the present invention, there is provided a method for deriving a plethysmographic variability measure, the method comprising:

[0014] obtaining a plethysmographic signal of a subject, the plethysmographic signal comprising a sampled time series of plethysmographic sensor measurements spanning a time window, wherein the time window spans a plurality of pulse cycles of the subject;

[0015] Obtain an instantaneous pulse rate signal of a patient over the time window;

[0016] Apply a resampling algorithm to the sampled time series of the plethysmogram signal, wherein the resampling algorithm is adapted to perform sampling rate modification at least in part based on the instantaneous pulse rate across the time window, and wherein the resampling algorithm resamples the plethysmogram signal across the time window at a resampling rate that is a function of the instantaneous pulse rate;

[0017] Apply a discrete time domain to frequency domain transform to the resampled plethysmogram signal waveform to obtain a frequency spectrum;

[0018] Identify a set of peaks of predetermined frequency components in the frequency spectrum;

[0019] Calculate a plethysmogram variability metric based on the frequency amplitudes of the identified peaks in the frequency spectrum.

[0020] Optionally, the method may further include generating a data output for derivation based on the calculated plethysmogram variability metric.

[0021] Regarding the plethysmogram signal, the discussion in this document may more specifically relate to a photoplethysmogram (PPG) signal. However, the principles of the present invention are more broadly applicable to any type of plethysmogram signal and plethysmogram sensor, including for example a speckle plethysmogram (SPG) sensor. Accordingly, any reference to a PPG sensor, PPG signal, or PPG variability metric should be construed to cover use with any plethysmogram sensor or signal.

[0022] For reasons that will become clear in the following description, the resampling operation has the effect of substantially eliminating the frequency modulation (FM) contribution to plethysmogram variability within the discrete frequency transform. Only amplitude modulation (AM) is closely related to fluid responsiveness, and thus eliminating FM significantly improves the accuracy and reliability of the resulting plethysmogram variability metric as an indicator of fluid responsiveness.

[0023] Embodiments of the present invention differ from known algorithms in employing adaptive resampling and deriving a variability metric through frequency domain analysis. This is different from existing variability algorithms that use the time domain and peak / valley identification.

[0024] The discrete frequency domain transform converts a finite sequence of equally spaced samples of a function into a sequence of the same length of equally spaced samples of an output frequency spectrum. Thus, the discrete transform assumes equally spaced input samples. By modifying the sampling density of the input function (PPG signal) in a manner related to the FM contribution, the FM contribution can be eliminated in the frequency domain transform.

[0025] In some embodiments, the discrete time domain to frequency domain transform can be a discrete Fourier transform, such as a fast Fourier transform (FFT).

[0026] The plethysmographic variability in this context can mainly refer to the variability during breathing or ventilation, i.e., respiratory plethysmographic variability or respiration-associated plethysmographic variability.

[0027] The resampling algorithm is adapted to perform a modification of the sampling rate of the plethysmographic signal across a time window to derive a resampled plethysmographic signal, and wherein the sample density of the resampled signal across the time window is a function of the instantaneous pulse rate over the time window.

[0028] Therefore, the resampling algorithm is an adaptive resampling algorithm. Adaptive resampling means a resampling where the resampling rate or density is adjustable or variable, or can vary according to a defined function.

[0029] The time window can be a moving time window. This allows, for example, obtaining and outputting a real-time plethysmographic variability metric from the method, i.e., a plethysmographic variability metric as a function of time.

[0030] Although the PPG sensor is mentioned above, in other examples, an SPG sensor can be used. The effects of the present invention remain the same.

[0031] In some embodiments, the method can additionally include performing a filtering operation before resampling.

[0032] Here, obtaining the plethysmographic signal can include: obtaining a red sensor signal waveform and an infrared sensor signal waveform from a plethysmographic sensor. The method can include applying an adaptive digital filter to the red sensor signal waveform and the infrared sensor signal waveform from the plethysmographic sensor. An adaptive digital filter means a digital filter having dynamically adjustable or variable coefficients.

[0033] In some embodiments, the coefficients of the adaptive digital filter can be configured based on a metric related to the SpO2 value for the patient. In some examples, the metric can be the ratio: This will be further explained later.

[0034] In some embodiments, the plethysmographic signal can be a plethysmographic signal obtained from a plethysmographic sensor disposed at a central body location of the patient, such as on the patient's forehead. Measurements at the central location may provide a high-quality signal and avoid the presence of waveform changes that occur when blood travels from the heart to the peripheral measurement site.

[0035] In some embodiments, the instantaneous pulse rate signal can be calculated based on the inter-beat time intervals in the plethysmography signal.

[0036] Optionally, in some embodiments, obtaining the instantaneous pulse rate signal can further include using the respiration signal for the subject. The respiratory cycle is related to the pulse rate and can thus be used to enhance the reliability or resolution of the instantaneous pulse rate signal.

[0037] The instantaneous pulse rate signal is provided as an input to the resampling algorithm and is used to define the resampling rate or density across a time window.

[0038] The adaptive resampling algorithm described above resamples the plethysmography signal at an increasing rate when the instantaneous pulse rate is high and at a decreasing rate when the instantaneous pulse rate is low.

[0039] The resampling algorithm can include, for example: defining a set of (re)sampling points with a sampling density proportional to the instantaneous pulse rate, and subsequently defining the plethysmography signal value for each sampling point based on interpolation from the originally measured plethysmography sampling points.

[0040] In some embodiments, the resampling operation includes: defining the number of resampling points for a time window; distributing the resampling points across the time window, the density of the resampling points depending on the instantaneous pulse rate as a function of time across the time window; and determining the plethysmography signal value for each resampling point based on interpolation of the sampling time series of the plethysmography signal obtained originally.

[0041] In some embodiments, the number of resampling points can be set to be equal to the number of sampling points in the original plethysmography signal, such that the average sampling rate over the time window remains constant.

[0042] Alternatively, the number of resampling points can vary. In some embodiments, the number of resampling points can be a function of the average respiration rate across the time window.

[0043] One way to do this is to make the number of sampling points over the time window spanning each respiratory cycle proportional to the duration of the respiratory cycle.

[0044] For example, in some embodiments, the resampling algorithm includes: receiving a respiratory signal for an object, the respiratory signal representing parameters related to respiratory cycle phases; identifying a time window in a plethysmogram signal that spans a single respiratory cycle; defining the number of resampling points for the time window in proportion to the duration of the single respiratory cycle; distributing the resampling points across the time window according to an instantaneous pulse rate as a function of time across the time window; and determining the plethysmogram signal value for each resampling point based on interpolation of a sampled time series of measurements from the originally obtained plethysmogram signal.

[0045] In some embodiments, the aforementioned set of predetermined frequency component peaks may correspond to: a frequency component corresponding to the pulse rate of the object (PPG(f)| PR ), a frequency component corresponding to the sum of the pulse rate and the respiratory rate (PPG(f)| PR+RR ), and a frequency component corresponding to the pulse rate minus the respiratory rate (PGG(f)| PR-RR ).

[0046] In some embodiments, a plethysmographic variability metric (PPGV) may be calculated based on evaluating the following equation:

[0047]

[0048] where (PPG(f)| PR ) is the amplitude of the frequency component corresponding to the pulse rate of the object, (PPG(f)| PR+RR ) is the amplitude of the frequency component corresponding to the sum of the pulse rate and the respiratory rate, and (PGG(f)| PR-RR ) is the amplitude of the frequency component corresponding to the pulse rate minus the respiratory rate.

[0049] In some embodiments, identifying peaks may simply include: identifying all peaks in the spectrum that exceed a threshold amplitude, assuming that the spectrum of the resampled plethysmogram signal contains only peaks at these frequency points. In simple cases, this assumption is indeed correct.

[0050] However, in some cases, it may be more difficult to identify relevant peaks, or there may be additional peaks due to systematic noise sources in the signal. This is particularly relevant for patients with autonomous breathing with variations in tidal volume and respiratory rate.

[0051] To improve peak identification, in some embodiments, the method includes: obtaining a respiratory signal for an object, the respiratory signal representing parameters related to respiratory cycle phases. The frequency component peak set may be identified or selected from a spectrum based on applying a peak identification algorithm that uses the respiratory signal as an input. The respiratory signal may be used to calculate the respiratory rate RR. As described above, the pulse rate PR has been obtained as part of the method.

[0052] In some embodiments, obtaining a respiratory signal (for any purpose described in this document) may include: receiving a temperature signal from a temperature sensor placed adjacent to the object's airway and mapping oscillations in the temperature sensor signal to oscillations in the object's respiratory cycle to derive the respiratory signal.

[0053] In some embodiments, the temperature sensor may be integrated with a plethysmography sensor in a common sensor unit.

[0054] Another aspect of the invention is a computer program product comprising computer program code configured to cause a processor to perform a method according to any embodiment described in this document or according to any claim of this application when run by the processor.

[0055] Another aspect of the invention is a processing device comprising: an input / output section; and one or more processors operatively coupled to the input / output section.

[0056] The one or more processors are adapted to: obtain a plethysmography signal for an object, the plethysmography signal comprising a sampled time series of plethysmography sensor measurements over a time window, and the time window spanning a plurality of pulse cycles of the object; obtain an instantaneous pulse rate signal of the patient over the time window; apply a resampling algorithm to the sampled time series of the plethysmography signal, wherein the resampling algorithm is adapted to perform sampling rate modification at least in part based on the instantaneous pulse rate across the time window, wherein the resampling algorithm resamples the plethysmography signal across the time window at a resampling rate that is a function of the instantaneous pulse rate; apply a discrete time domain to frequency domain transform to the resampled plethysmography signal waveform to obtain a spectrum; identify a set of predetermined frequency component peaks in the spectrum; calculate a plethysmography variability metric based on the frequency amplitudes of the identified peaks in the spectrum; and preferably generate a data output based on the calculated plethysmography variability metric.

[0057] Another aspect of the invention is a system comprising: a processing device as outlined above or according to any embodiment in this document; and a plethysmography sensor, such as a PPG sensor.

[0058] In some embodiments, the system may further include sensors for measuring parameters related to the phases of the respiratory cycle (i.e., respiratory signals). In some embodiments, the sensors include a temperature sensor for placement adjacent to the airway of the subject, and optionally, wherein the sensors include a sensing unit that includes both a plethysmography sensor and a temperature sensor.

[0059] These and other aspects of the invention will be apparent and elucidated with reference to the (one or more) embodiments described hereinafter. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] For a better understanding of the present invention, and for a clearer illustration of how it may be implemented, reference will now be made, by way of example only, to the accompanying drawings, in which:

[0061] Figure 1 A curve relating stroke volume to cardiac preload is shown;

[0062] Figure 2 An example PPG signal as a function of time over two respiratory cycles is shown;

[0063] Figure 3 Different types of signal modulation that may be present in the PPG signal are illustrated;

[0064] Figure 4 A flowchart of an example method according to one or more embodiments of the present invention is shown;

[0065] Figure 5 is a block diagram of an example processing device and system according to one or more embodiments of the present invention;

[0066] Figure 6 A flowchart of an example method according to one or more embodiments is shown;

[0067] Figure 7 A resampling operation is illustrated, in which the sampling density varies as a function of the instantaneous pulse rate;

[0068] FIG. 8 shows the identification of frequency peaks in the spectrum of a resampled PPG signal;

[0069] Figure 9 An example optional filtering operation is illustrated;

[0070] Figure 10 A processing flow of an example method according to one or more embodiments is shown;

[0071] Figure 11 A processing flow of an example method according to one or more additional embodiments is shown; and

[0072] Figure 12 The processing flow of an example method according to one or more additional embodiments is shown. DETAILED DESCRIPTION

[0073] The present invention will be described with reference to the accompanying drawings.

[0074] It should be understood that the detailed description and specific examples, although indicating exemplary embodiments of the apparatus, systems and methods, are intended for illustrative purposes only and are not intended to limit the scope of the present invention. These and other features, aspects and advantages of the apparatus, systems and methods of the present invention will be better understood from the following description, the appended claims and the drawings. It should be understood that the drawings are merely schematic and are not drawn to scale. It should be understood that the same reference numerals are used throughout the drawings to indicate the same or similar components.

[0075] The present invention provides a method for deriving a measure of plethysmographic variability that better indicates fluid responsiveness. This is achieved by preprocessing the plethysmographic signal using an adaptive resampling algorithm that resamples the signal using a sampling density that is a function of the instantaneous pulse rate. When the plethysmographic signal is subsequently transformed using a discrete frequency domain transform such as an FFT, the resulting spectrum reflects the variability of the samples, where the FM contribution is greatly eliminated. This provides variability information that more reflects the AM contribution, which is the dominant contribution associated with respiration. This in turn provides variability information that more closely indicates fluid responsiveness.

[0076] The examples and embodiments discussed herein relate specifically to PPG signals and PPG sensors. However, embodiments of the present invention can use any type of plethysmographic sensor, including for example a speckle plethysmographic sensor (SPG). Thus, it should be understood that any reference herein to a PPG signal, PPG sensor or PPG variability measure can be more generally replaced by a reference to a plethysmographic signal, plethysmographic sensor or plethysmographic variability measure without loss of technical effect.

[0077] Embodiments of the present invention provide a plethysmographic variability measure that is a more accurate and reliable indicator of fluid responsiveness. The proposed variability measure is valid for all classes of patients. Variability measurements in the prior art were limited to patients on mechanical ventilation. To improve the outcome of mechanical ventilation, the use of low tidal volumes has become more common in clinical practice. Embodiments of the present invention allow measurements to be made at low tidal volumes where the signal-to-noise ratio of the variability signal was previously low and thus the quality was limited. Variability measurements at low tidal volumes also enable embodiments of the present invention to be applied to spontaneously breathing patients. Spontaneously breathing patients also exhibit variations in tidal volume and respiratory rate. Embodiments of the present invention provide a valid variability measure in situations where the respiratory volume and respiratory rate can vary.

[0078] As described above, the plethysmographic signal exhibits modulations in its waveform that are related to the progression of the respiratory cycle. These are referred to as respiratory modulations. Both the instantaneous frequency and the amplitude of the plethysmographic signal are modulated as a function of the progression of the respiratory cycle.

[0079] Figure 3 Respiratory modulations of an exemplary photoplethysmogram (PPG) signal are illustrated. Three types of modulations in the signal can be distinguished: baseline wander (BW), amplitude modulation (AM), and frequency modulation (FM). Figure 3 The signal trace 22 in shows a PPG signal without modulation. The signal trace 24 shows a PPG signal with only baseline wander (BW) modulation. Baseline wander means the modulation in the baseline of the signal (i.e., the DC offset). The signal trace 26 shows a PPG signal with only amplitude modulation (AM). The signal trace 28 shows a PPG signal with only frequency modulation (FM). It is generally expected that a real PPG signal exhibits a mixture of all these types of modulations. Figure 3 They are shown separately for illustrative purposes.

[0080] The BW in the PPG signal mainly reflects blood volume changes due to respiratory variations in arterial blood pressure and vasoconstriction caused by respiration transferring blood to the veins. Due to sympathetic nervous system activity, there may also be baseline oscillations with a frequency lower than the respiratory frequency. PPG sensor movement may also affect the tissue blood volume in the detection area and cause irregular fluctuations in the PPG baseline.

[0081] The FM of the PPG signal is also known as respiratory sinus arrhythmia. Several mechanisms contribute to FM. These include, for example, the baroreflex-mediated changes in heart rate when stroke volume changes. Other contributing factors can include an increase in heart rate during the stretching of the sinoatrial node.

[0082] The AM of the PPG signal mainly reflects the changes in preload during respiration. Thus, in light of the above discussion, it will be immediately appreciated that the AM of the PPG signal is the most relevant type of modulation for measuring fluid responsiveness. Ideally, the PPG variability index (PPGV) to be used as a measure of fluid responsiveness will be based as much as possible on the amplitude modulation (AM) of the underlying PPG signal.

[0083] Embodiments of the present invention propose a method for processing a PPG signal to derive a PPG variability index, which is based on selectively suppressing other types of modulations and generating a variability signal that mainly indicates the PPG AM modulation.

[0084] Methods and algorithms applied to calculate such PPG variability metrics should ideally be able to distinguish AM from other types of modulation and also ideally be able to distinguish changes caused by artifacts (such as motion artifacts). Additionally, changes caused by venous pulsations are ideally also suppressed from the PPGV index.

[0085] Another important factor to consider is the delay between airway pressure and the consequent effects on different types of modulation. Airway pressure has an impact on arterial blood pressure and this manifests as the BW component after a certain time delay. Airway pressure also has an effect on AM after a certain time delay and thus occurs by exerting pressure on the vena cava inflow at the right ventricle, and this in turn affects the outflow at the left ventricle and the aortic pulse pressure. The delay between airway pressure and the AM effect in the PPG signal can be referred to as the pulmonary conduction time.

[0086] Existing state-of-the-art methods and algorithms for calculating PPV and PPG variability are based on identifying peaks in the time domain.

[0087] Different methods for calculating PPG variability are described in the following paper: Acosta et al., Biomedical Signal Processing and Control 60(2020)101947. This describes a frequency-domain algorithm that recovers PPV values with a signal-to-noise ratio six times lower than time-domain algorithms. The algorithm is based on the convolution components of cardiac and respiratory oscillations. Although this method takes into account the pulmonary conduction time and enables PPV measurements at low tidal volumes, this method has the limitation that it only processes BW and AM in the PPG signal. When both AM and FM are present, PPV cannot be accurately determined because these modulations are mixed in a complex way in the Fourier transform (such as FFT) spectrum for the AM and FM values that occur in clinical practice.

[0088] Figure 4 The steps of an example computer-implemented method 100 according to one or more embodiments are outlined in block diagram form. The steps will be outlined before being further explained in the form of example embodiments.

[0089] Method 100 is for deriving a metric of plethysmogram variability. Variability in the context of the present invention can refer to variability during respiration or ventilation. Ultimately, the aim is to derive a plethysmogram variability metric that reflects fluid responsiveness (associated therewith).

[0090] The method includes obtaining 112 a plethysmogram or plethysmographic signal of an object, which includes a sampled time series of plethysmogram sensor measurements across a time window, and the time window spans multiple pulse cycles of the object. In some embodiments, this can be a PPG signal.

[0091] The method further includes obtaining the instantaneous pulse rate signal of a patient over the time window 114. As will be explained later, this can be obtained from the plethysmogram signal itself, optionally in combination with a separate measurement source such as a respiratory sensor.

[0092] The method further includes applying a resampling algorithm 166 to the sampled time series of the plethysmogram signal. The resampling algorithm is adapted to perform a sampling rate modification, wherein the sampling rate modification is at least partially based on the instantaneous pulse rate. The resampling algorithm resamples the plethysmogram signal at a resampling rate that is a function of the instantaneous pulse rate. In other words, the signal resamples the plethysmogram over the time window, and wherein the sampling density over the time window is a function of the instantaneous pulse rate over the time window.

[0093] The method further includes applying a discrete time domain to frequency domain transform 118 (such as a Fourier transform) to the resampled plethysmogram signal waveform to obtain a frequency spectrum. The effect of the above resampling is to substantially eliminate FM modulation in the signal after the discrete frequency domain transform. The discrete frequency domain transform assumes equally spaced sampling points. Therefore, applying the discrete transform to a plethysmogram signal whose sampling density varies as a function of the pulse rate effectively eliminates frequency modulation in the signal.

[0094] The method further includes identifying a set of peaks of predetermined frequency components 120 in the frequency spectrum.

[0095] The method further includes calculating a plethysmogram variability metric 122 based on the amplitudes of the identified peaks in the frequency spectrum.

[0096] The method may further include generating a data output based on the calculated PPG variability metric. For example, the data output may be transmitted to a user interface or a data storage device or an external server.

[0097] Regarding resampling, this is an adaptive resampling, wherein the resampling algorithm is adapted to perform a sampling rate modification of the plethysmogram signal over the time window to derive the resampled plethysmogram signal, and wherein the sampling rate modification is a function of the instantaneous pulse rate over the time window.

[0098] The adaptive resampling algorithm resamples the plethysmogram signal at an increasing rate when the instantaneous pulse rate is high and at a decreasing rate when the instantaneous pulse rate is low.

[0099] The time series may span a moving time window such that a real-time plethysmogram variability index is derived as a function of time.

[0100] Although the photoplethysmography (PPG) signal is mentioned in the examples presented herein, embodiments of the present invention will equally apply to any other type of plethysmography signal, such as the speckle plethysmography (SPG) signal.

[0101] As described above, the method may also be embodied in hardware, for example in the form of a processing device configured to execute a method according to any example or embodiment described in this document or according to any claim of this application. For further assistance in understanding, Figure 5 A schematic representation of an example processing device 32 configured to execute a method according to one or more embodiments of the present invention is presented. The processing device is shown in the context of a system 30 that includes the processing device. The processing device represents an aspect of the present invention alone. The system 30 is another aspect of the present invention. The system provided need not include all of the illustrated hardware components; it may include only a subset of them.

[0102] The processing device includes one or more processors 36, which are configured to execute a method according to what is outlined above or according to any embodiment described in this document or any claim of this application. In the illustrated example, the processing device also includes a communication interface or input / output section 34.

[0103] As Figure 5 illustrated by the example of, the system 30 may also include a user interface 52, and the user interface 52 includes a display unit. This can be used to display the calculated plethysmogram variability metric.

[0104] The system may also include a plethysmogram sensor for generating a plethysmogram signal. In some embodiments, this may be a PPG sensor. An SPG sensor is an alternative. The plethysmogram sensor may be adapted to be mounted or attached at a central body location of a patient, such as the patient's forehead or the patient's nose.

[0105] The system 30 may also include a sensor 56 for generating a respiratory signal indicative of respiration. For example, the respiratory signal may represent a respiratory cycle phase. In some examples, this is a temperature sensor for placement adjacent to an object's airway, and wherein the processing device is adapted to map oscillations in the temperature sensor signal to oscillations in the object's respiratory cycle to derive the respiratory signal. In some embodiments, the respiratory sensor 56 may be integrated in the same sensor unit as the plethysmography sensor 54.

[0106] Of course, it is not necessary for the system 30 to include all of these hardware components. A system according to the present invention may be provided that does not include these components or includes only one or more of these components.

[0107] The input / output unit 34 is adapted to receive the aforementioned plethysmogram signal. Alternatively, it can receive from a data storage device or from an intermediate communication device (such as a hub server or a network node).

[0108] The system 30 may further include a memory 38 for storing computer program code (i.e., computer-executable code) that is configured to cause one or more processors 36 of the processing device 32 to execute the method as described above, or a method according to any embodiment described in the present disclosure or according to any claim.

[0109] As previously mentioned, the present invention can also be embodied in software form. Accordingly, another aspect of the present invention is a computer program product including code modules that are configured to cause a processor to execute a method according to any example or embodiment of the present invention described in this document or according to any claim of this patent application when run on the processor.

[0110] According to a set of preferred embodiments, the method 100 further includes applying a filtering operation to the obtained plethysmogram signal before resampling 116.

[0111] Figure 6 Steps of an additional exemplary method according to one or more embodiments are outlined, where such a filtering operation 132 is included as an additional step. The method is otherwise the same as the method outlined in Figure 4 as outlined.

[0112] More specifically, according to some embodiments, the filtering operation 132 can be implemented in the following manner. Here, for the sake of brevity, a photoplethysmogram (PPG) sensor will be specifically referred to, but other types of plethysmogram signals can be used.

[0113] The aforementioned obtaining 112 of the PPG signal may include obtaining a red sensor signal waveform and an infrared sensor signal waveform from the PPG sensor. Accordingly, the PPG signal consists of two components. Then, the method may further include applying an adaptive digital filter to the red sensor signal waveform and the infrared sensor signal waveform from the PPG sensor. Then the filtered PPG signal (i.e., the PPG signal having the filtered red and IR components) is used for the remainder of the method.

[0114] Regarding the coefficients for the adaptive digital filter, preferably, the coefficients of the adaptive digital filter are configured based on a metric related to or indicating the SpO2 value for the patient. This will be explained in more detail below.

[0115] As described above, the method involves a resampling operation, wherein the plethysmogram signal is resampled at a resampling rate that varies as a function of the instantaneous pulse rate over a sampling time window. This is now discussed in more detail. Particular reference will be made to the photoplethysmogram (PPG) signal, but the method is more generally applicable.

[0116] Resampling involves obtaining a signal indicative of the instantaneous pulse rate pulse(t) as a function of time over the course of a measurement time window. This can be calculated based on the peak-to-peak separation in the PPG signal itself or based on an external measurement source.

[0117] Then, the resampling operation involves processing the PPG signal over the measurement window to resample the signal at a resampling rate that varies over the time window as a function of the instantaneous pulse rate across the time window. Another way of expressing this is that the resampling operation involves determining a resampling density function s(t), which is a function of time over the time window, where the resampling density function is a function of the instantaneous pulse rate, i.e., is related to the instantaneous pulse rate. In other words, when the instantaneous pulse rate is high, the resampling density is high, and when the instantaneous pulse rate is low, the resampling density is low. Once the resampling density function is determined, new (re)sampling points are defined across the time window according to the resampling density. Then, the PPG signal value for each new resampling point is set based on interpolation of the true PPG signal values from the originally obtained PPG signal. In this way, if the resampled PPG signal values are to be plotted for equally spaced sampling points, the frequency modulation in the PPG signal is compensated for. The corollary is that in the discrete Fourier transform of the resampled signal (which assumes equally spaced sampling points), the FM is substantially eliminated.

[0118] Thus, in summary, in at least some embodiments, the resampling operation 116 includes: defining the number of resampling points for a time window; distributing the resampling points across the time window, the sample density depending on the instantaneous pulse rate as a function of time across the time window; and determining the PPG signal value for each resampling point based on interpolation of a sampling time series of measurements from the originally obtained PPG signal.

[0119] Resampling is performed by Figure 7A diagram, where the gray level indicates the resampling density or resampling rate as a function of time over a time window. Darker values are associated with a higher resampling rate, and lighter values are associated with a lower resampling rate. The waveform 62 is an example PPG signal. In this example, the instantaneous pulse rate can be inferred from the inter-beat interval, where a smaller interval indicates a higher pulse rate. It can be seen how the resampling rate (or resampling density) is higher (darker gray value) for portions of the signal with a lower instantaneous pulse rate (larger inter-beat interval).

[0120] Accordingly, the proposed resampling algorithm resamples the PPG signal at an increasing rate (increasing sampling density) when the instantaneous pulse rate is high and at a decreasing rate when the pulse rate is low. The average sampling rate over the time window can be maintained at a preset value, e.g., the same as in the original PPG signal. Thus, for example, the sampling rate can be made lower than the original signal in some regions and higher than the original signal in other regions. The time interval between successive (re)sampling points is chosen to be proportional to the instantaneous inter-beat interval. The PPG signal intensity for each new (re)sampling point is obtained by interpolation from the original samples. In this way, if the resampled PPG signal were (hypothetically) plotted for equidistant sampling points and were not present in the discrete Fourier transform of the resampled PPG signal (which assumes equally spaced samples), the FM is compensated for.

[0121] As described above, the method involves identifying a predefined set of frequency components in the spectrum of the resampled PPG signal.

[0122] To obtain the spectrum, a fast Fourier transform (FFT) (or other discrete frequency domain transform) is applied to transform the resampled PPG signal into the frequency domain.

[0123] The discrete transform assumes equally spaced sampling points. Thus, the effect of the discrete transform on a resampled data sequence of samples with intervals that vary as a function of the pulse rate is to effectively remove the PPG frequency modulation (FM) in the discrete Fourier transform. This is because FM is mainly caused by changes in the pulse rate. This is beneficial because, as described above, only the amplitude modulation component of the PPG variability is connected to fluid responsiveness. Thus, in cases where the goal is a PPG variability index that most closely reflects fluid responsiveness, it is desirable to remove the FM component of the PPG variability.

[0124] Regarding the predefined frequency components to be identified, these can be understood as follows.

[0125] In the absence of frequency modulation (FM) in the PPG signal, the spectrum of the PPG signal has peaks at frequencies corresponding to the respiratory rate (RR), the pulse rate (PR), the sum of the pulse rate and the respiratory rate (PR+RR), and the difference between the pulse rate and the respiratory rate (PR-RR).

[0126] In Figure 8A indicates these peaks in the spectrum.

[0127] Based on these, a PPG variability metric (PPGV) that mainly reflects amplitude modulation and thus reflects fluid responsiveness can be calculated from the following equation:

[0128]

[0129] where, (PPG(f)|PR) is the amplitude of the frequency component corresponding to the pulse rate (PR) of the subject, (PPG(f)|PR+RR) is the amplitude of the frequency component corresponding to the sum of the pulse rate and the respiratory rate (PR+RR), and

[0130] (PGG(f)|PR-RR) is the amplitude of the frequency component corresponding to the pulse rate minus the respiratory rate (PR-RR).

[0131] Therefore, this equation effectively gives the value of the ratio of the sum of the PR±RR frequency amplitude peaks to the PR frequency amplitude peak.

[0132] Changes in the pulmonary conduction time affect the phase of the frequency domain peaks. However, the amplitude of the frequency domain peaks remains insensitive to the pulmonary conduction time. The presence of small residual FM in the PPG signal is tolerable and has no significant impact on accuracy. This is because this contributes to equal and opposite changes in the PR-RR and PR+RR peaks, with the result that the average value remains unchanged and still reflects AM.

[0133] Relatively speaking, at a larger FM ratio, AM, transpulmonary delay, and FM are mixed in a complex manner in the spectrum, and the average values of PR-RR and PR+RR can no longer be used to extract the PPG variability index reflecting AM.

[0134] In addition, if the AM contribution to the PPG variation has not been properly removed, additional peaks appear in the spectrum of the PPG signal. In particular, peaks can also be observed at PR±n.RR, where n≥2. This is illustrated in Figure 8B The above adaptive resampling removes the FM contribution, thus removing these additional peaks. Therefore, if the resampling has successfully removed the FM modulation in the Fourier-transformed PPG signal, the additional peaks at PR±n.RR, where n≥2, should not be present.

[0135] In some embodiments, the method may include detecting whether the spectrum contains a frequency component peak at a frequency of PR±n.RR, where n≥2. This can be used as a check for successful elimination of FM modulation in the PPG signal. If an additional peak is detected, the resampling operation may be rerun again, or additional filtering may be performed, or the relevant signal portion (i.e., for the relevant current time window) may be discarded, and the method may be repeated for the next time window. The absence of a peak at PR±n.RR (where n≥2) can thus be used as feedback for the proper suppression of FM by the adaptive resampling algorithm.

[0136] In some embodiments, to implement the step of selecting or identifying relevant peaks in the spectrum, additional use may be made of the respiration signal of the subject, based on which the respiration rate for the subject can be determined. The pulse rate signal that has been acquired as an earlier part of the method may also be used. Then the set of frequency component peaks may be selected from the spectrum based on applying a peak identification algorithm that uses the respiration signal and the pulse rate signal as inputs. The respiration signal may be obtained from a respiration sensing module, which may be an airflow sensing module located at the airway, such as a temperature sensor located at the airway. Alternatively, the PPG sensor signal itself may be used to derive the respiration rate by well-known methods.

[0137] Alternatively, the method may employ a method in which it is assumed that the spectrum of the resampled signal contains peaks only at four frequencies RR, PR, PR - RR, and PR+RR, and thus simply identifies the amplitudes of the entire set of peaks present in the spectrum. Which peak corresponds to which of the four physiological markers can be determined by the order of the peaks: namely, the first peak is RR (usually the lowest frequency), the second peak is PR - RR, the third peak is PR, and the fourth peak is PR+RR.

[0138] As previously mentioned, in some embodiments, a filtering operation is additionally applied to the acquired PPG signal before the resampling operation. Now the details of an example implementation of such a filtering option will be discussed.

[0139] Preferably, the filtering operation employs the use of an adaptive digital filter, where the coefficients of the digital filter are configured according to the input of the filter. The coefficients may be configured based on a metric related to or indicative of the SpO2 value for the patient.

[0140] The above acquisition 112 of the PPG signal may include obtaining a red sensor signal waveform and an infrared sensor signal waveform from a PPG sensor.

[0141] The method may also include applying an adaptive digital filter to the red sensor signal waveform and the infrared sensor signal waveform from the PPG sensor. The filtered PPG signal (i.e., the PPG signal with filtered red and IR components) is then used for the remainder of the method.

[0142] Figure 9 Shows the red signal PPG for cleaning the PPG sensor signal R (t) and the infrared signal PPG IR (t) of the block diagram of the processing steps of an example adaptive filter operation 70.

[0143] Receive an input indicating a metric related to or associated with the SpO2 value for the subject. This can be calculated from the unfiltered PPG signal using known techniques. An example is further described below.

[0144] Reference Figure 9 , the red signal PPG R (t) and the infrared signal PPG IR (t) the AC component is normalized by its DC component. In other words, the AC component is extracted from each signal and divided by the DC component. This is indicated by Figure 9 the processing steps 72a and 72b in.

[0145] Thereafter, the difference 74a and the sum 74b signals of the two normalized signals are calculated.

[0146] The difference signal 74a is calculated as:

[0147]

[0148] where, and PPG n IR is the PPG signal sample n of the PPG IR signal, and PPG n R is the PPG signal sample n of the PPG R signal.

[0149] The ratio r SpO2 is a common metric derived in pulse oximetry and is related to or indicative of SpO2. The actual SpO2 value is typically calculated from the r SpO2 value using a sensor type-specific calibration equation. However, for the purpose of the filter, the ratio r SpO2 is sufficient.

[0150] The sum signal 74b is calculated as:

[0151] sum n =PPGn IR +c·PPG n R

[0152] where c is a constant that can be tuned according to desire.

[0153] The differential spectrum calculated in this way from 74a reflects the non-arterial and noise components and is used to determine the digital filter coefficients. The sum signal 74b is obtained by adding the fraction c of the red signal to the infrared signal. This fraction can be zero, but preferably c is set to obtain the highest signal-to-noise ratio of the sum signal. The clean PPG(t) is obtained by passing the sum signal through the digital filter 76 and using the differential signal to define the coefficients.

[0154] The result of the filtering operation is to leave a time-domain PPG signal from which the non-arterial components have been removed.

[0155] Reference will now be made to Figure 10 illustrate an example implementation of a method according to at least one set of embodiments. It should be recognized that not all features of this particular implementation are necessary for the overall inventive concept, and various details are presented by way of illustration of at least one example implementation. In particular, as already mentioned, the filtering operation 70 is optional and can therefore be omitted from the Figure 10 embodiment if preferred.

[0156] Figure 10 A block diagram outlining the steps of an example method for calculating a PPG variability metric indicative of fluid responsiveness is shown.

[0157] As indicated at 70, the method includes receiving a PPG signal for an object as input, the PPG signal including a red PPG signal PPG R (t) and an infrared PPG signal PPG IR (t).

[0158] The method may also include a processing operation for filtering the red and infrared signals. This leaves a time-domain PPG signal from which the non-arterial components have been removed. This can be done, for example, using the adaptive filter 70, such as the filter described above with reference to Figure 9 described filter. Figure 10 Examples of include SpO2 algorithms 82 for calculating a metric related to or indicative of a patient's SpO2 value. In this example, this is calculated based on the input red and infrared PPG signals. For example, in the Figure 9 adaptive filter of , the metric related to or indicative of SpO2 is the ratio: Thus, in some embodiments, the SpO2 algorithm 82 may output the ratio r SpO2Calculated value.

[0159] The SpO2 algorithm 82 can, for example, utilize a runtime window (e.g., a few seconds) to calculate the SpO2 metric. This is common for the calculation of SpO2.

[0160] If the filter 70 is omitted, the SpO2 algorithm 82 can also be omitted.

[0161] In parallel with the SpO2 algorithm 82, a pulse algorithm 84 can additionally be run, which is configured to calculate the instantaneous pulse rate. This is also calculated using the red and infrared PPG signals based on determining the inter-beat intervals within the red and / or infrared PPG signals. The pulse algorithm can operate with a runtime window of, for example, a few seconds. This time-domain algorithm 84 determines the inter-beat intervals (instantaneous pulse rate). As already discussed, the instantaneous pulse rate is used as an input for the adaptive resampling operation (providing information about the FM modulation depth of the PPG signal 70).

[0162] Due to the noise in the PPG signal, the instantaneous inter-beat intervals can sometimes be inaccurate, and thus optionally a smoothing operation can additionally be applied to overcome this problem.

[0163] After the optional filtering operation 70, the filtered signal is provided as an input to an adaptive resampling algorithm 86, which is adapted to resample the signal at a resampling rate that is a function of the instantaneous pulse rate. Thus, the output of the above-described pulse algorithm 84 is provided as an input to the adaptive resampling algorithm 86. As described above, the adaptive resampling algorithm removes or at least significantly reduces the FM contribution to the PPG signal variability. As described above, this is the purpose of the overall method, since only the AM contribution to the PPG variability mainly reflects the fluid responsiveness. The details of the adaptive resampling operation have been described above and will not be repeated here.

[0164] The resampled PPG signal is processed by a discrete time-domain to frequency-domain transform 88 (such as a fast Fourier transform (FFT)). This results in a spectrum for the resampled PPG signal. As described above, since the FFT assumes equally spaced samples, the Fourier transform of the signal resampled based on the pulse rate results in a spectrum that essentially removes the FM modulation.

[0165] A predefined set of peaks is extracted from the spectrum and used for calculating 90 the PPG variability metric. In particular, peaks at least at frequencies corresponding to the pulse rate (PR) and the pulse rate (PR) + / - the respiratory rate (RR) can be used. For example, the PPG variability metric PPGV can be calculated as:

[0166]

[0167] where (PPG(f)|PR) is the amplitude of the frequency component corresponding to the pulse rate (PR) of the subject, (PPG(f)|PR+RR) is the amplitude of the frequency component corresponding to the sum of the pulse rate and the respiration rate (PR+RR), and

[0168] (PGG(f)|PR-RR) is the amplitude of the frequency component corresponding to the pulse rate minus the respiration rate (PR-RR). The details thereof have been described above.

[0169] Also as described above, in some embodiments, the presence of additional peaks at PR+ / -2RR in the spectrum can be used as a check for successful FM cancellation from the PPG signal, and wherein, if a peak at PR+ / -2RR is detected above a certain threshold amplitude, adaptive resampling can be repeated. This is schematically indicated by the feedback path 92. This feature is optional.

[0170] In Figure 11 is schematically shown a processing flow according to a further example method in accordance with one or more embodiments.

[0171] This method is the same as the method described above Figure 10 except that an additional input of the respiration signal 102 of the subject obtained from a separate respiration measurement source is provided. Again, not all elements are necessary. In particular, at least the filtering operation 70, the SpO2 algorithm 82, and the feedback path 92 are not required and can be omitted.

[0172] The separate respiration signal can serve any of a variety of different purposes.

[0173] In some embodiments, the separate respiration signal 102 can be used to improve the accuracy of the above-described peak selection process. This has been discussed above.

[0174] In some embodiments, and as Figure 11 indicated, the respiration signal 102 can be used to improve the reliability of the instantaneous pulse rate calculation 84.

[0175] Specifically, deriving an instantaneous pulse rate may include: determining an inter-beat interval from the PPG signal 70, correlating the inter-beat interval with a respiratory signal, and deriving a final instantaneous pulse rate metric based on a combination of the respiratory signal and the PPG signal. In particular, it may be difficult to calculate a real-time instantaneous pulse rate using only the PPG inter-beat interval, which can account for pulse rate variations. This is because the sampling interval is relatively large compared to the scale at which pulse rate variations may occur. However, it is also known that real-time short-scale variations in a person's pulse rate are correlated with the respiratory cycle. Thus, by using a respiratory signal, it may be possible to better detect variations in the pulse rate at shorter scales. For example, one method is to use the PPG inter-beat interval times over a measurement window to calculate a first sequence of instantaneous pulse rate sample points. Then, by assuming that the pulse rate will vary in relation to the respiratory cycle, the set of sample points can be further supplemented with additional sample points calculated from the respiratory signal. In other words, an initial instantaneous pulse rate signal calculated solely from the PPG signal can be modulated in relation to the modulation in the subject's respiratory cycle. Thus, a high-quality instantaneous pulse rate reference signal can be provided to the adaptive resampling algorithm 86.

[0176] Regarding the source of the respiratory signal 102, for a mechanically ventilated patient, a volume or pressure signal from the ventilator can be used. For an autonomously breathing patient, respiratory measurements can be derived, for example, from a temperature sensor (such as a thermistor) located at the patient's airway (such as the patient's nasal airway). In some embodiments, the temperature sensor can be incorporated into a nasal SpO2 sensor. The temperature sensor responds to the temperature difference in the respiratory air during inhalation and exhalation. A capnography sensor is another option for providing respiratory input.

[0177] Thus, in summary, according to one or more embodiments, obtaining an instantaneous pulse rate signal may include: obtaining a respiratory signal for a subject and processing the respiratory signal in combination with the PPG signal to derive an instantaneous pulse rate signal. In some embodiments, obtaining the respiratory signal may include receiving a temperature signal from a temperature sensor placed adjacent to the subject's airway and mapping oscillations in the temperature sensor signal to oscillations in the subject's respiratory cycle to derive the respiratory signal.

[0178] The processing flow according to an additional exemplary method according to one or more embodiments is schematically shown in Figure 12 in.

[0179] This method can be the same as the above Figure 10 or Figure 11The method is the same, except that the resampling operation 86 and the peak identification operation can be supplemented in the manner described below. Also, not all elements of the method flow shown are necessary. For example, at least the filtering operation 70, the SpO2 algorithm 82, the feedback path 92, and the use of the respiratory signal 102 in the calculation of the instantaneous pulse rate signal are not required and can be omitted.

[0180] For spontaneously breathing patients, the respiratory cycle duration varies between breaths in an irregular manner. This results in a more distributed intensity across the spectrum than in ventilated patients and makes the derived PPGV values less accurate. To overcome this problem, a two-stage adaptive resampling can be implemented.

[0181] The respiratory signal 102 is provided as an input to the adaptive resampling algorithm. To implement the two-stage resampling, in the first step, the total number of (re)sampling points of the PPG signal over a time window spanning a single respiratory cycle is determined based on the duration of the respiratory cycle. In other words, the number of PPG sampling points over the respiratory cycle is made proportional to the duration of the respiratory cycle. In the second step, the sampling points during each given respiratory cycle are distributed over the respiratory cycle according to the instantaneous pulse rate, i.e., the sample density is made a function of the instantaneous pulse rate over the respiratory cycle.

[0182] Subsequently, based on interpolation from the original PPG samples, the PPG signal intensity for each of the distributed (re)sampling time points is determined.

[0183] In cases where the average pulse rate and the average respiratory rate are close in frequency, it becomes more difficult to identify the RR, PR, and PR±RR peaks in the spectrum. To solve this problem, a peak identification algorithm 104 can be added, which is configured to determine the spectral amplitudes of these peaks. The algorithm receives the respiratory signal 102 as an input and calculates the average respiratory rate over two or more respiratory cycles. Thereby, the RR, PR, and PR±RR frequencies can be calculated unambiguously, and the amplitude values corresponding to these frequencies can be read out to provide an input for calculating the PPGV metric in the manner described above.

[0184] The above two-stage adaptive resampling and the above additional peak identification process are functionally bound to each other, and thus an exemplary embodiment can include one or both of these optional features or neither of these optional features.

[0185] Although the above examples have involved the use of photoplethysmography (PPG) signals, any plethysmography signal can actually be used. For example, the method can be equally well applicable to speckle plethysmography (SPG) signals. The latter is similar to PPG technology, except that a coherent laser source is used instead of an LED, and an image sensor is used instead of a single photodetector for speckle (interference) detection. Studies have shown that SPG provides a higher signal-to-noise ratio than PPG and obtains a waveform more similar to pressure. Both can provide advantages when obtaining a higher quality variability index for fluid responsiveness.

[0186] If an SPG signal is used, the above optional adaptive filter 70 and SpO2 algorithm 82 can be omitted. All other features are fully compatible with the use of SPG signals.

[0187] In combination with any of the above embodiments, a further improvement in the reliability of the PPG variability metric can be obtained by normalizing the variability metric by a measure of the object's respiratory volume. This has the effect of compensating for patient-specific respiratory volumes as well as differences in respiratory volumes during ventilation compared to spontaneous breathing. If a Fourier transform is used in the method, this normalization can be achieved by dividing the PPG variability metric by the amplitude of the respiratory rate (RR) peak in the spectrum of the resampled PPG signal.

[0188] The above embodiments of the present invention employ a processing device. The processing device can generally include a single processor or multiple processors. It can be located in a single containing device, structure, or unit, or it can be distributed among multiple different devices, structures, or units. Thus, a reference to the processing device being adapted or configured to perform a particular step or task can correspond to the performance of that step or task by any one or more of the multiple processing components, either individually or in combination. Those skilled in the art will understand how such a distributed processing device can be implemented. The processing device includes a communication module or input / output section for receiving data and outputting the data to other components.

[0189] One or more processors of the processing device can be implemented in various ways using software and / or hardware to perform the various required functions. The processor generally employs one or more microprocessors that can be programmed using software (such as microcode) to perform the required functions. The processor can be implemented as a combination of dedicated hardware for performing some functions and one or more programmed microprocessors and associated circuits for performing other functions.

[0190] Examples of circuits that can be employed in various embodiments of the present disclosure include, but are not limited to, conventional microprocessors, application specific integrated circuits (ASICs), and field programmable gate arrays (FPGAs).

[0191] In various embodiments, a processor may be associated with one or more storage media such as volatile and non-volatile computer memories such as RAM, PROM, EPROM, and EEPROM. The storage media may be encoded with one or more programs that perform the required functions when run on one or more processors and / or controllers. The various storage media may be fixed within the processor or controller or may be transportable such that one or more programs stored thereon may be loaded into the processor.

[0192] By studying the drawings, the disclosure, and the appended claims, those skilled in the art will be able to understand and realize variations of the disclosed embodiments when practicing the claimed invention. In the claims, the word "comprising" does not exclude other elements or steps, and the words "a" or "one" do not exclude a plurality.

[0193] A single processor or other unit may implement the functions of several items recited in the claims.

[0194] Although specific measures are recited in mutually different dependent claims, this does not indicate that a combination of these measures cannot be used to advantage.

[0195] A computer program may be stored / distributed on a suitable medium, such as an optical storage medium or a solid-state medium provided together with or as part of other hardware, but the computer program may also be distributed in other forms, such as via the Internet or other wired or wireless telecommunication systems.

[0196] If the term "adapted to" is used in a claim or the specification, it should be noted that the term "adapted to" is intended to be equivalent to the term "configured to".

[0197] Any reference signs in the claims should not be construed as limiting the scope.

Claims

1. A method (100) for deriving plethysmographic variability metrics, the method comprising: obtaining (112) a plethysmographic signal for an object, the plethysmographic signal comprising a sampled time series of plethysmographic sensor measurements across a time window, and the time window spanning a plurality of pulse cycles of the object; obtaining (114) an instantaneous pulse rate signal of the object across the time window; applying (116) a resampling algorithm to the sampled time series of the plethysmographic signal, wherein the resampling algorithm is adapted to perform sampling rate modification at least in part based on the instantaneous pulse rate across the time window, and wherein the resampling algorithm resamples the plethysmographic signal across the time window at a resampling rate that is a function of the instantaneous pulse rate; applying (118) a discrete time domain to frequency domain transform to the resampled plethysmographic signal waveform to obtain a frequency spectrum; identifying (120) a set of peaks of predetermined frequency components in the frequency spectrum; calculating (122) a plethysmographic variability metric based on the frequency amplitudes of the identified peaks in the frequency spectrum; and generating a data output based on the calculated plethysmographic variability metric.

2. The method according to claim 1, wherein Obtaining the plethysmographic signal comprises: obtaining a red sensor signal waveform and an infrared sensor signal waveform from the plethysmographic sensor; and applying a digital filter to the red sensor signal waveform and the infrared sensor signal waveform from the plethysmographic sensor; Preferably, wherein the coefficients of the digital filter are configured based on a metric related to the SpO2 value for the patient.

3. The method according to claim 2, wherein, The plethysmographic signal is obtained from a plethysmographic sensor disposed at a central body location of the patient, the central body location being, for example, the forehead of the patient.

4. The method according to any one of claims 1 to 3, wherein Obtaining the instantaneous pulse rate signal comprises: detecting the inter-beat time intervals in the plethysmographic signal.

5. The method according to any one of claims 1 - 4, wherein, The resampling algorithm comprises: defining a number of resampling points for the time window; distributing the resampling points across the time window according to the instantaneous pulse rate as a function of time across the time window; determining a plethysmographic signal value for each resampling point based on interpolation of the sampled time series of measurements from the obtained plethysmographic signal.

6. The method according to any one of claims 1-5, wherein, The resampling algorithm comprises: receiving a respiratory signal for the object, the respiratory signal representing a parameter related to a respiratory cycle phase; identifying a time window in the plethysmographic signal that spans a single respiratory cycle; defining a number of resampling points for the time window in proportion to the duration of the single respiratory cycle; distributing the resampling points across the time window according to the instantaneous pulse rate as a function of time across the time window; determining a plethysmographic signal value for each resampling point based on interpolation of the sampled time series of measurements from the obtained plethysmographic signal.

7. The method according to any one of claims 1-6, wherein, The set of peaks of predetermined frequency components corresponds to: Frequency component corresponding to the pulse rate of the object (PPG(f)| PR ), The frequency component corresponding to the sum of the pulse rate and the respiration rate (PPG(f)| PR+RR ), and The frequency component corresponding to the pulse rate minus the respiration rate (PGG(f)| PR-RR ).

8. The method according to claim 7, wherein The plethysmographic variability metric (PPGV) is calculated based on evaluating the following equation:

9. The method according to claim 7 or 8, Among them, The method includes obtaining a respiratory signal for the object, the respiratory signal representing a parameter related to a respiratory cycle phase; and wherein a set of frequency component peaks is selected from the spectrum based on applying a peak identification algorithm that uses the respiratory signal as an input.

10. The method according to claim 9, wherein, Obtaining the respiratory signal includes: receiving a temperature signal from a temperature sensor placed adjacent to the airway of the object and mapping oscillations in the signal of the temperature sensor to oscillations in the respiratory cycle of the object to derive the respiratory signal.

11. A computer program product comprising computer program code configured to, when run by a processor, cause the processor to perform the method according to any one of claims 1 - 10.

12. A processing device (32) comprising: an input / output section (34); one or more processors (36) operatively coupled to the input / output section and adapted to: obtain a plethysmographic signal for an object, the plethysmographic signal comprising a sampled time series of plethysmographic sensor measurements over a time window, and the time window spanning a plurality of pulse cycles of the object; obtain an instantaneous pulse rate signal for the patient over the time window; apply a resampling algorithm to the sampled time series of the plethysmographic signal, wherein the resampling algorithm is adapted to perform a sampling rate modification based at least in part on the instantaneous pulse rate over the time window, and wherein the resampling algorithm resamples the plethysmographic signal over the time window at a resampling rate that is a function of the instantaneous pulse rate; apply a discrete time domain to frequency domain transform to the resampled plethysmographic signal waveform to obtain a spectrum; identify a set of predetermined frequency component peaks in the spectrum; calculate a plethysmographic variability metric based on the frequency amplitudes of the identified peaks in the spectrum; and generate a data output based on the calculated plethysmographic variability metric.

13. A system (30) comprising: the processing device (32) according to claim 12; and a plethysmographic sensor (54), such as a PPG sensor.

14. The system (30) according to claim 13, further comprising a sensor (56) for measuring a parameter related to a respiratory cycle phase.

15. The system (30) according to claim 14, wherein, The sensor includes a temperature sensor for placement adjacent to the airway of the object, and optionally, wherein the sensor includes a sensing unit that includes both the plethysmographic sensor and the temperature sensor.

Images