Methods and systems for assessing disease using dynamic analysis of cardiac and photoplethysmographic signals

Abstract

The illustrative methods and systems facilitate one or more dynamic analyses that can characterize and identify synchronization between acquired cardiac signals and photoplethysmography signals to predict / estimate the presence, location, and / or severity of abnormal cardiovascular conditions or diseases, including, for example, but not limited to, coronary artery disease, heart failure (including but not limited to indicators of disease or condition, such as abnormal left ventricular end-diastolic pressure disease), and pulmonary hypertension. In some embodiments, statistical properties of the synchronization between the cardiac signal and the photoplethysmography signal are evaluated. In some embodiments, statistical properties of a histogram of synchronization between the cardiac signal and the photoplethysmography signal are evaluated. In some embodiments, statistical and / or geometric properties of a Poincare map of synchronization between the cardiac signal and the photoplethysmography signal are evaluated.

Description

[0001] This International PCT Application claims priority to and the benefit of U.S. Provisional Patent Application No. 62 / 863,005, filed on June 18, 2019, entitled “Method and System to Assess Disease Using Dynamical Analysis of Cardiac and Photoplethysmographic Signals,” and U.S. Provisional Patent Application No. 62 / 862,991, filed on June 18, 2019, entitled “Method and System to Assess Disease Using Dynamical Analysis of Biophysical Signals,” each of which is incorporated herein by reference in its entirety. Technical Field

[0002] The present disclosure generally relates to non-invasive methods and systems for characterizing one or more physiological systems and their associated functions, activities, and abnormalities. More specifically, in one aspect, the present disclosure relates to non-invasive methods that utilize cardiac measurements and photoplethysmography-related measurements, alone or in combination with other types of measurements of physiological phenomena and systems, to predict and / or detect the presence, severity, and / or location of cardiovascular, pulmonary, and cardiopulmonary diseases, progression, or conditions, and the like. Background Art

[0003] As described in more detail below, the term "biophysical signal" encompasses any physiological signal from which information can be derived. Without wishing to be limiting, a biophysical signal is characterized in part by the energy form employed by the signal (e.g., electrical, acoustic, chemical, thermal, magnetic, optical, etc.), one or more physiological systems (e.g., circulatory / cardiovascular, neural, respiratory, etc.) that are the source of the signal and / or are associated with the signal, the associated organ system, tissue type, cell type, cellular components such as organelles, and the like, including combinations thereof. Biophysical signals can be acquired passively, actively, or both.

[0004] Biophysical signals are typically acquired in conjunction with or through invasive or minimally invasive techniques (e.g., catheterization) and / or the use of radiation (e.g., nuclear imaging), exercise / stress (e.g., treadmill or nuclear stress testing), and / or the administration of pharmacological and / or other agents (e.g., vasodilators, contrast agents). These various approaches can moderately or significantly increase the cost of acquiring such signals, as they may require administration in specialized settings, often using expensive equipment that requires patient travel, and sometimes even requires an overnight stay in a hospital or hotel, for example. Some of these approaches increase the patient's risk of side effects, such as infection or allergic reactions. Some expose the patient to undesirable doses of radiation. And, in the case of exercise or treadmill testing, for example, they can trigger moderate to severe adverse events (e.g., myocardial infarction) that would otherwise not occur. Furthermore, these various approaches often increase the time required to determine the health, disease, or condition of the patient whose biophysical signals are being characterized, often requiring weeks or months for patients who already have or are at risk of moderate or even severe health conditions. This can lead to decreased productivity and increased overall healthcare costs for society. Such delays can also cause emotional trauma to patients (which itself can be detrimental to their health), their families, friends, and other caregivers who cater to the patients’ needs.

[0005] Thus, it would be desirable to obtain information from biophysical signals that minimizes or even eliminates the need to use invasive and / or minimally invasive techniques, radiation, exercise / stress, and / or the use of pharmacological and / or other agents so that the assessment (e.g., prediction and / or detection) of the presence, severity, and (in some cases) location of various diseases, pathologies, or conditions in mammalian or non-mammalian organisms can be accomplished more safely, at a lower cost, and / or in a shorter time than current methods and systems.

[0006] The methods and systems described herein address this need and can be used for a variety of clinical and even research needs in a variety of settings—from the hospital to the emergency room, laboratory, battlefield, or remote settings, at the point of care with the patient's primary care physician or other caregiver, or even at home. Without limitation, the following description provides example methods and systems for use in the context of cardiac or cardiovascular-related disease states and conditions, particularly various forms of pulmonary hypertension (PH), various forms of coronary artery disease (CAD), and various forms of heart failure. Summary of the Invention

[0007] Example methods and systems facilitate one or more dynamic analyses that can characterize and identify synchronization between acquired cardiac signals and photoplethysmography signals to predict and / or estimate the presence, severity, and / or location (if applicable) of abnormal cardiovascular conditions or diseases, including, but not limited to, coronary artery disease, abnormal left ventricular end-diastolic pressure (LVEDP), pulmonary hypertension and its subtypes, heart failure (HF), and the like, as discussed herein. In some embodiments, statistical properties of synchronization between one or more cardiac signals and one or more photoplethysmography signals are evaluated. In some embodiments, statistical properties of a histogram of synchronization between one or more cardiac signals and one or more photoplethysmography signals are evaluated. In some embodiments, statistical properties and / or geometric properties of a Poincare map of synchronization between one or more cardiac signals and one or more photoplethysmography signals are evaluated. One or more cardiac signals and one or more photoplethysmography signals are synchronously acquired (interchangeably herein with the term "simultaneously acquired") for various synchronization assessments disclosed herein.

[0008] The terms "synchrony" and "synchronization" refer to the physiological relationship between one or more signals of a first modality (e.g., a cardiac signal) and a signal of a second modality (e.g., a photoplethysmography signal). For example, cardiac electrical activity detected by electrodes or sensors of a measurement system stimulates muscles, causing the left ventricle to eject oxygenated blood into the body. Some of this blood then flows to the fingertip, where its oxygenation level is measured by one or more photoplethysmography sensors. The time lag between maximum left ventricular electrical activity (e.g., corresponding to the R-peak in the cardiac signal) and peak fingertip oxygenation can be defined as the "pulse transit time" (PTT), a time measure. Because the physiological synchronization between cardiac electrical activity (e.g., measured using cardiac biopotential signals) and pulsatile blood oxygenation (e.g., measured using PPG signals) can vary, PTT can vary from beat to beat. The Poincare synchronization techniques and corresponding features disclosed herein characterize, among other things, variations in synchronization.

[0009] The term "simultaneous acquisition" means that a data point acquired from a first modality (e.g., a channel of a cardiac signal) at time n has a corresponding data point from a second modality (e.g., a channel of a photoplethysmography signal), or even a third or more modalities at time n. The timing or temporal accuracy of such multimodal signal acquisition is typically dictated by, for example, signal acquisition device circuitry, firmware, etc. In the embodiments disclosed herein, signals acquired from different modalities are acquired with high temporal accuracy (e.g., minimal temporal skew). In some embodiments, simultaneous signal acquisition / data point acquisition for different modalities is performed, for example, via one or more circuits located within a single integrated hardware component or signal acquisition device, or even within a single printed circuit board or component thereof. In other embodiments, simultaneous signal acquisition / data point acquisition for different modalities is performed via one or more circuits located on different signal acquisition devices with a common / shared clock, signal acquisition trigger, and / or other components. Furthermore, various configurations of circuitry, other hardware, and accessories (e.g., leads, electrodes, PPG sensors, etc.) within and between signal acquisition devices can achieve this temporal accuracy.

[0010] As used herein, "cardiac signal" refers to one or more signals related to the structure, function, and / or activity of the cardiovascular system, including the electrical / electrochemical conduction aspects of such signals, which, for example, cause myocardial contraction. In some embodiments, the cardiac signal may include electrocardiographic signals, such as those acquired via an electrocardiogram (ECG), or signals of other modalities.

[0011] As used herein, "photoplethysmography" refers to a signal waveform acquired from an optical sensor that corresponds to measured changes in the absorption of light by oxygenated and deoxygenated hemoglobin, such as light having wavelengths in the red and infrared spectrums. In some embodiments, the photoplethysmography signal comprises a raw signal acquired by a pulse oximeter or photoplethysmography (PPG). In some embodiments, the photoplethysmography signal is acquired from custom or dedicated equipment or circuitry (including off-the-shelf devices) configured to acquire such signal waveforms for use in diagnosing a disease or abnormal condition. Photoplethysmography signals typically include red photoplethysmography signals (e.g., electromagnetic signals in the visible light spectrum primarily having wavelengths of approximately 625 to 740 nanometers) and infrared photoplethysmography signals (e.g., electromagnetic signals extending from the nominal red edge of the visible light spectrum up to approximately 1 mm), although different combinations of other light spectrums (e.g., near-infrared, blue, and green) may be used, depending on the type and / or mode of PPG employed.

[0012] "Biophysical signals" are not limited to cardiac signals, neurological signals, or photoplethysmography signals, but encompass any physiological signal from which information can be derived. Without limitation to the examples, biophysical signals can be divided into types or categories that may include, for example, electrical signals (e.g., certain cardiac and nervous system-related signals that can be observed, identified, and / or quantified by techniques such as measuring voltage / potential, impedance, resistivity, conductivity, current, etc., in various domains such as the time domain and / or frequency domain), magnetic signals, electromagnetic signals, optical signals (e.g., signals that can be observed, identified, and / or quantified by techniques such as reflectance, interference, spectroscopy, absorbance, transmittance, visual observation, photoplethysmography, etc.), acoustic signals, chemical signals, mechanical signals (e.g., signals related to fluid flow, pressure, motion, vibration, displacement, strain), thermal signals, and electrochemical signals (e.g., signals that can be associated with the presence of certain analytes such as glucose). In some cases, biophysical signals may be described in the context of physiological systems (e.g., respiratory, circulatory (cardiovascular, pulmonary), neural, lymphatic, endocrine, digestive, excretory, muscular, skeletal, renal / urinary / excretory, immune, epidermal / exocrine, and reproductive) or organ systems (e.g., signals that may be unique to the heart and lungs when they work together), or they may be described in the context of tissues (e.g., muscle, adipose, neural, connective tissue, bone), cells, organelles, molecules (e.g., water, proteins, fats, carbohydrates, gases, free radicals, inorganic ions, minerals, acids and other compounds, elements, and their subatomic components). Unless otherwise specified, the term "biophysical signal acquisition" generally refers to any passive or active means of acquiring biophysical signals from a physiological system (e.g., a mammalian or non-mammalian organism). Passive and active biophysical signal acquisition generally refers to the observation of natural or induced electrical, magnetic, optical, and / or acoustic emissions from body tissues. Non-limiting examples of passive and active biophysical signal acquisition means include, for example, observing the natural emissions of body tissue in the form of voltage / potential, current, magnetism, light, sound, and other non-active means, and in some cases, inducing such emissions. Non-limiting examples of passive and active biophysical signal acquisition means include, for example, ultrasound, radio waves, microwaves, infrared and / or visible light (e.g., for pulse oximetry or photoplethysmography), visible light, ultraviolet light, and other active means of interrogating body tissue that do not involve ionizing energy or radiation (e.g., X-rays). Active biophysical signal acquisition may involve excitation-emission spectroscopy (including, for example, excitation-emission fluorescence). Active biophysical signal acquisition may also involve transmitting ionizing energy or radiation (e.g., X-rays) (also referred to as "ionizing biophysical signals") to body tissue.Passive and active biophysical signal acquisition approaches can be performed in conjunction with invasive procedures (e.g., via surgery or invasive radiological intervention protocols) or can be performed non-invasively (e.g., via imaging).

[0013] The methods and systems described in various embodiments herein are not limited thereto and may be used in any context involving another physiological system or a living body's system, organ, tissue, cell, etc. By way of example only, two types of biophysical signals that may be used in a cardiovascular context include cardiac signals that may be acquired by conventional electrocardiogram (ECG / EKG) equipment, bipolar broadband biopotential (cardiac) signals that may be acquired from other devices (such as those described herein), and signals that may be acquired by various volumetric descriptive techniques such as photoplethysmography.

[0014] In the context of the present disclosure, techniques are described for acquiring and analyzing biophysical signals, particularly for diagnosing the presence, location (if applicable), and / or severity of certain disease states or conditions in, associated with, or affecting the cardiovascular (or cardiac) system, including, for example, pulmonary hypertension, coronary artery disease, and heart failure (e.g., left-sided or right-sided heart failure).

[0015] Pulmonary hypertension, heart failure, and coronary artery disease are three diseases / conditions related to the cardiovascular or cardiac system. Pulmonary hypertension (PH) generally refers to high blood pressure in the arteries of the lungs and can encompass a range of conditions. PH often has a complex and multifactorial etiology and has an insidious clinical onset of varying severity. PH may develop complications such as right heart failure and, in many cases, is fatal. The World Health Organization (WHO) classifies PH into five groups or types. The first PH group in the WHO classification is pulmonary arterial hypertension (PAH). PAH is a chronic disease with no current cure that, among other things, causes the walls of the pulmonary arteries to tighten and harden. PAH requires at least cardiac catheterization for diagnosis. PAH is characterized by pulmonary artery vasculopathy, defined during cardiac catheterization as a mean pulmonary artery pressure of 25 mmHg or higher. One form of pulmonary hypertension is called idiopathic pulmonary hypertension—PAH that occurs without a clear cause. In addition, PAH subtypes include hereditary PAH, drug- and toxin-induced PAH, and PAH associated with other systemic diseases (such as connective tissue diseases, HIV infection, portal hypertension, and congenital heart disease). PAH encompasses all causes of narrowing of the pulmonary vascular structures. In PAH, the gradual narrowing of the pulmonary artery bed is due to an imbalance in vasoactive mediators (including prostacyclin, nitric oxide, and endothelin-1). This can lead to increased right ventricular afterload, right heart failure, and premature death. The second group of PH, according to the WHO classification, is pulmonary hypertension due to left-sided heart disease. This group is typically characterized by problems on the left side of the heart. Over time, these problems lead to changes in the pulmonary arteries. Specific subgroups include left ventricular systolic dysfunction, left ventricular diastolic dysfunction, valvular disease, and congenital cardiomyopathies and obstructions caused by non-valvular conditions. Treatment of group II PH often focuses on the underlying problem (e.g., surgery to replace the heart valve, various medications, etc.). The third group of PH, according to the WHO classification, is large and diverse and is often related to lung disease or hypoxia. Subgroups include chronic obstructive pulmonary disease, interstitial lung disease, sleep-disordered breathing, alveolar hypoventilation disorders, chronic high-altitude exposure, and developmental lung disease. The fourth PH group, classified by the WHO as chronic thromboembolic pulmonary hypertension, is caused when a blood clot enters or forms in the lungs, blocking blood flow through the pulmonary arteries. The fifth PH group, classified by the WHO as a subgroup, includes rare conditions that cause PH, such as hematologic disorders, systemic diseases (such as sarcoidosis affecting the lungs), metabolic disorders, and other conditions. The mechanisms of PH in this fifth group are poorly understood.

[0016] All forms of PH can be difficult to diagnose during a routine physical exam because the most common symptoms of PH (shortness of breath, fatigue, chest pain, edema, palpitations, and dizziness) are associated with many other conditions. Blood tests, chest X-rays, electrocardiograms and echocardiograms, pulmonary function tests, exercise tolerance tests, and nuclear scans are all widely used to help physicians diagnose specific forms of PH. As mentioned above, the "gold standard" for diagnosing PH, and particularly PAH, is cardiac catheterization of the right side of the heart to directly measure pressure in the pulmonary arteries. If PAH is suspected, one of several investigations may be performed to confirm the condition, such as an electrocardiogram, chest x-ray, and pulmonary function tests. An electrocardiogram will often reveal right-sided heart strain, and a chest x-ray will show evidence of significant pulmonary artery or cardiomegaly. However, a normal electrocardiogram and chest x-ray do not rule out the diagnosis of PAH. Further testing may be necessary to confirm the diagnosis and determine the cause and severity. For example, blood tests, exercise testing, and overnight oximetry testing may be performed. Imaging tests may also be performed. Examples of imaging tests include isotope perfusion lung scanning, high-resolution computed tomography, computed tomography pulmonary angiography, and magnetic resonance pulmonary angiography. If these (and possibly other) non-invasive tests support a diagnosis of PAH, right heart catheterization is typically required to confirm the diagnosis by directly measuring pulmonary pressures. This also allows for measurement of cardiac output and estimation of left atrial pressure using pulmonary artery wedge pressure. While non-invasive techniques exist to determine the presence of PAH in a subject, these techniques cannot reliably confirm a diagnosis of PAH unless invasive right heart catheterization is performed. Aspects and embodiments of methods and systems for assessing PH are disclosed in commonly owned U.S. Patent Application No. 16 / 429,593, the entire contents of which are incorporated herein by reference.

[0017] Heart failure affects nearly 6 million people in the United States alone, with more than 870,000 diagnosed annually. The term "heart failure" (sometimes called congestive heart failure or CHF) generally refers to a chronic, progressive condition or progression in which the heart muscle cannot pump enough blood to meet the body's needs due to weakening or stiffening of the heart muscle, or defects that impair normal circulation. This can lead to symptoms such as a back-up of blood and fluid into the lungs, swelling, fatigue, dizziness, fainting, a rapid and / or irregular heartbeat, a dry cough, nausea, and shortness of breath. Common causes of heart failure are coronary artery disease (CAD), high blood pressure, cardiomyopathy, arrhythmias, kidney disease, heart defects, obesity, smoking, and diabetes. Common types of heart failure include diastolic heart failure (DHF), left or left-sided heart failure / disease (also known as left ventricular heart failure), right or right-sided heart failure / disease (also known as right ventricular heart failure), and systolic heart failure (SHF).

[0018] Left-sided heart failure is further divided into two main types: systolic failure (or heart failure with reduced ejection fraction or reduced left ventricular function) and diastolic failure / dysfunction (or heart failure with preserved ejection fraction or preserved left ventricular function). Procedures and techniques commonly used to determine whether a patient has left-sided heart failure include cardiac catheterization, X-rays, echocardiograms, electrocardiograms (EKGs), electrophysiology studies, radionucleotide imaging, and various treadmill tests, including those that measure peak VO2. Ejection fraction (EF) is a measurement expressed as the percentage of blood pumped out by the heart's ventricles (in the case of left-sided heart failure, the left ventricle) with each contraction and is most often obtained noninvasively via echocardiography. A normal left ventricular ejection fraction (LVEF) ranges from approximately 55% to approximately 70%.

[0019] When systolic failure occurs, the left ventricle cannot contract forcefully enough to keep blood circulating throughout the body, preventing it from properly supplying the body with blood. As the left ventricle pumps harder to compensate, it becomes weaker and thinner. As a result, blood flows backward into organs, leading to fluid accumulation in the lungs and / or swelling elsewhere in the body. Echocardiography, magnetic resonance imaging, and nuclear medicine scans (e.g., multi-gated acquisition) are techniques used to noninvasively measure ejection fraction (EF), expressed as the percentage of blood volume pumped by the left ventricle relative to its filling volume, to help diagnose systolic failure. Specifically, a left ventricular ejection fraction (LVEF) value below 55% indicates that the heart's pumping capacity is below normal, and in severe cases, measurements can drop below approximately 35%. Generally, when these LVEF values ​​are below normal, a diagnosis of systolic failure can be made or can help diagnose systolic failure.

[0020] When diastolic heart failure occurs, the left ventricle becomes stiff or thickened, losing its ability to relax properly. This, in turn, means the heart's lower left chamber cannot properly fill with blood. This reduces the amount of blood pumped to the body. Over time, this can cause blood to accumulate in the left atrium and then in the lungs, leading to fluid congestion and heart failure symptoms. In these conditions, LVEF values ​​tend to remain within a normal range. Therefore, other tests, such as invasive catheterization, can be used to measure left ventricular end-diastolic pressure (LVEDP) to aid in the diagnosis of diastolic heart failure and other forms of heart failure with preserved EF. Typically, LVEDP is measured directly by placing a catheter in the left ventricle or indirectly by measuring pulmonary capillary wedge pressure through a catheter placed in the pulmonary artery. Such catheterization techniques, by their nature, increase the risk of infection and other complications for patients and are often expensive. Therefore, non-invasive methods and systems for determining or estimating LVEDP when diagnosing the presence and / or severity of diastolic heart failure and countless other forms of heart failure with preserved EF are desirable. Furthermore, non-invasive methods and systems for diagnosing the presence and / or severity of diastolic heart failure, as well as myriad other forms of heart failure with preserved EF, that do not necessarily include determination or estimation of abnormal LVEDP are desirable. Embodiments of the present disclosure address all of these needs.

[0021] Right-sided heart failure often develops as a result of left-sided heart failure, when a weakened and / or stiff left ventricle loses its ability to effectively pump blood to the rest of the body. As a result, fluid is forced back up through the lungs, weakening the right side of the heart and leading to right-sided heart failure. This backflow can back up in the veins, causing fluid to swell in the legs, ankles, gastrointestinal tract, and liver. In other cases, certain lung diseases (such as chronic obstructive pulmonary disease and pulmonary fibrosis) can lead to right-sided heart failure, even if the left side of the heart is functioning normally. Procedures and techniques commonly used to determine if a patient has left-sided heart failure include blood tests, cardiac CT scans, cardiac catheterization, X-rays, coronary angiography, echocardiograms, electrocardiograms (EKGs), myocardial biopsies, pulmonary function studies, and various forms of stress testing, such as treadmill tests.

[0022] Pulmonary hypertension is closely associated with heart failure. As mentioned above, PAH (first WHO PH group) can lead to increased right ventricular afterload, right heart failure, and premature death. PH caused by left heart failure (second WHO PH group) is considered the most common cause of PH.

[0023] Ischemic heart disease (also known as cardiac ischemia or myocardial ischemia) and related conditions or pathologies can also be assessed or diagnosed using the techniques disclosed herein. Ischemic heart disease is a disease or group of conditions characterized by reduced blood flow to the myocardium, typically caused by coronary artery disease (CAD). CAD is closely related to heart failure and is its most common cause. CAD typically develops when atherosclerosis (hardening or stiffening of the lining and accumulation of plaque within it, often accompanied by abnormal inflammation) develops in the lining of the coronary arteries that supply the myocardium, or heart muscle. Over time, CAD can also weaken the heart muscle and lead to conditions such as angina, myocardial infarction (cardiac arrest), heart failure, and arrhythmias. Arrhythmias are abnormal heart rhythms that can include any variation in the normal sequence of electrical conduction in the heart and, in some cases, can lead to cardiac arrest. The evaluation of PH, heart failure, CAD, and other diseases and / or conditions can be complex and utilizes numerous invasive techniques and tools to assess the presence and severity of these conditions. Furthermore, the commonality among the symptoms of these diseases and / or conditions, as well as the fundamental connection between the respiratory and cardiovascular systems—as they work together to supply oxygen to the body's cells and tissues—represents a complex physiological interconnectedness that can be exploited to improve the detection and ultimately treatment of such diseases and / or conditions. Traditional methods of evaluating these biophysical signals in this setting remain significant challenges in providing healthcare providers with the tools to accurately detect / diagnose the presence or absence of such diseases and conditions.

[0024] For example, in electrocardiography—the field of cardiology in which the heart's electrical activity is analyzed to gain information about its structure and function—it has been observed that significant ischemic heart disease can alter the ventricular conduction properties of the myocardium in the perfusion bed downstream of coronary artery narrowing or occlusion. Pathology can manifest at different locations in the heart and at varying stages of severity, making accurate diagnosis challenging. Furthermore, the conductive properties of the myocardium can vary from person to person, and other factors, such as measurement variability associated with the placement of measurement probes and parasitic losses associated with such probes and their associated components, can also affect the biophysical signals captured during cardiac electrophysiological examinations. Furthermore, when the conductive properties of the myocardium are captured as relatively long cardiac phase gradient signals, they can exhibit complex, nonlinear variability that cannot be effectively captured by traditional modeling techniques.

[0025] As described above, the exemplary methods and systems facilitate one or more dynamic analyses that can characterize and identify synchronization between acquired cardiac signals and photoplethysmography signals to predict and / or estimate the presence, severity, and / or location of abnormal cardiovascular conditions or diseases, including but not limited to coronary artery disease, abnormal left ventricular end-diastolic pressure disease (LVEDP), pulmonary hypertension and its subtypes, heart failure (HF), etc., as discussed herein. In some embodiments, the dynamic features include a statistical or dynamic analysis of the phase relationship between the acquired set of one or more cardiac signals and the acquired set of one or more photoplethysmography signals. In some embodiments, the dynamic features include a statistical or dynamic analysis of the variance between landmarks between the acquired set of one or more cardiac signals and the acquired set of one or more photoplethysmography signals. In some embodiments, the dynamic features include a statistical or dynamic analysis of the variance of landmarks determined in the acquired set of one or more photoplethysmography signals, where the landmarks are defined by the cardiac signals.

[0026] In one aspect, a method for non-invasively assessing a disease state or abnormal condition of a subject is disclosed, the method comprising obtaining, by one or more processors (e.g., from a stored database or from a measurement system), a first biophysical signal dataset associated with saturation of oxyhemoglobin or deoxyhemoglobin of the subject, including a red photoplethysmography signal and an infrared photoplethysmography signal; obtaining, by the one or more processors (e.g., from a stored database or from a measurement system), a second biophysical signal dataset associated with a cardiac signal of the subject (e.g., from a phase space recorder or from an ECG); device); determining, by the one or more processors, one or more dynamic characteristics of synchrony between the first biophysical signal dataset associated with the saturation of the oxyhemoglobin and / or the deoxyhemoglobin and the second biophysical signal dataset associated with the cardiac signal; and determining, by the one or more processors, an estimate of the presence, severity, and / or location (if applicable) of a disease state based on the determined one or more dynamic characteristics of synchrony, wherein the disease state includes the presence, severity, and / or location (if applicable) of coronary artery disease (e.g., significant coronary artery disease) or abnormal left ventricular end-diastolic pressure.

[0027] In some embodiments, the presence and / or severity of a disease or condition can be assessed based on an assessment of left ventricular end-diastolic pressure (LVEDP), including abnormal LVEDP.

[0028] In some embodiments, the disease state or condition comprises significant coronary artery disease.

[0029] In some embodiments, the disease state or condition comprises pulmonary hypertension.

[0030] In some embodiments, the disease state or condition comprises pulmonary arterial hypertension (PAH).

[0031] In some embodiments, the disease state or condition comprises pulmonary hypertension due to left heart disease.

[0032] In some embodiments, the disease state or condition comprises a rare disease that causes pulmonary hypertension.

[0033] In some embodiments, the disease state or condition comprises left ventricular heart failure or left-sided heart failure.

[0034] In some embodiments, the disease state or condition comprises right ventricular heart failure or right-sided heart failure.

[0035] In some embodiments, the disease state or condition comprises systolic heart failure (SHF).

[0036] In some embodiments, the disease state or condition comprises diastolic heart failure.

[0037] In some embodiments, the disease state or condition comprises ischemic heart disease.

[0038] In some embodiments, the disease state or condition comprises a cardiac arrhythmia.

[0039] In some embodiments, the method further includes determining, by the one or more processors, one or more second estimates of the presence, location, and / or severity of the two or more diseases or conditions.

[0040] In some embodiments, the synchronization dynamics of the first and second biophysical signal data sets (e.g., Example PM#1) include a statistical evaluation of cardiac signal values ​​at landmarks defined by both the red photoplethysmography signal and the infrared photoplethysmography signal.

[0041] In some embodiments, a landmark defined by both the red photoplethysmography signal and the infrared photoplethysmography signal is defined at a time when values ​​of the red photoplethysmography signal and the infrared photoplethysmography signal intersect.

[0042] In some embodiments, the synchronized dynamic characteristics of the first biophysical signal dataset and the second biophysical signal dataset (e.g., according to Example #2) include a statistical evaluation of values ​​of one of the red photoplethysmography signal and the infrared photoplethysmography signal at landmarks defined in the cardiac signal.

[0043] In some embodiments, the landmark defined in the cardiac signal comprises a correlation peak associated with ventricular depolarization.

[0044] In some embodiments, the landmarks defined in the cardiac signal include correlation peaks associated with ventricular repolarization or atrial depolarization.

[0045] In some embodiments, the dynamic characteristics of the synchrony of the first biophysical signal dataset and the second biophysical signal dataset (e.g., Example #3) include a statistical evaluation of the time intervals between i) a first set of landmarks and ii) a second set of landmarks, wherein the first set of landmarks is defined between the red photoplethysmography signal and the infrared photoplethysmography signal, and the second set of landmarks is defined in the cardiac signal.

[0046] In some embodiments, the second set of landmarks defined in the cardiac signal includes correlation peaks in the cardiac signal associated with ventricular depolarization.

[0047] In some embodiments, the second set of landmarks defined in the cardiac signal includes correlation peaks in the cardiac signal associated with ventricular repolarization or atrial depolarization.

[0048] In some embodiments, a first set of landmarks defined by both the red photoplethysmography signal and the infrared photoplethysmography signal is defined at times when values ​​of the red photoplethysmography signal and the infrared photoplethysmography signal intersect.

[0049] In some embodiments, the synchronized dynamic characteristics of the first biophysical signal dataset and the second biophysical signal dataset (e.g., Example #4) include a statistical evaluation of a phase relationship between a period of one of the red photoplethysmography signal and the infrared photoplethysmography signal and a period of the cardiac signal.

[0050] In some embodiments, the method further includes causing, by the one or more processors, generation of a visualization of an estimate of the presence, severity, and / or localization (if applicable) of the disease state, wherein the generated visualization is rendered and displayed on a display of a computing device (e.g., a computing workstation; a surgical, diagnostic, or instrumentation device) and / or presented in a report (e.g., an electronic report).

[0051] In some embodiments, the method further includes determining, by the one or more processors, a histogram of the synchrony of the first biophysical signal dataset and the second biophysical signal dataset; 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, skewness, and kurtosis, wherein the extracted first statistical parameter is used to determine the estimate of the presence, severity, and / or location (if applicable) of the disease state.

[0052] In some embodiments, the method further includes determining, by the one or more processors, a Poincare map of the synchrony of the first biophysical signal dataset and the second biophysical signal dataset; and extracting a second statistical parameter of the Poincare map, wherein the second statistical parameter of the histogram is selected from the group consisting of mean, mode, median, skewness, and kurtosis, wherein the extracted second statistical parameter is used to determine the estimate of the presence, severity, and / or location (if applicable) of the disease state.

[0053] In some embodiments, the method further includes determining, by the one or more processors, a Poincare map of the synchrony of the first biophysical signal dataset and the second biophysical signal dataset; and extracting geometric properties of ellipses (e.g., major / minor diameters of the ellipses, angles of the ellipses) of the clusters fitted to the Poincare map, wherein the extracted geometric properties of the ellipses are used to determine the estimate of the presence, severity, and / or location (if applicable) of the disease state.

[0054] In some embodiments, the Poincare map is generated by iteratively plotting a parameter associated with the synchrony of the first biophysical signal dataset and the second biophysical signal dataset at index x and index x+1 on the x-axis, and iteratively plotting the parameter at index x and index x-1 on the y-axis.

[0055] In some embodiments, the parameter is a time interval between a landmark of the cardiac signal (eg, an R-peak) and a crossing point between the red photoplethysmography signal and the infrared photoplethysmography signal.

[0056] In some embodiments, the parameter is an amplitude signal value of the cardiac signal at a crossing landmark defined between the red photoplethysmography signal and the infrared photoplethysmography signal.

[0057] In some embodiments, the parameter is an amplitude signal value of the photoplethysmography signal at a landmark defined in the cardiac signal.

[0058] In another aspect, a system (e.g., for non-invasively assessing a disease state or abnormal symptom in a subject) is disclosed, the system comprising a processor; and a memory having instructions stored thereon, wherein execution of the instructions by the processor causes the processor to: obtain (e.g., from a stored database or from a measurement system) a first biophysical signal dataset associated with saturation of oxyhemoglobin or deoxyhemoglobin for the subject, including a red photoplethysmography signal and an infrared photoplethysmography signal; obtain (e.g., from a stored database or from a measurement system) a second biophysical signal dataset associated with a cardiac signal for the subject (e.g., from a phase space recorder or from an ECG acquisition); determine one or more synchronized dynamic characteristics between the first biophysical signal dataset associated with saturation of the oxyhemoglobin and / or deoxyhemoglobin and the second biophysical signal dataset associated with the cardiac signal; and determine an estimate of the presence of a disease state (e.g., wherein the disease state comprises coronary artery disease (e.g., significant coronary artery disease) or a disease or condition associated with abnormal left ventricular end-diastolic pressure) based on the determined one or more synchronized dynamic characteristics.

[0059] In some embodiments, the synchronization dynamics of the first biophysical signal dataset and the second biophysical signal dataset (e.g., Example PM#1) include a statistical evaluation of cardiac signal values ​​at landmarks defined by both the red photoplethysmography signal and the infrared photoplethysmography signal.

[0060] In some embodiments, a landmark defined by both the red photoplethysmography signal and the infrared photoplethysmography signal is defined at a time when values ​​of the red photoplethysmography signal and the infrared photoplethysmography signal intersect.

[0061] In some embodiments, the synchronized dynamic characteristics of the first biophysical signal dataset and the second biophysical signal dataset (e.g., according to Example #2) include a statistical evaluation of values ​​of one of the red photoplethysmography signal and the infrared photoplethysmography signal at landmarks defined in the cardiac signal.

[0062] In some embodiments, the landmark defined in the cardiac signal comprises a correlation peak associated with ventricular depolarization.

[0063] In some embodiments, the landmarks defined in the cardiac signal include correlation peaks associated with ventricular repolarization or atrial depolarization.

[0064] In some embodiments, the synchronized dynamic characteristics of the first biophysical signal dataset and the second biophysical signal dataset (e.g., Example #3) include a statistical evaluation of time intervals between i) a first set of landmarks defined between the red photoplethysmography signal and the infrared photoplethysmography signal and ii) a second set of landmarks defined in the cardiac signal.

[0065] In some embodiments, the second set of landmarks defined in the cardiac signal includes correlation peaks in the cardiac signal associated with ventricular depolarization.

[0066] In some embodiments, the second set of landmarks defined in the cardiac signal includes correlation peaks in the cardiac signal associated with ventricular repolarization or atrial depolarization.

[0067] In some embodiments, a first set of landmarks defined by both the red photoplethysmography signal and the infrared photoplethysmography signal is defined at times when values ​​of the red photoplethysmography signal and the infrared photoplethysmography signal intersect.

[0068] In some embodiments, the synchronized dynamics of the first biophysical signal dataset and the second biophysical signal dataset (e.g., Example #4) includes a statistical evaluation of a phase relationship between a period of one of the red photoplethysmography signal and the infrared photoplethysmography signal and a period of the cardiac signal.

[0069] In some embodiments, execution of the instructions by the processor further causes the processor to generate a visualization of the estimate for the presence of the disease state, wherein the generated visualization is rendered and displayed on a display of a computing device (e.g., a computing workstation; a surgical, diagnostic, or instrumentation device) and / or presented in a report (e.g., an electronic report).

[0070] In some embodiments, execution of the instructions by the processor further causes the processor to determine a histogram of the synchrony of the first biophysical signal dataset and the second biophysical signal dataset; and extract 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, skewness, and kurtosis, wherein the extracted first statistical parameter is used to determine the estimate for the presence of the disease state.

[0071] In some embodiments, execution of the instructions by the processor further causes the processor to determine a Poincare map of the synchrony of the first biophysical signal dataset and the second biophysical signal dataset; and extract a second statistical parameter of the Poincare map, wherein the second statistical parameter of the histogram is selected from the group consisting of mean, mode, median, skewness, and kurtosis, wherein the extracted second statistical parameter is used to determine the estimate for the presence of the disease state.

[0072] In some embodiments, execution of the instructions by the processor further causes the processor to determine a Poincare map of the synchrony of the first biophysical signal dataset and the second biophysical signal dataset; and to extract geometric properties of ellipses (e.g., major / minor diameters of the ellipses, angles of the ellipses) of the clusters fitted into the Poincare map, wherein the extracted geometric properties of the ellipses are used to determine the estimate for the presence of the disease state.

[0073] In some embodiments, the Poincare map is generated by iteratively plotting a parameter associated with the synchrony of the first biophysical signal dataset and the second biophysical signal dataset at index x and index x+1 on the x-axis, and iteratively plotting the parameter at index x and index x-1 on the y-axis.

[0074] In some embodiments, the parameter is a time interval between a landmark of the cardiac signal (eg, an R-peak) and a crossing point between the red photoplethysmography signal and the infrared photoplethysmography signal.

[0075] In some embodiments, the parameter is an amplitude signal value of the cardiac signal at a crossing landmark defined between the red photoplethysmography signal and the infrared photoplethysmography signal.

[0076] In some embodiments, the parameter is an amplitude signal value of the photoplethysmography signal at a landmark defined in the cardiac signal.

[0077] In some embodiments, the system further comprises a measurement system configured to acquire one or more photoplethysmography signals.

[0078] In some embodiments, the system further comprises a measurement system configured to acquire one or more cardiac signals.

[0079] In some embodiments, the system further comprises a first measurement system configured to acquire one or more photoplethysmography signals and a second measurement system configured to acquire one or more cardiac signals.

[0080] In another aspect, a system is disclosed, comprising a processor; and a memory having instructions stored therein, wherein execution of the instructions by the processor causes the processor to perform any of the above methods.

[0081] In another aspect, a computer-readable medium having instructions stored therein is disclosed, wherein execution of the instructions by a processor causes the processor to perform any of the above methods. 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 method and system.

[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 provided for illustrative purposes only, illustrate novel and non-obvious aspects of the present invention. The accompanying drawings include the following figures:

[0084] Figure 1 is a schematic diagram of an example system configured to non-invasively assess dynamic characteristics of a physiological system to predict and / or estimate the presence, severity, and / or location (as applicable) of a disease or condition in such physiological system, or an indicator indicative of one of the foregoing, in accordance with an exemplary embodiment.

[0085] Figure 2A and 2B According to an exemplary embodiment, Figure 1 Example of a photoplethysmography signal acquired by a measurement system.

[0086] Figure 2C A method for collecting Figure 2A Example sensor configuration for a photoplethysmography signal.

[0087] Figure 2D According to another exemplary embodiment, a method for collecting Figure 2A Another example sensor configuration for a photoplethysmography signal.

[0088] Figure 2E and 2F The high frequency noise is removed. Figure 2A The power spectral density of the photoplethysmography signal.

[0089] Figure 3A According to an exemplary embodiment, Figure 1 Example cardiac signals (e.g., biopotential signals) are example biophysical signals acquired by the measurement system.

[0090] 3B is an example measurement system, such as the system depicted in FIG. 1 , configured to non-invasively measure biophysical signals to be used to assess dynamic properties of a physiological system to predict and / or estimate the presence, severity, and / or location (if applicable) of a disease or condition in such physiological system, or an indicator indicative of one of the foregoing, in accordance with an exemplary embodiment.

[0091] Figure 3C According to an exemplary embodiment, Figure 3B Example use of the measurement system on patients in a clinical setting.

[0092] Figure 3D According to an exemplary embodiment Figure 3B Example placement of surface electrodes of a measurement system on the patient's chest and back to collect Figure 3A Diagram of heart signals.

[0093] Figure 4A A three-dimensional phase space diagram of a photoplethysmography signal acquired via an infrared sensor according to an exemplary embodiment is shown.

[0094] Figure 4B According to an exemplary embodiment, Figure 4A A two-dimensional projection of the same data.

[0095] Figure 5A 、 5B 5C illustrate example dynamic characteristics of synchronization between acquired photoplethysmography signals and cardiac signals according to an exemplary embodiment.

[0096] Figure 5D It is shown that according to an exemplary embodiment, Figure 5C Example fitted ellipse features extracted from the Poincare map.

[0097] Figure 5E An example Poincare map is shown for a data set acquired from CAD-negative patients (ie, patients without CAD).

[0098] Figure 5F An example Poincare map is shown for a data set collected from CAD-positive patients (ie, patients with some form of CAD).

[0099] Figure 5G and 5H Further shown is a crossing landmark on three cardiac signals acquired via a phase space recorder, which may be used to trigger analysis of the photoplethysmography signal in a Poincare map, in accordance with an exemplary embodiment.

[0100] Figure 5I 、 5J5K and 5L illustrate another example dynamic characteristic of synchronization between an acquired photoplethysmography signal and a cardiac signal according to an exemplary embodiment.

[0101] Figure 6A 、 6B 6C illustrate another set of example dynamic characteristics of synchrony between acquired photoplethysmography signals and cardiac signals according to an exemplary embodiment.

[0102] Figure 6D Histogram and Poincare map results for CAD-negative patients are shown according to an exemplary embodiment.

[0103] Figure 6E Shown are histogram and Poincare map results for CAD-positive patients according to an exemplary embodiment.

[0104] Figure 7A 、 7B 7C illustrate yet another example dynamic characteristic of synchronization between an acquired photoplethysmography signal and a cardiac signal according to an exemplary embodiment.

[0105] Figure 7D Histogram and Poincare map results for CAD-negative patients are shown according to an exemplary embodiment.

[0106] Figure 7E Shown are histogram and Poincare map results for CAD-positive patients according to an exemplary embodiment.

[0107] Figure 7F 、 7G 7H and 7I illustrate yet another example dynamic characteristic of synchronization between an acquired photoplethysmography signal and a cardiac signal according to an exemplary embodiment.

[0108] Figure 8A 、 8B 8C illustrate another set of example dynamic characteristics of synchrony between acquired photoplethysmography signals and cardiac signals according to an exemplary embodiment.

[0109] Figure 8D An analysis of a phase difference generated between an acquired infrared photoplethysmography signal and an acquired cardiac signal for a CAD-negative patient is shown in accordance with an exemplary embodiment.

[0110] Figure 8E An analysis of a phase difference generated between an acquired infrared photoplethysmography signal and an acquired cardiac signal for a CAD-positive patient is shown in accordance with an exemplary embodiment.

[0111] Figure 9Experimental results from a study according to an exemplary embodiment are shown that indicate the clinical predictive value of specific dynamic features extracted from Poincare and phase analysis of photoplethysmography signals and cardiac signals, which clinical predictive value indicates the presence or absence of a disease or abnormal condition or an indicator of one of the above.

[0112] Figure 10A and 10B ROC curves for significant CAD classification using the trained elastic net model for two datasets are shown, respectively, according to an exemplary embodiment.

[0113] Figure 10C and 10D ROC curves for elevated or abnormal LVEDP classification using the trained XGBoost model on two datasets are shown, respectively, according to an exemplary embodiment.

[0114] Figure 10E and 10F ROC curves are shown for correctly classifying the presence of significant CAD and elevated or abnormal LVEDP, respectively, using an elastic net model subsequently trained using only Poincare map-based features and a larger training dataset, according to an exemplary embodiment.

[0115] Figure 10G and 10H Shown respectively Figure 10E and 10F Feature contribution of the classifier model for CAD and LVEDP classification.

[0116] Figures 11A-11F Experimental results of training a classifier to predict elevated LVEDP according to an exemplary embodiment are shown.

[0117] Figure 12 An example computing environment is shown in which example embodiments of the analysis system and aspects thereof may be implemented. DETAILED DESCRIPTION

[0118] Each and every feature described herein and each combination of two or more of these features is included within the scope of the present invention, provided that the features included in this combination are not mutually inconsistent.

[0119] While the present disclosure relates to the beneficial evaluation of biophysical signals (e.g., raw or pre-processed photoplethysmography signals, cardiac signals, etc.) in the diagnosis and treatment of cardiac-related pathologies and conditions, such evaluations may also be applied to the diagnosis and treatment (including surgical, minimally invasive, and / or pharmacological treatments) of any pathology or condition of any relevant system in a living organism in which the biophysical signals are involved. In the cardiac (or cardiovascular) setting, the evaluations may be applied to the diagnosis and treatment of coronary artery disease (CAD) and diseases and / or conditions associated with abnormal left ventricular end-diastolic pressure (LVEDP). The evaluations may be applied to the diagnosis and treatment of any number of therapies (alone or in combination), such as placement of stents in coronary arteries, performing atherectomy, angioplasty, prescribing medications, and / or prescribing exercise, nutritional and other lifestyle changes, and the like. Other cardiac-related pathologies or conditions that can be diagnosed include, for example, arrhythmias, congestive heart failure, valvular failure, pulmonary hypertension (e.g., pulmonary hypertension, pulmonary hypertension due to left heart disease, pulmonary hypertension due to lung disease, pulmonary hypertension due to chronic thrombosis, and pulmonary hypertension due to other diseases (e.g., blood or other disorders), and other cardiac-related pathologies, conditions, and / or diseases. In some embodiments, the assessment can be applied to neurological-related pathologies and conditions. Non-limiting examples of neurological-related diseases, pathologies or conditions that may be diagnosed include, for example, epilepsy, schizophrenia, Parkinson's disease, Alzheimer's disease (and all other forms of dementia), autism spectrum disorders (including Asperger's syndrome), attention deficit hyperactivity disorder, Huntington's disease, muscular dystrophy, depression, bipolar disorder, brain / spinal cord tumors (malignant and benign), movement disorders, cognitive disorders, language disorders, various psychiatric disorders, brain / spinal cord / nerve injuries, chronic traumatic encephalopathy, cluster headaches, migraines, neuropathies (various forms, including peripheral neuropathy), phantom limb / pain, chronic fatigue syndrome, acute and / or chronic pain (including back pain, failed back surgery syndrome, etc.), movement dysfunction, anxiety disorders, conditions caused by infections or exogenous factors (such as Lyme disease, encephalitis, rabies), narcolepsy and other sleep disorders, post-traumatic stress disorder, neurological conditions / effects associated with stroke, aneurysm, 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 the reference list and discussed in the disclosure provided herein. Citation and / or discussion of such references are provided solely to clarify the description of the disclosed technology, and it should not be considered that any such reference is "prior art" with respect to any aspect 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, "Scikit-learn: Machine learning in Python" by 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)). All references cited and discussed in this application document are incorporated herein by reference in their entirety to the same extent as if each reference were incorporated by reference individually.

[0121] Example System

[0122] Figure 1 is a schematic diagram of an example system 100 configured to non-invasively assess dynamic characteristics of a physiological system to predict and / or estimate (e.g., determine) the presence, location, and / or severity of a disease or condition, or an indicator indicative thereof, in such physiological system, according to an exemplary embodiment. As used herein, the term "predict" refers to foretelling a future event (e.g., the potential development of a disease or condition), while the term "estimate" may refer to quantifying some metric based on available information, such as the presence, location, and / or severity of a disease or condition, or an indicator indicative thereof. The operations of predicting and estimating may generally be referred to as determining.

[0123] As used herein, "physiological system" may refer to the cardiovascular system, pulmonary system, renal system, nervous system, and other functional systems and subsystems of the body. In the context of the cardiovascular system, system 100 facilitates studying the complex, nonlinear dynamics of the heart over many cardiac cycles.

[0124] exist Figure 1In the present invention, a non-invasive measurement system 102 (shown as “measurement system” 102 ) acquires two or more sets of biophysical signals (shown as sets 104 a and 104 b ) from a subject 108 (shown at locations 108 a and 108 b ) via measurement probes 106 (shown as probes 106 a, 106 b, and probes 124 a - 124 f ) to generate a biophysical signal dataset 110 including a first type and a second type (shown as 110 a and 110 b ).

[0125] A first type of photoplethysmography signal is acquired from a subject (e.g., a finger of the subject) at location 108a via probes 106a, 106b to generate a raw photoplethysmography signal dataset 110a based on photoplethysmography signal 104a. In some embodiments, raw photoplethysmography signal dataset 110a includes one or more photoplethysmography signals associated with changes in measured optical absorption of oxygenated and / or deoxygenated hemoglobin.

[0126] A second type is collected from the subject 108 via the probes 124a-124f to generate a cardiac signal dataset 110b based on the cardiac signal 104b. In some embodiments, the cardiac signal dataset 110b includes data associated with biopotential signals collected across multiple channels. In some embodiments, the cardiac signal dataset 110b includes, for example, broadband biopotential signals collected via a phase space recorder, as described in U.S. patent application publication No. 2017 / 0119272, entitled “Method and Apparatus for Wide-Band Phase Gradient Signal Acquisition,” the entire contents of which are incorporated herein by reference. In some embodiments, the cardiac signal dataset includes, for example, bipolar broadband biopotential signals collected via a phase space recorder, as described in U.S. patent application publication No. 2018 / 0249960, entitled “Method and Apparatus for Wide-Band Phase Gradient Signal Acquisition,” the entire contents of which are incorporated herein by reference. In other embodiments, the cardiac signal dataset 110b includes one or more biopotential signals acquired from a conventional electrocardiogram (ECG / EKG) device (eg, a Holter device, a 12-lead ECG, etc.).

[0127] Example photoplethysmography signal

[0128] Figure 2A and 2B1 shows an example of a photoplethysmography signal 104a (also referred to herein as a PPG signal) acquired via the measurement system 102 according to an exemplary embodiment. Specifically, Figure 2A A first signal waveform 302 is shown that includes a red photoplethysmography signal associated with a level of absorption of the red spectrum by deoxyhemoglobin from a patient's finger. In some embodiments, the red photoplethysmography signal has an associated wavelength spanning 660 nm. Figure 2A Also shown is a second signal waveform 304 (also referred to herein as a red photoplethysmography signal) representing the absorption level of the infrared spectrum (e.g., having a wavelength spanning 940 nm) by oxyhemoglobin from the patient's finger. Other spectra can be acquired as needed. Furthermore, measurements can be taken at other locations on the body. In FIG2A , the x-axis shows time (in seconds) and the y-axis shows signal amplitude in millivolts (mV). Figure 2B Shows a larger time scale (x-axis) Figure 2A signal to include additional data in the waveform. Figure 2C An example sensor configuration for acquiring a photoplethysmography signal 104a is shown in FIG. 1 , according to an exemplary embodiment; other configurations are possible. Figure 2C In the embodiment, the transmission system includes a light source (eg, a red LED and an infrared LED) and a phototransistor (eg, a red detector and an infrared detector); the phototransistor is far away from the light source.

[0129] Figure 2D Another example sensor configuration for acquiring a photoplethysmography signal 104 according to another example embodiment is shown. Figure 2D In the embodiment, the system also includes a light source (eg, a red LED and an infrared LED) and a phototransistor (eg, a red detector and an infrared detector); however, the phototransistor is close to the light source to measure the reflectivity.

[0130] Photoplethysmography is a technique used to optically measure changes in the volume of blood-perfused tissue (e.g., skin, subcutaneous tissue, cartilage) when light (emitted from an LED or other light source, typically at a specific wavelength) is incident upon it. The intensity of the light after it passes through the tissue (e.g., fingertip, earlobe, etc.) is then recorded by a photodetector to provide a photoplethysmography signal. The amount of light absorbed depends on the amount of blood perfusing the interrogated tissue. Changes in the absorbed light are observable in the photoplethysmography signal and can provide valuable information about cardiac activity, lung function, their interactions, and the function of other physiological systems

[13] .

[0131] In some embodiments, the measurement system 102 includes custom or dedicated equipment or circuitry (including off-the-shelf equipment) configured to acquire such signal waveforms for the purpose of diagnosing a disease or abnormal condition. In other embodiments, the measurement system 102 includes a pulse oximeter or optical photoplethysmography device that can output the acquired raw signal for analysis. Indeed, in some embodiments, the acquired waveform 104 can be analyzed to calculate Figure 1 The oxygen saturation level of the blood is shown as an “SpO2 reading.” However, for the example analysis application, only the waveform is processed and utilized.

[0132] Figure 2E and 2F Shown separately Figure 2A and 2B The power spectrum density of the photoplethysmography signal after removing high-frequency noise. Figure 2E and 2F In the graph, the x-axis represents frequency (in Hz) and the y-axis represents the logarithmic power of the signal.

[0133] Photoplethysmography signal 104 can be considered a measurement of the state of a dynamic system in the body, similar to a cardiac signal. The behavior of a dynamic system may be affected by the actions of the cardiac and respiratory systems. Presumably, any system distortion (e.g., due to disease or abnormal conditions) may manifest itself in the dynamics of photoplethysmography signal 104 through some interaction mechanism or mechanisms.

[0134] In some embodiments, the acquired photoplethysmography signal 104 is downsampled to 250 Hz. Other frequency ranges may be used. In some embodiments, the acquired photoplethysmography signal 104 is processed to remove baseline drift and / or filter noise and / or mains frequency.

[0135] The acquired photoplethysmography signal 104 can be embedded in some higher dimensional space (e.g., phase space embedding) to reconstruct the manifold (phase space) created by the underlying dynamical system. An example three-dimensional visualization of the acquired photoplethysmography signal 104 (shown as 104c) and its two-dimensional projection are shown in FIG. Figure 4A and 4B middle. Specifically, Figure 4AA 3D phase space plot of the acquired photoplethysmography signal 104 acquired via an infrared sensor is shown. The axes are converted voltage values. The colors are chosen to illustrate the coherent structure within this geometric object. The dynamic characteristics of the PPG are calculated based on the embedding represented by the graph. A description of embedding can be found in Sauer et al., Embedology, Jour. (Statistical Physics, Vol. 65:3-4, pp. 579-616 (November 1991)). Figure 4B Its two-dimensional projection is shown, and the Figure 4A Same axis.

[0136] Example heart signal

[0137] Electrocardiogram (ECG) signals measure the action potential of cardiac tissue (i.e., cardiomyocytes). Within the context of this disclosure, various lead configurations are available for obtaining these signals in mammalian tissue, particularly humans. In an exemplary configuration, seven leads are used. This configuration results in three orthogonal channels / signals: for example, X, Y, and Z, corresponding to the coronal, sagittal, and transverse planes, respectively.

[0138] As described above, in some embodiments, the cardiac signal dataset 110b includes data associated with biopotential signals acquired on multiple channels. In some embodiments, the cardiac signal dataset 110b includes broadband biopotential signals, such as signals acquired by a phase-space recorder, such as the phase-space recorder described in U.S. Application Publication No. 2017 / 0119272, entitled “Method and Apparatus for Wide-Band Phase Gradient Signal Acquisition,” the entire contents of which are incorporated herein by reference. In some embodiments, the cardiac signal dataset includes, for example, bipolar broadband biopotential signals acquired by a phase-space recorder, such as the phase-space recorder described in U.S. Application Publication No. 2018 / 0249960, entitled “Method and Apparatus for Wide-Band Phase Gradient Signal Acquisition,” the entire contents of which are incorporated herein by reference. In other embodiments, the cardiac signal dataset 110b includes one or more biopotential signals acquired from a conventional electrocardiogram (ECG / EKG) device (eg, a Holter device, a 12-lead ECG, etc.).

[0139] In some embodiments, a phase space recorder as described in U.S. Patent Application Publication No. 2017 / 0119272 is configured to simultaneously acquire photoplethysmography signals 104a and cardiac signals 104b. Thus, in some embodiments, measurement system 102b is configured to acquire two types of biophysical signals.

[0140] Figure 3A An example cardiac signal (e.g., a biopotential signal) is shown as a Figure 1 Example biophysical signals acquired by a measurement system. The signals are shown with baseline drift and high-frequency noise removed. In some embodiments, cardiac signal 104b is acquired using a phase-space recorder device (e.g., as described in U.S. Patent Application Publication No. 2017 / 0119272). Signal 104b includes bipolar biopotential measurements acquired via three channels to provide three signals 302, 304, and 306 (also referred to as channel "x," channel "y," and channel "z," or coronal, sagittal, and transverse planes, respectively). In FIG3A , the x-axis shows time in seconds, and the y-axis shows signal amplitude in millivolts (mV).

[0141] Figure 3B 2 is a diagram of a phase space recorder device configured to acquire a cardiac signal 104b, such as described in U.S. Patent Application Publication No. 2017 / 0119272. The phase space recorder device is further configured to also acquire a photoplethysmography signal 104a. Figure 3C Shown Figure 3B Example placement of the measurement system on a human patient.

[0142] Still refer to Figure 3B The non-invasive measurement system 102b is configured to transmit the acquired cardiac signal dataset 110b, or a dataset derived therefrom or processed therefrom, to a repository 112 (e.g., a storage area network), for example, via a communication system and / or network, or via a direct connection, which is accessible to a non-invasive biophysical signal evaluation system. The non-invasive biophysical signal evaluation system 114 (shown as analysis engine 114) is configured to analyze dynamic characteristics of the acquired photoplethysmography signal.

[0143] In a cardiac and / or electrocardiographic context, the measurement system 102 is configured to capture cardiac-related biopotential signals or electrophysiological signals from a mammalian subject (e.g., a human) as a biopotential cardiac signal dataset. In some embodiments, the measurement system 102 is configured to acquire broadband cardiac phase gradient signals as biopotential signals, current signals, impedance signals, magnetic signals, ultrasound or acoustic signals, etc. The term "broadband" with respect to the acquired signals and their corresponding datasets refers to signals whose frequency range is significantly greater than the Nyquist sampling rate of the highest dominant frequency of the physiological system of interest. For a cardiac signal typically having a primary frequency component between about 0.5 Hz and about 80 Hz, the broadband cardiac phase gradient signal or broadband cardiac biophysical signal includes cardiac frequency information at a frequency selected from the group consisting of: a range between about 0.1 Hz and about 1 kHz, a range between about 0.1 Hz and about 2 kHz, a range between about 0.1 Hz and about 3 kHz, a range between about 0.1 Hz and about 4 kHz, a range between about 0.1 Hz and about 5 kHz, a range between about 0.1 Hz and about 6 kHz, a range between about 0.1 Hz and about 7 kHz, a range between about 0.1 Hz and about 8 kHz, a range between about 0.1 Hz and about 9 kHz, a range between about 0.1 Hz and about 10 kHz, and a range between about 0.1 Hz and greater than 10 kHz (e.g., a range between 0.1 Hz and 50 kHz or a range between 0.1 Hz and 500 kHz). In addition to capturing the primary frequency components, broadband acquisition also helps capture other frequencies of interest. Examples of such frequencies of interest may include QRS frequency profiles (which can be as high as 250 Hz). The term "phase gradient," as used with respect to acquired signals and corresponding datasets, refers to signals acquired at different vantage points on the body to observe phase information for a set of different events / functions of the physiological system of interest. Following signal acquisition, the term "phase gradient" refers to the preservation of phase information through the use of non-distorting signal processing and preprocessing hardware, software, and techniques (e.g., phase-linear filters and signal processing operators and / or algorithms).

[0144] In some embodiments, the cardiac signal dataset 110b includes broadband biopotential signals, which are acquired, for example, via a phase space recorder described in U.S. patent application publication number 2017 / 0119272, entitled “Method and Apparatus for Wide-Band Phase Gradient Signal Acquisition,” the entire contents of which are incorporated herein by reference. In some embodiments, the cardiac signal dataset 110b includes bipolar broadband biopotential signals, which are acquired, for example, via a phase space recorder described in U.S. patent application publication number 2018 / 0249960, entitled “Method and Apparatus for Wide-Band Phase Gradient Signal Acquisition,” the entire contents of which are incorporated herein by reference. In other embodiments, the cardiac signal dataset 110b includes one or more biopotential signals acquired from a conventional electrocardiogram (ECG / EKG) device (e.g., a Holter device, a 12-lead ECG, etc.).

[0145] In some embodiments, a phase space recorder as described in U.S. Patent Application Publication No. 2017 / 0119272 is configured to simultaneously acquire the photoplethysmography signal 104a and the cardiac signal 104b. Thus, in some embodiments, the measurement system 102b is configured to acquire two types of biophysical signals.

[0146] In a neurological context, measurement system 102 is configured to capture neurologically relevant biopotential or electrophysiological signals from a mammalian subject (e.g., a human) as a neurobiophysical signal dataset. In some embodiments, measurement system 102 is configured to acquire broadband neural phase gradient signals as biopotential signals, current signals, impedance signals, magnetic signals, ultrasonic or acoustic signals, optical signals, etc. Examples of measurement system 102 are described in U.S. Patent Application Publication Nos. 2017 / 0119272 and 2018 / 0249960, each of which is incorporated herein by reference in its entirety.

[0147] In some embodiments, the measurement system 102 is configured to capture broadband biopotential and biophysical phase gradient signals as unfiltered mammalian electrophysiological signals, such that the spectral components of the signals are not altered. In fact, in such embodiments, the broadband biopotential and biophysical phase gradient signals are captured, converted, and even analyzed without filtering (e.g., by hardware circuitry and / or digital signal processing techniques, etc.) (e.g., prior to digitization) that would otherwise affect the phase linearity of the biophysical signal of interest. In some embodiments, the broadband biopotential and biophysical phase gradient signals are captured with a microvolt or sub-microvolt resolution that is at, below, or significantly below the noise floor of conventional electrocardiogram (ECG), electroencephalogram (EEG), and other biophysical signal acquisition instruments. In some embodiments, the broadband biopotential and biophysical signals are sampled simultaneously (with a time deviation or "lag" of less than about 1 microsecond and, in other embodiments, with a time deviation or lag of no more than about 10 femtoseconds). Notably, the illustrative system minimizes nonlinear distortions (e.g., those that may be introduced by certain filters) in the acquired broadband phase gradient signals so as not to affect the information therein.

[0148] Figure 3D is an example placement of surface electrodes 106a-106g on a patient to collect Figure 3A Specifically, Figure 3D An example placement of surface electrodes 106a-106g on the chest and back of a patient is shown to acquire biopotential signals associated with broadband cardiac phase gradient signals in accordance with an exemplary embodiment. Figure 3D In the left pane of FIG, surface electrodes 106a-106g are shown placed on the chest and back areas of the patient. Figure 3D In the right pane, a side view of the placement of the surface electrodes 106a-106g is shown.

[0149] exist Figure 3D In the example configuration shown, surface electrodes 106a-106g are positioned on the patient's skin at: i) a first location near the right anterior axillary line, corresponding to the 5th intercostal space; ii) a second location near the left anterior axillary line, corresponding to the 5th intercostal space; iii) a third location near the left sternal border, corresponding to the 1st intercostal space; iv) a fourth location near the left sternal border, below the sternum, and lateral to the patient's xiphoid process; v) a fifth location near the left sternal border, corresponding to the 3rd intercostal space; vi) a sixth location near the patient's back, directly opposite the fifth location, on the left side of the patient's spine; and viii) a seventh location near the right upper quadrant, corresponding to the 2nd intercostal space along the left axillary line. A common lead (shown as "CMM") is also shown. In other embodiments of the present disclosure, the locations of the various surface electrodes may vary, as other electrode configurations may be useful.

[0150] refer to Figure 1 The non-invasive measurement system 102 is configured with circuitry and computing hardware, software, firmware, middleware, etc. to acquire cardiac signals and photoplethysmography signals to generate a biophysical signal dataset 110. In other embodiments, the non-invasive measurement system 102 includes a first device (not shown) for acquiring cardiac signals and includes a second device (not shown) for acquiring photoplethysmography signals.

[0151] Still refer to Figure 1 The non-invasive measurement system 102 is configured to transmit the acquired biophysical signal dataset 110, or a dataset derived therefrom or processed therefrom, to a repository 112 (e.g., a storage area network) (not shown), such as via a communication system and / or network, or via a direct connection, which is accessible to a non-invasive biophysical signal evaluation system. The non-invasive biophysical signal evaluation system 114 (shown as an analysis engine 114) is configured to analyze, for example, the dynamic characteristics of the acquired photoplethysmography signal.

[0152] In some embodiments, the analysis engine 114 includes a machine learning module 116 that is configured to evaluate a set of features determined from the acquired biophysical signals by one or more feature extraction modules (e.g., 118, 120) to determine clinically significant features. Once features are extracted from the PPG signal or cardiac signal, any type of machine learning can be used. Examples of embodiments of the machine learning module 116 are configured to implement decision trees, random forests, SVMs, neural networks, linear models, Gaussian processes, nearest neighbors, SVMs, naive Bayes, but are not limited thereto. In some embodiments, the machine learning module 116 can be implemented as described in the following U.S. patent applications: 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,” the entire contents of which are incorporated herein by reference. The photoplethysmography signal can be combined with other acquired photoplethysmography signals for use in a training dataset or validation dataset by the machine learning module 116 in evaluating a set of assessed dynamic features. The photoplethysmography signal has an associated tag 122 for a given disease state or abnormal condition, or an indicator indicative thereof. If an assessed dynamic feature (e.g., from 118 or 120) is determined to be clinically significant, it can then be used as a predictor for the given disease state or abnormal condition, or an indicator indicative thereof.

[0153] In some embodiments, the analysis engine 114 includes a pre-processing module, for example, configured to remove baseline drift from the acquired photoplethysmography signal and / or normalize it.

[0154] In some embodiments, the system 100 includes a healthcare provider portal to display, for example, in a report, the scores or various outputs of the analysis engine 114 when predicting and / or estimating the presence, severity, and / or location (if applicable) of a disease or abnormal condition, or an indicator of one of the above. In some embodiments, the physician or clinician portal 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 comply with various privacy laws and regulations, such as the U.S. Health Insurance Portability and Accountability Act of 1996 (HIPAA). A further description of an example healthcare provider portal is provided in U.S. Patent No. 10,292,596, entitled "Method and System for Visualization of Heart Tissue at Risk," which is incorporated herein by reference in its entirety. Although in some embodiments, the portal is configured to present patient medical information to healthcare professionals, in other embodiments, the healthcare provider portal can be accessed by patients, researchers, academics, and / or other portal users.

[0155] Synchronicity assessment between cardiac signals and raw photoplethysmography signals

[0156] Still refer to Figure 1 In some embodiments, the dynamic feature extraction module 118 is configured to evaluate one or more nonlinear dynamic characteristics of synchronization between the one or more acquired photoplethysmography signals 104a and the one or more acquired cardiac signals 104b. Several examples of synchronization are disclosed.

[0157] The electrophysiological activity of the heart is a nonlinear process that combines with the electromechanical feedback of the myocytes to produce very complex nonlinear responses

[26] . Nonlinear statistics related to the nonlinear dynamics and chaos of the heart can be used to study and characterize whether these behaviors are normal (in response to external conditions) or caused by disease. Synchronicity features based on the dynamics observed in the heart and photoplethysmography signals can encode the health state of the heart and be used to train machine learning models to predict various disease states.

[0158] In Poincare mapping, one can define a map X using a trigger (e.g., intersection with Σ) n+1 = P(X n ), then we can calculate the Poincare point set {X0, X1, ... . . , X n}Perform geometric and / or statistical analysis to derive more information about the system.

[0159] Synchronous Feature Example PM#1

[0160] Figure 5A 、 5B 5C illustrate example dynamic characteristics of synchronization between acquired photoplethysmography signals and cardiac signals according to an exemplary embodiment. Figure 5A 、 5B In 5C , synchrony was determined by histogram / Poincare map analysis using landmarks defined by the photoplethysmography signal.

[0161] Specifically, Figure 5A In the first graph 502, cross-over landmarks 504a and 504b are shown defined between the red photoplethysmography signal 302 and the infrared photoplethysmography signal 304. The second graph 506 shows the cross-over landmarks 504a and 504b associated with one of the channels of the acquired cardiac signal 104b. Figure 5A In the graph, the x-axis represents time in seconds and the y-axis represents the signal amplitude in millivolts (mV).

[0162] Figure 5B A histogram of the distribution of values ​​of the cardiac signal 104b at the intersection landmarks 504a and 504b is shown. Specifically, Figure 5B Distributions 508, 510 corresponding to amplitude values ​​of the cardiac signal at respective first and second sets of crossing landmarks (504a, 504b) are shown. In FIG5B, the x-axis of the histogram represents signal amplitude (in mV) and the y-axis represents frequency / counts.

[0163] In some embodiments, the dynamic feature extraction module 118 is configured to generate a histogram (e.g., as shown in FIG. Figure 5Bgenerated) and extracting statistical and geometric properties from the generated histogram. In some embodiments, the extracted histogram features include, for example, mode, standard deviation, skewness, kurtosis, and mutual information, but are not limited thereto. Mode refers to the set of data values ​​that occur most frequently in a data set. Skewness refers to a measure of the asymmetry of the probability distribution of a data set about its mean. Kurtosis refers to the sharpness of the peak of the distribution curve. In some embodiments, mutual information is used to quantify the probabilistic dependence of information in the collected signal, which is determined by first calculating the probability normalization of the histogram of each time series, and then constructing a two-dimensional histogram of the two time series and normalizing it. The mutual information I(X, Y) between two random variables X and Y can be the reduction in the uncertainty of one random variable, for example, X given another variable Y as defined in Equation 1.

[0164]

[0165] In Equation 5, p(., .) is the probability distribution of the specified variable.

[0166] Figure 5C A Poincare map 512 of the values ​​of cardiac signal 104b at intersections 504a and 504b is shown. That is, the Poincare map records the values ​​of cardiac signal 104b, or at least one of its channels, based on a trigger defined by intersections 504a and 504b. In FIG5C , the x-axis and y-axis each show the difference in amplitude values ​​between cycles of the cardiac signal.

[0167] In some embodiments, to generate the Poincare map 512, the system draws / generates two-dimensional point pairs [x i , x i+1 ] (e.g., (x1, x2), (x2, x3), etc.) relative to the amplitude value of the cardiac signal at the intersection landmark point formed between the photoplethysmography signals [x i , x i-1 ] (e.g., (x0, x1), (x1, x2), etc.).

[0168] In some embodiments, the dynamic feature extraction module 118 is configured to generate the Poincare map 512. After generating the Poincare map 512, in some embodiments, the dynamic feature extraction module 118 is configured to generate geometric objects from the map data. Figure 5CIn some embodiments, the dynamic feature extraction module 118 determines an ellipse 511 based on an ellipse fitting operation of data associated with the cluster (e.g., 510a). In some embodiments, based on the fitted ellipse, the dynamic feature extraction module 118 is configured to determine geometric parameters such as, but not limited to, the length of the semi-axis "a" (514), the length of the semi-axis "b" (516), the length along the major axis (518), and the length along the minor axis (520), as shown in FIG. Figure 5D shown.

[0169] In some embodiments, the dynamic feature extraction module 118 may extract other parameters, such as void area, surface area, porosity, perimeter, density, etc.

[0170] In practice, the synchronization between an acquired photoplethysmography signal (e.g., where the acquired raw signal is processed only to remove baseline drift and high-frequency noise) and a cardiac signal based on triggers defined in the photoplethysmography signal can be used to assess the presence, severity, and / or localization (as applicable) of coronary artery disease (CAD), pulmonary hypertension, various forms of heart failure, and other diseases and conditions. In the context of CAD, Figure 5E shows an example Poincare map for a dataset acquired from a CAD-negative patient (i.e., a patient without CAD). Figure 5F shows an example Poincare map for a dataset acquired from a CAD-positive patient (i.e., a patient with some form of CAD). Clearly, the Poincare maps for the CAD-negative and CAD-positive patients in Figures 5E and 5F differ. Figure 5E and 5F An example Poincare map generated based on the amplitude values ​​of the cardiac signal at landmarks defined by the photoplethysmography signal is shown. On the x-axis and y-axis, the Poincare map shows the signal amplitude of the cardiac signal (e.g., normalized with high frequency and baseline drift removed) at a first index x-1 and a second index x on the x-axis and a second index x and a third index x+1 on the y-axis. In practice, in the Poincare map, time and data position represented by the index values ​​are synonymous and are used interchangeably herein. In addition, other indices or time increments may be used. That is, each evaluation parameter (e.g., signal amplitude) at a given time / data point is shown in the Poincare map relative to the next time / data point (e.g., [x - 1]). i-1 , x] for [x, x i+ 1 ]). Thus, the Poincare map facilitates analysis of changes in a given parameter between cycles in an acquired dataset (e.g., changes in the lowest peak landmark). Similar analyses can be applied to any of the parameters and features discussed here.

[0171] Figure 5G and5H Further illustrated are intersection landmarks (504a, 504b) associated with three cardiac signals acquired via a phase-space recorder, which can be used to trigger analysis of the photoplethysmography signal in a Poincare map. For example, an example phase-space recorder and its associated probe positions are described in U.S. Patent Application Publication No. 2018 / 0249960, entitled “Method and Apparatus for Wide-Band Phase Gradient Signal Acquisition,” which is incorporated herein by reference in its entirety. FIG5G shows a dataset for a CAD-negative patient. FIG5H shows a dataset for a CAD-positive patient.

[0172] Specifically, in Figure 5G In the image, crossing landmarks from the photoplethysmography signal are observed that generally correspond to the P wave (522) and T wave (524) of the cardiac signal 104b (shown as 104b_channelx, 104b_channely, and 104b_channelz). The P wave generally corresponds to atrial depolarization associated with atrial contraction and atrial systole. The T wave generally corresponds to ventricular repolarization.

[0173] On the contrary, Figure 5H As shown, for CAD-positive patients, it can be observed that the crossing landmarks from the photoplethysmography signal have shifted relative to the P and T waves. Poincare mapping (e.g., as described with respect to FIG. 5C ) facilitates quantification and, in some embodiments, visualization of the resulting shifts, which can be interpreted as signs of disease (here, coronary artery disease or CAD).

[0174] exist Figure 5G and 5H , amplitude values ​​of the cardiac signal at landmarks in the photoplethysmography signal are plotted in pairs (eg, a first time x-1 and a second time x on the x-axis and a second time x and a third time x+1 on the y-axis).

[0175] Figure 5I 、 5J 5K and 5L illustrate another example of the dynamic characteristics of the synchronization between the acquired photoplethysmography signal and the cardiac signal according to an exemplary embodiment. Specifically, FIG. 5I shows a histogram of the distribution of the Px1 and Px2 Poincare points for a healthy patient (i.e., a CAD-negative patient), such as shown in FIG. 5A , and FIG. 5J shows a Poincare map of the same data and an ellipse fit to geometrically characterize the data distribution. FIG. 5K shows a histogram of the distribution of the Px1 and Px2 Poincare points for an unhealthy patient (i.e., a CAD-positive patient), such as shown in FIG. 5A . x1and P x2 Figure 5L shows a histogram of the distribution of the Poincare points, and the corresponding Poincare map and ellipse fit for the same data.

[0176] Statistical properties of these distributions (eg, mean, median, deviation, kurtosis, etc.) and geometric properties of the bounding ellipse (eg, major and minor diameters and slope) can be calculated and used as features.

[0177] Table 1 provides a list of example synchronization feature extraction parameters associated with Poincare map analysis PM#1 as well as their corresponding descriptions.

[0178] Table 1

[0179]

[0180]

[0181] Synchronicity Feature Example PM#2

[0182] Figure 6A 、 6B 6C illustrate another set of example dynamic characteristics of synchronization between the acquired photoplethysmography signal and the cardiac signal according to an exemplary embodiment. Figure 6A 、 6B and 6C , synchrony was determined by histogram / Poincare map analysis using landmarks in the cardiac signal.

[0183] Figure 6A One channel of an acquired cardiac signal data set is shown for an acquired red photoplethysmography signal 302 and an infrared photoplethysmography signal 304. The x-axis represents the time domain (in the form of index counts of the data set) and the y-axis represents the acquired signal amplitude (in millivolts);

[0184] Figure 6B A histogram 604 shows the amplitude magnitude values ​​of one of the photoplethysmography signals at the estimated peak (also known as the R-peak) of the QRS waveform of the cardiac signal. Here, the magnitude of the infrared photoplethysmography signal is shown. In other embodiments, the amplitude magnitudes of both the red and infrared photoplethysmography signals are recorded and analyzed for statistical and geometric characteristics. In FIG6B , the x-axis of the histogram represents signal amplitude (in mV) and the y-axis represents frequency / counts.

[0185] In some embodiments, the dynamic feature extraction module 118 is configured to generate a histogram and extract statistical properties from the generated histogram, such as but not limited to mode, scale, skewness, kurtosis, and mutual information, for example, as described with respect to Figure 5Bdiscussed.

[0186] Figure 6C A Poincare map 606 of the amplitude magnitude values ​​of the acquired infrared photoplethysmography signal 304 at the estimated R-peak of the acquired cardiac signal is shown. Figure 6C , amplitude values ​​of the photoplethysmography signal at landmarks in the cardiac signal are plotted in pairs (eg, at a first time x-1 and a second time x on the x-axis and at a second time x+1 on the y-axis).

[0187] After generating the Poincare map 606, in some embodiments, the dynamic feature extraction module 118 is configured to generate geometric objects from the data. Figure 6C In FIG. 6 , the dynamic feature extraction module 118 determines an ellipse 611 based on an ellipse fitting operation of the data associated with the cluster (e.g., 510a). In some embodiments, based on the fitted ellipse, the dynamic feature extraction module 118 is configured to determine geometric parameters, such as the semi-axis length "a" (514), the semi-axis length "b" (516), the length along the major axis (518), and the length along the minor axis (520), for example, as described with respect to FIG. Figure 5D In some embodiments, the dynamic feature extraction module 118 can extract other parameters, such as void area, surface area, porosity, perimeter, density, etc.

[0188] Indeed, synchronization between the acquired raw photoplethysmography signal and the cardiac signal based on triggers defined in the cardiac signal can be used to assess the presence, severity and / or location of coronary artery disease, pulmonary hypertension, heart failure, and other diseases, disorders, and related conditions.

[0189] In some embodiments, to generate the Poincare map 512, module 118 draws / generates two-dimensional point pairs [x i ,x i+1 ] (e.g., (x1, x2), (x2, x3), etc.) relative to the amplitude value of a given photoplethysmography signal (e.g., a red photoplethysmography signal or an infrared photoplethysmography signal) at a landmark of the cardiac signal (e.g., at one of channels "x", "y", or "z") at a point [x i-1 , x i ] (for example, (x0, x1), (x1, x2)).

[0190] Figure 6D 1 shows the histogram and Poincare map results for CAD negative patients according to an exemplary embodiment. Specifically, Figure 6DA histogram 608 and a Poincare map 610 for a CAD-negative patient are shown, where the histogram 608 and Poincare map 610 are generated based on the amplitude magnitude values ​​of the acquired red photoplethysmography signal at the R-peak of one of the acquired cardiac signals 104 b, and the histogram 612 and Poincare map 614 are generated based on the amplitude magnitude values ​​of the acquired infrared photoplethysmography signal at the R-peak of one of the acquired cardiac signals 104 b. In histograms 608 and 612, the x-axis of the histograms represents signal amplitude (in mV) and the y-axis represents frequency / count. In the Poincare maps 610, 610a, 614, and 614a, the x-axis and y-axis each show the amplitude values ​​of the photoplethysmography signal at landmarks in the cardiac signal plotted in pairs (e.g., a first time x-1 and a second time x on the x-axis, and a second time x and a third time x+1 on the y-axis). Plots can also be made for index values ​​represented for a given data set.

[0191] Figure 6E 1 shows the histogram and Poincare map results for CAD-positive patients according to an exemplary embodiment. Specifically, Figure 6E A histogram 616 and a Poincare map 618 for a CAD-positive patient are shown, where the histogram 616 and Poincare map 618 are generated based on the amplitude magnitude values ​​of the acquired red photoplethysmography signal at the R-peak of one of the acquired cardiac signals 104 b, and the histogram 620 and Poincare map 622 are generated based on the amplitude magnitude values ​​of the acquired infrared photoplethysmography signal at the R-peak of one of the acquired cardiac signals 104 b. In histograms 616 and 620, the x-axis of the histograms represents signal amplitude (in mV) and the y-axis represents frequency / count. In the Poincare maps 618, 618a, 622, and 622a, the x-axis and y-axis each represent the amplitude values ​​of the photoplethysmography signal at landmarks in the cardiac signal plotted in pairs (e.g., a first time x-1 and a second time x on the x-axis, and a first time x and a third time x+1 on the y-axis). Plots can also be performed for index values ​​represented for a given data set.

[0192] Table 2 provides a list of example synchronization feature extraction parameters associated with the Poincare map analysis PM#2 as well as their corresponding descriptions.

[0193] Table 2

[0194]

[0195]

[0196] Synchronicity Feature Example PM #3

[0197] Figure 7A 、 7B 7C illustrate yet another example dynamic characteristic of synchronization between an acquired photoplethysmography signal and a cardiac signal according to an exemplary embodiment. Figure 7A 、 7B and 7C , synchrony is determined using a phase relationship between one or more landmarks in the cardiac signal and one or more landmarks in the photoplethysmography signal via histogram / Poincare map analysis.

[0198] Specifically, Figure 7A In the first graph 702, it is shown as follows Figure 5A Depicted are intersection landmarks 504a and 504b defined between the red photoplethysmography signal 302 and the infrared photoplethysmography signal 304. A second graph 704 illustrates intersection landmarks 504a and 504b associated with one of the channels of the acquired cardiac signal 104b. Graph 704 also illustrates the intersection landmarks 504a and 504b of the phase difference R-peak 602 (denoted as "TP" 706 and "TT" 708) of the cardiac signal 104b and the photoplethysmography signal. The x-axis represents the time domain (in the form of data set index counts), and the y-axis represents the acquired signal amplitude (in millivolts).

[0199] Figure 7B A histogram showing the distribution of phase relationships between the cardiac signal 104b and the corresponding crossing landmarks 504a and 504b is shown. Specifically, Figure 7B Distributions 710, 712 corresponding to the phase relationship between the R-peak of the cardiac signal and the first and second sets of crossing landmarks (504a, 504b) are shown. In FIG7B, the x-axis of the histogram represents signal amplitude (in bin number) and the y-axis represents frequency / count.

[0200] In some embodiments, the dynamic feature extraction module 118 is configured to generate a histogram (e.g., as shown in FIG. Figure 5B The method further comprises the steps of: generating a histogram and extracting statistical and geometric properties from the generated histogram. In some embodiments, the extracted histogram features include, for example, mode, standard deviation, skewness, kurtosis, and mutual information, but are not limited thereto. The term "mode" as used herein refers to the set of data values ​​that occur most frequently in a data set. The term "skewness" as used herein refers to a measure of the asymmetry of the probability distribution of a data set about its mean. The term "kurtosis" as used herein refers to the sharpness of the peak of a distribution curve.

[0201] Figure 7C A Poincare map 714 is shown showing the phase relationship between the cardiac signal 104b and each of the intersection landmarks 504a and 504b. Figure 7C , time values ​​of TP intervals and TT intervals defined between the photoplethysmography signal and the cardiac signal are plotted in pairs (eg, a first time x-1 and a second time x on the x-axis and a second time x and a third time x+1 on the y-axis).

[0202] exist Figure 7C , amplitude values ​​(eg, in bits) of the photoplethysmography signal at landmarks of the cardiac signal are plotted in pairs (eg, a first time x-1 and a second time x on the x-axis and a second time x and a third time x+1 on the y-axis).

[0203] That is, to generate the Poincare map 714, the system is indexed relative to the TT interval / time point [x i , x i+1 ] (e.g. (x1, x2), (x2, x3) etc.) Draw / generate 2D point pairs [x] with TP interval index / time i , x i+1 ] (e.g. (x1, x2), (x2, x3), etc.).

[0204] In some embodiments, the dynamic feature extraction module 118 is configured to generate a Poincare map 714. After generating the Poincare map 714, in some embodiments, the dynamic feature extraction module 118 is configured to generate geometric objects from the map data. Figure 7C In some embodiments, the dynamic feature extraction module 118 determines the ellipse 716 based on an ellipse fitting operation of the data associated with the cluster (e.g., 712a). In some embodiments, based on the fitted ellipse, the dynamic feature extraction module 118 is configured to determine geometric parameters such as, but not limited to, Figure 5D Shown are the length of semi-axis "a" (514), the length of semi-axis "b" (516), the length along the major axis (518), and the length along the minor axis (520).

[0205] In some embodiments, the dynamic feature extraction module 118 may extract other parameters, such as void area, surface area, porosity, perimeter, density, etc.

[0206] In practice, synchronization between one or more acquired photoplethysmography signals and one or more cardiac signals, based on phase relationships between landmarks in the photoplethysmography signal and the cardiac signal, can be used to assess the presence, severity, and / or location (as applicable) of coronary artery disease (CAD), pulmonary hypertension, heart failure, and other diseases and conditions. Figure 7DAn example Poincare map for a data set acquired from a CAD-negative patient is shown. Figure 7E shows an example Poincare map for a data set acquired from a CAD-positive patient.

[0207] Figure 7D 1 shows the histogram and Poincare map results for CAD negative patients according to an exemplary embodiment. Specifically, Figure 7D A histogram 716 and a Poincare map 718 for a CAD-negative patient are shown, generated based on the phase relationship (eg, TP and TT) between the acquired infrared photoplethysmography signal and the R-peak of one of the acquired cardiac signals 104b.

[0208] Figure 7E 1 shows the histogram and Poincare map results for CAD-positive patients according to an exemplary embodiment. Specifically, Figure 7E A histogram 720 and a Poincare map 722 for a CAD-positive patient are shown, generated from the phase relationship (e.g., based on TP and TT intervals) between the acquired infrared photoplethysmography signal and the R-peak of one of the acquired cardiac signals 104b. The Poincare maps 718a and 722a further show fitted ellipses in the respective graphs 718 and 722.

[0209] Figure 7F 、 7G , 7H and 7I illustrate yet another example dynamic characteristic of synchronization between the acquired photoplethysmography signal and the cardiac signal according to an exemplary embodiment. Specifically, Figure 7F is a graph showing the time intervals between predefined landmarks in the photoplethysmography signal and predefined landmarks in the cardiac signal for healthy patients (ie, CAD negative patients) Figure 7A Figure 7G shows the Poincare map of the same data along with an ellipse fit to geometrically characterize the data distribution. Figure 7H is the time interval between a predefined landmark in the photoplethysmography signal and a predefined landmark in the cardiac signal for an unhealthy patient (i.e., a CAD-positive patient) ( Figure 7A Figure 7I shows the corresponding Poincare map and ellipse fit for the same data.

[0210] Table 3 provides a list of example synchronization feature extraction parameters associated with Poincare map analysis PM#3 as well as their corresponding descriptions.

[0211] Table 3

[0212]

[0213]

[0214] Synchronicity Feature Example #4

[0215] Figure 8A 、 8B 8C illustrate another set of example dynamic characteristics of synchronization between one or more acquired photoplethysmography signals and one or more cardiac signals according to an exemplary embodiment. Figure 8A 、 8B In and 8C, synchrony is determined by phase analysis using landmarks in the cardiac signal.

[0216] Figure 8A The phase 802 of the cardiac signal 104b, defined over one full revolution / cycle from one R-peak (eg, 602) to the next R-peak (eg, 602), is shown superimposed on the underlying cardiac signal data set used to generate the phase data.

[0217] Figure 8B The phases 804, 806 of the red and infrared photoplethysmographic signals defined over a full revolution / cycle are shown. The phases of the red and infrared photoplethysmographic signals are shown as -π to π (y-axis) using a Hilbert transform; the x-axis is time (index count of the data set). Figure 8B As shown, the two phases are the red photoplethysmography signal and the infrared photoplethysmography signal, and the two signals are consistent, indicating that the two are synchronized.

[0218] Figure 8C The phase difference between the cardiac signal and one of the photoplethysmography signals is shown, as determined by the difference between the period of the photoplethysmography signal and the period of the cardiac signal. In FIG8C , the x-axis represents time (index count of the data set) and the y-axis represents the magnitude value of the calculated difference.

[0219] In fact, the synchronization between the acquired raw photoplethysmography signal and the cardiac signal based on the phase difference between the cardiac signal and the photoplethysmography signal can be used to assess the presence, severity and / or location (if applicable) of coronary artery disease, pulmonary hypertension, heart failure and other diseases, disorders and related conditions.

[0220] Figure 8D 1 shows an analysis of the phase difference generated between the acquired infrared photoplethysmography signal 304 and the acquired cardiac signal 104b for a CAD negative patient (ie, a patient with a negative diagnosis of coronary artery disease). Specifically, Figure 8DGraph 808 shows a period 802 of a cardiac signal and a period 804 of an infrared photoplethysmography signal. In graph 808, the x-axis is time (expressed in index counts of the data set), while the y-axis represents phase (expressed in radians). Graph 810 shows the calculated lag between periods 802 and 804. In some embodiments, the time / index lag is calculated by cross-correlation between the two signals. The lag is the time interval required for one signal to be offset relative to the other to produce maximum (or minimum) cross-correlation. In graph 810, the x-axis is time (expressed in index counts of the data set), and the y-axis represents the cross-correlation value (unitless).

[0221] Plot 812 shows a frequency analysis of the difference data from 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 shows the difference between infrared photoplethysmography signal 304 and cardiac signal 104b. In plot 814, the x-axis is time (in data set index counts). Plot 816 shows a filtered version of the difference data from plot 814. Plot 818 shows a histogram of the filtered difference data from plot 816. In histogram 818, the x-axis of the histogram shows the difference amplitude (derived from the difference data, in bins), and the y-axis shows frequency / counts.

[0222] Figure 8E 1 shows an analysis of the phase difference generated between the acquired infrared photoplethysmography signal 304 and the acquired cardiac signal 104b for a CAD-positive patient. Specifically, Figure 8E Graph 820 shows a period 802 of a cardiac signal and a period 804 of an infrared photoplethysmography signal. In graph 820, the x-axis is time (in data set index counts), and the y-axis is phase (in radians). Graph 822 shows the calculated lag between periods 802 and 804. Graph 824 shows a frequency analysis of the difference data from graph 822. In graph 822, the x-axis is time (in data set index counts), and the y-axis shows the cross-correlation value (unitless). In graph 824, the x-axis is frequency (in Hz), and the y-axis is the relative amplitude of the signals. Graph 826 shows the filtered difference between infrared photoplethysmography signal 304 and cardiac signal 104b. In graph 826, the x-axis is time (in data set index counts). Graph 828 shows a histogram of the filtered difference data from graph 824. In histogram 828, the x-axis of the histogram represents the magnitude of the difference (derived from the difference data, in bins) and the y-axis represents the frequency / count.

[0223] Table 4 provides a list of example synchronicity feature extraction parameters associated with Phase Analysis #4 as well as their corresponding descriptions.

[0224] Table 4

[0225]

[0226] Machine learning-based classifiers

[0227] Machine learning techniques predict outcomes based on input datasets. For example, machine learning techniques are used to recognize patterns and images, supplement medical diagnoses, and more. Some machine learning techniques rely on a set of features generated using a training dataset (i.e., a dataset of observations where the outcome to be predicted is known for each observation). Each feature represents some measurable aspect of the observed data, thereby generating and tuning one or more predictive models. For example, observed signals (e.g., cardiac, plethysmographic, or other biophysical signals from multiple subjects, or any combination thereof) can be analyzed to gather frequency, mean, and other statistical information about these signals. Machine learning techniques can use these features to generate and tune a model that classifies or associates these features with one or more conditions, such as certain forms of cardiovascular disease or conditions, including, for example, coronary artery disease, heart failure, and pulmonary hypertension. This model can then be applied to data (e.g., biophysical data from one or more individuals) to detect and / or understand the presence and severity of one or more diseases or conditions (e.g., as described herein), which might not otherwise be detected or understood to the same degree. Traditionally, in the context of cardiovascular disease, these features are manually selected from conventional electrocardiogram signals and combined by data scientists in collaboration with domain experts.

[0228] Examples of machine learning embodiments include, but are not limited to, decision trees, random forests, SVMs, neural networks, linear models, Gaussian processes, nearest neighbors, SVMs, and naive Bayes. In some embodiments of the present disclosure, machine learning techniques may be implemented, such as those 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,” the entire contents of which are incorporated herein by reference.

[0229] Experimental Results and Other Examples

[0230] 9 illustrates experimental results from a study indicating the clinical predictive value of specific dynamic features extracted from Poincare and phase analysis of one or more photoplethysmography signals (red photoplethysmography signals and infrared photoplethysmography signals) and one or more cardiac signals, indicating the presence, severity, and / or localization (if applicable) of a disease or condition, or an indicator of one of the foregoing, according to an exemplary embodiment.

[0231] In this study, candidate features were evaluated using t-tests, mutual information, or area under curves (AUC). T-tests were performed for the null hypothesis of normal LVEDP and the null hypothesis of negative coronary artery disease. The t-test is a statistical test that determines whether there is a difference between the means of two samples from two populations with unknown variances. The output of the t-test is a dimensionless quantity called the p-value. Small p-values ​​(typically ≤ 0.05) indicate strong evidence against the null hypothesis. This study used random sampling with replacement (bootstrapping) to generate the test set.

[0232] Mutual information is used to assess the dependence of elevated or abnormal LVEDP or significant coronary artery disease on some set of features. Mutual information is an information-theoretic measure of the mutual dependence between two random variables. MI is normalized by the number of bins, and high MI and low MI are calculated as The selected feature has a high value greater than 1.0 and a low value greater than 1.0.

[0233] Table 1 provides a description of the synchronization extraction parameters for each evaluation in Figure 9 associated with Poincare map analysis PM#1. Table 2 provides a description of the synchronization extraction parameters for each evaluation in Figure 9 associated with Poincare map analysis PM#2. Table 3 provides a description of the synchronization extraction parameters for each evaluation in Figure 9 associated with Poincare map analysis PM#3.

[0234] Table 4 provides the data associated with Phase Analysis Example #4. Figure 9 Description of each evaluated synchronization extraction parameter. Parameters can be configured as double variables.

[0235] Experimental results of the characteristics of Poincare map analysis #1

[0236] As mentioned above, Table 1 provides the data associated with the Poincare Map analysis PM#1. Figure 9 9 illustrates that various geometric and statistical features extracted from the Poincare map according to Poincare map analysis PM#1 as described herein have potential clinical relevance in predicting and / or estimating the presence, severity, and / or location (as applicable) of coronary artery disease and elevated or abnormal LVEDP (which may indicate the presence, absence, and / or severity of the disease and / or condition).

[0237] Specifically, FIG9 (and reproduced in Tables 1-A and 1-B) shows that the major and minor diameters (shown as "dXDmj" and "dXDmn") of the ellipse generated from the Poincare map PM#1 of the PSR / ECG "x" channel have t-test p-values ​​of 0.012 and 0.003, respectively, in predicting and / or estimating the presence, absence, and / or severity of elevated or abnormal LVEDP (which may indicate the presence, absence, and / or severity of a disease or condition). Furthermore, FIG9 shows that the minor diameter (shown as "dZDmn") of the ellipse generated from the Poincare map PM#1 of the PSR / ECG "z" channel has a t-test p-value of 0.037 in predicting and / or estimating the presence, absence, and / or severity of elevated or abnormal LVED (which may indicate the presence, absence, and / or severity of a disease or condition). Small p-values ​​(usually ≤ 0.05) indicate strong evidence against the null hypothesis (ie, no elevated or abnormal LVEDP exists; normal data sets do not have elevated or abnormal LVEDP).

[0238]

[0239]

[0240] Furthermore, Figure 9 (and reproduced in Table 1-C) shows that the tilt angle alpha (denoted as "dYAlpha" and "dZAlpha") of the ellipse from the Poincare map analysis of PM#1 for the PSR / ECG "y" and "z" channels has a t-test p-value of 0.049 for predicting the presence, location (if applicable), and / or severity of coronary artery disease, and a t-test p-value of 0.039 for predicting the presence and / or severity of elevated or abnormal LVED (which may indicate the presence and / or severity of a disease or condition). Small p-values ​​(typically ≤ 0.05) indicate strong evidence against the null hypothesis (i.e., the absence of elevated or abnormal LVEDP).

[0241]

[0242] Furthermore, Figure 9 (and reproduced in Table 1-D) shows that the mean amplitude of the PSR / ECG "x" channel at the first intersection / crossover point of the photoplethysmography signal (e.g., in Poincare map analysis PM#1) (denoted as "dXMean1") has a t-test p-value of 0.00064 and an AUC of 0.548, respectively; and a t-test p-value of 0.011 and an AUC of 0.518 for predicting and / or estimating the presence, location, and / or severity of coronary artery disease in certain populations. Small p-values ​​(typically ≤ 0.05) indicate strong evidence against the null hypothesis (i.e., the absence of elevated or abnormal LVEDP); an AUC greater than 0.5 is highly significant in indicating the presence of CAD (defined as an angiographic stenosis greater than 70% or a streamlined flow fraction less than 0.80).

[0243]

[0244] Furthermore, FIG9 (and reproduced in Tables 1-E, 1-F, and 1-G) shows that the standard deviations of the distributions of the PSR / ECG "x" channel (e.g., in Poincare Map Analysis PM #1) triggered by the first and second crossing landmarks of the photoplethysmography signal (denoted as "dXStd1" and "dXStd2") have corresponding t-test p-values ​​of 0.037 and 0.042 in predicting and / or estimating the presence, absence, and / or severity of elevated or abnormal LVED (which can indicate the presence, absence, and / or severity of a disease or condition). Furthermore, FIG9 shows that the standard deviation of the distributions of the PSR / ECG "y" channel (e.g., in Poincare Map Analysis PM #1) triggered by the second crossing landmark of the photoplethysmography signal (denoted as "dYStd2") has a mutual information value of 1.143 in predicting / estimating the presence, location (if applicable), and / or severity of coronary artery disease. Mutual information greater than 1.0 is considered significant; p-values ​​less than 0.05 are considered significant.

[0245]

[0246]

[0247]

[0248] also, Figure 9 (And reproduced in Table 1-H) shows that the kurtosis of the distribution of the PSR / ECG "y" channel (e.g., in Poincare Map Analysis PM #1) triggered by the second crossing landmark of the photoplethysmography signal (shown as "dYKurt2") has a mutual information of 1.061 when predicting and / or estimating the presence, location (if applicable), and / or severity of coronary artery disease. A mutual information greater than 1.0 is considered significant.

[0249]

[0250] also, Figure 9 The kurtosis of the PSR / ECG "z" channel distribution (e.g., in Poincare map analysis PM #1) triggered by the second crossing landmark of the photoplethysmography signal (shown as "dZKurt2") (and reproduced in Table 1-1) is shown to have a mutual information of 1.076 when predicting and / or estimating the presence, absence, and / or severity of elevated or abnormal LVED (which may indicate the presence, absence, and / or severity of a disease or condition); and a mutual information of 1.192 when predicting and / or estimating the presence, location (if applicable), and / or severity of coronary artery disease. A mutual information greater than 1.0 is considered significant.

[0251]

[0252] also, Figure 9 (and reproduced in Table 1-J) shows that the modes of the distributions of the PSR / ECG "y" and "z" channels (e.g., in Poincare Map Analysis PM#1) triggered by the second crossing landmark of the photoplethysmography signal (shown as "dYMode2" and "dZMode2") have mutual information values ​​of 1.104 and 1.036 when predicting and / or estimating the presence, location (if applicable), and / or severity of coronary artery disease. A mutual information value greater than 1.0 is considered significant.

[0253]

[0254] also, Figure 9 (And reproduced in Tables 1-K and 1-L) show that the modes of the distributions of the PSR / ECG "z" channel (e.g., in Poincare map analysis PM#1) triggered by the first and second crossing landmarks of the photoplethysmography signal (shown as "dZSkew1" and "dZSkew2") have mutual information values ​​of 1.094 and 1.058, respectively, for predicting and / or estimating the presence, location (if applicable), and / or severity of coronary artery disease. A mutual information value greater than 1.0 is considered significant.

[0255]

[0256]

[0257] also, Figure 9 (and reproduced in Table 1-M) shows that the relative difference between the standard deviation and median absolute deviation (MAD) of the distribution of PSR / ECG "y" channel data triggered at the second crossing landmark of the photoplethysmography signal (e.g., in Poincare map analysis PM#1) (shown as "dYRelStdMAD2") has a t-test p-value of 0.042 and a mutual information value of 1.048 in predicting and / or estimating the presence, absence and / or severity of elevated or abnormal LVED (which may indicate the presence, absence and / or severity of a disease or condition).

[0258]

[0259] also, Figure 9(And reproduced in Tables 1-N) shows the relative difference between the standard deviation and median absolute deviation (MAD) of the distribution of PSR / ECG "z" channel data triggered at the first crossing landmark of the photoplethysmography signal (e.g., in Poincare map analysis PM #1) (shown as "dZRelStdMAD1") in predicting and / or estimating the presence, absence, and / or severity of elevated or abnormal LVED (which may indicate the presence, absence, and / or severity of a disease or condition). A p-value less than 0.05 is considered significant; a mutual information value greater than 1.0 is considered significant.

[0260]

[0261] Experimental results of the characteristics of Poincare map analysis #2

[0262] As mentioned above, Table 2 provides the data associated with the Poincare map analysis PM#2. Figure 9 Figure 9 shows various geometric and statistical features extracted from the Poincare map of PM#2 according to the Poincare map analysis, which, as described herein, have potential clinical relevance in predicting the presence of coronary artery disease and elevated or abnormal LVED (which may indicate the presence, absence, and / or severity of the disease or condition).

[0263] Specifically, FIG9 (and reproduced in Tables 2-A and 2-B) shows that the major diameters of the ellipses generated in the Poincare map derived from the amplitudes of the infrared and red photoplethysmography signals at the R-peak of the cardiac signal (denoted as "dDmjL" and "dDmjU") have t-test p-values ​​of 0.031 and 0.007, respectively, for predicting and / or estimating the presence, absence, and / or severity of elevated or abnormal LVED (which may indicate the presence, absence, and / or severity of a disease or condition). Furthermore, FIG9 shows that the major diameters of the ellipses generated in the Poincare map derived from the amplitudes of the infrared photoplethysmography signals at the R-peak of the cardiac signal (denoted as "dDmjL") have a t-test p-value of 0.035, a mutual information value of 1.104, and an AUC of 0.502 for predicting and / or estimating the presence, location (if applicable), and / or severity of coronary artery disease.

[0264]

[0265]

[0266] Furthermore, FIG9 (and reproduced in Table 2-C) shows that the short path of the generated ellipse (denoted as "dDmnU") in the Poincare map derived from the amplitude of the red photoplethysmography signal at the R-peak of the cardiac signal has a t-test p-value of 0.0380 in predicting and / or estimating the presence and / or severity of elevated or abnormal LVED (which may indicate the presence and / or severity of a disease or condition). Mutual information values ​​greater than 1.0 are considered significant; AUC values ​​greater than 1.0 are considered significant; and p-values ​​less than 0.05 are considered significant.

[0267]

[0268] Furthermore, Figure 9 (and reproduced in Tables 2-D and 2-E) shows that the inclination angle alpha of the ellipse in the Poincare map (e.g., according to example Poincare map analysis PM#2) derived from the amplitude of the infrared and red photoplethysmography signals at the R-peak of the cardiac signal (shown as "dAlphaL" and "dAlphaU") has mutual information values ​​of 1.043 and 1.03, respectively, for predicting and / or estimating the presence, location (if applicable), and / or severity of coronary artery disease. Mutual information values ​​greater than 1.0 are considered significant.

[0269]

[0270]

[0271] also, Figure 9 (and reproduced in Table 2-F) shows that the kurtosis of the histogram of the infrared photoplethysmography signal at the R peak of the cardiac signal (shown as "dKurtL") has a mutual information value of 1.171 in predicting and / or estimating the presence, location (if applicable), and / or severity of elevated or abnormal LVED (which may indicate the presence and / or severity of a disease or condition). Mutual information values ​​greater than 1.0 are considered significant.

[0272]

[0273] Furthermore, FIG9 (and reproduced in Tables 2-G and 2-H) shows that the mean values ​​of the histograms of the infrared and red photoplethysmography signals at the R-peak of the cardiac signal (denoted as "dMeanL" and "dMeanU") have corresponding t-test p-values ​​of 0.033 and 0.003, respectively, in predicting and / or estimating the presence, location (if applicable), and / or severity of elevated or abnormal LVED (which may indicate the presence and / or severity of a disease or symptom). Further, Figure 9The histogram means of the infrared and red photoplethysmography signals at the R peak of the cardiac signal (denoted as "dMeanL" and "dMeanU") were shown to have corresponding mutual information values ​​of 1.012 and an AUC value of 0.516, respectively, and a mutual information value of 1.091 for predicting and / or estimating the presence, location (if applicable), and / or severity of coronary artery disease. A mutual information value greater than 1.0 is significant, and an AUC value greater than 1.0 is significant.

[0274]

[0275]

[0276] Furthermore, FIG9 (and reproduced in Tables 2-I and 2-J) shows that the modes of the histograms of the infrared and red photoplethysmography signals at the R-peak of the cardiac signal (denoted as "dModeLP" and "dModeUP") have t-test p-values ​​of 0.024 and 0.004, respectively, in predicting and / or estimating the presence, absence, and / or severity of elevated or abnormal LVED (which may indicate the presence, absence, and / or severity of a disease or condition). Furthermore, Figure 9 The mode of the histogram of the infrared photoplethysmography signal at the R-peak of the cardiac signal (shown as "dModeLP") has an AUC value of 0.507 in predicting the presence of coronary artery disease. An AUC value greater than 1.0 is considered significant, and a p value less than 0.05 is considered significant.

[0277]

[0278]

[0279] Furthermore, FIG9 (and reproduced in Table 2-K) shows that the standard deviation of the histogram of the red photoplethysmography signal at the R-peak of the cardiac signal (denoted as "dStdU") has an AUC value of 0.511 in predicting and / or estimating the presence, location (if applicable), and / or severity of coronary artery disease. An AUC value greater than 1.0 is considered significant.

[0280]

[0281] Experimental results of the characteristics of Poincare map analysis #3

[0282] As mentioned above, Table 3 provides the data associated with the Poincare Map analysis PM#3. Figure 9Figure 9 illustrates various geometric and statistical features extracted from the Poincare map of PM#3 based on the Poincare map analysis, which, as described herein, have potential clinical relevance in predicting and / or estimating the presence, location (if applicable), and / or severity of coronary artery disease and elevated or abnormal LVED (which may indicate the presence, absence, and / or severity of the disease or condition).

[0283] Specifically, Figure 9 (and reproduced in Tables 3-A and 3-B) shows that the major axis of an ellipse (displayed as "dDmjLUXR") generated in a Poincare map derived from the difference in time intervals TT and TP between i) the R-peak in the cardiac signal and ii) the intersection landmarks between the acquired red and infrared photoplethysmography signals has an AUC value of 0.501 when used to predict and / or estimate the presence, location (if applicable), and / or severity of coronary artery disease. Furthermore, FIG9 shows that the minor axis of an ellipse (displayed as "dDmLUXR") generated in a Poincare map derived from the difference in time intervals TT and TP between the intersection landmarks between i) the R-peak in the cardiac signal and ii) the intersection landmarks between the acquired red and infrared photoplethysmography signals has a t-test p-value of 0.02 when used to predict and / or estimate the presence, location (if applicable), and / or severity of elevated or abnormal LVED (which may indicate the presence and / or severity of a disease or condition). A p-value less than 0.05 was considered significant, and an AUC value greater than 0.5 was considered significant.

[0284]

[0285]

[0286] also, Figure 9 (and reproduced in Tables 3-C and 3-D) show that the means of the TP and TT time intervals (i.e., the time intervals between the R-peak of the PSR / ECG "x" channel and the respective first and second crossing landmarks between the acquired red and infrared photoplethysmography signals) (shown as "dMeanLURP1" and "dMeanLURP2") have t-test p-values ​​of 0.013 and 0.02, respectively, in predicting and / or estimating the presence, location (if applicable), and / or severity of elevated or abnormal LVED (which can indicate the presence and / or severity of a disease or condition).

[0287]

[0288]

[0289] also, Figure 9 (and reproduced in Tables 3-E and 3-F) show that the modes of the TP and TT time intervals (i.e., the time intervals between the R-peak of the PSR / ECG "x" channel and the respective first and second crossing landmarks between the acquired red and infrared photoplethysmography signals) (shown as "dModeLURP1" and "dModeLURP2") have t-test p-values ​​of 0.013 and 0.028, respectively, in predicting and / or estimating the presence, location (if applicable), and / or severity of elevated or abnormal LVED (which may indicate the presence and / or severity of a disease or condition).

[0290]

[0291]

[0292] In addition, Figure 9 (and reproduced in Table 3-G) shows that the skewness of the TP time interval (i.e., the time interval between the R-peak of the PSR / ECG "x" channel and the first intersection landmark between the acquired red and infrared photoplethysmographic signals) (shown as "dSkewLURP1") in predicting and / or estimating the presence, location (if applicable), and / or severity of coronary artery disease has a t-test p-value of 0.034.

[0293]

[0294] also, Figure 9 (and reproduced in Table 3-H) shows that the standard deviation of the TT time interval (the time interval between the R-peak of the PSR / ECG X channel and the second occurrence of the Poincare map analysis PM#3 landmark) (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, location (if applicable), and / or severity of coronary artery disease.

[0295]

[0296] also, Figure 9 (and reproduced in Table 3-I) shows that the standard deviation of the TT time interval (the time interval between the R-peak of the PSR / ECG "x" channel and the second intersection landmark between the acquired red photoplethysmography signal and the infrared photoplethysmography signal) (shown as "dRelMeanMedDiffLURP1") has an AUC value of 0.5 in predicting and / or estimating the presence, location (if applicable), and / or severity of coronary artery disease.

[0297]

[0298] Experimental results of characteristics of phase analysis #4

[0299] As mentioned above, Table 4 provides the data associated with Phase Analysis Example #4. Figure 9 FIG9 also illustrates that various geometric and statistical features extracted from Phase Analysis #4, as described above, have potential clinical relevance in predicting and / or estimating the presence, location (if applicable), and / or severity of coronary artery disease and elevated or abnormal LVED (which may indicate the presence and / or severity of the disease or condition).

[0300] Specifically, Figure 9 (and reproduced in Table 4-A) shows that the median of the phase difference distribution (shown as "dPhiDiffXL1Med") belonging to the first distribution after the phase difference between the photoplethysmography signal and the cardiac signal is divided into two parts (a first part with a higher mean and a second part with a lower mean) has a t-test p-value of 0.015 in predicting and / or estimating the presence, location (if applicable), and / or severity of coronary artery disease. A p-value of less than 0.05 is considered significant.

[0301]

[0302] also, Figure 9 (and reproduced in Table 4-B) shows that the standard deviation of the phase difference distribution (shown as "dPhiDiffXL2Std") belonging to the second distribution after dividing the phase difference between the photoplethysmography signal and the cardiac signal into two parts (a first part with a higher mean and a second part with a lower mean) has an AUC value of 0.502 in predicting and / or estimating the presence, location (if applicable), and / or severity of coronary artery disease. An AUC greater than 0.5 is considered significant.

[0303]

[0304] Furthermore, FIG9 (and reproduced in Table 4-C) shows that the mean of the distribution of phase differences between the photoplethysmography signal and the cardiac signal (denoted as "dPhiDiffXLMean") has a t-test p-value of 0.26 for predicting and / or estimating the presence, location (if applicable), and / or severity of coronary artery disease. A p-value of less than 0.05 is considered significant.

[0305]

[0306] In addition, FIG9 shows (and reproduced in Table 4-D) that pulse transit time (i.e., the time difference (lag) between the phase of the PSR / ECG "x" channel and the phase of the infrared photoplethysmography signal) (shown as "dPTT") has a t-test p-value of 0.045 in predicting and / or estimating the presence, location (if applicable), and / or severity of elevated or abnormal LVED (which can indicate the presence and / or severity of a disease or condition). A p-value of less than 0.05 is considered significant.

[0307]

[0308] Coronary Artery Disease - Learning Algorithm Development Research

[0309] The Coronary Artery Disease - Learning Algorithm Development (CADLAD) study, which acquired photoplethysmography signals along with cardiac signals to support the development and testing of machine learning algorithms, has not yet been conducted.

[0310] In this study, paired clinical data was used to guide the design and development of the preprocessing, feature extraction, and machine learning phases of the development process. Specifically, the collected clinical research data was divided into multiple cohorts: a training cohort, a validation cohort, and a verification cohort. Each acquired dataset was first preprocessed to clean and normalize the data. Following preprocessing, a set of features was extracted from the signal, where each set of features was paired with a representation of the true condition (e.g., a binary classification of the presence or absence of significant CAD or a scored classification of the presence of significant CAD in a given coronary artery).

[0311] In some embodiments, the evaluation system (e.g., 114, 114a, 114b) automatically and iteratively explores feature combinations in various functional permutations with the goal of finding those combinations that successfully match predictions based on the features. To avoid overfitting the solution to the training data, a validation set is used as a comparator. Once candidate predictors are developed, they are manually applied to a validation dataset to evaluate predictor performance against data that was not used to generate the predictor. If the dataset is large enough, the performance of the selected predictor on the validation set will approach the performance of the predictor on new data.

[0312] The study also developed and evaluated a machine learning-based prediction model that employs nonlinear dynamics and chaos to extract physically meaningful and significant features from cardiac biopotential and photoplethysmography signal data. Traditional features based on linear representation of the signals are unable to detect the more complex and nonlinear patterns hidden in the signals. In the study, three types of features were developed by employing nonlinear dynamics: (i) features based on the dynamics of the cardiac system represented by the biopotential signal, (ii) features based on the dynamics represented by the PPG signal, and (iii) features that characterize the synchrony between the two dynamics.

[0313] For the first two groups, invariant measures of the dynamics, such as the Lyapunov exponent (LE), the fractal dimension (D2), and the entropy rate (K2), are calculated. The 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 little memory of the past. For a given dynamic system, as the value of the Lyapunov exponent becomes larger, the time range of past information that can be used to predict the future becomes shorter. Entropy (KS) (or Kolmogorov Sinai entropy K2 [31, 32]) represents the rate of change of entropy over time. The fractal dimension (D2) characterizes the topological properties of the attractor in the phase space, which can be used to combine the geometric information (fractal) of the attractor to reveal more information about the dynamics and how the dynamics evolve on it

[33] . Examples of attractors for acquired cardiac and photoplethysmographic signals are shown in Figure 4A and 4B shown.

[0314] Nonlinear dynamics and chaos theory can be systematically used to explain the complexity of linear systems and provide tools for quantitatively analyzing their behavior

[19] . Linear systems can generate responses that are exponentially growing / decaying or periodic oscillations, or a combination thereof, where any irregular patterns in the responses can be attributed to the irregularity or randomness of the inputs to these systems. Linear systems are simplifications of reality, and most dynamic systems, whether natural or artificial, are inherently nonlinear and will produce complex irregular behaviors even in the absence of any random source. These behaviors are often referred to as deterministic chaos. Nonlinear dynamics and chaos tools have been used to explain a variety of complex biological and physiological phenomena [20, 21, 22, 23], for example, to classify atrial fibrillation

[24] and characterize heart rate variability

[25] , each of which is incorporated herein by reference in its entirety. These dynamic features are further described in U.S. Provisional Patent Application No. 62 / 862,991, filed on June 18, 2019, entitled “Method and System to Assess Disease Using Dynamical Analysis of Biophysical Signals,” which is incorporated herein by reference in its entirety.

[0315] Other invariant measures of the dynamics can be used as feature sets. Deterministic dynamical systems that exhibit chaotic behavior often have invariant properties that are independent of when the observation is made and therefore independent of the evolution of the system.

[0316] For the synchronization feature set, three types of Poincare maps were defined, and the resulting sets were statistically and geometrically characterized. Several machine learning models were then trained using the computed feature sets matched to the appropriate labels. Models were selected based on their individual AUC performance on a holdout test set. Cross-validation and grid search were performed to tune the hyperparameters used in classifier training. In this study, the developed elastic net model was observed to have AUCs of 0.78 and 0.61 on two test datasets for CAD classification. Furthermore, the developed XGboost model was observed to have AUCs of 0.86 and 0.63 on two test datasets. This study demonstrates an efficient and cost-effective method for leveraging advanced nonlinear feature extraction processes from non-invasive modalities for machine learning operations to predict disease or abnormal symptoms.

[0317] Elastic Net, Lasso, or Ridge classifiers are often suitable for smaller datasets with a large number of features because they can be tuned to prevent overfitting. The Elastic Net is a hybrid of the Lasso and Ridge, including an absolute value penalty (Lasso) and a squared penalty (Ridge). Each penalty has a hyperparameter that can be optimized to produce a stronger model. Both the Lasso and Ridge have only one hyperparameter, making optimization more limited.

[0318] Data Description In this study, two cohorts of human subjects, with a mean age of 63 years (Group A) and 28 years (Group B), were recruited for data collection. Cohorts were selected after undergoing an eligibility screening process. For the older cohort, CAD signature and LVEDP values ​​were determined using corresponding gold standard tests, while the younger cohort was considered healthy according to clinical criteria. That is, the younger cohort did not have CAD and their LVEDP values ​​were not abnormally high or elevated.

[0319] Cardiac signals (as biopotential signals) and photoplethysmography signals were acquired as time series data from each subject in Groups A and B. Data for both signal modalities were acquired within 3.5 minutes, with the entire process averaging approximately 10 minutes per subject. Each cardiac signal was collected using a phase-space recorder as described with respect to Figures 3A-3E at an 8 kHz sampling rate (i.e., 8,000 samples per second for each of the six channels over 210 seconds). Three differential input pairs, along with a reference lead, were arranged orthogonally across the subject's chest. The acquired signals were subjected to baseline drift removal and filtering to remove power line and high-frequency noise for feature extraction.

[0320] During the same period as cardiac signals were collected from the subjects, photoplethysmography signals were collected using the same phase-space recorder at a sampling rate of 500 Hz. Light absorption data for the red and infrared channels were recorded at a rate of 500 samples per second during the same 210-second period. These photoplethysmography and cardiac signals were acquired simultaneously for each subject. The jitter in the data (intermodal jitter) was less than approximately 10 microseconds (μs). The jitter between cardiac signal channels was approximately 10 femtoseconds (fs).

[0321] CAD feature research. This study defined severe coronary artery occlusion as patients with a stenosis greater than 70% or exceeding a functional threshold for flow restriction [14, 15]. For group A, patients with double-vessel disease (i.e., two vessels with lesions meeting this definition) were considered disease-positive, and non-disease cases were defined as healthy control subjects who underwent invasive catheterization to evaluate for coronary artery disease but did not have any coronary artery lesions. Table 5 lists the number of positive and negative cases in the coronary artery disease dataset, which was used to develop the coronary artery disease signature for this study. Table 5 further shows the mean age and sex composition associated with the subjects in the dataset. This study used invasive coronary angiography (the “gold standard” for coronary artery disease) as the ground truth indicator. In coronary angiography, fluoroscopy is used to image the coronary arteries after the injection of a radiopaque contrast agent. With coronary angiography, stenosis (blockage) in the arteries can be detected, and patients are subsequently labeled as CAD-positive or CAD-negative.

[0322]

[0323] The results of the study indicate that synchronization between the photoplethysmography signal and the cardiac signal, as represented by synchronization features from analysis between the photoplethysmography signal and the cardiac signal as described herein, can be used to predict the presence or absence of significant coronary artery disease.

[0324] LVDEP feature feasibility study Left ventricular end-diastolic pressure (LVEDP) is an invasively obtained hemodynamic measurement that describes filling pressures on the left side of the heart in patients undergoing cardiac catheterization. LVEDP is a key parameter for the hemodynamic evaluation of patients with systolic or diastolic LV dysfunction, both of which are associated with decreased LV compliance. Alterations in the pressure-volume relationship that result in markedly elevated filling pressures are a hallmark of cardiomyopathy

[10] .

[0325] Measurement of filling pressures can be used to assess risk stratification and the development of appropriate treatment strategies. In addition, LVEDP provides important prognostic information, as elevated LVEDP has been identified as an independent predictor of adverse outcomes in the setting of acute myocardial infarction

[16] , cardiogenic shock

[17] , postoperative success of cardiac surgery

[18] , and percutaneous cardiac intervention. Table 6 lists the number of LVEDP-positive and -negative cases used in the evaluation of LVEDP characteristics in the dataset in Table 5.

[0326]

[0327] The results of the study indicate that synchronization between the photoplethysmography signal and the cardiac signal, as represented by synchronization features from analysis between the photoplethysmography signal and the cardiac signal as described herein, can be used to predict the presence or absence of abnormal LVEDP.

[0328] Machine Learning Classifier Analysis In the study, a feature set extracted from the acquired dataset was extracted and evaluated during machine-based classifier analysis. The feature set included 94 synchronization features defined between the photoplethysmography signal and the cardiac signal (e.g., synchronization analysis based on Poincare maps 1, 2, and 3), as well as 6 features of phase analysis #4, etc. (e.g., dynamic features, etc.). The feature set including the synchronization features was paired with the corresponding CAD or LVEDP labels and provided as input to the machine learning model. The feature set included 36 other dynamic features related to the cardiac signal (i.e., biopotential signal), and 29 additional dynamic features related to the photoplethysmography signal 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,” filed concurrently with this application (which claims priority to U.S. Provisional Patent Application Ser. No. 62 / 863,005, filed Jun. 18, 2019), the entire contents of which are incorporated herein by reference.

[0329] In the classifier analysis, the data for CAD and LVEDP were each divided into a training-validation set and a test set. Table 7 shows the composition of the training-validation and test datasets used for machine learning model training and evaluation. As mentioned above, information for Groups A and B of the CAD dataset is listed in Table 5, and information for Groups A and B of the LVEDP dataset is listed in Table 6.

[0330]

[0331] The training-validation set was used to train and fine-tune the candidate machine learning models using 5-fold cross-validation. Table 8 lists the classifiers used for training and model selection for both the CAD and LVEDP datasets in this study. The pipeline for data scaling, model training, grid search, and model evaluation was implemented in Python using the Scikit-learn package

[36] .

[0332]

[0333] To find the optimal hyperparameter set for each model, we performed a grid search within a predefined hyperparameter range. Using the average AUC as the performance metric, we selected the optimal hyperparameter set for each model. We then trained the selected models on the entire training-validation set and ranked them based on their AUC performance on a holdout test set.

[0334] In this study, the elastic net model and support vector classifier model were found to be most predictive for significant CAD, while the XGBoost model and elastic net model were found to be most predictive for elevated or abnormal LVEDP status. Table 9 shows the predictive performance of the elastic net model and support vector classifier model in predicting significant CAD. Table 10 shows the predictive performance of the elastic net model and XGBoost model in predicting significant CAD status.

[0335]

[0336]

[0337] Figure 10A and 10B The ROC curves for the classification of significant CAD using the trained elastic net model on test 1 and test 2 are shown, respectively. Figures 10C and 10D show the ROC curves for the classification of abnormal LVEDP using the trained XGBoost model on test 1 and test 2, respectively.

[0338] like Figure 10A 、 10B As shown in Figures 10C and 10D, the CAD and LVEDP classification tasks can achieve relatively good AUC performance. In the case of CAD prediction, AUC = 0.78 was observed in Test 1 and AUC = 0.61 in Test 2. For the prediction of elevated or abnormal LVEDP, AUC = 0.86 was observed in Test 1 and AUC = 0.63 in Test 2.

[0339] The model was trained on the datasets from Groups A and B (older and younger subjects, respectively), as described in Tables 5 and 6. The use of the Group B dataset reinforced the training on the Group A dataset and allowed the model to learn from diseased subjects that they were healthy. Consequently, the model trained for this task demonstrated better performance in Test 1 (which included subjects from both Groups A and B) than in Test 2 (which included subjects from Group A alone). Furthermore, because the dataset was skewed toward non-diseased cases, as shown in Tables 5 and 6, the trained model in this study performed better at detecting CAD-negative subjects. It is expected that model performance in Test 2 will improve with a more balanced dataset between diseased and non-diseased cases. XGBoost performance can also be improved by performing a more refined hyperparameter search and stronger regularization.

[0340] Further improvements to the second elastic net model can be made using only the synchronicity feature set and a larger dataset.

[0341] Figure 10E Shown is an ROC curve for correctly classifying the presence of significant CAD using a subsequently trained elastic net model using only the synchrony feature set, according to an exemplary embodiment.

[0342] Figure 10F Shown is a ROC curve for correctly classifying the presence of elevated or abnormal LVEDP using a subsequently trained elastic net model using only the synchrony feature set, according to an exemplary embodiment.

[0343] Figure 10E and 10F It is shown that the synchronous feature set in combination with features can be used to achieve classification with high specificity and sensitivity.

[0344] Figure 10G and 10H Shown respectively Figure 10E and 10F Feature contributions of the classifier model for CAD and LVEDP classification. In Figure 10G, features are divided into three subgroups (PM1, PM2, and PM3) based on the Poincare map (PM) used to generate the features. Figure 10H shows the absolute values ​​of the differences in feature contributions used in the LVEDP and CAD classification models. Features with larger differences indicate that they are more disease-specific. Table 1 includes some features for PM1, as shown in Figures 10G and 10H. Table 2 includes some features for PM2, as shown in Figures 10G and 10H. Table 3 includes some features for PM3, as shown in Figures 10G and 10H.

[0345] Table 11 lists the cumulative feature contributions in each subgroup PM1-PM3, as shown in Figure 10G and 10H. As shown in Table 11, PM3 features have the largest contribution in CAD and LVEDP classification, while PM1 has the lowest contribution. Table 11 shows the sum of feature contributions of the Poincare map used in the elastic net model for CAD and LVEDP classification.

[0346]

[0347] Although the elastic net classifier was found to be the best-performing model for classifying CAD and LVEDP, the contributions of synchronization features differed between the two diseases. The absolute values ​​of the differences in feature contributions are plotted in Figure 11. This reflects the utility of synchronization features, which have different distributions between subjects with LVEDP and CAD. These PMs were developed based on triggers and other information deemed useful from a signaling perspective; while current work has characterized their utility in disease assessment, the underlying physiological characteristics captured by these features remain unclear, and future work will explore this mechanistic aspect.

[0348] Study on LVDEP characteristic performance A second LVDEP-related study was conducted with the prediction of elevated LVEDP as the primary outcome. This study also investigated as secondary outcomes (i) the diagnostic sensitivity of the machine learning predictor in three subgroups of progressively elevated LVEDP (≥ 20 mmHg, ≥ 25 mmHg, and ≥ 30 mmHg) and (ii) the predictive performance of the predictor in an age- and sex-prone cohort.

[0349] A second LVDEP-related study retrospectively developed and evaluated a machine-learning predictor using a dataset collected from the Cardiac Phase Space Analysis Study in the manner described herein (i.e., using a phase space recorder as described with respect to Figures 3A-3E). Biopotential (cardiac) and photoplethysmography signals were acquired from 1,919 consecutive subjects recruited from 21 centers immediately before elective angiography. Data were collected for a comparison group (control group) of 634 healthy subjects without cardiovascular disease, recruited from two of the 21 sites, in the same manner as described for the 1,919 subjects.

[0350] Data for both signal modalities was acquired over a 3.5-minute period, with the entire process taking approximately 10 minutes. Biopotential signals were collected at an 8 kHz sampling rate (8,000 samples per second for each of the six channels over 210 seconds). Three differential input pairs, along with a reference lead, were arranged orthogonally across the patient's chest. The acquired signals were then filtered for baseline drift, power line noise, and high-frequency noise before feature extraction.

[0351] Among 1919 symptomatic subjects who underwent elective angiography, 256 subjects were found to have an LVEDP ≥ 20 mmHg at catheterization; these 256 subjects formed the study cohort. As previously described, patients were referred for angiography for evaluation of symptoms, and elevated or abnormal LVEDP (when present) was determined in each patient during cardiac catheterization by direct measurement of LV pressure during ventriculography.

[0352] To develop the machine learning predictor, cross-validation was performed for 100 iterations, with 70% of the subjects used for training and 30% for testing. Subjects were grouped to stratify the set by disease prevalence (LVEDP ≥ 20 mmHg), but otherwise the division was random. The training subject features were input into an elastic net model configured with an increasing regularization penalty to reduce overfitting. After training, the model was applied to the validation subjects to assess diagnostic performance.

[0353] Figures 11A-11F Experimental results for training a classifier to predict elevated LVEDP, according to an exemplary embodiment, are shown. FIG11A shows the receiver operating characteristic (ROC) curve for the classification of elevated LVEDP ≥ 20 mmHg. Classification was based on an elastic net model. As shown in FIG11A , the machine-learned cardiac phase-space predictor provided robust prediction of elevated LVEDP ≥ 20 with an area under the curve (AUC) of 0.97. The predictor algorithm also performed with increasing diagnostic sensitivity with increasing LVEDP. The algorithm maintained high fidelity even after age and gender biases were matched, with an area under the curve (AUC) of 0.88 for predicting LVEDP ≥ 20. The ROC curve, which includes AUC, sensitivity, specificity, PPV, and NPV values, was calculated using the R package ROC.

[0354] Figure 11BShown is a receiver operating characteristic (ROC) curve illustrating the diagnostic performance of the machine learning approach in a propensity-matched secondary analysis (age and sex) for predicting LVEDP ≥ 20 mmHg. In this analysis, subjects with elevated LVEDP were propensity-matched with subjects without elevated LVEDP according to sex and age within 5 years prior to stratification into training and testing datasets. Matching, training, and testing were then performed in 100 iterations to capture dataset heterogeneity.

[0355] Figure 11C shows the sensitivity results of the machine learning approach for LVEDP ≥ 20 mmHg, LVEDP ≥ 25 mmHg, and LVDEP ≥ 30 mmHg in additional secondary analyses. Figures 11D, 11E, and 11F show ROC curves, respectively, illustrating the diagnostic performance of the machine learning approach for predicting LVEDP ≥ 20 mmHg, LVEDP ≥ 25 mmHg, and LVDEP ≥ 30 mmHg in propensity-matched analyses (age and sex).

[0356] Healthcare Provider Portal

[0357] refer to Figure 1 (as well as Figure 1 A and 1B), in some embodiments, system 100 (e.g., 100a, 100b) includes a healthcare provider portal to display assessments of disease states or symptoms (e.g., related to abnormal LVEDP and / or the presence of coronary artery disease and / or pulmonary hypertension, etc.) in a report. In some embodiments, the report is structured as an angiography-equivalent report. In some embodiments, the physician or clinician portal 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 HIPAA (and its equivalents) and various other privacy requirements. An example healthcare provider portal is provided in U.S. patent application Ser. No. 15 / 712,104, entitled "Method and System for Visualization of Heart Tissue at Risk," which is incorporated herein by reference in its entirety. While in some embodiments, the portal is configured to present patient medical information to healthcare professionals, in other embodiments, the healthcare provider portal can be accessed by patients, other caregivers, family members, researchers, academics, and / or others. The portal can be used to address a variety of clinical and even research needs in a variety of settings – from the hospital to the emergency room, laboratory, battlefield, or remote settings, at the point of care with the patient’s primary care physician or other caregiver, or even at home.

[0358] Sample computing environment

[0359] Figure 12 An example computing environment is shown in which example embodiments of the analysis system 114 and aspects thereof may be implemented, for example, in one or more devices.

[0360] The computing device environment is only one example of a suitable computing environment and is not intended to suggest any limitation as to scope of use or functionality.

[0361] Numerous other general-purpose or special-purpose computing device 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, and distributed computing environments including any of the foregoing.

[0362] Computer-executable instructions executed by a computer, such as program modules, can be used. Generally, program modules include routines, programs, objects, components, data structures, etc., which perform specific tasks or implement specific abstract data types. A distributed computing environment can be used when tasks are performed by remote processing devices connected through a communication 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).

[0363] Reference Figure 12 , an example system for implementing aspects described herein includes a computing device, such as computing device 1000. In its most basic configuration, computing device 1000 typically includes at least one processing unit 1002 and memory 1004. Depending on the specific configuration and type of computing device, memory 1004 may be volatile (such as random access memory (RAM)), non-volatile (such as read-only memory (ROM), flash memory, etc.), or some combination of the two. This most basic configuration is as follows Figure 12 As shown by the dotted line 1006.

[0364] The computing device 1000 may have additional features / functionality. For example, the computing device 1000 may include additional storage devices (removable and / or non-removable), including but not limited to magnetic or optical disks or tapes. Such additional storage devices may be stored in the computer system. Figure 12 Shown are removable storage 1008 and non-removable storage 1010 .

[0365] The 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, including both volatile and nonvolatile media, removable and non-removable media.

[0366] Computer storage media includes volatile and nonvolatile, 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 device 1008, and non-removable storage device 1010 are all examples of computer storage media. Computer storage media include, but are not limited to, RAM, ROM, electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, CD-ROM, digital versatile disk (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store the desired information and that can be accessed by computing device 1000. Any such computer storage media can be part of computing device 1000.

[0367] Computing device 1000 may include communication connections 1012 that allow the device to communicate with other devices. Computing device 1000 may also include input devices 1014, such as a keyboard, mouse, pen, voice input device, touch input device, etc. (alone or in combination). Output devices 1016 (such as a display, speakers, printer, vibration mechanism, etc.) may also be included, alone or in combination. All of these devices are well known in the art and need not be discussed in detail here.

[0368] It should be understood that the various techniques described herein may be implemented in conjunction with hardware components or software components, or, where appropriate, through a combination of both. Example types of hardware components that may be used include field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), and the like. 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 a tangible medium (e.g., a floppy disk, CD-ROM, hard drive, or any other machine-readable storage medium), wherein, when the program code is loaded into and executed by a machine (e.g., a computer), the machine becomes an apparatus for practicing the presently disclosed subject matter.

[0369] Although example implementations may refer to using the presently disclosed subject matter in the context of one or more standalone computer systems, the subject matter is not limited thereto and may be implemented in any computing environment, such as a network or distributed computing environment. Furthermore, various aspects of the presently disclosed subject matter may be implemented in or across multiple processing chips or devices, and storage may similarly be implemented across multiple devices. For example, these devices may include personal computers, network servers, handheld devices, and wearable devices.

[0370] 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 acts described above are disclosed as example forms of implementing the claims.

[0371] Other examples of processes that can be used with the example methods and systems are described in the following: U.S. Patent No. 9,289,150, entitled “Non-invasive Method and System for Characterizing Cardiovascular Systems”; U.S. Patent No. 9,655,536, entitled “Non-invasive Method and System for Characterizing Cardiovascular Systems”; U.S. Patent No. 9,968,275, entitled “Non-invasive Method and System for Characterizing Cardiovascular Systems”; U.S. Patent No. 8,923,958, entitled “System and Method for Evaluating an Electrophysiological Signal”; U.S. Patent 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. Patent 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. Patent No. 9,737,229, entitled “Noninvasive Electrocardiographic Method for Estimating Mammalian Cardiac Chamber Size and Mechanical Function”; U.S. Patent No. 10,039,468, entitled “Noninvasive Electrocardiographic Method for Estimating Mammalian Cardiac Chamber Size and Mechanical Function”;U.S. Patent No. 9,597,021, entitled “Noninvasive Method for Estimating Glucose, Glycosylated Hemoglobin and Other Blood Constituents”; U.S. Patent No. 9,968,265, entitled “Method and System for Characterizing Cardiovascular Systems From Single Channel Data”; U.S. Patent No. 9,910,964, entitled “Methods and Systems Using Mathematical Analysis and Machine Learning to Diagnose Disease”; U.S. Patent Application 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” Acquisition”; U.S. Patent Application Publication No. 2018 / 0000371, entitled “Non-invasive Method and System for Measuring Myocardial Ischemia, Stenosis Identification, Localization and Fractional Flow Reserve Estimation”; International Application 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. Patent No. 10,292,596, entitled “Method and System for Visualization of Heart Tissue at Risk”;U.S. Patent Application No. 16 / 402,616, entitled “Method and System for Visualization of Heart Tissue at Risk”; U.S. Patent Application Publication No. 2018 / 0249960, entitled “Method and System for Wideband Phase Gradient Signal Acquisition”; U.S. Patent Application No. 16 / 232,801, entitled “Method and System to Assess Disease Using Phase Space Volumetric Objects”; International PCT Application No. IB / 2018 / 060708, entitled “Method and System to Assess Disease Using Phase Space Volumetric Objects”; U.S. Patent Application Publication No. US2019 / 0117164, entitled “Methods and Systems of De-Noising Magnetic-Field Based Sensor Data of Electrophysiological Signals”; U.S. Patent Application No. 16 / 232,586, entitled “Method and System to Assess Disease Using PhaseSpace Tomography and U.S. patent application entitled “Machine Learning”; International application PCT application number PCT / IB2018 / 060709, entitled “Method and System to Assess Disease Using Phase Space Tomography and Machine Learning”; U.S. patent application number 16 / 445,158, entitled “Methods and Systems to Quantify and Remove Asynchronous Noise in Biophysical Signals”;U.S. Patent Application No. 16 / 725,402, entitled “Method and System to Assess Disease Using Phase Space Tomography and Machine Learning” (Attorney Docket No. 10321-034pv1, claiming priority to U.S. Provisional Application No. 62 / 784,984); U.S. Patent Application No. 16 / 429,593, entitled “Methods and Systems to Assess Pulmonary Hypertension Using Phase Space Tomography and Machine Learning”; U.S. Patent Application No. 16 / 725,416, entitled “Method and System for Automated Quantification of Signal Quality”; U.S. Patent Application No. 16 / 725,430, entitled “Method and System to Configure and Use Neural Network To Assess Medical Disease”; U.S. Patent Application 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 No. 15 / 653,431, entitled “Discovering Genomes to Use in Machine Learning Techniques,” each of which is incorporated herein by reference in its entirety.

[0372] Unless otherwise expressly stated, it is not intended that any method described herein be construed as requiring that its steps be performed in a specific order. Therefore, if a method claim does not actually recite the order in which its steps are to be followed, or if the steps are not otherwise specifically stated in the claims or specification to be limited to a particular order, then in no way should that order be inferred. This applies to any possible non-express basis for interpretation, including: logical issues regarding the arrangement of steps or operational flow; plain meaning derived from grammatical organization or punctuation; or the number or type of embodiments described in the specification.

[0373] Although the methods and systems have been described in conjunction with certain embodiments and specific examples, it is not intended to limit the scope to the particular embodiments set forth, as the embodiments herein are intended in all respects to be illustrative rather than restrictive.

[0374] The methods, systems, and processes described herein can be used to generate stenosis and FFR outputs for use with procedures such as placement of vascular stents within a blood vessel, such as an artery, of a living (e.g., human) subject, and other interventional and surgical systems 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 the number of stents to be placed in a living (e.g., human) subject during a procedure, including their optimal deployment location within a given blood vessel.

[0375] 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: electrocardiogram (ECG) datasets, electroencephalogram (EEG) datasets, gamma synchronization signal datasets, respiratory function signal datasets, pulse oximetry signal datasets, perfusion data signal datasets, quasi-periodic biosignal datasets, fetal ECG datasets, blood pressure signals, cardiac magnetic field datasets, and heart rate signal datasets.

[0376] Example analyses can be used for the diagnosis and treatment of cardiac-related pathologies and conditions and / or neurological-related pathologies and conditions. Such assessments can be applied to the diagnosis and treatment (including surgical, minimally invasive, and / or pharmacological treatments) of any pathology or condition of any relevant system in a living organism in which biophysical signals are implicated. A cardiac example is the diagnosis of CAD, as well as other diseases or conditions disclosed herein, and their treatment (alone or in combination) through various therapies, such as coronary artery stent placement, atherectomy, angioplasty, medication prescription, and / or exercise prescription, nutritional and other lifestyle changes. Other cardiac-related pathologies or conditions that can be diagnosed include, for example, arrhythmias, congestive heart failure, valvular 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 thrombosis, and pulmonary hypertension due to other conditions (e.g., hematologic or other diseases)), and other cardiac-related pathologies, conditions, and / or diseases. Non-limiting examples of diagnosable neurological-related diseases, pathologies or conditions include, for example, epilepsy, schizophrenia, Parkinson's disease, Alzheimer's disease (and all other forms of dementia), autism spectrum disorders (including Asperger's syndrome), attention deficit hyperactivity disorder, Huntington's disease, muscular dystrophy, depression, bipolar disorder, brain / spinal cord tumors (malignant and benign), movement disorders, cognitive disorders, language disorders, various psychiatric disorders, brain / spinal cord / nerve injuries, chronic traumatic encephalopathy, cluster headaches, migraines, neuropathies (various forms, including peripheral neuropathy), phantom limb / pain, chronic fatigue syndrome, acute and / or chronic pain (including back pain, failed back surgery syndrome, etc.), movement dysfunction, anxiety disorders, conditions caused by infections or exogenous factors (such as Lyme disease, encephalitis, rabies), narcolepsy and other sleep disorders, post-traumatic stress disorder, neurological conditions / effects associated with stroke, aneurysm, hemorrhagic injury, etc., tinnitus and other hearing-related diseases / conditions, and vision-related diseases / conditions.

[0377] The following patents, applications, and publications listed below and throughout this document are hereby incorporated by reference in their entirety.

[0378] References

[0379] [1] I. Kononenko, “Machine learning for medical diagnosis: history, current status and prospects”, Artificial Intelligence in Medicine, 23 (1) 89–109 (2001).

[0380] [2] BA Mobley, E. Schechter, WE Moore, PA McKee, JE Eichner, “Prediction of coronary artery stenosis by artificial neural networks”, Artificial Intelligence in Medicine, 18 (3) 187–203 (2000).

[0381] [3] VL Patel, EH Shortliffe, M. Stefanelli, P. Szolovits, MR Berthold, R.Bellazzi, A. Abu-Hanna, “The coming of age of artificial intelligence in medicine”, Artificial Intelligence in Medicine, 46 (1) 5-17 (2009).

[0382] [4] V. Jahmunah, SL Oh, V. Rajinikanth, EJ Ciaccio, KH Cheong, UR Acharya, et al., “Automatic detection of schizophrenia using nonlinear signal processing methods,” Artificial Intelligence in Medicine, vol. 100, 101698 (September 2019).

[0383] [5] AM Tai, A. Albuquerque, NE Carmona, M. Subramanieapillai, DS Cha, M. Sheko, Y. Lee, R. Mansur, RS McIntyre, “Machine learning and big data: impact on disease modeling and treatment discovery in psychiatry,” Artificial Intelligence in Medicine, 101704 (2019).

[0384] [6] GK Hansson, “Inflammation, atherosclerosis, and coronary artery disease,” New England Journal of Medicine, 352 (16) 1685–1695 (2005).

[0385] [7] WG Members, D. Lloyd-Jones, RJ Adams, TM Brown, M. Carnethon, S. Dai, G. De Simone, TB 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).

[0386] [8] GA Mensah, DW Brown, “Overview of the Burden of Cardiovascular Disease in the United States,” Health Affairs, 26(1) 38-48 (2007).

[0387] [9] Y. N. Reddy, A. El-Sabbagh, R. A. Nishimura, “Comparison of pulmonary artery wedge pressure and left ventricular end-diastolic pressure for assessment of left-sided filling pressures,” JAMA Cardiology, 3 (6) 453–454 (2018).

[0388]

[10] M.J. Kern, T. Christopher, “Hemodynamic Examination Series II: LVEDP,” Catheterization and Cardiovascular Diagnosis, 44 (1) 70–74 (1998).

[0389]

[11] J.-H. Park, T. H. Marwick, “Uses and limitations of e / e' for echocardiographic assessment of left ventricular filling pressures,” Journal of Cardiovascular Ultrasound, 19 (4) 169–173 (2011).

[0390]

[12] SR Ommen, RA Nishimura, CP Appleton, F. Miller, JK Oh, MM Redfield, A. Tajik, “Clinical application of Doppler echocardiography and tissue Doppler imaging for estimation of left ventricular filling pressures: a comparative simultaneous Doppler-catheterization study,” Circulation, 102(15)1788–1794 (2000).

[0391]

[13] J. Allen, “Photoplethysmography and its application to clinical physiological measurements,” Physiological Measurements, 28(3) R1 (2007).

[0392]

[14] SD Fihn, JM Gardin, J. Abrams, K. Berra, JC Blankenship, AP Dallas, PS Douglas, JM Foody, TC Gerber, AL Hinderliter, et al., “2012 accf / aha / acp / aats / pcna / scai / sts guidelines for the diagnosis and management of patients with stable ischemic heart disease,” Journal of the American College of Cardiology, 60 (24) 2564–2603 (2012).

[0393]

[15] GN Levine, ER Bates, JC Blankenship, SR Bailey, J.A. Bittl, B. Cercek, CE Chambers, SG Ellis, RA Guyton, SM Hollenberg, et al., “2011 accf / aha / scai guidelines for percutaneous coronary intervention: executive summary,” Journal of the American College of Cardiology, 58 (24) 2550–2583 (2011).

[0394]

[16] LM Mielniczuk, GA Lamas, GC Flaker, G. Mitchell, SC Smith, BJ Gersh, SD Solomon, LA Moy´e, JL Rouleau, JD Rutherford, et al., “Left ventricular end-diastolic pressure and risk of subsequent heart failure in patients after acute myocardial infarction,” Congestive Heart Failure, 13(4) 209–214 (2007).

[0395]

[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).

[0396]

[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 independent of left ventricular ejection fraction,” BJA: British Journal of Anaesthesia, 97 (3) 292–297 (2006).

[0397]

[19] S. H. Strogatz, “Nonlinear dynamics and chaos: applications in physics, biology, chemistry, and engineering”, CRC Press, (2018).

[0398]

[20] AL Goldberger, DR Rigney, BJ West, “Chaos and Fractals in Human Physiology”, Scientific American, 262 (2) 42-49 (1990).

[0399]

[21] AL Goldberger, “Nonlinear dynamics, fractals, and chaos: applications to cardiac electrophysiology,” Annals of Biomedical Engineering, 18 (2) 195–198 (1990).

[0400]

[22] L. Glass, A. Beuter, D. Larocque, “Time delays, oscillations and chaos in physiological control systems,” Mathematical Biosciences, 90 (1-2) 111–125 (1988).

[0401]

[23] L. Glass, L. Glass, “Synchrony and rhythmic processes in physiology”, Nature, 410 (6825) 277(2001).

[0402]

[24] M.I. Owis, A.H. Abou-Zied, A.-B. Youssef, Y.M. Kadah, “Feature study based on nonlinear dynamics modeling for electrocardiogram arrhythmia detection and classification,” IEEE Transactions on Biomedical Engineering, 49 (7)733–736 (2002).

[0403]

[25] A. Voss, S. Schulz, R. Schroeder, M. Baumert, P. Caminal, “A method based on the derivation of nonlinear dynamics for the analysis of heart rate variability, Philosophical Transactions of the Royal Society A: Mathematical,” Physical and Engineering Sciences, 367 (1887) 277–296 (2008).

[0404]

[26] L. Glass, P. Hunter, A. McCulloch, “Heart theory: biomechanics, biophysics, and the nonlinear dynamics of cardiac function,” Springer Science & Business Media, (2012).

[0405]

[27] P. Billingsley, Ergodic Theory and Information, vol. 1, Wiley, New York, 1965.

[0406]

[28] T. Sauer, J. A. Yorke, M. Casdagli, “Embeddings,” Journal of Statistical Physics, 65 (3-4) 579–616 (1991).

[0407]

[29] A. Chatterjee, “Introduction to eigenorthogonal decomposition,” Current Science, 808–817 (2000).

[0408]

[30] A. Wolf, JB Swift, H.L. Swinney, J.A. Vastano, “Determination of Lyapunov exponents from time series”, Physics D: Nonlinear Phenomena, 16 (3) 285–317 (1985).

[0409]

[31] AN Kolmogorov, “Entropy per unit time as a metric invariant of automorphisms”, Proceedings of the Russian Academy of Sciences, vol. 124, pp. 754-755 (1959).

[0410]

[32] P. Grassberger, I. Procaccia, “Estimation of K-entropy from chaotic signals”, Physical Review A 28(4) 2591 (1983).

[0411]

[33] J. Theiler, “An efficient algorithm for estimating the correlation dimension from a set of discrete points,” Physical Review, A36 (9) 4456 (1987).

[0412]

[34] A. Pikovsky, J. Kurths, M. Rosenblum, J. Kurths, “Synchrony: A universal concept in nonlinear science”, vol. 12, Cambridge University Press (2003).

[0413]

[35] D. Dubin, “Rapid Interpretation of the Electrocardiogram: An Interactive Course,” Cover Publishing Company (2000).

[0414]

[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).

[0415]

[37] T. Chen, C. Guestrin, “Xgboost: A scalable tree boosting system,” in Proceedings of the 22nd acm-sigkdd international conference on knowledge discovery and data mining, ACM, pp. 785–794 (2016).

[0416]

[38] H. Zou, T. Hastie, “Regularization and variable selection via elastic nets,” Journal of the Royal Statistical Society: Series B (Statistical Methods), 67 (2) 301–320 (2005).

Claims

1. A method for non-invasively assessing a disease state or abnormal condition in a subject, the method comprising: obtaining, by one or more processors, a first biophysical signal dataset associated with saturation of oxyhemoglobin or deoxyhemoglobin of a subject, comprising a red photoplethysmography signal and an infrared photoplethysmography signal acquired over a plurality of cardiac cycles of the subject; obtaining, by the one or more processors, a second biophysical signal dataset associated with a cardiac signal of the subject, wherein the cardiac signal has been acquired over the plurality of cardiac cycles, and wherein the red photoplethysmography signal, the infrared photoplethysmography signal, and the cardiac signal are acquired by a surface sensor placed on the subject; determining, by the one or more processors, one or more values ​​associated with one or more synchrony dynamics characterizing a physiological relationship between the first biophysical signal dataset associated with saturation of the oxyhemoglobin and / or the deoxyhemoglobin and the second biophysical signal dataset associated with the cardiac signal, the one or more synchrony dynamics of the first biophysical signal dataset and the second biophysical signal dataset comprising a statistical evaluation of cardiac signal values ​​at landmarks defined by both the red photoplethysmography signal and the infrared photoplethysmography signal; as well as determining, by the one or more processors, an estimate of the presence and / or severity of a disease state based on the determined one or more values ​​associated with the one or more synchronicity dynamics, The method further comprises: determining, by the one or more processors, a Poincare map of synchrony of the first biophysical signal dataset and the second biophysical signal dataset; as well as Geometric properties of ellipses fitted to clusters in the Poincare map are extracted, wherein the extracted geometric properties of the ellipses are used to determine the estimate of the presence and / or severity of the disease state.

2. The method of claim 1, wherein the disease or condition is diagnosable based on an assessment indication and / or estimate of the presence, absence and / or severity of elevated or abnormal left ventricular end-diastolic pressure (LVEDP).

3. The method of claim 1 or 2, wherein the disease state or condition comprises coronary artery disease.

4. The method of claim 1 or 2, wherein the disease state or condition comprises pulmonary hypertension.

5. The method of claim 1 or 2, wherein the disease state or condition comprises pulmonary hypertension.

6. The method of claim 1 or 2, wherein the disease state or condition comprises pulmonary hypertension due to left heart disease.

7. The method of claim 1 or 2, wherein the disease state or condition comprises a rare disease that causes pulmonary hypertension.

8. The method of claim 1 or 2, wherein the disease state or condition comprises left ventricular heart failure or left-sided heart failure.

9. The method of claim 1 or 2, wherein the disease state or condition comprises right ventricular heart failure or right-sided heart failure.

10. The method of claim 1 or 2, wherein the disease state or condition comprises systolic heart failure.

11. The method of claim 1 or 2, wherein the disease state or condition comprises diastolic heart failure.

12. The method of claim 1 or 2, wherein the disease state or condition comprises ischemic heart disease.

13. The method of claim 1 or 2, wherein the disease state or condition comprises a cardiac arrhythmia.

14. The method of claim 1 or 2, wherein a landmark defined by both the red photoplethysmography signal and the infrared photoplethysmography signal is defined at a time when values ​​of the red photoplethysmography signal and the infrared photoplethysmography signal intersect.

15. The method according to claim 1 or 2, wherein: The one or more synchrony dynamic characteristics of the first and second biophysical signal datasets include statistical evaluations of values ​​of one of the red photoplethysmography signal or the infrared photoplethysmography signal at landmarks defined in a cardiac signal.

16. The method of claim 15, wherein the landmarks defined in the cardiac signal include correlation peaks associated with ventricular depolarization.

17. The method of claim 15, wherein the landmarks defined in the cardiac signal include correlation peaks associated with ventricular repolarization or atrial depolarization.

18. The method of claim 1 or 2, wherein the one or more synchrony dynamic characteristics of the first biophysical signal dataset and the second biophysical signal dataset comprise a statistical evaluation of time intervals between i) a first set of landmarks and ii) a second set of landmarks, wherein the first set of landmarks is defined between the red photoplethysmography signal and the infrared photoplethysmography signal, and the second set of landmarks is defined in the cardiac signal.

19. The method of claim 18, wherein the second set of landmarks defined in the cardiac signal comprises correlation peaks in the cardiac signal associated with ventricular depolarization.

20. The method of claim 18, wherein the second set of landmarks defined in the cardiac signal includes correlation peaks in the cardiac signal associated with ventricular repolarization or atrial depolarization.

21. The method of claim 18, wherein a first set of landmarks defined by both the red photoplethysmography signal and the infrared photoplethysmography signal is defined at times when values ​​of the red photoplethysmography signal and the infrared photoplethysmography signal intersect.

22. A method according to claim 1 or 2, wherein the one or more synchronization dynamic characteristics of the first biophysical signal dataset and the second biophysical signal dataset include a statistical evaluation of the phase relationship between the period of one of the red photoplethysmography signal or the infrared photoplethysmography signal and the period of the cardiac signal.

23. The method according to claim 1 or 2, further comprising: Generating, by the one or more processors, a visualization of the estimate of the presence, absence, and / or severity of the disease state, wherein the generated visualization is rendered and displayed on a display of a computing device and / or presented in a report.

24. The method according to claim 1 or 2, further comprising: determining, by the one or more processors, a histogram of synchrony between the first biophysical signal dataset and the second biophysical signal dataset; as well as 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, skewness, and kurtosis, wherein the extracted first statistical parameter is used to determine the estimated value for the presence and / or severity of the disease state.

25. The method according to claim 24, further comprising: determining, by the one or more processors, a Poincare map of synchrony of the first biophysical signal dataset and the second biophysical signal dataset; as well as Extracting a second statistical parameter of the Poincare map, wherein the second statistical parameter of the histogram is selected from the group consisting of mean, mode, median, skewness, and kurtosis, wherein the extracted second statistical parameter is used to determine the estimate for the presence and / or severity of the disease state.

26. The method of claim 1, wherein the Poincare map is generated by iteratively plotting parameters associated with the synchrony of the first biophysical signal dataset and the second biophysical signal dataset at a first index x-1 and a second index x on the x-axis, and iteratively plotting the parameters at the second index x and a third index x+1 on the y-axis.

27. The method according to claim 26, wherein The parameter is a time interval between a landmark of the cardiac signal and a crossing point between the red photoplethysmography signal and the infrared photoplethysmography signal.

28. The method according to claim 26, wherein The parameter is an amplitude signal value of the cardiac signal at a crossing landmark defined between the red photoplethysmography signal and the infrared photoplethysmography signal.

29. The method of claim 26, wherein the parameter is an amplitude signal value of a photoplethysmography signal at a landmark defined in the cardiac signal.

30. A system comprising: processor; as well as A memory storing instructions, wherein execution of the instructions by the processor causes the processor to perform the method according to any one of claims 1-29.

31. A computer-readable medium having instructions stored thereon, wherein execution of the instructions by a processor causes the processor to perform the method according to any one of claims 1-29.

32. A system comprising: processor; as well as a memory having stored thereon instructions, wherein execution of the instructions by the processor causes the processor to: obtaining a first biophysical signal dataset associated with saturation of oxyhemoglobin or deoxyhemoglobin of a subject, comprising a red photoplethysmography signal and an infrared photoplethysmography signal acquired over a plurality of cardiac cycles of the subject; obtaining a second biophysical signal dataset associated with a cardiac signal of the subject, wherein the cardiac signal has been acquired over the plurality of cardiac cycles, and wherein the red photoplethysmography signal, the infrared photoplethysmography signal, and the cardiac signal are acquired by a surface sensor placed on the subject; determining one or more values ​​associated with one or more synchrony dynamics characterizing a physiological relationship between the first biophysical signal dataset associated with saturation of the oxyhemoglobin and / or the deoxyhemoglobin and the second biophysical signal dataset associated with the cardiac signal, the one or more synchrony dynamics of the first biophysical signal dataset and the second biophysical signal dataset comprising a statistical evaluation of cardiac signal values ​​at landmarks defined by both the red photoplethysmography signal and the infrared photoplethysmography signal; as well as determining an estimate of the presence and / or severity of a disease state based on the determined one or more values ​​associated with the one or more synchronicity dynamic characteristics, Wherein execution of the instructions by the processor further causes the processor to determine a Poincare map of the synchrony of the first biophysical signal dataset and the second biophysical signal dataset; and to extract geometric properties of ellipses that fit to the clusters in the Poincare map, wherein the extracted geometric properties of the ellipses are used to determine the estimate of the presence and / or severity of the disease state.

33. The system of claim 32, wherein a landmark defined by both the red photoplethysmography signal and the infrared photoplethysmography signal is defined at a time when values ​​of the red photoplethysmography signal and the infrared photoplethysmography signal intersect.

34. The system according to any one of claims 32-33, wherein: The synchrony dynamics of the first and second biophysical signal datasets include statistical evaluations of values ​​of one of the red photoplethysmography signal or the infrared photoplethysmography signal at landmarks defined in a cardiac signal.

35. The system of claim 34, wherein the landmarks defined in the cardiac signal include correlation peaks associated with ventricular depolarization.

36. The system of claim 34, wherein the landmarks defined in the cardiac signal include correlation peaks associated with ventricular repolarization or atrial depolarization.

37. A system according to claim 32 or 33, wherein the one or more synchrony dynamic characteristics of the first biophysical signal dataset and the second biophysical signal dataset include a statistical evaluation of the time intervals between i) a first set of landmarks and ii) a second set of landmarks, wherein the first set of landmarks is defined between the red photoplethysmography signal and the infrared photoplethysmography signal, and the second set of landmarks is defined in the cardiac signal.

38. The system of claim 37, wherein the second set of landmarks defined in the cardiac signal includes correlation peaks in the cardiac signal associated with ventricular depolarization.

39. The system of claim 37, wherein the second set of landmarks defined in the cardiac signal includes correlation peaks in the cardiac signal associated with ventricular repolarization or atrial depolarization.

40. The system of claim 37, wherein a first set of landmarks defined by both the red photoplethysmography signal and the infrared photoplethysmography signal is defined at times when values ​​of the red photoplethysmography signal and the infrared photoplethysmography signal intersect.

41. A system according to claim 32 or 33, wherein the one or more synchronization dynamic characteristics of the first biophysical signal dataset and the second biophysical signal dataset include a statistical evaluation of the phase relationship between the period of one of the red photoplethysmography signal or the infrared photoplethysmography signal and the period of the cardiac signal.

42. A system according to claim 32 or 33, wherein execution of the instructions by the processor further causes the processor to generate a visualization of an estimate of the presence, absence and / or severity of the disease state, wherein the generated visualization is presented and displayed on a display of a computing device and / or presented in a report.

43. A system according to claim 32 or 33, wherein execution of the instructions by the processor further causes the processor to determine a histogram of the synchrony of the first biophysical signal dataset and the second biophysical signal dataset; and extract a first statistical parameter of the histogram, wherein the first statistical parameter of the histogram is selected from a group consisting of mean, mode, median, skewness, and kurtosis, wherein the extracted first statistical parameter is used to determine the estimated value for the presence or absence and / or severity of the disease state.

44. A system according to claim 43, wherein execution of the instructions by the processor further causes the processor to determine a Poincare map of the synchrony of the first biophysical signal dataset and the second biophysical signal dataset; and extract a second statistical parameter of the Poincare map, wherein the second statistical parameter of the histogram is selected from the group consisting of mean, mode, median, skewness, and kurtosis, and wherein the extracted second statistical parameter is used to determine the estimated value for the presence or absence and / or severity of the disease state.

45. The system of claim 32, wherein the Poincare map is generated by iteratively plotting a parameter associated with the synchrony of the first biophysical signal dataset and the second biophysical signal dataset at a first index x-1 and a second index x on the x-axis, and iteratively plotting the parameter at the second index x and a third index x+1 on the y-axis.

46. ​​The system of claim 45, wherein: The parameter is a time interval between a landmark of the cardiac signal and a crossing point between the red photoplethysmography signal and the infrared photoplethysmography signal.

47. The system of claim 45, wherein: The parameter is an amplitude signal value of the cardiac signal at a crossing landmark defined between the red photoplethysmography signal and the infrared photoplethysmography signal.

48. The system of claim 45, wherein the parameter is an amplitude signal value of a photoplethysmography signal at a landmark defined in the cardiac signal.

49. The system of claim 32 or 33, further comprising: A measurement system is configured to acquire one or more photoplethysmography signals.

50. The system of claim 32 or 33, further comprising: A measurement system is configured to acquire one or more cardiac signals.

51. The system of claim 32 or 33, further comprising: a first measurement system configured to acquire one or more photoplethysmography signals; as well as The second measurement system is configured to acquire one or more cardiac signals.

Images