METHOD AND SYSTEM TO ASSESS DISEASE USING DYNAMICAL ANALYSIS OF CARDIAC AND PHOTOPLETHYSMOGRAPHIC SIGNALS
20220304585 · 2022-09-29
Inventors
- Mehdi Paak (Toronto, CA)
- Timothy William Fawcett Burton (Toronto, CA)
- Shyamlal Ramchandani (Kingston, CA)
Cpc classification
A61B5/7239
HUMAN NECESSITIES
Y02A90/10
GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
A61B5/7246
HUMAN NECESSITIES
A61B5/7264
HUMAN NECESSITIES
A61B5/318
HUMAN NECESSITIES
A61B5/02416
HUMAN NECESSITIES
A61B5/349
HUMAN NECESSITIES
A61B5/7275
HUMAN NECESSITIES
A61B5/0205
HUMAN NECESSITIES
A61B5/0245
HUMAN NECESSITIES
International classification
A61B5/00
HUMAN NECESSITIES
A61B5/0205
HUMAN NECESSITIES
A61B5/0245
HUMAN NECESSITIES
A61B5/1455
HUMAN NECESSITIES
A61B5/318
HUMAN NECESSITIES
Abstract
The exemplified methods and systems facilitate one or more dynamical analyses that can characterize and identify synchronicity between acquired cardiac signals and photoplethysmographic signals to predict/estimate presence, non-presence, localization, and/or severity of abnormal cardiovascular conditions or disease, including, for example, but not limited to, coronary artery disease, heart failure (including but not limited to indicators of disease or conduction such as abnormal left ventricular end-diastolic pressure disease), and pulmonary hypertension, among others. In some embodiments, statistical properties of the synchronicity between cardiac signals and photoplethysmographic signals are evaluated. In some embodiments, statistical properties of histogram of synchronicity between cardiac signals and photoplethysmographic signals are evaluated. In some embodiments, statistical and/or geometric properties of Poincaré map of synchronicity between cardiac signals and photoplethysmographic signals are evaluated.
Claims
1. A method for non-invasively assessing a disease state or abnormal condition of a subject, the method comprising: obtaining, by one or more processors, a first biophysical signal data set of a subject associated with saturation of oxygenated or deoxygenated hemoglobin, including a red photoplethysmographic signal and an infrared photoplethysmographic signal; obtaining, by the one or more processors, a second biophysical signal data set of the subject associated with a cardiac signal; determining, by the one or more processors, one or more synchronicity dynamical properties between the first biophysical signal data set associated with the saturation of oxygenated and/or deoxygenated hemoglobin and the second biophysical signal data set associated with the cardiac signal; and determining, by the one or more processors, an estimated value for presence of a disease state based on the determined one or more synchronicity dynamical properties.
2. The method of claim 1, wherein the disease or condition can be diagnosed based on assessed indication and/or estimate of presence, non-presence, and/or severity of elevated or abnormal left ventricular end-diastolic pressure (LVEDP).
3. The method of claim 1, wherein the disease state or condition is selected from the group consisting of coronary artery disease, pulmonary hypertension, pulmonary arterial hypertension, pulmonary hypertension due to left heart disease, rare disorders that lead to pulmonary hypertension, left ventricular heart failure or left-sided heart failure, right ventricular heart failure or right-sided heart failure, systolic heart failure, diastolic heart failure, ischemic heart disease, and arrhythmia.
4. The method of claim 1, wherein the synchronicity dynamical property of the first and second biophysical signal data sets comprises a statistical assessment of values of the cardiac signal at a landmark defined by both the red photoplethysmographic signal and the infrared photoplethysmographic signal.
5. The method of claim 4, wherein the landmark defined by both the red photoplethysmographic signal and the infrared photoplethysmographic signal is defined at a time where the values of the red photoplethysmographic signal and the infrared photoplethysmographic signal intersects.
6. The method of claim 1, wherein the synchronicity dynamical property of the first and second biophysical signal data sets comprises a statistical assessment of values of one of the red photoplethysmographic signal or the infrared photoplethysmographic signal at a landmark defined in the cardiac signal.
7. The method of claim 6, wherein the landmark defined in the cardiac signal includes an associated peak associated with ventricular depolarization.
8. The method of claim 4, wherein the landmark defined in the cardiac signal includes an associated peak associated with ventricular repolarization or atrial depolarization.
9. The method of claim 1, wherein the synchronicity dynamical property of the first and second biophysical signal data sets comprises a statistical assessment of time intervals between i) a first set of landmarks defined between the red photoplethysmographic signal and the infrared photoplethysmographic signal and ii) a second set of landmarks defined in the cardiac signal.
10. The method of claim 9, wherein the second set of landmarks defined in the cardiac signal includes associated peaks in the cardiac signal associated with ventricular depolarization.
11. The method of claim 9, wherein the second set of landmarks defined in the cardiac signal includes associated peaks in the cardiac signal associated with ventricular repolarization or atrial depolarization.
12. The method of claim 9, wherein the first set of landmarks defined by both the red photoplethysmographic signal and the infrared photoplethysmographic signal are defined at times where the values of the red photoplethysmographic signal and the infrared photoplethysmographic signal intersect.
13. The method of claim 1, wherein the synchronicity dynamical property of the first and second biophysical signal data sets comprises a statistical assessment of phase relations between periods of one of the red or infrared photoplethysmographic signals and periods of the cardiac signal.
14. The method of claim 1 further comprising: causing, by the one or more processors, generation of a visualization of the estimated value for the presence of the disease state, wherein the generated visualization is rendered and displayed at a display of a computing device and/or presented in a report.
15. The method of claim 1 further comprising: determining, by the one or more processors, a histogram of the synchronicity of the first and second biophysical signal data sets; and extracting a first statistical parameter of the histogram, wherein the first statistical parameter of the histogram is selected from the group consisting of mean, mode, median, skew, kurtosis, wherein the extracted first statistical parameter is used in the determining of the estimated value for the presence of the disease state.
16. The method of claim 1 further comprising: determining, by the one or more processors, a Poincaré map of the synchronicity of the first and second biophysical signal data sets; and extracting a second statistical parameter of the Poincaré map, wherein the second statistical parameter of the histogram is selected from the group consisting of mean, mode, median, skew, kurtosis, wherein the extracted second statistical parameter is used in the determining of the estimated value for the presence of the disease state.
17. The method of claim 1, further comprising: determining, by the one or more processors, a Poincaré map of the synchronicity of the first and second biophysical signal data sets; and extracting a geometric property of an eclipse fitted to a cluster in the Poincaré map, wherein the extracted geometric property of the eclipse is used in the determining of the estimated value for the presence of the disease state.
18. The method of claim 17, wherein the Poincaré map is generated by iteratively plotting in an x-axis a parameter associated with the synchronicity of the first and second biophysical signal data sets at a first index x−1 and a second index x and in a y-axis the parameter at the second index x and a third index x+1.
19. The method of claim 18, wherein the parameter is a time interval between a landmark of a cardiac signal and a crossover between the red and infrared photo-photoplethysmographic signals.
20. The method of claim 17, wherein the parameter is an amplitude signal value of a cardiac signal at a crossover landmark defined between the red and infrared photo-photoplethysmographic signals.
21-43. (canceled)
Description
BRIEF DESCRIPTION OF THE DRAWINGS
[0082] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments and together with the description, serve to explain the principles of the methods and systems.
[0083] Embodiments of the present invention may be better understood from the following detailed description when read in conjunction with the accompanying drawings. Such embodiments, which are for illustrative purposes only, depict novel and non-obvious aspects of the invention. The drawings include the following figures:
[0084]
[0085]
[0086]
[0087]
[0088]
[0089]
[0090]
[0091]
[0092]
[0093]
[0094]
[0095]
[0096]
[0097]
[0098]
[0099]
[0100]
[0101]
[0102]
[0103]
[0104]
[0105]
[0106]
[0107]
[0108]
[0109]
[0110]
[0111]
[0112]
[0113]
[0114]
[0115]
[0116]
[0117]
DETAILED SPECIFICATION
[0118] Each and every feature described herein, and each and ever combination of two or more of such features, is included within the scope of the present invention provided that the features included in such a combination are not mutually inconsistent.
[0119] While the present disclosure is directed to the beneficial assessment of biophysical signals. e.g., raw or pre-processed photoplethysmographic signals, cardiac signals, etc., in the diagnosis and treatment of cardiac-related pathologies and conditions, such assessment can be applied to the diagnosis and treatment (including, surgical, minimally invasive, and/or pharmacologic treatment) of any pathologies or conditions in which a biophysical signal is involved in any relevant system of a living body. In the cardiac (or cardiovascular) context, the assessment can be applied to the diagnosis and treatment of coronary artery disease (CAD) and diseases and/or conditions associated with an abnormal left ventricular end-diastolic pressure (LVEDP). The assessment can be applied for the diagnosis and treatment of any number of therapies, alone or in combination, such as the placement of a stent in a coronary artery, performance of an atherectomy, angioplasty, prescription of drug therapy, and/or the prescription of exercise, nutritional and other lifestyle changes, etc. Other cardiac-related pathologies or conditions that may be diagnosed include, e.g., arrhythmia, congestive heart failure, valve failure, pulmonary hypertension (e.g., pulmonary arterial hypertension, pulmonary hypertension due to left heart disease, pulmonary hypertension due to lung disease, pulmonary hypertension due to chronic blood clots, and pulmonary hypertension due to other disease such as blood or other disorders), as well as other cardiac-related pathologies, conditions and/or diseases. In some embodiments, the assessment may be applied to neurological-related pathologies and conditions. Non-limiting examples of neurological-related diseases, pathologies or conditions that may be diagnosed include. e.g., epilepsy, schizophrenia. Parkinson's Disease. Alzheimer's Disease (and all other forms of dementia), autism spectrum (including Asperger syndrome), attention deficit hyperactivity disorder, Huntington's Disease, muscular dystrophy, depression, bipolar disorder, brain/spinal cord tumors (malignant and benign), movement disorders, cognitive impairment, speech impairment, various psychoses, brain/spinal cord/nerve injury, chronic traumatic encephalopathy, cluster headaches, migraine headaches, neuropathy (in its various forms, including peripheral neuropathy), phantom limb/pain, chronic fatigue syndrome, acute and/or chronic pain (including back pain, failed back surgery syndrome, etc.), dyskinesia, anxiety disorders, conditions caused by infections or foreign agents (e.g., Lyme disease, encephalitis, rabies), narcolepsy and other sleep disorders, post-traumatic stress disorder, neurological conditions/effects related to stroke, aneurysms, hemorrhagic injury, etc.; tinnitus and other hearing-related diseases conditions and vision-related diseases conditions
[0120] Some references, which may include various patents, patent applications, and publications, are cited in a reference list and discussed in the disclosure provided herein. The citation and/or discussion of such references is provided merely to clarify the description of the disclosed technology and is not an admission that any such reference is “prior an” to any aspects of the disclosed technology described herein. In terms of notation. “|n|” corresponds to the nth reference in the list. For example, |36| refers to the 36th reference in the list, namely F. Pedreuosa. G. Varoquaux, A. Gramfort, V. Michel. B. Thirion, O. Grisel. M. Blondel. P. Prettenhofer. R. Weiss. V. Dubourg, et al., “Scikit-learn: Machine learning in python.” Journal of machine learning research 12, 2825-2831) (October 2011). All references cited and discussed in this specification are incorporated herein by reference in their entireties and to the same extent as if each reference was individually incorporated by reference.
[0121] Example System
[0122]
[0123] As noted herein. “physiological systems” can refer to the cardiovascular system, the pulmonary system, the renal system, the nervous system, and other functional systems and sub-systems of the body. In the context of the cardiovascular system, system 100 facilitates the investigation of complex, nonlinear dynamical properties of the heart over many heart cycles.
[0124] In
[0125] The first type is acquired via probes 106a. 106b from the subject at location 108a (e.g., a finger of the subject) to generate a raw photoplethysmographic signal data set 110a from photoplethysmographic signal(s) 104a. In some embodiments, the raw photoplethysmographic signal data set 110a includes one or more photoplethysmographic signals) associated with measured changes in light absorption of oxygenated and/or deoxygenated hemoglobin.
[0126] The second type is acquired via probes 124a-124f from subject 108 to generate a cardiac signal data set 110b from cardiac signals 104b. In some embodiments, cardiac signal data set 110b includes data associated with biopotential signals acquired across a plurality of channels. In some embodiments, cardiac signal data set 110b includes w ide-band biopotential signals. e.g., acquired via a phase-space recorder, as described in U.S. Patent Publication No. 2017.0119272, entitled “Method and Apparatus for Wide-Band Phase Gradient Signal Acquisition.” which is incorporated by reference herein in its entirety. In some embodiments, the cardiac signal data set includes bipolar wide-band biopotential signals. e.g., acquired via a phase-space recorder, as described in U.S. Patent Publication No. 2018/0249960, entitled “Method and Apparatus for Wide-Band Phase Gradient Signal Acquisition.” Which is incorporated by reference herein in its entirety. In other embodiments, the cardiac signal data set HO includes one or more biopotential signals acquired from conventional electrocardiogram (ECG/EKG) equipment (e.g., Holter device, 12 lead ECG, etc.).
[0127] Example Photoplethysmographic Signals
[0128]
[0129]
[0130] Photoplethysmography is used to optically measure variations of the volume of blood perfusing tissue (e.g., cutaneous, subcutaneous, cartilage) into which light is emitted, typically at a specific wavelength, from a LED or other source. The intensity of this light after passing through the tissue (e.g., fingertip, earlobe, etc.) is then registered by a photodetector to provide the photoplethysmographic signals. The amount of light absorbed depends on the volume of the blood perfusing the interrogated tissue. The variation in light absorbed is observable in the photoplethysmographic signal and can provide valuable information with regard to cardiac activity, pulmonary function, their interactions, and other physiological system functions |13|.
[0131] In some embodiments, measurement system 102 comprises custom or dedicated equipment or circuitry (including off-the-shelf devices) that are configured to acquire such signal waveforms for the purpose diagnosing disease or abnormal conditions. In other embodiments, measurement system 102 comprises pulse oximeter or optical photoplethysmographic device that can output acquired raw signals for analysis. Indeed, in some embodiments, the acquired waveform 104 may be analyzed to calculate the level of oxygen saturation of the blood shown in
[0132]
[0133] Photoplethysmographic signal(s) 104 may be considered measurements of the state of a dynamical system in the body, similar to cardiac signals. The behavior of the dynamical system may be influenced by the actions of the cardiac and respiratory systems. It is postulated that any system aberrations (due. e.g., to a disease or abnormal condition) may manifest itself in the dynamics of photoplethysmographic signal(s) 104 via some interaction mechanism or mechanisms.
[0134] In some embodiments, the acquired photoplethysmographic signal(s) 104 are down-sampled to 250 Hz. Other frequency ranges may be used. In some embodiments, the acquired photoplethysmographic signal(s) 104 are processed to remove baseline wander and/or to filter for noise and/or mains frequencies.
[0135] The acquired photoplethysmographic signal(s) 104 may be embedded in some higher dimensional space (e.g., phase space embedding) to reconstruct the manifold (phase space) the underlying dynamical system creates. An example three-dimensional visualization and its two-dimensional projection of acquired photoplethysmographic signal(s) 104 (shown as 104c) are shown in
[0136] Example Cardiac Signals
[0137] Electrocardiographic signals measure the action potentials of cardiac tissue (i.e., cardiomyocytes). There are various configurations of leads that can be used in a mammalian body, and in particular humans, to obtain these signals in the context of the present disclosure. In an example configuration, seven leads are used. This con figuration results in three orthogonal channels signals: e.g., X, Y and Z, corresponding to the coronal, sagittal and transverse planes, respectively.
[0138] As discussed above, in some embodiments, cardiac signal data set 110 includes data associated with biopotential signals acquired across a plurality of channels. In some embodiments, cardiac signal data set 110b includes wide-band biopotential signals, e.g., signals acquired via a phase-space recorder such as described in U.S. Patent Publication No. 2017/0119272, entitled “Method and Apparatus for Wide-Band Phase Gradient Signal Acquisition.” which is incorporated by reference herein in its entirety. In some embodiments, the cardiac signal data set includes bipolar wide-band biopotential signals. e.g., acquired via a phase-space recorder such as described in U.S. Patent Publication No. 2018/0249960, entitled “Method and Apparatus for Wide-Band Phase Gradient Signal Acquisition.” which is incorporated by reference herein in its entirety. In other embodiments, the cardiac signal data set Mb includes one or more biopotential signals acquired from conventional electrocardiogram (ECG/EKG) equipment (e.g., Holter device, 12 lead ECG, etc.).
[0139] The phase space recorder as described in 2017/0119272, in some embodiments, is configured to concurrently acquire photoplethysmographic signals 104a along with cardiac signal 104b. Thus, in some embodiments, measurement system 102b is configured to acquire mo types of biophysical signals.
[0140]
[0141]
[0142] Referring still to
[0143] In the cardiac and/or electrocardiography contexts, measurement system 102 is configured to capture cardiac-related biopotential or electrophysiological signals of a mammalian subject (such as a human) as a biopotential cardiac signal data set. In some embodiments, measurement system 102 is configured to acquire a wide-band cardiac phase gradient signals as a biopotential signal, a current signal, an impedance signal, a magnetic signal, an ultrasound or acoustic signal, etc. The term “wide-band” in reference to an acquired signal, and its corresponding data set, refers to the signal having a frequency range that is substantially greater than the Nyquist sampling rate of the highest dominant frequency of a physiological system of interest. For cardiac signals, which typically has a dominant frequency components between about 0.5 Hz and about 80 Hz, the wide-band cardiac phase gradient signals or wide-hand cardiac biophysical signals comprise cardiac frequency information at a frequency selected from the group consisting between about 0.1 Hz and 1 KHz, between about 0.1 Hz and about 2 KHz, between about 0.1 Hz and about 3 KHz, between about 0.1 Hz and about 4 KHz, between about 0.1 Hz and about 5 KHz between about 0.1 Hz and about 6 KHz, between about 0.1 Hz and about 7 KHz, between about 0.1 Hi and about 8 KHz, between about 0.1 Hz and about 9 KHz, between about 0.1 Hz and about 10 KHz, and between about 0.1 Hz and greater than 10 KHz (e.g., 0.1 Hz to 50 KHz or 0.1 Hz to 500 KHz). In addition to capturing the dominant frequency components, the wide-hand acquisition also facilitates capture of other frequencies of interest. Examples of such frequencies of interest can include QRS frequency profiles (which can have frequency ranges up to 250 Hi), among others. The term “phase gradient” in reference to an acquired signal, and corresponding data set, refers to the signal being acquired at different vantage points of the body to observe phase information for a set of distinct events functions of the physiological system of interest. Following the signal acquisition, the term “phase gradient” refers to the preservation of phase information via use of non-distorting signal processing and pre-processing hardware, software, and techniques (e.g., phase-linear filters and signal-processing operators and/or algorithms).
[0144] In some embodiments, cardiac signal data set 110b includes wide-band biopotential signals, such as, e.g., those acquired via a phase-space recorder as described in U.S. Patent Publication No. 2017/0119272, entitled “Method and Apparatus for Wide-Band Phase Gradient Signal Acquisition.” which is incorporated by reference herein in its entirety. In some embodiments, the cardiac signal data set includes bipolar wide-hand biopotential signals. e.g., acquired via a phase-space recorder such as described in U.S. Patent Publication No. 2018/0249960, entitled “Method and Apparatus for Wide-Band Phase Gradient Signal Acquisition.” which is incorporated by reference herein in its entirety. In other embodiments, the cardiac signal data set 110b includes one or more biopotential signals acquired from conventional electrocardiogram (ECG/EKG) equipment (e.g., Holter device, 12 lead ECG, etc.).
[0145] The phase space recorder as described in U.S. Patent Publication No. 2017/0119272, in some embodiments, is configured to concurrently acquire photoplethysmographic signals 104a along with cardiac signal 104b. Thus, in some embodiments, measurement system 102b is configured to acquire two types of biophysical signals.
[0146] In the neurological context, measurement system 102 is configured to capture neurological-related biopotential or electrophysiological signals of a mammalian subject (such as a human) as a neurological biophysical-signal data set. In some embodiments, \ measurement system 102 is configured to acquire wide-hand neurological phase gradient signals as a biopotential signal, a current signal, an impedance signal, a magnetic signal, an ultrasound or acoustic signal, an optical signal, etc. An example of measurement system 102 is described in U.S. Patent Publication No. 2017/0119272 and in U.S. Patent Publication No. 2018/0249960, which is incorporated by reference herein in its entirety.
[0147] In some embodiments, the measurement system 102 is configured to capture wide-band biopotential biophysical phase gradient signals as unfiltered mammalian electrophysiological signals such that the spectral component(s) of the signals are not altered. Indeed, in such embodiments, the wide-band biopotential biophysical phase gradient signals are captured, converted, and even analyzed without having been filtered (via, e.g., hardware circuitry and/or digital signal processing techniques, etc.) (e.g., prior to digitization) that otherwise can affect the phase linearity of the biophysical signal of interest. In some embodiments, the wide-band biopotential biophysical phase gradient signals are captured in micros oft or sub-microvolt resolutions that are at, below, or significantly below, the noise floor of conventional electrocardiographic, electroencephalographic, and other biophysical-signal acquisition instruments. In some embodiments, the wide-band biopotential biophysical signals are simultaneously sampled having a temporal skew or “lag” of less than about 1 microsecond, and in other embodiments, having a temporal skew or lag of not more than about 10 femtoseconds. Notably, the exemplified embodiments minimize non-linear distortions (e.g., those that can be introduced via certain filters) in the acquired NN ide-band phase gradient signal to not affect the information therein.
[0148]
[0149] In the example configuration shown in
[0150] Referring to
[0151] Referring still to
[0152] In some embodiments, analytic engine 114 includes a machine learning module 116 configured to assess a set of features determined via one or more feature extraction modules (e.g. 118, 120) from the acquired biophysical signal(s) to determine features of clinical significance. Once the features have been extracted from the PPG signal(s) or cardiac signal(s), then any type of machine learning can be used. Examples of embodiments of machine learning module 116 is configured to implement, but not limited to, decision trees, random forests. SVMs, neural networks, linear models. Gaussian processes, nearest neighbor. SVMs. Naïve Bayes. In some embodiment, machine learning module 116 may be implemented. e.g., as described in U.S. patent application Ser. No. 15/653,433, entitled “Discovering Novel Features to Use in Machine Learning Techniques, such as Machine Learning Techniques for Diagnosing Medical Conditions”: and U.S. patent application Ser. No. 15/653,431, entitled “Discovering Genomes to Use in Machine Learning Techniques”: each of which are incorporated by reference herein in its entirety. The photoplethysmographic signal(s) may be combined with other acquired photoplethysmographic signal(s) to be used in a training data set or validation data set for the machine learning module 116 in the evaluation of a set of assessed dynamical features. The photoplethysmographic signal(s) may have an associated label 122 for a given disease state or abnormal condition, or an indicator of one. If determined to be of clinical significance, an assessed dynamical features (e.g., from 118 or 120) may be subsequently used as a predictor for the given disease or abnormal condition, or an indicator of one.
[0153] In some embodiments, analytic engine 114 includes a pre-processing module. e.g., configured to normalize and/or remove baseline wander from the acquired photoplethysmographic signal(s).
[0154] In some embodiments, system 101) includes a healthcare provider portal to display. e.g., in a report, score or various outputs of the analytic engine 114 in predicting and/or estimating presence, non-presence, severity, and/or localization (where applicable) of a disease or abnormal condition, or an indicator of one. The physician or clinician portal, in some embodiments, is configured to access and retrieve reports from a repository (e.g., a storage area network). The physician or clinician portal and/or repository can be compliant with various privacy laws and regulations such as the U.S. Health Insurance Portability and Accountability act of 19% (HIPAA). Further description of an example healthcare provider portal is provided in U.S. Pat. No. 10,292,596, entitled “Method and System for Visualization of Heart Tissue at Risk”, which is incorporated by reference herein in its entirety. Although in certain embodiments, the portal is configured for presentation of patient medical information to healthcare professionals, in other embodiments, the healthcare provider portal can be made accessible to patients, researchers, academics, and/or other portal users.
[0155] Synchronicity Evaluation Between Cardiac Signal and Raw Photoplethysmographic Signals
[0156] Referring still to
[0157] The electrophysiological activity of the heart is a nonlinear process which in conjunction with the myocytes mechano-electrical feedback produces very complex nonlinear responses |26|. These behaviors whether normal (reaction to extrinsic conditions) or due to a disease can be studied and characterized using nonlinear statistics related to the nonlinear dynamics and chaoticity of the heart. Synchronicity features that are based on dynamics observed in cardiac and photoplethysmographic signals may encode the health state of the heart and are used to train a machine learning model for prediction for various disease state of condition.
[0158] In a Poincarè map, the mapping X.sup.n+1=P(X.sub.n) may be defined using triggers (e.g., intersection with Σ), and the set of Poincaré points {X.sub.0, X.sub.1, . . . , X.sub.n} can then be analyzed geometrically and/or statistically to deduce more information about the system.
Synchronicity Features Example PM #1
[0159]
[0160] Specifically,
[0161]
[0162] In some embodiments, dynamical feature extraction module 118 is configured to generate a histogram (e.g., as generated per
I(X,Y)=Σ.sub.y∈YΣ.sub.x∈XP(x,y)log(p(x,y)/p(x)p(y)) (Equation 1)
[0163] In Equation 5, p( . . . ) is the probability distribution over the specified variables.
[0164]
[0165] In some embodiments, to generate the Poincaré map 512, the system plots generates a 2D pairs of points (e.g., (x.sub.1, x.sub.2), (x.sub.2, x.sub.3), etc.) against the points (e.g., (x.sub.0, x.sub.1). (x.sub.1, x.sub.2), etc.) of the amplitude values of a cardiac signal at the cross-over landmark points formed between photoplethysmographic signals.
[0166] In some embodiments, dynamical feature extraction module 118 is configured to generate Poincaré map 512. Following generation of Poincaré map 512, dynamical feature extraction module 118, in some embodiments, is configured to generate a geometric object from the map data. In
[0167] The dynamical feature extraction module 118, in some embodiments, may extract other parameters such as void area, surface area, porosity, perimeter length, density, among others.
[0168] Indeed, synchronicity between acquired photoplethysmographic signals (e.g., NN here acquired raw signals are merely processed to remove baseline wander and high frequency noise) and a cardiac signal based on triggers defined in the photoplethysmographic signal may be used to assess for the presence, non-presence, severity, and/or localization (where applicable) of coronary artery disease (CAD), pulmonary hypertension, heart failure in Various forms, among other diseases and conditions. In the CAD context.
[0169]
[0170] Specifically, in
[0171] In contrast, as shown in
[0172] In
[0173]
[0174] Statistical properties of these distributions (e.g., mean, median, deviation, kurtosis etc.) and the geometrical properties of the encompassing ellipse (e.g., major and minor diameters and tilt) may be computed and used as features.
[0175] Table 1 provides a list of example synchronicity feature extracted parameters associated with Poincaré map analysis PM #1 as their corresponding description.
TABLE-US-00001 TABLE 1 Parameter name Description dXDmj Major diameter of ellipse from Poincaré map PM#1 for the PSR/ECG “X” channel dXDmn Minor diameter of ellipse from Poincaré map PM#1 for the PSR/ECG “X” channel. dZDmn Minor diameter of ellipse from Poincaré map analysis PM#1 on the PSR/ECG “Z” channel. dYAlpha Tilt angle, alpha, of the ellipse from Poincaré map analysis PM#1 on the PSR/ECG “Y” channel. dZAIpha Tilt angle, alpha, of ellipse from Poincaré map analysis PM#1 on the PSR/ECG “Z” channel dXMean1 Amplitude mean of the PSR/ECG “X” channel at the first intersection crossover points of photoplethysmographic signals (e.g., in Poincaré map analysis PM#1). dXStd1 Standard deviation of the distribution of the PSR/ECG “X” channel triggered by the first crossover landmarks of photoplethysmographic signals (e.g., in Poincaré map analysis PM#1) dXStd2 Standard deviation of the distribution of the PSR/ECG “X” channel triggered by second crossover landmarks of photoplethysmographic signals (e.g., in Poincaré map analysis PM#1). dYStd2 Standard deviation of distribution of the PSR/ECG “Y” channel data triggered at second crossover landmarks of photoplethysmographic signals (e.g., in Poincaré map analysis PM#1) dYKurt2 Kurtosis of distribution of the PSR/ECG “Y” channel data triggered at second crossover landmarks of photoplethysmographic signals (e.g., in Poincaré map analysis PM#1). dZKurt2 Kurtosis of distribution of the PSR/ECG “Y” channel data triggered at second crossover landmarks of photoplethysmographic signals (e.g., in Poincaré map analysis PM#1). dYMode2 Mode of distribution of the PSR/ECG “Y” channel data triggered at second crossover landmarks of photoplethysmographic signals (e.g., in Poincaré map analysis PM#1). dZMode2 Mode of distribution of the PSR/ECG Z″ channel data triggered at the second crossover landmarks of photoplethysmographic signals (e.g., in Poincaré map analysis PM#1). dZSkew1 Kurtosis of distribution of the PSR/ECG “Z” channel data triggered at first crossovers of photoplethysmographic signals (e.g., in Poincaré map analysis PM#1) dZSkew2 Kurtosis of distribution of the PSR/ECG “Z” channel data triggered at second crossovers of photoplethysmographic signals (e.g., in Poincaré map analysis PM#1). dYRelStdMAD2 Relative difference between the standard deviation and median absolute deviation (MAD) of distribution of the PSR/ECG “Y” channel data triggered at second crossovers of photoplethysmographic signals (e.g., in Poincaré map analysis PM#1) dZRelStdMAD1 Relative difference between the standard deviation and median absolute deviation (MAD) of distribution of the PSR/ECG “Z” channel data triggered at first crossovers of photoplethysmographic signals (e.g., in Poincaré map analysis PM#1)
Synchronicity Features Example PM #2
[0176]
[0177]
[0178]
[0179] In some embodiments, the dynamical feature extraction module 118 is configured to generate a histogram and extract statistical properties, such as, but not limited to modes, scale, skewness, kurtosis, and mutual information, from the generated histogram. e.g., as discussed in relation to
[0180]
[0181] Following generation of the Poincaré map 606, the dynamical feature extraction module 118, in some embodiments, is configured to generate a geometric object from the data. In
[0182] Indeed, synchronicity between acquired raw photoplethysmographic signal and cardiac signal based on triggers defined in the cardiac signal may be used to assess for presence, non-presence, severity, and/or localization of coronary artery disease, pulmonary hypertension, heart failure, among other disease, conditions, and associated conditions.
[0183] In some embodiments, to generate Poincaré map 512, module 118 plots/generates a 2D pairs of points |x.sub.i, x.sub.i+1| (e.g., (x.sub.1, x.sub.2), (x.sub.2, x.sub.3) etc.) against the points |x.sub.i−1, x.sub.i| (e.g., (x.sub.0, x.sub.1). (x.sub.1, x.sub.2)) of the amplitude values of a given photoplethysmographic signal (e.g., the red or the infrared photoplethysmographic signal) at a landmark of a cardiac signal (e.g., at one of channel “x”. “y”, or “z”).
[0184]
[0185]
[0186] Table 2 provides a list of example synchronicity feature extracted parameters associated with Poincaré map analysis PM #2 as their corresponding description.
TABLE-US-00002 TABLE 2 dDmjL Major diameter of the ellipse in Poincaré map derived from amplitude of infrared photo-photoplethysmographic signal at R-peaks of a cardiac signal (e.g., per example Poincaré map analysis PM#2). dDmjU Major diameter of the ellipse in Poincaré map derived from amplitude of red photo-photoplethysmographic signal at R-peaks of a cardiac signal (e.g., per example Poincaré map analysis PM#2) dDmnU Minor diameter of the ellipse in Poincaré map derived from amplitude of red photo-photoplethysmographic signal at R-peak of a cardiac signal (e.g., per example Poincaré map analysis PM#2). dAlphaL Tilt angle, alpha, of the ellipse in Poincaré map derived from amplitude of the infrared photo-photoplethysmographic signal at R-peak of a cardiac signal (e.g., per example Poincaré map analysis PM#2). dAlphaU Tilt angle, alpha, of the ellipse in Poincaré map derived from amplitude of the red photo-photoplethysmographic signal at R-peak of a cardiac signal (e.g., per example Poincaré map analysis PM#2). dKurL Kurtosis of histogram of Poincaré map analysis “PM2” of infrared photoplethysmographic signal. dMeanL Mean value of histogram of Poincaré map analysis “PM2” of infrared photoplethysmographic signal. dMeanU Mean value of histogram of Poincaré map analysis PM#2 for red photoplethysmographic signal dModeLP Mode of the distribution (histogram) of Poincaré map analysis PM#2 for infrared photoplethysmographic signal. dModeUP Mode of the distribution (histogram) of Poincaré map analysis PM#2 for infrared photoplethysmographic signal. dStdU Standard deviation of histogram of Poincaré map analysis PM#2 for the red photoplethysmographic signal.
Synchronicity Features Example PM #3
[0187]
[0188] Specifically,
[0189]
[0190] In some embodiments, dynamical feature extraction module 118 is configured to generate a histogram (e.g., as generated per
[0191]
[0192] In
[0193] That is, to generate the Poincaré map 714, the system plots/generates a 2D pairs of points |x.sub.i, x.sub.i+1| (e.g., (x.sub.1, x.sub.2). (x.sub.2, x.sub.3) etc.) of the TP interval index/time against the points |x.sub.i, x.sub.i+1| (e.g., (x.sub.1, x.sub.2). (x.sub.2, x.sub.3) etc.) of the TT interval index/time.
[0194] In some embodiments, dynamical feature extraction module 118 is configured to generate Poincaré map 714. Following generation of Poincaré map 714, the dynamical feature extraction module 118, in some embodiments, is configured to generate a geometric object from the map data. In
[0195] The dynamical feature extraction module 118, in some embodiments, may extract other parameters such as void area, surface area, porosity, perimeter length, density, among others.
[0196] Indeed, synchronicity between one or more acquired raw photoplethysmographic signals and one or more cardiac signals based on phase relations between landmarks in the photoplethysmographic signals and in the cardiac signal may be used to assess for presence, non-presence, severity, and/or localization (where applicable) of coronary artery disease, pulmonary hypertension, heart failure, among other disease, conditions, and associated conditions.
[0197]
[0198]
[0199]
[0200] Table 3 provides a list or example synchronicity feature extracted parameters associated with Poincaré map analysis PM #3 as their corresponding description.
TABLE-US-00003 TABLE 3 dDmjLUXR Major diameter of the ellipse in Poincaré map derived from differences in time intervals TT and TP between i) R-peaks in cardiac signals and ii) crossover landmarks between the acquired red and infrared photo-photoplethysmographic signals (e.g., per example Poincaré map analysis PM#3). dDmnLUXR Minor diameter of the ellipse in Poincaré map derived from differences in time intervals TT and TP between i) R-peaks in cardiac signals and ii) crossover landmarks between the acquired red and infrared photo-photoplethysmographic signals (e.g.. per example Poincaré map analysis PM#3). dMeanLURP1 Mean of TP time interval (time interval between R-peak of the PSR/ECG X channel and the first crossover landmarks between the red and infrared photo-photoplethysmographic signals in the Poincaré map analysis PM#3 landmarks). dMeanLURP2 Mean of TT time interval (time interval between R-peak of the PSR ECG X channel and the second crossover landmarks between the red and infrared photo-photoplethysmographic signals in the Poincaré map example analysis PM#3 landmarks). dModeLURP1 Mode of TP time interval (time interval between R-peak of the PSR/ECG X channel and the first occurrence of the Poincaré map analysis PM#3 landmarks) for the red photoplethysmographic signal. dModeLURP2 Mode of TT time interval (time interval between R-peak of the PSR/ECG X channel and the second occurrence of the Poincaré map analysis PM#3 landmarks for the red photoplethysmographic signal). dSkewLURP1 Skew of TP time interval (time interval between R-peak of the PSR/ECG X channel and the first occurrence of the Poincaré map analysis PM#3 landmarks). dStdLURP2 Standard deviation of TT time interval (time interval between R-peak of the PSR/ECG X channel and the second occurrence of the Poincaré map analysis PM#3 landmarks). dRelMeanMedDiffLURP1 Ratio of mean-med/mean for two histograms: one for TP and one for TT, derived from Poincaré map analysis PM#3 (e.g., per FIG. 7B).
Synchronicity Features Example #4
[0201]
[0202]
[0203]
[0204]
[0205] Indeed, synchronicity between acquired raw photoplethysmographic signals and cardiac signal based on phase differences between the cardiac signal and the photoplethysmographic signal(s) may be used to assess for presence, non-presence, severity, and/or localization (where applicable) of coronary artery disease, pulmonary hypertension, heart failure, among other disease, conditions, and associated conditions.
[0206]
[0207] Plot 812 shows a frequency analysis of the difference data of plot 810. In plot 812, the x-axis is frequency (in Hz) and the y-axis is the relative amplitude of the signal. Plot 814 shoes a difference between the infrared photoplethysmographic signal 304 and the cardiac signal 104b. In plot 814, the x-axis is lime (in index count of the data set). Plot 816 shows a filtered version of the difference data of plot 814. Plot 818 shows a histogram of the filtered difference data of plot 816. In the histogram 818, the x-axis of the histogram shows difference amplitude (in bins derived from the difference data) and the y-axis shows the frequency/count.
[0208]
[0209] Table 4 provides a list of example synchronicity feature extracted parameters associated with Phase analysis #4 as their corresponding description.
TABLE-US-00004 TABLE 4 dPhiDiffXL1Med Median value of the phase difference distribution belonging to part 1 distribution after the phase difference between a photoplethysmographic signal and cardiac signal are split into two parts: part 1 with higher mean mean dPhiDiffXL2Med Median value of the phase difference distribution belonging to part 1 distribution after the phase difference between a photoplethysmographic signal and cardiac signal are split into two parts: here, part 2 with lower mean. dPhiDiffXL1Std Standard deviation of phase difference distribution belonging to Part 2 distribution after the phase difference between a photoplethysmographic signal and a cardiac signal are split into two parts: here, part 1 with higher mean. dPhiDiffXL2Std Standard deviation of phase difference distribution belonging to Part 2 distribution after the phase difference between a photoplethysmographic signal and a cardiac signal are split into two parts: part 1 with higher mean and part 2 with lower mean. dPhiDiffXLMean Mean value of the whole distribution for phase difference distribution between a photoplethysmographic signal and a cardiac signal. dPTT Pulse transit time: time difference (lag) between the phase of the PSR/ECG X channel and phase of the infrared photoplethysmographic signal.
[0210] Machine-Learning Based Classifier
[0211] Machine learning techniques predict outcomes based on sets of input data. For example, machine learning techniques are used to recognize patterns and images, supplement medical diagnoses, and so forth. Some machine learning, techniques rely on a set of features generated using a training set of data (i.e., a data set of observations, in each of Which an outcome to be predicted is known), each of which represents some measurable aspect of observed data, to generate and tune one or more predictive models. For example, observed signals (e.g., cardiac, plethysmographic, or other biophysical signals from a number of subjects, alone or in any number of combinations) may be analyzed to collect frequency, average values, and other statistical information about these signals. A machine learning technique may use these features to generate and tune a model that classifies or relates these features to one or more conditions, such as some form of cardiovascular disease or condition, including. e.g., coronary artery disease, heart failure, pulmonary hypertension, etc., and then apply that model to data, such biophysical data of one or more humans, to detect and/or to gain an understanding of the presence, non-presence, severity of one or more diseases or conditions (such as described herein) that might otherwise not be detectable or understandable to the same degree. Conventionally, in the context of cardiovascular disease, these features are manually selected from conventional electrocardiographic signals and combined by data scientists working with domain experts.
[0212] Examples of embodiments of machine learning include, but are not limited to, decision trees, random forests. SVMs, neural networks, linear models. Gaussian processes, nearest neighbor. SVMs and Naïve Bayes. In some embodiments of the present disclosure, machine learning techniques may be implemented. e.g., as described in U.S. patent application Ser. No. 15/653,433, entitled “Discovering Novel Features to Use in Machine Learning Techniques, such as Machine Learning Techniques for Diagnosing Medical Conditions”: and U.S. patent application Ser. No. 15/653,431, entitled “Discovering Genomes to Use in Machine Learning Techniques”: each of which is incorporated by reference herein in its entirety
Experimental Results and Other Embodiments
[0213]
[0214] In the study, candidate features were evaluated using t-test, mutual information, or AUC. T-tests were conducted against a null-hypothesis of normal LVEDP and null hypothesis of negative coronary artery disease. A t-test is a statistical test that can determine if there is a difference between two sample means from two populations with unknown variances. The output of the t-test is a dimensionless quantity known as a p-value. A small p-value (typically ≤0.05) indicates strong evidence against the null hypothesis. The study used random sampling with replacement (bootstrapping) to generate test sets.
[0215] Mutual information were conducted to assessed dependence of elevated or abnormal LVEDP or significant coronary artery disease on certain feature sets. Mutual information refers to an information theoretic measure of the mutual dependence between two random variables. MI is normalized by number of bins and the high and low MI are calculated as a high and a low of
A selected feature has a high that is greater than 1.0 and a low that is greater than 1.0.
[0216] Table 1 provides a description of each of the assessed synchronicity extracted parameters of
[0217] Table 4 provides a description of each of the assessed synchronicity extracted parameters of
[0218] Experimental Results for Features of Poincaré Map Analysis #1
[0219] As discussed above. Table 1 provides a description of each of the assessed synchronicity extracted parameters of
[0220] Specifically,
TABLE-US-00005 TABLE I-A Feature Name Disease State Gender t-test p-value dXDmj LVEDP Female 0.012 Description: Major diameter of ellipse from Poincaré map PM#1 for the PSR/ECG “X” channel
TABLE-US-00006 TABLE 1-B Feature Name Disease State Gender t-test p-value dXDmn LVEDP Female 0.003 dZDmn LVEDP Both Genders 0.037 Description: Minor diameter of ellipse from Poincaré map PM#1 for the PSR/ECG “X” and “Z” channels
[0221] Further.
TABLE-US-00007 TABLE I-C Feature Name Disease State Gender t-test p-value dYAlpha CAD Both Genders 0.049 dZAlpha LVEDP Male 0.039 Description: Tilt angle, alpha, of the ellipse from Poincaré map analysis PM#1 on the PSR/ECG “Y” and “Z” channels
[0222] In addition.
TABLE-US-00008 TABLE I-D Feature Name Disease State Gender t-test p-value ROC-AUC dXMean1 CAD Female 0.00064 0.548 CAD Female 0.011 0.518 Description: Amplitude mean of PSR/ECG “X” channel at the first intersection crossover points of photoplethysmographic signals (e.g., in Poincaré map analysis PM#1)
[0223] In addition.
TABLE-US-00009 TABLE I-E Feature Name Disease State Gender t-test p-value dXStd1 LVEDP Female 0.037 Description: Standard deviation of the distribution of the PSR/ECG “X” channel triggered by the first crossover landmarks of photoplethysmographic signals (e.g., in Poincaré map analysis PM#1)
TABLE-US-00010 TABLE I-F Feature Name Disease State Gender t-test p-value dXStd2 LVEDP Both Genders 0.042 Description: Standard deviation of the distribution of the PSR/ECG “X” channel triggered by second crossover landmarks of photoplethysmographic signals (e.g., in Poincaré map analysis PM#1)
TABLE-US-00011 TABLE I-G Feature Name Disease State Gender Mutual Information dYStd2 CAD Male 1.143 Description: Standard deviation of distribution of the PSR/ECG “Y” channel data triggered at second crossover landmarks of photoplethysmographic signals (e.g., in Poincaré map analysis PM#1)
[0224] Further.
TABLE-US-00012 TABLE I-H Feature Name Disease State Gender Mutual Information dYKurt2 CAD Male 1.061 Description: Kurtosis of distribution of the PSR ECG “Y” channel data triggered at second crossover landmarks of photoplethysmographic signals (e.g., in Poincaré map analysis PM#1)
[0225] Further.
TABLE-US-00013 TABLE I-I Feature Name Disease State Gender Mutual Information dZKurt2 LVEDP Both Genders 1.076 CAD Female 1.192 Description: Kurtosis of distribution of the PSR ECG “Y” channel data triggered at second crossover landmarks of photoplethysmographic signals (e g., in Poincaré map analysis PM#1)
[0226] Further.
TABLE-US-00014 TABLE I-J Feature Name Disease State Gender Mutual Information dYMode2 CAD Both Genders 1.104 dZMode2 CAD Male 1.036 Description: Mode of distribution of the PSR ECG “Y” and “Z” channel data triggered at the second crossover landmarks of photoplethysmographic signals (e.g . in Poincaré map analysis PM# 1)
[0227] Further.
TABLE-US-00015 TABLE I-K Feature Name Disease State Gender Mutual Information dZSkew 1 CAD Female 1.094 Description: Kurtosis of distribution of the PSR ECG “Z” channel data triggered at first crossover landmarks of photoplethysmographic signals (e.g.. in Poincaré map analysis PM#1)
TABLE-US-00016 TABLE I-L Feature Name Disease State Gender Mutual Information dZSkew2 CAD Both Gender 1.058 Description: Kurtosis of distribution of the PSR ECG “Z” channel data triggered at second crossover landmarks of photoplethysmographic signals (e g., in Poincaré map analysis PM#1)
[0228] In addition.
TABLE-US-00017 TABLE I-M Disease t-test Mutual Feature Name State Gender p-value Information dYRelStdMAD2 CAD Male 0.042 1.048 Description: Relative difference between the standard deviation and median absolute deviation (MAD) of distribution of the PSR/ECG “Y” channel data triggered at second crossover landmarks of photoplethysmographic signals (e.g., in Poincaré map analysis PM#1)
[0229] Further.
TABLE-US-00018 TABLE 1-N Feature Name Disease State Gender t-test p-value dZRelStdMAD1 LVEDP Female 0.041 Description: Relative difference between the standard deviation and median absolute deviation (MAD) of distribution of the PSR/ECG “Z” channel data triggered at first crossover landmarks of photoplethysmographic signals (e.g., in Poincaré map analysis PM#1)
[0230] Experimental Results for Features of Poincaré Map Analysis #2
[0231] As discussed above. Table 2 provides a description of each of the assessed synchronicity extracted parameters of
[0232] Specifically,
TABLE-US-00019 TABLE 2-A Feature Name Disease State Gender t-test p-value Mutual Information ROC-AUC dDmjL CAD Female 0.035 1.104 0.502 LVEDP Both Genders 0.031 n/s n/s Description: Major diameter of the ellipse in Poincaré map derived from amplitude of infrared photo-photoplethysmographic signal at R-peaks of a cardiac signal (e.g., per example Poincaré map analysis PM#2)
TABLE-US-00020 TABLE 2-B Feature Name Disease State Gender t-test p-value dDmjU LVEDP Both Genders 0.007 Description: Major diameter of the ellipse in Poincaré map derived from amplitude of red photo-photoplethysmographic signal at R-peaks of a cardiac signal (e.g., per example Poincaré map analysis PM#2)
[0233] Further,
TABLE-US-00021 TABLE 2-C Feature Name Disease State Gender t-test p-value dDmnU LVEDP Both Genders 0.038 Description: Minor diameter of the ellipse in Poincaré map derived from amplitude of red photo-photoplethysmographic signal at R-peak of a cardiac signal (e.g., per example Poincaré map analysis PM#2)
[0234] In addition.
TABLE-US-00022 TABLE 2-D Feature Name Disease State Gender Mutual Information dAlphaL CAD Female 1.043 Description: Tilt angle, alpha, of the ellipse in Poincaré map derived from amplitude of the infrared photo-photoplethysmographic signal at R-peak of a cardiac signal (e.g., per example Poincaré map analysis PM#2)
TABLE-US-00023 TABLE 2-E Feature Name Disease State Gender Mutual Information dAlphaU CAD Both Genders 1.03 Description: Tilt angle, alpha, of the ellipse in Poincaré map derived from amplitude of the red photo-photoplethysmographic signal at R-peak of a cardiac signal (e.g., per example Poincaré map analysis PM#2)
[0235] In addition.
TABLE-US-00024 TABLE 2-F Feature Name Disease State Gender Mutual Information dKurtL LVEDP Both Genders 1.171 Description: Kurtosis of histogram of Poincaré map analysis “PM2” of infrared photoplethysmographic signal
[0236] In addition.
TABLE-US-00025 TABLE 2-G Feature Name Disease State Gender t-test p-value Mutual Information ROC-AUC dMeanL CAD Female n/s 1.012 0.516 LVEDP Male 0.033 n/s n/s Description: Mean value of histogram of Poincaré map analysis “PM2” of the infrared photoplethysmographic signals
TABLE-US-00026 TABLE 2-H Mutual Feature Name Disease State Gender t-test p-value Information dMeanU CAD Female n/s 1.091 LVEDP Both Genders 0.003 n/s Description: Mean value of histogram of Poincaré map analysis PM#2 for red photoplethysmographic signal
[0237] In addition.
TABLE-US-00027 TABLE 2-I Feature Name Disease State Gender t-test p-value ROC-AUC dModeLP CAD Both Genders n/s 0.507 LVEDP Both Genders 0.024 n/s Description: Mode of the distribution (histogram) of Poincaré map analysis PM#2 for the infrared photoplethysmographic signal
TABLE-US-00028 TABLE 2-J Feature Name Disease State Gender t-test p-value dModeUP LVEDP Both Genders 0.004 Description: Mode of the distribution (histogram) of Poincaré map analysis PM#2 for the infrared photoplethysmographic signal
[0238] In addition.
TABLE-US-00029 TABLE 2-K Feature Name Disease State Gender ROC-AUC dStdU CAD Female 0.511 Description: Standard deviation of histogram of Poincaré map analysis PM#2 for the red photoplethysmographic signal
[0239] Experimental Results for Features of Poincaré Map Analysis #3
[0240] As discussed above. Table 3 provides a description of each of the assessed synchronicity extracted parameters of
[0241] Specifically,
TABLE-US-00030 TABLE 3-A Feature Name Disease State Gender ROC-AUC dDmjLUXR CAD Both Genders 0.501 Description: Major diameter of the ellipse in Poincaré map derived from differences in time intervals TT and TP between i) R-peaks in cardiac signals and ii) crossover landmarks between the acquired red and infrared photo-photoplethysmographic signals (e.g., per example Poincaré map analysis PM#3)
TABLE-US-00031 TABLE 3-B Feature Name Disease State Gender t-test p-value dDmnLUXR LVEDP Female 0.02 Description: Minor diameter of the ellipse in Poincaré map derived from differences in time intervals TT and TP between i) R-peaks in cardiac signals and ii) crossover landmarks between the acquired red and infrared photo-photoplethysmographic signals (e.g., per example Poincaré map analysis PM#3)
[0242] In addition.
TABLE-US-00032 TABLE 3-C Feature Name Disease State Gender t-test p-value dMeanLURP1 LVEDP Both Genders 0.013 Description: Mean of TP time interval (time interval between R-peak of the PSR/ECG X channel and the first crossoxer landmarks between the red and infrared photo-photoplethysmographic signals in the Poincaré map analysis PM#3 landmarks)
TABLE-US-00033 TABLE 3-D Feature Name Disease State Gender t-test p-value dMeanLURP2 LVEDP Male 0.02 Description: Mean of TT time interval (lime interval between R-peak of the PSR/ECG X channel and the second crossover landmarks between the red and infrared photophotoplethysmographic signals in the Poincaré map example analysis PM#3 landmarks)
[0243] In addition.
TABLE-US-00034 TABLE 3-E Feature Name Disease State Gender t-test p-value dModeLURP1 LVEDP Both Genders 0.013 Description: Mode of TP time interval (time interval between R-peak of the PSR/ECG X channel and the first occurrence of the Poincaré map analysis PM#3 landmarks) for the red photoplethysmographic signal
TABLE-US-00035 TABLE 3-F Feature Name Disease State Gender t-test p-value dModeLURP2 LVEDP Male 0.028 Description: Mode of TT time interval (time interval between R-peak of the PSR/ECG X channel and the second occurrence of the Poincaré map analysis PM#3 landmarks for the red photoplethysmographic signal)
[0244] In addition.
TABLE-US-00036 TABLE 3-G Feature Name Disease State Gender t-test p-value dSkewLURP1 CAD Both Genders 0.034 Description: Skew of TP time interval (time interval between R-peak of the PSR/ECG X channel and the first occurrence of the Poincaré map analysis PM#3 landmarks)
[0245] In addition. (and reproduced in Table 3-H) shoes that the standard deviation of TT time interval (time interval between R-peak of the PSR/ECG X channel and the second occurrence of the Poincaré map analysis PM #3 landmarks) (shown as “dStdLURP2”) has a mutual information value of 1.486 and an AUC value of 0.541 in predicting and/or estimating the presence, non-presence, localization (where applicable), and/or severity of presence of coronary artery disease.
TABLE-US-00037 TABLE 3-H Feature Disease Mutual Name State Gender Information ROC-AUC dStdLURP2 CAD Both Genders 1.486 0.541 Description Stantard deviation of TT time interval (time interval between R-peak of the PSR/ECG X channel and the second occurrence of the Poincaré map analysis PM#3 landmarks)
[0246] In addition.
TABLE-US-00038 TABLE 3-I Feature Name Disease State Gender ROC-AUC dRelMeanMedDiffLURP1 CAD Both Genders 0.5 Description: Ratio of mean-med/mean for two histograms: one for TP and one for TT, derived from Poincaré map analysis PM#3 (e.g., per FIG. 7B).
[0247] Experimental Results for Features of Phase Analysis #4
[0248] As discussed above. Table 4 provides a description of each of the assessed synchronicity extracted parameters of
[0249] Specifically,
TABLE-US-00039 TABLE 4-A Feature Name Disease State Gender t-test p-value dPhiDiffXL1Med CAD Both Genders 0.015 Description: Median value of the phase difference distribution belonging to part 1 distribution after the phase difference between a photoplethysmographic signal and cardiac signal are split into two parts: part 1 with higher mean.
[0250] In addition.
TABLE-US-00040 TABLE 4-B Feature Name Disease State Gender ROC-AUC dPhiDiffXL2Std CAD Male 0.502 Description: Standard deviation of phase difference distribution belonging to Part 2 distribution after the phase difference between a photoplethysmographic signal and a cardiac signal are split into two parts: here, part 2 with lower mean.
[0251] In addition.
TABLE-US-00041 TABLE 4-C Feature Name Disease State Gender t-test p-value dPhiDiffXLMean CAD Both Genders 0.026 Description: Mean value of the whole distribution for phase difference distribution between a photoplethysmographic signal and a cardiac signal
[0252] In addition.
TABLE-US-00042 TABLE 4-D Feature Name Disease State Gender t-test p-value dPTT LVEDP Female 0.045 Description: Pulse transit time: time difference (lag) between the phase of the PSR/ECG X channel and phase of the infrared photoplethysmographic signal
[0253] Coronary Artery Disease—Learning Algorithm Development Study
[0254] A “Coronary Artery Disease—Learning Algorithm Development” (CADLAD) study was untaken that acquired photoplethysmographic signals and cardiac signals to support the development and testing of the machine-learned algorithms.
[0255] In the study, paired clinical data were used to guide the design and development of the pre-processing, feature extraction, and machine learning phase of the development. That is, the collected clinical study data are split into cohorts: a training cohort, a validation cohort, and a verification cohort. In the study, each acquired data set is first pre-processed to clean and normalize the data. Following the pre-processing processes, a set of features are extracted from the signals in which each set of features is paired with a representation of the true condition—for example, the binary classification of the presence or absence of significant CAD or the scored classification of the presence of significant CAD in a given coronary artery.
[0256] The assessment system (e.g., 114, 114a. 114b), in some embodiments, automatically and iteratively explores combinations of features in various functional permutations with the aim of finding those combinations which can successfully match a prediction based on the features. To avoid overfitting of the solutions to the training data, the validation set is used as a comparator. Once candidate predictors have been developed, they are then manually applied to a verification data set to assess the predictor performance against data that has not been used at all to generate the predictor. Provided that the data sets are sufficiently large, the performance of a selected predictor against the verification set will be close to the performance of that predictor against new data.
[0257] The study also developed and evaluated machine learning-based predictive models that employ nonlinear dynamics and chaos for extracting physically meaningful and significant features from the cardiac biopotential and photoplethysmographic signal data. Traditional features based on linear characterizations of the signals are not capable of detecting more complex and nonlinear patterns hidden in the signals. In the study, by employing nonlinear dynamics, three categories of features were developed: (i) features based on the dynamics of the cardiac system represented by biopotential signal. (ii) features based on the dynamics represented by the PPG signals and (iii) features characterizing the synchronicity between the two dynamics.
[0258] For the first two sets, invariant measures of the dynamics such as Lyapunov exponent (LE), fractal dimension (D2) and rate of entropy (K2) were computed. Lyapunov exponent is a global measure that characterizes the strength of the exponential divergence |30|. For chaotic systems, the maximum Lyapunov exponent is a positive number which indicates that the system has less memory of the past. For a given dynamical system, as Lyapunov exponent value becomes larger, the time horizon over which the past information can be used to predict the future becomes shorter. Entropy (KS) (or Kohnogoroy Sinai entropy K2 |31, 32|) represents the rate of change of entropy with time. Fractal dimension (D2) characterizes the topological property of an attractor in phase space and can be used to reveal more about the dynamics in combining the geometric information of the attractor (fractality) and how the dynamics evolve on it |33|. An example of an attractor of the acquired cardiac and photoplethysmographic signals are shown in
[0259] Nonlinear dynamics and chaos theory systematically can be used to explain the complexity of linear system systems and provides tools to quantitatively analyze their behavior |19|. Linear systems can generate responses which grow decay exponentially or oscillate periodically or a combination thereof in which any irregular pattern in the response may be ascribed to irregularity or randomness in the inputs to these systems. Linear systems are a simplification of reality, and most dynamical systems whether natural or man-made are inherently nonlinear which can produce complex irregular behavior even without any source of randomness. These behaviors are often called deterministic chaos. Nonlinear dynamics and chaos tools have been used to explain various complex biological and physiological phenomena |20, 21, 22, 23|: for example, to classify atrial fibrillations |24| and to characterize heart rate variability |25|, each of where is incorporated by reference here in its entirety. Further description of these dynamical features are described in U.S. Provisional Patent Application No. 62/862,991, filed Jun. 18, 2019, entitled “Method and System to Assess Disease Using Dynamical Analysis of Biophysical Signals.” which is incorporated by reference herein in its entirety.
[0260] Other invariant measure of dynamics may be used as a feature set. Deterministic dynamical systems that exhibit chaotic behavior often possess invariant properties which do not depend on when the observations are made and are thus independent of the evolution of the system.
[0261] For the synchronicity feature sets, three types of Poincaré maps were defined and the resulting sets were characterized statistically and geometrically. The computed features set matched with appropriate labels were then used to train several machine learning models. Model were selected based on its respective AUC performance on a holdout test set. The study performed cross validation and grid searches to tune the hyperparameters used in the classifier training. In the study, a developed Elastic Net model was observed to have AUCs of 0.78 and 0.61 in CAD classification on two tested data sets. And a developed XGboost model was observed to have AUCs of 0.86 and 0.63 on two tested data sets. The study demonstrated an efficient and cost-effective means of using advanced nonlinear feature extraction processes of non-invasive modalities for machine learning operations for disease or abnormal condition prediction.
[0262] Elastic Net. Lasso. or Ridge classifiers are generally suited for smaller datasets with a large number of features because they can be adjusted to prevent overfitting. Elastic Net is a hybrid of Lasso and Ridge, where both the absolute value penalization (Lasso) and squared penalization (Ridge) are included. For each penalty, hyperparameters exist that can be optimized to generate stronger models. Lasso and Ridge only have a single hyperparameter each which makes optimization more limited.
[0263] Data description. In the study, two human subject cohorts with an average age of 63 (group A) and 28 (group B) were recruited for data collection. Subjects of the cohorts were selected after undergoing a qualification screening process. For the older group, the CAD labels and LVEDP values were determined by the corresponding gold standard tests while the younger group was considered to be healthy by clinical criteria. That is, the younger group did not have CAD and their LVEDP values were not abnormally high or elevated.
[0264] From each subject in groups A and B, cardiac signals (as biopotential signals) and photoplethysmographic signals as time series data were acquired. Data of both signal modalities were acquired over 3.5 minutes, and the entire procedure took about 10 minutes per subject on average. The cardiac signals were each collected at a sampling rate of 8 KHz (i.e., 8,000 samples per seconds for each of 6 channels collected over 210 seconds) using a phase space recorder as described in relation to
[0265] During the same procedure in which the cardiac signals were collected from a subject, photoplethysmographic signals were collected at a sampling rate of 500 Hz using the same phase space recorder. Photo-absorption data of red and infrared channels were each recorded at 500 samples per second over the same 210 second period. These photoplethysmographic and cardiac signals were simultaneously acquired for each subject. Jitter (inter-modality jitter) in the data was less than about 10 microseconds (μs). Jitter among the cardiac signal channels were around 10 femtoseconds (fs).
[0266] CAD Feature Study. The study used a definition for significant coronary occlusions as either having greater than 70% stenosis or a patient that passed a functional threshold for blood flow limitation |14, 15|. For group A, two-vessel disease (i.e., two vessels with lesions meeting this definition) was considered as being disease-positive, and non-disease cases were defined as healthy control subjects that had underwent invasive catheterization for evaluation of coronary artery disease but did not have any coronary lesions. Table 5 lists the number of positive and negative cases in the coronary artery disease data set used in the development of coronary artery disease features for the study. Table 5 further shows the average age and gender composition associated with the subjects in the data set. The study used invasive coronary angiography, the “gold standard” for coronary artery disease, as the ground truth metric. In coronary angiography, fluoroscopy is used to image coronary arteries following an injection of a radiopaque contrast agent. With coronary angiography, stenoses (blockages) in the arteries may be detected and patients are subsequently labelled as CAD-positive or CAD-negative.
TABLE-US-00043 TABLE 5 Age Group Total Positive Negative Male Female IQR A 1211 463 748 61% 39% 56-71 B 358 0 358 42% 58% 21-31 Total 1569 463 1106 57% 43% —
[0267] The study results show that synchronicity between photoplethysmographic signals and cardiac signals, represented by way of synchronicity features from analysis between photoplethysmographic signals and cardiac signals as described herein, can be used to predict the presence or non-presence of significant coronary artery disease.
[0268] LVDEP Feature Feasibility Study Left ventricular end diastolic pressure (LVEDP) is an invasively-obtained hemodynamic measurement used to describe the heart's left-sided filling pressures in patients undergoing cardiac catheterization. LVEDP is a critical parameter in the hemodynamic evaluation of patients with either systolic or diastolic LV dysfunction, which are both associated with decreased LV compliance. Alterations in the pressure-volume relationships that result in markedly elevated filling pressures are the hallmark of cardiomyopathies |10|.
[0269] Measurement of filling pressures may be used to assess risk stratification and the development of an appropriate treatment strategy. Furthermore. LVEDP provides important prognostic information, as elevated LVEDP has been established as an independent predictor of adverse outcomes in the setting of acute myocardial infarct |16|, cardiogenic shock |17|, post-procedural success of cardiac surgery |18|, and percutaneous cardiac interventions. Table 6 lists the number of positive and negative cases of LVEDP used in the evaluation of LVEDP features in the data set of Table 5.
TABLE-US-00044 TABLE 6 Age Group Total High Low Male Female IQR A 470 211 259 60% 40% 57-71 B 418 0 418 42% 58% 21-35 Total 888 211 677 52% 48%-
[0270] The study results show that the synchronicity between photoplethysmographic signals and cardiac signals, represented by way of synchronicity features from analysis between photoplethysmographic signals and cardiac signals as described herein, can be used to predict presence or non-presence of abnormal LVEDP.
[0271] Machine-Learning Classifier Analysis. In the study, feature sets extracted from the acquired data set, including 94 synchronicity features defined between the photoplethysmographic signals and cardiac signals (e.g., per synchronicity analysis of Poincare maps 1, 2 and 3), as well as 6 features of the phase analysis #4, among others (e.g., dynamical features, etc.), were extracted and assessed in a machine-based classifier analysis. The feature sets including the synchronicity features were paired with the corresponding CAD or LVEDP labels and provided as input to the machine learning models. The feature sets included 36 other dynamical features associated with cardiac signals (i.e., biopotential signals) and 29 further dynamical features associated with photoplethysmographic signals were also evaluated. These features are described in U.S. patent application Ser. No. ___/______, entitled “Method and System to Assess Disease Using Dynamical Analysis of Biophysical Signals”, concurrently filed with the instant application (and claimed priority to U.S. Provisional Patent Application No. 62/863,005, filed Jun. 18, 2019), which is incorporated by reference herein in its entirety.
[0272] In the classifier analysis, the data for CAD and LVEDP were each split into a training-validation set and a test set. Table 7 shows the composition of the training-validation and test data sets for the machine learning model training and evaluation. As noted above, information about Groups A and B for the CAD data sets are listed in Table 5 and information about Groups A and B for the LVEDP data sets are listed in Table 6.
TABLE-US-00045 TABLE 7 ML Data set Composition Train-validation 80% A + 50% B Test 1 20% A + 50% B Test 2 20% A
[0273] The training-validation set is used to train and fine-tune candidate machine learning models using 5-fold cross validation. Table 8 lists the classifiers used in the study for training and model selection for both the CAD and LVEDP data sets. The pipeline for data scaling, model training, grid search and model evaluation was implemented in Python using the Scikit-learn package |36|.
TABLE-US-00046 TABLE 8 1 Gradient tree boosting (XGBClassifier) [37] 2 K nearest neighbors classifier (KNeighborsClassifier 3 support vector classifier (SVC) 4 Random forest classifier 5 Logistic regression 6 Elastic net (ElasticNet) [38]
[0274] To lied an optimal set of hyperparameters for each model, the study performed a grid search over a pre-defined range of hyperparameters. Using average AUC as the performance metric, the best hyper-parameters set is selected for each model. The selected models are then trained on the entire training-validation set and their AUC performance on the holdout test sets is ranked.
[0275] In the study, the Elastic Net model and the support vector classifier model were found to be most predictive for significant CAD predictions, and the XGBoost model and Elastic Net model were found to be most predictive for an elevated or abnormal LVEDP state. Table 9 shoes the predictive performance of the Elastic Net model and the support vector classifier model to predict significant CAD. Table 10 shows the predictive performance of the Elastic Net model and the XGBoost model to predict a significant CAD state.
TABLE-US-00047 TABLE 9 AUC Rank Model Training Test 1 Test 2 1 Elastic net 0.71 0.78 0.61 2 Linear SVC 0.75 0.65 0.52
TABLE-US-00048 TABLE 10 AUC Rank Model Training Test 1 Test 2 1 XGBoost 1.0 0.86 0.63 2 ElasticNet 0.79 0.84 0.51
[0276]
[0277] As shown in
[0278] The models were trained on both Groups A and B (older and younger subjects, respective)) data sets, as described in relation to Tables 5 and 6. The use of the Group B data set augments the training of the Group A data set and allows the models to learn very healthy from diseased subjects. As a result, the model honed for this task exhibits better performance on Test 1 (which contains subjects from both Groups A and B) as compared to Test 2 (which is from Group A only). Further, because the acquired data sets are skewed toward non-diseased cases, as shown in Tables 5 and 6, consequently, the trained model better in this study at detecting CAD-negative subjects. It is expected that with data sets that are more balanced between diseased and non-diseased cases, model performance for Test 2 would be improved. XGBoost performance may also be improved by performing a more refined hyperparameter search and stronger regularization.
[0279] Further improvements to a second Elastic Net model was made using only the synchronicity feature sets and with a larger data set.
[0280]
[0281]
[0282]
[0283]
[0284] Table 11 lists the accumulative feature contributions in each of the sub-groups PM1-PM3 as denoted in
TABLE-US-00049 TABLE 11 Target Poincaré Map 1 Poincaré Map 2 Poincaré Map 3 CAD 0.116 0.207 0.677 LVEDP 0.214 0.325 0.461
[0285] Although Elastic Net classifiers were found to be the best performing model to classify CAD and LVEDP, synchronization features contribute differently across the two diseases. The absolute value of the difference in the feature contribution is plotted in
[0286] LVDEP Feature Performance Study. A second LVDEP-related study was conducted to predict, as a primary outcome, an elevated LVEDP. This study also investigated as secondary outcomes (i) the diagnostic sensitivity of the machine-learned predictor among three sub-groups of increasing LVEDP (≥20 mmHg. ≥25 mmHg. and ≥30 mmHg) and (ii) the predictive performance of the predictor within an age and gender propensity matched cohort.
[0287] The second LVDEP-related study used data sets collected in the manner described herein (i.e., using a phase space recorder as described in relation to
[0288] Data of both signal modalities were acquired over 3.5 minutes and the entire procedure took about 10 minutes. The biopotential signals were collected with a sampling rate of 8 KHz (i.e., 8.000 samples per seconds for each of 6 channels over 2 It) seconds). Three differential input pairs w ere arranged orthogonally at the patient's thorax along with a reference lead. The acquired signals were used for feature extraction after removing baseline wander and filtering powerline and high frequency noise.
[0289] Out of the 1.919 symptomatic subjects who underwent elective angiography, 256 subjects were found on catheterization to have an LVEDP ≥20 mmHg: these 256 subjects formed the study cohort. As noted, the patients were referred to angiography for the evaluation of symptoms, and elevated or abnormal LVEDP w as determined for each patient, when present, during cardiac catheterization with direct LV pressure measurements during ventriculography.
[0290] To develop the machine learned predictors, cross-validation was performed over 100 iterations, with 70% of the subjects used for training and 30% for testing. The subjects were divided to stratify prevalence of disease (LVEDP ≥20 mmHg) across the sets, but division was otherwise random. The training subjects features were inputted to an Elastic Net model configured with added regularization penalties to reduce overfilling. Once trained, the model was applied to the validation subjects to assess diagnostic performance.
[0291]
[0292]
[0293]
[0294] Healthcare Provider Portal
[0295] Referring to
[0296] Example Computing Device
[0297]
[0298] The computing device environment is only one example of a suitable computing environment and is not intended to suggest any limitation as to the scope of use or functionality.
[0299] Numerous other general-purpose or special purpose computing devices environments or configurations may be used. Examples of well-known computing devices, environments, and/or configurations that may be suitable for use include, but are not limited to, personal computers, server computers, handheld or laptop devices, mobile phones, wearable devices, multiprocessor systems, microprocessor-based systems, network personal computers (PCs), minicomputers, mainframe computers, embedded systems, distributed computing environments that include any of the above systems or devices, and the like.
[0300] Computer-executable instructions, such as program modules, being executed by a computer may be used. Generally, program modules include routines, programs, objects, components, data structures. etc. that perform particular tasks or implement particular abstract data types. Distributed computing environments may be used where tasks are performed by remote processing devices that are linked through a communications network or other data transmission medium. In a distributed computing environment, program modules and other data may be located in both local and remote computer storage media including memory storage devices.
[0301] With reference to
[0302] Computing device 1000 may have additional features/functionality. For example, computing device 1000 may include additional storage (removable and/or non-removable) including, but not limited to, magnetic or optical disks or tape. Such additional storage is illustrated in
[0303] Computing device 1000 typically includes a variety of computer readable media. Computer readable media can be any available media that can be accessed by the device 1000 and includes both volatile and non-volatile media, removable and non-removable media.
[0304] Computer storage media include volatile and non-volatile, and removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data. Memory 1004, removable storage 1008, and non-removable storage 1010 are all examples of computer storage media. Computer storage media include, but are not limited to, RAM, ROM, electrically erasable program read-only memory (EEPROM), flash memory or other memory technology. CD-ROM, digital versatile disks (DVD) or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to store the desired information and which can be accessed by computing device 1000. Any such computer storage media may be part of computing device 1000.
[0305] Computing device 1000 may contain communication connection(s) 1012 that allow the device to communicate with other devices. Computing device 1000 may also have input device(s) 1014 such as a keyboard, mouse, pen, voice input device, touch input device, etc., singly or in combination. Output device(s) 1016 such as a display, speakers, printer, vibratory mechanism. etc. may also be included singly or in combination. All these devices are well known in the art and need not be discussed at length here.
[0306] It should be understood that the various techniques described herein may be implemented in connection with hardware components or software components or, where appropriate, with a combination of both. Illustrative types of hardware components that can be used include Field-programmable Gate Arrays (FPGAs). Application-specific Integrated Circuits (ASICs), Application-specific Standard Products (ASSPs). System-on-a-chip systems (SOCs). Complex Programmable Logic Devices (CPLDs), etc. The methods and apparatus of the presently disclosed subject matter, or certain aspects or portions thereof, may take the form of program code (i.e., instructions) embodied in tangible media, such as floppy diskettes. CD-ROMs, hard drives, or any other machine-readable storage medium where, when the program code is loaded into and executed by a machine, such as a computer, the machine becomes an apparatus for practicing the presently disclosed subject matter.
[0307] Although example implementations may refer to utilizing aspects of the presently disclosed subject matter in the context of one or more stand-alone computer systems, the subject matter is not so limited, but rather may be implemented in connection with any computing environment, such as a network or distributed computing environment. Still further, aspects of the presently disclosed subject matter may be implemented in or across a plurality of processing chips or devices, and storage may similarly be effected across a plurality of devices. Such devices might include personal computers, network servers, handheld devices, and wearable devices, for example.
[0308] Although the subject matter has been described in language specific to structural features and/or methodological acts, it is to be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or acts described above. Rather, the specific features and ads described above are disclosed as example forms of implementing the claims.
[0309] Further examples of processing that may be used with the exemplified method and system are described in: U.S. Pat. No. 9,289,150, entitled “Non-invasive Method and System for Characterizing Cardiovascular Systems”: U.S. Pat. No. 9,655,536, entitled “Non-invasive Method and System for Characterizing Cardiovascular Systems”: U.S. Pat. No. 9,968,275, entitled “Non-invasive Method and System for Characterizing Cardiovascular Systems”: U.S. Pat. No. 8,923,958, entitled “System and Method for Evaluating an Electrophysiological Signal”: U.S. Pat. No. 9,408,543, entitled “Non-invasive Method and System for Characterizing Cardiovascular Systems and All-Cause Mortality and Sudden Cardiac Death Risk”: U.S. Pat. No. 9,955,883, entitled “Non-invasive Method and System for Characterizing Cardiovascular Systems and All-Cause Mortality and Sudden Cardiac Death Risk”: U.S. Pat. No. 9,737,229, entitled “Noninvasive Electrocardiographic Method for Estimating Mammalian Cardiac Chamber Site and Mechanical Function”: U.S. Pat. No. 10,039,468, entitled “Noninvasive Electrocardiographic Method for Estimating Mammalian Cardiac Chamber Size and Mechanical Function”: U.S. Pat. No. 9,597,021, entitled “Noninvasive Method for Estimating Glucose. Glycosylated Hemoglobin and Other Blood Constituents”: U.S. Pat. No. 9,968,265, entitled “Method and System for Characterizing Cardiovascular Systems From Single Channel Data”: U.S. Pat. No. 9,910,964, entitled “Methods and Systems Using Mathematical Analysis and Machine Learning to Diagnose Disease”: U.S. Patent Publication No. 2017/0119272, entitled “Method and Apparatus for Wide-Band Phase Gradient Signal Acquisition”: PCT Publication No. WO2017/033164, entitled “Method and Apparatus for Wide-Band Phase Gradient Signal Acquisition”: U.S. Patent Publication No 2018/0000371, entitled “Non-invasive Method and System for Measuring Myocardial Ischemia, Stenosis Identification. Localization and Fractional Flow Reserve Estimation”: PCT Publication No. WO2017/221221, entitled “Non-invasive Method and System for Measuring Myocardial Ischemia. Stenosis Identification, Localization and Fractional Flow Reserve Estimation”: U.S. Pat. No. 10,291,596, entitled “Method and System for Visualization of Heart Tissue at Risk”: U.S. patent application Ser. No. 16/402,616, entitled “Method and System for Visualization of Heart Tissue at Risk”: U.S. Patent Publication No. 2018/0249960, entitled “Method and System for Wide-band Phase Gradient Signal Acquisition”: U.S. patent application Ser. No. 16/232,801, entitled “Method and System to Assess Disease Using Phase Space Volumetric Objects”: PCT Application No. IB/2018/060708, entitled “Method and System to Assess Disease Using Phase Space Volumetric Objects”: U.S. Patent Publication No. US2019/0117164, entitled “Methods and Systems of De-Noising Magnetic-Field Based Sensor Data of Electrophysiological Signals”: U.S. patent application Ser. No. 16/232,586, entitled “Method and System to Assess Disease Using Phase Space Tomography and Machine Learning”: PCT Application No. PCT/IB2018/060709, entitled “Method and System to Assess Disease Using Phase Space Tomography and Machine Learning”: U.S. patent application Ser. No. 16/445,158, entitled “Methods and Systems to Quantify and Remove Asynchronous Noise in Biophysical Signals”: U.S. patent application Ser. No. 16/725,401 entitled “Method and System to Assess Disease Using Phase Space Tomography and Machine Learning” (having attorney docket no. 10321-034pv1 and claiming priority to U.S. Patent Provisional Application No. 62/784,984): U.S. patent application Ser. No. 16/429,593, entitled “Method and System to Assess Pulmonary Hypertension Using Phase Space Tomography and Machine Learning”: U.S. patent application Ser. No. 16/725,416, entitled “Method and System for Automated Quantification of Signal Quality”: U.S. patent application Ser. No. 16/725,430, entitled “Method and System to Configure and Use Neural Network To Assess Medical Disease”: U.S. patent application Ser. No. 15/653,433, entitled “Discovering Novel Features to Use in Machine Learning Techniques, such as Machine Learning Techniques for Diagnosing Medical Conditions”: U.S. patent application Ser. No. 15/653,431, entitled “Discovering Genomes to Use in Machine Learning Techniques”, each of which is incorporated by reference herein in its entirety.
[0310] Unless otherwise expressly stated, it is in no way intended that any method set forth herein be construed as requiring that its steps be performed in a specific order. Accordingly, where a method claim does not actually recite an order to be followed by its steps or it is not otherwise specifically stated in the claims or descriptions that the steps are to be limited to a specific order, it is no way intended that an order be inferred, in any respect. This holds for any possible non-express basis for interpretation, including matters of logic with respect to arrangement of steps or operational flow; plain meaning derived from grammatical organization or punctuation: the number or type of embodiments described in the specification.
[0311] While the methods and systems have been described in connection with certain embodiments and specific examples, it is not intended that the scope be limited to the particular embodiments set forth, as the embodiments herein are intended in all respects to be illustrative rather than restrictive.
[0312] The methods, systems and processes described herein may be used generate stenosis and FFR outputs for use in connection with procedures such as the placement of vascular stems within a vessel such as an artery of a living (e.g., human) subject, and other interventional and surgical system or processes. In one embodiment, the methods, systems and processes described herein can be configured to use the FFR/stenosis outputs to determine and/or modify, intra operation, a number of stents to be placed in a living (e.g., human), including their optimal location of deployment within a given vessel, among others.
[0313] Examples of other biophysical signals that may be analyzed in whole, or in part, using the example methods and systems include, but are not limited to, an electrocardiogram (ECG) data set, an electroencephalogram (EEG) data set, a gamma synchrony signal data set: a respiratory function signal data set: a pulse oximetry signal data set: a perfusion data signal data set: a quasi-periodic biological signal data set: a fetal ECG data set: a blood pressure signal: a cardiac magnetic field data set, and a heart rate signal data set.
[0314] The example analysis can be used in the diagnosis and treatment of cardiac-related pathologies and conditions and/or neurological-related pathologies and conditions, such assessment can be applied to the diagnosis and treatment (including, surgical, minimally invasive, and/or pharmacologic treatment) of any pathologies or conditions in which a biophysical signal is involved in any relevant system of a living body. One example in the cardiac context is the diagnosis of CAD, and other diseases and conditions disclosed herein and its treatment by any number of therapies, alone or in combination, such as the placement of a stent in a coronary artery, performance of an atherectomy, angioplasty, prescription of drug therapy, and/or the prescription of exercise, nutritional and other lifestyle changes, etc. Other cardiac-related pathologies or conditions that may be diagnosed include. e.g., arrhythmia, congestive heart failure, valve failure, pulmonary hypertension (e.g., pulmonary arterial hypertension, pulmonary hypertension due to left heart disease, pulmonary hypertension due to lung disease, pulmonary hypertension due to chronic blood clots, and pulmonary hypertension due to other disease such as blood or other disorders), as well as other cardiac-related pathologies, conditions and/or diseases. Non-limiting examples of neurological-related diseases, pathologies or conditions that may be diagnosed include. e.g., epilepsy, schizophrenia. Parkinson's Disease. Alzheimer's Disease (and all other forms of dementia), autism spectrum (including Asperger syndrome), attention deficit hyperactivity disorder. Huntington's Disease, muscular dystrophy, depression, bipolar disorder, brain/spinal cord tumors (malignant and benign), movement disorders, cognitive impairment, speech impairment, various psychoses, brain/spinal cord/nerve injury, chronic traumatic encephalopathy, cluster headaches, migraine headaches, neuropathy (in its various forms, including peripheral neuropathy), phantom limb/pain, chronic fatigue syndrome, acute and/or chronic pain (including back pain, failed back surgery syndrome, etc.), dyskinesia, anxiety disorders, conditions caused by infections or foreign agents (e.g., Lyme disease, encephalitis, rabies), narcolepsy and other sleep disorders, post-traumatic stress disorder, neurological conditions/effects related to stroke, aneurysms, hemorrhagic injury, etc., tinnitus and other hearing-related diseases/conditions and vision-related diseases conditions.
[0315] The following patents, applications and publications as listed below and throughout this document are hereby incorporated by reference in their entirety herein.
LIST OF REFERENCES
[0316] |1|I. Kononenko. “Machine learning for medical diagnosis: history, state of the art and perspective.” Artificial Intelligence in medicine 23 (1) 89-109 (2001). [0317] |2| B. A. Mobley, E. Schechter. W. E. Moore. P. A. McKee. J. E. Eichner. “Predictions of coronary artery stenosis by artificial neural network.” Artificial Intelligence in Medicine 18 (3) 187-203 (2000). [0318] |3| V. L. Patel. E. H. Shortliffe. M. Stefanelli. P. Szolovits. M. R. Berthold, R. Bellazzi, A Abu-Hanna. “The coming of age of artificial intelligence in medicine,” Artificial intelligence in medicine 46 (1) 5-17 (2009). [0319] |4| V. Jahmunah. S. L. Oh. V. Rajinikanth. E. J. Ciaccio. K. H. Cheong. U. R. Acharya, et al., “Automated detection of schizophrenia using nonlinear signal processing methods.” Artificial Intelligence in Medicine. Vol. 100.101698 (September 2019). [0320] |5| A. M. Tai. A. Albuquerque. N. E. Carmona. M. Subramanieapillai, D. S. Cha, M. Sheko. Y. Lee. R. Mansur, R. S. McIntyre. “Machine learning and big data: Implications for disease modeling and therapeutic discovery in psychiatry.” Artificial Intelligence in Medicine 101704 (2019). [0321] |6| G. K. Hansson, “Inflammation, atherosclerosis, and coronary artery disease.” New England Journal of Medicine 352 (16) 1685-1695 (2005). [0322] |7| W. G. Members, D. Lloyd-Jones, R. J. Adams, T. Ni. Brown, M. Carnethon. S. Dai. G. De Simone. T. B. Ferguson. E. Ford. K. Furie, et al., “Executive summary: heart disease and stroke statistics 2010 update: a report from the American heart association.” Circulation 121 (7) 948-954 (2010). [0323] |8| G. A. Mensah. D. W. Brown. “An overview of cardiovascular disease burden in the united states.” Health affairs 26 (1) 38-48 (2007). [0324] |9| Y. N. Reddy. A. El-Sabbath R. A. Nishimura. “Comparing pulmonary arterial wedge pressure and left ventricular end diastolic pressure for assessment of left-sided filling pressures.” JAMA cardiology 3 (6) 453-454 (2018). [0325] |10| M. J. Kern. T. Christopher. “Hemodynamic rounds series II: the LVEDP.” Catheterization and cardiovascular diagnosis 44 (1) 70-74 (1998). [0326] |11| J.-H. Park. T. H. Marwick. “Use and limitations of e.e* to assess left ventricular filling pressure by echocardiography,” Journal of cardiovascular ultrasound 19 (4) 169-173 (2011). [0327] |12| S. R. Ommen. R. A. Nishimura. C. P. Appleton. F. Miller. J. K. Oh. M. M. Redfield. A. Tajik. “Clinical utility of doppler echocardiography and tissue doppler imaging in the estimation of left ventricular filling pressures: a comparative simultaneous doppler-catheterization study.” Circulation 102 (15) 1788-1794 (2000). [0328] |13| J. Allen. “Photoplethysmography and its application in clinical physiological measurement.” Physiological measurement 28 (3) RI (2007). [0329] |14| S. D. Fihn. J. M. Gardin. J. Abrams. K. Berra, J. C. Blankenship. A. P. Dallas. P. S. Douglas. J. M. Foody, T. C. Gerber. A. L. Hinderliter, et al., “2012 accf/aha/acp/aats/pcna/scai/sts guideline for the diagnosis and management of patients with stable ischemic heart disease.” Journal of the American College of Cardiology 60 (24) 2564-2603 (2012). [0330] |15| G. N. Levine. E. R. Bates, J. C. Blankenship. S. R. Bailey, J. A. Bittl. B. Cercek. C. E. Chambers. S. G. Ellis. R. A. Guyton. S. M. Hollenberg, et al., “2011 accf/aha/scai guideline for percutaneous coronary intervention: executive summary.” Journal of the American College of Cardiology 58 (24) 2550-2583 (2011). [0331] |16| L. Ni. Mielniczuk. G. A. Lamas. G. C. Flaker. G. Mitchell. S. C. Smith. B. J. Gersh. S. D. Solomon. L. A. Moy′e. J. L. Rouleau. J. D. Rutherford, et al., “Left ventricular end-diastolic pressure and risk of subsequent heart failure in patients following an acute myocardial infarction.” Congestive Heart Failure 13 (4) 209-214 (2007). [0332] |17| J. J. Russo, N. Aleksova. I. Pitcher. E. Couture, S. Parlow, M. Faraz, S. Visintini. T. Simard. P. Di Santo. R. Mathew, et al., “Left ventricular unloading during extracorporeal membrane oxygenation in patients with cardiogenic shock.” Journal of the American College of Cardiology 73 (6) 654-662 (2019). [0333] |18| R. Salem. A. Denault. P. Couture. S. Belisle. A. Fortier. M.-C. Guertin, M. Carrier. R. Martineau. “Left ventricular end-diastolic pressure is a predictor of mortality in cardiac surgery independently of left ventricular ejection fraction.” BJA: British Journal of Anaesthesia 97 (3) 292-297 (2006). [0334] |19| S. H. Strogatz, “Nonlinear dynamics and chaos: with applications to physics, biology, chemistry, and engineering,” CRC Press. (2018). [0335] |20| A. L. Goldberger, D. R. Rigney. B. J. West. “Chaos and fractals in human physiology.” Scientific American 262 (2) 42-49 (1990). [0336] |21| A. L. Goldberger. “Nonlinear dynamics, fractals and chaos: applications to cardiac electrophysiology,” Annals of biomedical engineering 18 (2) 195-198 (1990), [0337] |22| L. Glass, A. Beuter, D. Larocque. “Time delays, oscillations, and chaos in physiological control systems.” Mathematical Biosciences 90 (1-2) 111-125 (1988). [0338] |23| L. Glass. “Synchronization and rhythmic processes in physiology,” Nature 410 (6825) 277 (2001). [0339] |24| M. I. Owis. A. H. Abou-Zied, A.-B. Youssef. Y. M. Kadah. “Study of features based on nonlinear dynamical modeling in ecg arrhythmia detection and classification.” IEEE transactions on Biomedical Engineering 49 (7) 733-736 (2002). [0340] |25| A. Voss. S. Schulz. R. Schroeder, M. Baumert. P. Caminal. “Methods derived from nonlinear dynamics for analysing heart rate variability. Philosophical Transactions of the Royal Society A: Mathematical.” Physical and Engineering Sciences 367 (1887) 277-296 (2008). [0341] |26| L. Glass. P. Hunter. A. McCulloch. “Theory of heart: biomechanics, biophysics, and nonlinear dynamics of cardiac function.” Springer Science & Business Media. (2012). [0342] |27| P. Billingsley. “Ergodic theory and information,” Vol. 1. Wiley New York, 1965, [0343] |28| T. Sauer. J. A. Yorke. M. Casdagli. “Embedology.” Journal of statistical Physics 65 (3-4) 579-616 (1991). [0344] |29| A. Chatterjee. “An introduction to the proper orthogonal decomposition.” Current science 808-817 (2000). [0345] |30| A. Wolf. J. B. Swift. H L. Swinney, J. A. Vastano. “Determining Lyapunov exponents from a time series,” Physica D: Nonlinear Phenomena 16 (3) 285-317 (1985). [0346] |31| A. N. Kolmogorov “Entropy per unit time as a metric invariant of automorphisms,” Doklady of Russian Academy of Sciences. Vol. 124. pp. 754-755 (1959). [0347] |32| P. Grassberger. I. Procaccia. “Estimation of the kolmogorov entropy from a chaotic signal.” Physical review A 28 (4) 2591 (1983), [0348] |33| J. Theiler. “Efficient algorithm for estimating the correlation dimension from a set of discrete points.” Physical review A 36 (9) 4456 (1987). [0349] |34| A. Pikovsky. J. Kurths, M. Rosenblum. J. Kurths. “Synchronization: a universal concept in nonlinear sciences.” Vol. 12. Cambridge university press (2003). [0350] |35| D. Dubin. “Rapid interpretation of EKG's: an interactive course,” Cover Publishing Company (2000). [0351] |36| F. Pedregosa, G. Varoquaux, A. Gramfort. V. Michel. B. Thirion. O. Grisel. M. Blondel. P. Prettenhofer. R. Weiss, V. Dubourg, et al., “Scikit-learn: Machine learning in python.” Journal of machine learning research 12.2825-2830 (October 2011). [0352] |37| T. Chen. C. Guestrin. “Xgboost: A scalable tree boosting system.” Proceedings of the 22nd acm-sigkdd international conference on knowledge discovery and data mining. ACM. pp. 785-794 (2016). [0353] |38| H. Zou, T. Hastie. “Regularization and variable selection via the elastic net.” Journal of the royal statistical society: series B (statistical methodology) 67 (2) 301-320 (2005)