Method and system for cardiac cycle analysis

Abstract

A method and system for processing pulse plethysmography detects peaks, valleys, and amplitudes in a series of measurement periods and identifies measurement periods that do not meet selected quality criteria. Periods that do not meet the selected quality criteria are removed and the remaining periods are stitched together. The system and method help improve the quality and reliability of cardiovascular parameters, such as stroke volume and cardiac output, calculated using plethysmography data as input.

Description

[Technical Field]

[0001] The present invention relates to devices, methods, and systems for measuring cardiac output, stroke volume, and / or arterial blood pressure using data obtained by pulse plethysmography, including photoplethysmography. [Background technology]

[0002] Background technology Plethysmography involves measuring volume changes in all or part of the body. Volume changes in various parts of the body can be measured in different ways to detect or monitor various medical conditions by measuring various physiological parameters. For example, air cuff plethysmography (CPG) measures changes in the circumference of a limb or finger by recording changes in pressure in an air-filled cuff that surrounds the limb or finger. Based on volume changes, CPG is used to measure blood flow and provides useful information related to chronic venous disease. Impedance plethysmography, also known as dielectrography, involves measuring changes in electrical impedance between electrodes placed on opposite sides of a body part, indirectly measuring volume changes in the path of electrical current flowing between the electrodes.

[0003] Inductance plethysmography is a method in which an alternating current is applied through a loop of wire attached to the chest or abdomen, generating a magnetic field perpendicular to the orientation of the loop. A change in the area enclosed by the loop produces a backflow within the loop that is directly proportional to the rate of change of area. Unlike in inductance plethysmography, the current does not pass through the body. Respiratory inductance plethysmography is used to calculate chest and abdominal expansion, respiratory rate, breathing pattern, and tidal volume. Electromagnetic inductance plethysmography uses a vest containing conductive elements that surround the torso to quantify volume changes in the chest and abdomen as the conductive elements move in a magnetic field.

[0004] Photoplethysmography (PPG) measures changes in blood volume in tissue caused by blood pressure (BP) pulses through the tissue's vasculature. Blood volume changes are detected by measuring the amount of light transmitted or reflected to a sensor from a light source used to illuminate the skin. Transmissive PPG can also be obtained using endoscopic techniques. The shape of the PPG waveform varies from subject to subject and depends on the location and manner in which the pulse oximeter is contacted or attached to the body.

[0005] Recently, methods and devices have been developed for noninvasively measuring left ventricular stroke volume (SV), cardiac output (CO), and other physiological parameters, involving a type of physiological sensor data processing that enables these measurements. U.S. Patent No. 8,494,829 discloses a method for processing hemodynamic and electrodynamic sensor data to calculate SV. The hemodynamic and electrodynamic sensor data can be acquired noninvasively and processed using a model that represents hemodynamic and electrodynamic physiology and iteratively fuses data from the two sensors. U.S. Patent No. 8,494,829 is incorporated herein by reference in its entirety, particularly the description of the physical model in columns 5-20.

[0006] U.S. Patent No. 9,649,036 describes a method for analyzing PPG and electrodynamic signals using a digital signal processor with a dynamic state space model (DSSM) that includes a process model that relates PPG and internally derived electrodynamic signals to systolic contractions. U.S. Patent No. 9,649,036 is incorporated herein by reference, particularly as an example of a pulse oximeter (PO) with probabilistic data filtering in columns 15-19.

[0007] U.S. Patent No. 9,060,722 describes a method for estimating SV using sensor data from PPG using a dual estimation algorithm with a DSSM that mathematically represents the physiological process that generates SV and the data collected by PPG. U.S. Patent No. 9,060,722 is incorporated herein by reference, particularly for its description of the dual estimation algorithm and DSSM in columns 6-14. U.S. Patent No. 9,375,171 describes a method for estimating SV from BP and PPG data using a dual estimation DSSM with a hemodynamic process model.

[0008] International Publication WO 2010 / 0274102 A1 discloses a PO system and processing method for measuring SV and CO. The PO system includes a data processor configured to execute a method combining a stochastic processor and a physiological model of the cardiovascular system in a Data Storage Module (DSSM) that can remove contaminating noise and artifacts from the PO sensor output and measure blood oxygen saturation, heart rate (HR), SV, aortic pressure, and systemic pressure. The DSSM includes a mathematical model of the cardiovascular system that models the physiological processes that generate the pulse measured by the PO. One of the models includes variables and parameters including aortic pressure, radial artery pressure, peripheral resistance, aortic impedance, and blood density. The model uses data from the PO, including the measured HR.

[0009] EP 2512325 describes a method for processing PO data using a dynamic state-space model to extract estimates for one or more of total blood volume, SV, vasomotor tone, and autonomic tone. A microprocessor receives the data and places it within a dynamic state-space model that mathematically represents the physiological processes that produce the physiological parameters measured by PO.

[0010] While the above-described inventions can provide accurate and medically useful measurements of cardiovascular parameters including CO and SV, there are situations in which the measured plethysmographic waveform has characteristics that do not allow for measurement of SV and CO or that result in measurements of SV and CO with less than desirable accuracy. These measurements of SV, CO, and other cardiovascular parameters involve the use of plethysmographic waveform (PW) data as input to a computer model of the cardiovascular system.

[0011] The quality of the PW data used as input affects the quality of the output, including measured SV, CO, and / or other cardiovascular parameters, depending on the cardiovascular system model used. Movement, including plethysmography sensor or patient respiratory movement, introduces artifacts into the measured plethysmography waveform. Pulsatile variations in an individual's HR and SV, body movement, and changes in peripheral resistance over time complicate accurate measurement of SV and CO. In populations, various cardiac abnormalities, cardiovascular diseases, body mass index, body size, age, and many other physiological parameters can also complicate CO and SV measurements, as population PW data may have different shapes, making analysis of the input data for inclusion as inputs into cardiovascular models difficult. Measured PW data for a particular patient may be difficult or impossible to process using a cardiovascular system model because the shape of the PW may differ significantly from the average patient population. Summary of the Invention [Problem to be solved by the invention]

[0012] In summary, there are many situations in which the measured PW contains artifacts or has other characteristics that do not allow for the measurement of SV, CO, and other cardiovascular parameters, or that result in measurements of SV and CO with less accuracy than desired. [Means for solving the problem]

[0013] Embodiments of the present invention seek to mitigate, alleviate or eliminate one or more deficiencies, drawbacks or problems in the art, singly or in any combination, such as those identified above, preferably by providing systems and methods for processing PW data representative of blood flow to improve the reliability and accuracy of SV and CO measurements made using the PW data as input.

[0014] In one aspect, the present invention provides a computer-implemented method for analyzing cardiac cycles in a set of PW data and using the processed data to generate input data for a method or processor for measuring or calculating CO, SV, or other cardiovascular parameters.

[0015] In another aspect, the present invention provides an apparatus or system for analyzing cardiac cycles in a set of PW data and generating input data for a processor that uses the processed data to measure CO, SV, or other cardiovascular parameters, the apparatus comprising a data processor including instructions for carrying out the method.

[0016] In yet another aspect, the present invention provides a non-transitory computer-readable storage medium storing a program for causing a computer to perform a method for analyzing cardiac cycles in a set of PW data and using the processed data to generate input data for a processor that measures CO, SV, or other cardiovascular parameters.

[0017] The described methods and systems improve the accuracy and reliability of cardiovascular parameter measurements made using pulse plethysmography data by selecting portions of waveform data from the pulse plethysmography data to use in calculating cardiovascular parameters and selecting other portions of the waveform data to exclude from use in calculating cardiovascular parameters. The elements of the drawings are not necessarily to scale relative to each other, emphasis instead being placed upon clearly illustrating the principles of the present disclosure. Like reference characters designate corresponding parts throughout the several views of the drawings. [Brief explanation of the drawings]

[0018] [Figure 1] 1 is a flowchart of a general process that may be performed in an embodiment of a method for processing waveform data in accordance with the present invention. [Figure 2a] 1 is a flowchart of steps that may be performed in a periodic data quality assessment process. [Figure 2b] This is an example of PW data cycle quality assessment by identifying key cycle features. [Figure 3a] 1 is a flowchart of steps that may be performed in a coarse screening process. [Figure 3b] 1 is an example of PW data, approximate waveforms, and residuals determined during coarse sieving. [Figure 4a] 1 is a flowchart of steps that may be performed in a cardiac cycle analysis process. [Figure 4b] 10 is an example of two skewed Gaussian functions used to fit PW period data. [Figure 4c] 10 is an example of a PW period compared to an approximate period using two skewed Gaussian functions. [Figure 4d] 10 is an example of three skewed Gaussian functions used to fit PW period data. [Figure 4e] 10 is an example of a PW period compared to an approximate period using three skewed Gaussian functions. [Figure 4f] 10 is an example of four skewed Gaussian functions used to fit PW period data. [Figure 4g] 10 is an example of a PW period compared to an approximate period using four skewed Gaussian functions. [Figure 5] 1 is a flowchart showing the flow of data between components of a device or system for analyzing the cardiac cycle to measure CO, SV, or other cardiovascular parameters. DETAILED DESCRIPTION OF THE INVENTION

[0019] All technically specific terms used herein are intended to have their art-recognized meaning in the context of the description unless otherwise indicated. All non-technical specifically terms are intended to have their plain language meaning in the context of the description unless otherwise indicated.

[0020] The term "parameter" in the context of the cardiovascular system includes stroke volume (SV), cardiac output (CO), aortic pressure, venous pressure, arterial pressure, flow, vascular inertance, vascular compliance, vascular resistance, and the inertance, compliance, and resistance of a collection of blood vessels or a portion of the cardiovascular system. In the context of a computational model of the cardiovascular system, some physical factors may also be parameters in the modeling sense, in that they may correspond to fixed input values ​​into the computational model. Other physical factors, in the context of a computational model, correspond to variables whose values ​​are calculated by the computational model. Cardiovascular system parameters may correspond to variables of the computational model or may be calculated from one or more variables. Whether a cardiovascular system parameter corresponds to or is calculated from variables and / or parameters of the computational model depends on the equations used in the computational model.

[0021] As used herein, plethysmographic data is data collected from a human or non-human patient, the data including waveforms representing pressure versus time or volume versus time of blood flow in the patient.

[0022] Plethysmographic waveform data or PW data relates to measuring changes in blood volume over time, which can be translated into changes in blood pressure (BP) over time. Time-varying pulse pressure can be measured directly or indirectly from plethysmographic data. Plethysmographic waveforms can include measured pulse volume versus time or fraction of a cardiac cycle. Plethysmographic waveforms can include plethysmographically measured pulse pressure versus time or fraction of a cardiac cycle.

[0023] The most commonly used plethysmography method and apparatus is the pulse oximetry or pulse oximeter (PO) device, which uses the absorption of light at two wavelengths corresponding to the absorption maxima for oxygenated and deoxygenated forms of hemoglobin. Changes in absorption are related to changes in blood volume, and changes in blood pressure can result from changes in blood volume. Time-varying blood volume can additionally or alternatively be measured using ultrasound to measure blood vessel diameter. Time-varying pulse pressure waveforms may be measured directly using a partially inflated blood pressure cuff, a pressure transducer, a strain gauge, or a stretch sensor. Systems, devices, and methods are described herein using pressure as the time-varying parameter. This is convenient because PO data is typically converted to pressure versus time data.

[0024] Pressure need not be used as a time-varying variable in plethysmography data. Instead, PW periods may be in the form of, for example, volume versus time or absorption versus time. PO devices can use two wavelengths to compare the amount of oxygenated and deoxygenated hemoglobin and report the oxygenation rate of hemoglobin. While conventional PO data is used in the examples herein, the apparatus and methods herein may also use data collected using, for example, only changes in absorption by oxygenated hemoglobin or resulting changes in blood volume or BP. While PO data is often collected at the fingertip, it can additionally or optionally be measured at other extremities, such as the ear or toe, and reflection or scattering PO measurements can be made at locations on the body that are not compatible with transmission PO measurements.

[0025] In the following description, PO plethysmography data is used as an example of plethysmography data for measuring SV and CO. Other types of plethysmography data may be in the form of, or converted to, pulsatile blood flow waveforms of pressure versus time, volume versus time, or light absorption versus time. These may also be used, as volume, pressure, and absorption may be converted from one to the other, depending on the method of data collection. PO data may be obtained from a database of PO measurements made on the subject at a remote location and stored in memory. PO data may also come directly from the PO during measurement.

[0026] Plethysmography data processed as described may be used as input data for measuring, calculating, or estimating SV, CO, and other cardiovascular parameters by a method using a human cardiovascular system (HCS) model, which may include, for example, a dynamic state space model (DSSM). It is not necessary for human plethysmography data to be input into a computer-implemented method using a non-human animal cardiovascular system model to measure or calculate cardiovascular parameters. Examples of computer-implemented methods for measuring cardiovascular parameters are described in U.S. Pat. Nos. 8,494,829, 9,649,036, 9,060,722, 9,375,171, International Publication WO 2010 / 0274102, and EP 2512325.

[0027] HR, SV, CO, and other cardiovascular parameters are not constant over time and may vary from beat to beat. Therefore, any measurement of these types of parameters is not necessarily an average value. Constant variations in measured pulse pressure, cycle duration, and PW shape contribute to the measurement of SV, CO, and other cardiovascular parameters. In addition, subject movement and other physical disturbances can cause artifacts in the data. Applying quality metrics to the measured PW data can resolve issues arising from artifacts and contaminants in the measured PW data by minimizing the impact of artifacts and outliers on subsequent processing steps. For example, an algorithm can select individual sets of cycles and / or runs of cycles that combine a high ratio of cycle amplitude to cycle amplitude variability, minimal localized plethysmography or pressure wave amplitude variability, and minimal HR variability.

[0028] The present invention provides a method for processing PW data to select periods for use as input to a computer-implemented method for measuring cardiovascular parameters using plethysmographic pulse data. The method improves the accuracy of the measured cardiovascular parameters and extends the applicability of the computer-implemented method for measuring cardiovascular parameters to patients with abnormally shaped PWs. This is achieved by a combination of data quality assessment, data sieving, and curve fitting to select periods from the PW to use as input to the computer-implemented method for measuring cardiovascular parameters using plethysmographic pulse data.

[0029] FIG. 1 outlines a general process that may be performed in an embodiment of a method for processing PW data. The data consists of a series of waveforms corresponding to a series of pulses measured by a plethysmography measurement device, with each pulse resulting from a heartbeat, or cycle. The PW data, e.g., from a PO, is input to a software module configured to execute a cycle data quality (CDQ) assessment algorithm that uses PW cycle characteristics, such as wave amplitude, waveform, and period, to select or deselect each cycle for further processing. The PW data of selected cycles is input to a coarse cycle data sieve or sieve software module that compares the received CDQ data with the expected output of the CDQ data and roughly approximates the shape of each cycle with a first mathematical formula. The algorithm compares the coarse approximation to the CDQ data and rejects PW cycles based on the cumulative error between the measured and approximated cycle waveforms. The remaining PW cycles are input to a software module that executes a cardiac cycle analysis (CCA) algorithm. The CCA module receives the coarsely filtered PW data and fits a second mathematical equation to each cycle waveform. The shape of the mathematical representation of each cycle's waveform is compared to the shape of that cycle's corresponding measured waveform. Cycles with measured and fitted waveforms that differ by measurements above a threshold are removed, and the PW cycle data that passes through the CCA can be made available as input to a computer module that calculates cardiovascular parameters, in this case SV and CO, using the CCA output data as input to a human cardiovascular system (HCS) model.

[0030] CDQ assessment, cycle data screening, and cardiac cycle analysis function as quality measurement processes to identify measurement cycles that do not meet selected quality criteria associated with these processes. The remaining measurement cycles or mathematical representations of the remaining measurement cycle waveforms are useful as inputs to computational models that calculate cardiovascular parameters from the measured plethysmographic waveform data. Depending on the computational model used, the calculated cardiovascular parameters may include SV, CO, aortic pressure, vascular inertance, compliance, resistance, etc.

[0031] The sequence of operations shown in FIG. 1 pertains to one embodiment of a method for processing PW data. Other embodiments of such methods may include the same processes in a different order. For example, in one embodiment, a coarse or fine curve fit to a waveform sequence may be performed as a first step during CDQ assessment. Additionally or alternatively, a coarse or fine curve fit to a waveform sequence may be performed as a first step during coarse data sieving. These embodiments may use the fitted curve to perform method steps associated with CDQ or coarse data sieving. In method steps in which one or more periods are removed or eliminated by ablation, adjacent periods may be stitched together to regenerate a continuous sequence of periodic waveforms.

[0032] FIG. 2a is a flowchart of method steps that may be performed by one embodiment of a CDQ assessment software module. Data including a sequence of PW periods from a fingertip PO is received by the module, optionally along with patient identification data and / or patient parameters such as age, sex, weight, height, and medical condition. To prevent artificially or abnormally shaped PW periods from leading to abnormal detection of start, stop, peak, and valley features or optimizer failure, the ratio of standard deviation to mean PPG period time can be used to identify high-risk periods before these periods are processed. If any period produces a standard deviation / mean period time ratio greater than a set threshold, the entire period may be removed by a removal process in which the waveform of the period is removed along the time axis from the start to the stop or end of the removed period. The periods immediately before and after the discarded period are spliced ​​together by shifting the time of the first subsequent period to match the start time of the removed preceding period.

[0033] PW cycle start, stop, peak, and valley characteristics are detected using a slope threshold crossing technique that ensures that only one slope crossing feature is detected per cycle. This allows for accurate identification of adjacent cycles, regardless of the waveform structure within each cycle. Slope threshold crossing landmarks are adaptively thresholded using the PW slope (the first derivative of the volume / pressure / absorbance vs. time signal) and an exponentially weighted moving average of the absolute value of the PW slope. A strong positive slope during systole generates a characteristic threshold crossing that is independent of, for example, subject motion and respiratory artifacts, cycle duration variations, or cycle vertical drift caused by variable heart rate. The slope moving average can be used to derive a second threshold to measure the approximate time of cycle onset for each cycle. As long as slope crossing feature detection maintains a one-to-one characteristic with respect to PW cycles, PW slope crossing features are used to initialize and retrieve cycle minimum, or valley T, and cycle maximum, or peak P, values, ensuring that minimum and maximum values ​​are correctly determined.

[0034] Figure 2b illustrates that a cycle peak P occurs after the maximum positive slope point +S and before the maximum negative slope point −S, as determined from the first derivative of the PW slope. A cycle valley occurs after the maximum negative slope point −S and before the maximum positive slope point +S of the next cycle. The upper dashed curve in Figure 2b extends along the cycle peak P, and the lower dashed curve extends along the cycle valley T, and the amplitude of each cycle is determined from the difference. The lower dashed curve can typically serve as a baseline that rises and falls in unison with the patient's breathing. "X"s are used to indicate the locations of the valley T, the maximum positive slope point +S, the cycle peak P, and the maximum negative slope point −S. Any of these locations may be used as the end of one cycle and the beginning of the next cycle. In Figures 3b and 4b-4g, the end / start points are indicated by vertical dashed lines at the cycle valleys.

[0035] Because the average HR associated with each PW data set is known, there is a range of expected cycle durations associated with the average HR. Once peaks are identified, peak-to-peak periods that are incompatible with the average HR can be removed from the data set. Abnormal heartbeats, such as premature ventricular contractions (PVCs) and patient movement during data collection, can cause cycle peaks to fall outside an acceptable range of the average HR. Similarly, large variations in peak amplitudes can also indicate abnormal cardiac cycles, patient movement, and other artifacts. A threshold can be set to remove cycles whose amplitudes differ from adjacent cycles by more than an acceptable amount.

[0036] The shapes of adjacent periodic waveforms may also be compared to identify periods that differ significantly from adjacent periods in terms of waveform shape. Data representing each PW may be in the form of volume or pressure versus time, for example, at a rate of 75 points per second. The shapes of adjacent periods may be compared point-by-point, originating from a common landmark in the wave, and the point-by-point error may be accumulated to provide an indication of the difference between the shapes of adjacent periods. Periods that differ from adjacent periods by more than a threshold or tolerance may be removed from the data set before the data is passed to a coarse sieving process. Periods adjacent to a removed period are stitched together by shifting the start of the period immediately following the removed period along the time axis to the start time of the first period or the first of the consecutive periods removed.

[0037] FIG. 3a is a flowchart of method steps that may be performed by one embodiment of a coarse sieving process. During CDQ assessment, PWs may be removed for periods with amplitudes, shapes, durations, etc. outside of set tolerances. If the measured data is particularly poor quality, only a portion of the original data is passed to the coarse data sieving process. Before fitting one or more mathematical formulas in the sieving process, it can be determined whether a sufficient portion of the original measurements is of sufficient quality for further processing. A threshold can be set below which further processing is stopped and an error message generated to notify the user that the measured data is of insufficient quality. For example, in a three-minute PW measurement, the threshold can be set to 50%. If more than 50% of the periods are removed during CDQ assessment, the user can be notified and the measurement can be repeated. The user may have the option to override the message and proceed with the coarse sieving. The threshold may also be set to 100%, effectively eliminating the step and allowing processing to continue even if all periods have been removed. This functionality can be included as the final step of the CDQ assessment, rather than as the first step of the coarse sieving.

[0038] Once data sufficiency is established, a cycle start point is assigned to the PW data. The cycle start point CS can be set at the beginning of the systolic wave or at the cycle trough, as shown in Figure 3b. Alternatively, the cycle start CS can be consistently selected anywhere along the cycle waveform. The baseline rises and falls with respiration, but can be compensated or normalized to provide a flat baseline. Measured PW waveform data often contains low-frequency signal noise from sources including the effects of the subject's breathing and movement. Without this noise removal, skew-Gaussian basis functions alone cannot model the variable baseline, potentially preventing an effective modeling solution in subsequent steps. Therefore, this baseline can be modeled using shape-preserving piecewise cubic interpolation, for example, as provided by the Matlab® function interp1 ("pchip"). The valley point of the PW cycle is used as the input to the interpolation. The baseline interpolation output is added to a sum of skew-Gaussian basis functions to more completely model the PW waveform.

[0039] One or more waveform-representing mathematical equations are used to approximate the shape of the PW period. Preferably, one or more skewed Gaussian distribution functions are used, although other types of equations, such as Gaussian and half-sine distribution functions, alone or in combination, can be used. The least-squares solution is initialized with basis function parameter values ​​that are reasonably close to the global minimum or most desirable solution, and appropriate minimum and maximum bounds are set to prevent unwanted local minimum solutions. A feature detection algorithm in the CDQ assessment module is used to estimate an initial, minimum, and maximum set of parameters, which are then used to locate the cycle start, systolic peak, diastolic trough, and cycle end. These feature points, combined with heuristic problem solving based on subject population surveys, can be used to initialize the initial, minimum, and maximum parameters of the skewed Gaussian to optimal values ​​found, through testing and experimentation, to produce stable and accurate pulse PW modeling results. The initial, minimum, and maximum amplitudes of the basis functions are calculated directly from the measured amplitudes of the PW period. The initial, minimum, and maximum timings and widths (sigma) of the distorted Gaussian basis functions are based on a linear-in-time subdivision of each cycle period. The skewness is preferably initialized to zero.

[0040] For each period, the shape of the measured period is compared to the shape of the mathematically approximated period, and an error value is calculated that indicates how different the shapes of the measured and approximated periods are. Periods with an error value greater than a threshold are excluded from the data set. As an example, the shape difference can be measured using least-squares analysis using the Levenberg-Marquardt (LM) damped least-squares method. This method is provided with initial parameters, minimum parameters, maximum parameters, and an error function. The error function accepts a set of parameters and calculates the sum of a distorted Gaussian basis function and an interpolated PPG baseline. This waveform is then subtracted from the measured PPG waveform to produce an error vector that is returned to the LM solution for analysis and solution iterations.

[0041] Figure 3b shows an example of a measured PW period and an approximated, or modeled, period MS overlaid on top of each other. The cumulative residual error RE for each period is shown along the dashed baseline. In this example, each PW period is modeled as a strain Gaussian distribution function. The RE curves along the baseline can be seen to contain significant error. Depending on the threshold set for tolerance, periods with errors exceeding that value can be excluded from being transferred to the CCA module. Significant and subtle differences in shape detected using a single coarse strain Gaussian function can provide value for eliminating periods containing noise or artifacts not captured in the CDQ assessment. The vertical dashed lines indicate the period start CS for each period.

[0042] FIG. 4a is a flowchart of method steps that may be performed by one embodiment of a cardiac cycle analysis (CCA) process. The CCA algorithm, in this example, provides for modeling and decomposition of fingertip PPG (photoplethysmography) of a measured period (PW period) into a sum (superposition) of multiple parameterized strained Gaussian basis functions. One or more of these basis functions are used to model each period. The strained Gaussian basis functions are defined based on known Gaussian distribution functions, which are specified by three parameters: amplitude, mean, and sigma or standard deviation. A fourth parameter, skewness, is incorporated to allow the Gaussian distribution function to exhibit asymmetry that is measurable and consistent with a statistical measure of skewness. Thus, each strained Gaussian basis function is uniquely specified by four parameters: amplitude (PW amplitude units or vertical axis), position (time units or horizontal axis), sigma (time units or width), and skewness (dimensionless). One or more strained Gaussian basis functions are used to model a single measured PPG (PW) period. The skewed Gaussian basis functions are defined from negative infinity to positive infinity, but for efficiency the range of these functions is restricted in time to just before the start of each cycle, through systole and diastole, and beyond the start of the next cycle.

[0043] PPG cycles that exhibit a reflected wave, including the commonly observed dicrotic notch formation (DN, Figures 4b and 4d), require two or three distorted Gaussian basis functions and cannot be effectively modeled with only one basis function. PPG cycles that do not exhibit a dicrotic notch or a reflected wave can be accurately modeled with only two basis functions and rarely require more than two. Generally, the first distorted Gaussian basis function (WF1) adequately models the rise in BP (systolic) from the start of the cycle to the BP peak (systolic). The first (WF1) and second (WF2) distorted Gaussian basis functions generally model the diastolic decrease in BP. The third (WF3, Figures 4f and 4h) and fourth (WF4, Figure 4h) distorted Gaussian basis functions can model the second and third reflected waves that appear in some subject PPG data. Next, when waveforms WF1, WF2, and WF3 in FIG. 4d are combined with waveforms WF1, WF2, W3, and WF4 in FIG. 4f, the resulting modeled period MS is visually indistinguishable from the measured period in FIGS. 4c and 4g, and the residual error RE is close to zero.

[0044] To promote maximum cycle modeling accuracy, basis functions modeling adjacent cycles preferably allow overlap between cycles. For example, a strongly delayed reflected wave may overlap with the main wave (or systole) of the next cycle. This is necessary to simultaneously solve groups of three cycles, accurately model the central cycle, and allow the basis functions modeling the central cycle to span two adjacent cycles. This provides greater accuracy at the expense of additional processing time.

[0045] To facilitate efficient operation of the CCA algorithm, as well as effective testing and validation, it is preferable to operate on segmented time series data rather than on a single, continuous time series data of unlimited length. Because PPG periods often span segment boundaries, the CCA algorithm processes the current period along with the preceding and succeeding periods so that PPG periods that span period edges can be treated as complete periods to provide a seamless analysis without missing periods or samples. The preceding segment is provided back to the algorithm using a segment lag scheme to ensure that the current segment is complete and seamless, allowing the preceding segment data to be reused.

[0046] An optimization function such as lsqnonlin can be used to implement a least-squares method, such as the Levenberg-Marquardt (LM) damped least-squares method, to solve for an error value for comparing the shape of each measured and modeled (approximated) period. The method is provided with initial parameters, minimum parameters, maximum parameters, and an error function. The error function accepts a set of parameters and calculates the sum of a distorted Gaussian basis function and an interpolated PPG baseline. This waveform is then subtracted from the measured PPG waveform to produce an error vector that is fed back to the LM solution for further analysis and solution iterations.

[0047] After optimization is complete, the first distorted Gaussian basis function for each PPG cycle can then be calculated independently of the other basis functions (reflections) and other PPG cycles. This results in a pure distorted Gaussian function that can be normalized and analyzed to accurately determine the 1% maximum amplitude point at which a PPG cycle begins. The simplicity and noise-free nature of a single distorted Gaussian function allows this point to be determined with high precision and subsample accuracy. Accurate determination of the cycle start for each cycle is critical to the accuracy of the HCS model's calculation of cardiovascular parameter values.

[0048] FIG. 5 provides an example of how cardiac cycle analysis methods can be incorporated into a cardiac cycle analysis system. Data from a plethysmography device, such as a PO, is processed to remove cycles that do not meet set criteria, and the remaining cycles are used as input for the HCS model (solid arrow). The data may represent measured waveform cycles in the form of data points measured by the plethysmography device. Additionally or alternatively, the data may include a mathematical function that models the waveform of the measured cycle, also referred to as a modeling cycle. This may be particularly useful in the presence of high-frequency noise. Additional data may optionally be made available to the HCS model, for example, from a blood pressure monitor or electrocardiogram. The optional additional data may be used by the HCS model as input for cardiovascular parameters.

[0049] A system for noninvasively measuring a patient's cardiovascular parameters may include a pulse oximeter, a data collection, storage, and transmission device, a computer, and a means for reporting the value of the cardiovascular parameter to a user. The data collection, storage, and transmission device may be, for example, a laptop computer or tablet. The pulse oximeter is configured to communicate with the data collection, storage, and transmission device, collect plethysmography waveform data from the patient, and transmit the data to the data collection, storage, and transmission device. The data collection and transmission device receives the waveform data and transmits it to the computer. The computer may be the same device as the data collection, storage, and transmission device, but the computer may require more computing power and may be a remote cloud computing device. The computer includes a program of instructions that, when executed by the computer, causes the computer to process the waveform data as described above before providing the resulting output data to a computational model of the cardiovascular system configured to calculate the value of the cardiovascular parameter. The system further includes a blood pressure monitor that transmits blood pressure data from the subject to the data collection, storage, and transmission device so that the data can be used as input to the computational model of the cardiovascular system. Data can be collected and transmitted continuously or at intervals such as 1, 2, or 5 minutes. Other non-invasive monitoring devices can also be included to provide input data to the HCS model, such as electrocardiogram data.

[0050] The described methods and systems are not limited to physiological systems. For example, they may be applied to other non-physiologically compatible vascular systems that include positive displacement pumps with pulsatile flow and vascular parameters similar to those of the cardiovascular system of a human or non-human mammal. While the present invention has been described herein with reference to certain preferred embodiments, those skilled in the art will readily appreciate that other applications may be substituted for those described herein without departing from the spirit and scope of the present invention.

Claims

1. 1. A computer-implemented method for processing pulse plethysmography data, the data comprising a series of measurement periods, the method comprising: using a slope threshold crossing technique to detect peaks, valleys, and peak amplitudes for each measurement period of the series of measurement periods; applying a quality metrics process to the series of measurement periods to identify measurement periods that do not meet selected quality criteria; removing said cycles that do not meet selected quality criteria by ablation; splicing together the non-removed measurement periods to provide output data comprising a series of processed measurement periods; applying said quality metrics process, comparing the duration and amplitude of each measurement period with the peak amplitude and duration of an adjacent measurement period, or with the average peak amplitude and average duration of multiple measurement periods; and eliminating measurement periods having durations or amplitudes outside a tolerance range of discrepancy relative to the peak amplitudes and durations of the adjacent measurement periods or the average peak amplitudes and average durations of the plurality of measurement periods.

2. Applying at least one of the quality metrics processes comprises: comparing the shape of each measurement period to the shape of an adjacent measurement period or to an average shape of adjacent measurement periods; and removing measurement periods having shapes outside a tolerance range of shape mismatch with the adjacent measurement periods or an average of the adjacent measurement periods.

3. Applying at least one of the quality metrics processes comprises: modeling each measurement period as one or more mathematical functions to provide a series of modeled periods; comparing the shape of each measured period with the shape of a corresponding modeled period to determine a shape mismatch with the corresponding modeled period; and removing measured periods having shapes outside a tolerance range of the shape mismatch with the corresponding modeled periods.

4. 4. The method of claim 3, wherein the one or more mathematical functions comprise one or more of a half-sine function, a square wave function, a Gaussian function, a distorted Gaussian function, a Bessel function, a Fourier series, a Hankel function, a spherical Bessel function, a Struve function, and a Weber function.

5. The method of claim 3 or 4, wherein each of the one or more mathematical functions includes up to 10 parameters.

6. The method of claim 4 , wherein the one or more mathematical functions include at least two skewed Gaussian functions.

7. and further comprising calculating a value of the cardiovascular parameter, said calculating comprising: providing the output data, including the series of processed measurement periods, to a computational model of the cardiovascular system configured to calculate the value for the cardiovascular parameter; The method of claim 1 , further comprising causing the computational model to calculate and provide an output reporting the value for the cardiovascular parameter.

8. eliminating modeling periods corresponding to the removed measurement periods to provide a series of processed modeling periods; providing the series of processed modeled periods to a computational model of the cardiovascular system configured to calculate values ​​of cardiovascular parameters; The method of claim 3 , further comprising causing the computational model to calculate and provide an output reporting the value for the cardiovascular parameter.

9. The cardiovascular parameters include stroke volume, cardiac output, aortic pressure, inertance, compliance, resistance, 9. The method according to claim 7 or 8, wherein the measured value is at least one of arterial pressure and venous pressure.

10. The method of claim 7 or 8, wherein the computational model of the cardiovascular system is integrated into a dynamic state space model.

11. 1. A system for non-invasively measuring a cardiovascular parameter of a patient, comprising: a pulse oximeter, a data collection, storage and transmission device, a computer, and means for reporting the value of said cardiovascular parameter to a user; the pulse oximeter in communication with the data collection, storage, and transmission device; the pulse oximeter is configured to collect plethysmographic waveform data of the patient, the waveform data including a series of measurement periods; and transmit the plethysmographic waveform data to the data collection, storage, and transmission device; the data collection and transmission device is configured to receive the plethysmography waveform data and transmit the plethysmography waveform to the computer; The computer includes a program of instructions, which, when executed by the computer, causes the computer to: using a slope threshold crossing technique to detect peaks, valleys, and peak amplitudes for each measurement period of the series of measurement periods; applying a quality metrics process to the series of measurement periods to identify measurement periods that do not meet selected quality criteria, wherein applying the quality metrics process comparing the duration and amplitude of each measurement period with the peak amplitude and duration of an adjacent measurement period, or with the average peak amplitude and average duration of multiple measurement periods; removing measurement periods having durations or amplitudes outside a tolerance range of discrepancy relative to the peak amplitudes and durations of the adjacent measurement periods or the average peak amplitudes and average durations of the plurality of measurement periods; removing said cycles that do not meet selected quality criteria by ablation; splicing together the non-removed measurement periods to provide output data comprising a series of processed measurement periods and / or a series of processed modeling periods corresponding to the series of processed measurement periods; providing the output data to a computational model of the cardiovascular system configured to calculate the value for the cardiovascular parameter; causing the computational model to calculate and provide an output reporting the value for the cardiovascular parameter.

12. further comprising a blood pressure monitor; the blood pressure monitor is configured to communicate with the data collection, storage, and transmission device, collect blood pressure data of the patient, and transmit the blood pressure data to the data collection, storage, and transmission device; The system of claim 11 , wherein the computer includes software that calculates measurements for the cardiovascular parameters from the blood pressure data and the output data and reports the measurements for the cardiovascular parameters.

13. A program including instructions, which when executed by a computer: receiving pulse plethysmography data including a series of measurement periods; using a slope threshold crossing technique to detect peaks, valleys, and peak amplitudes for each measurement period of the series of measurement periods; applying a quality metrics process to the series of measurement periods to identify measurement periods that do not meet selected quality criteria, wherein applying the quality metrics process comparing the duration and amplitude of each measurement period with the peak amplitude and duration of an adjacent measurement period, or with the average peak amplitude and average duration of multiple measurement periods; removing measurement periods having durations or amplitudes outside a tolerance range of discrepancy relative to the peak amplitudes and durations of the adjacent measurement periods or the average peak amplitudes and average durations of the plurality of measurement periods; removing said cycles that do not meet selected quality criteria by ablation; splicing together the non-removed measurement periods to provide output data comprising a series of processed measurement periods and / or a series of processed modeling periods corresponding to the series of processed measurement periods; providing the output data to a computational model of the cardiovascular system configured to calculate a value of a cardiovascular parameter; and calculating and providing an output reporting the value for the cardiovascular parameter.

Images