Characterization of disease states using system dynamic control of ECG signals

Abstract

Disclosed are systems and methods that can differentiate between disease states using ECG data analyzed via dynamic control analysis. In some embodiments, the disease states are Hypertrophic Cardiomyopathy (HCM) and Left Ventricular Hypertrophy. In some embodiments, the exemplary system and method is configured to transform ECG data in the time domain into complex numbers in the frequency domain (s-domain). A pattern recognition or AI classifier can then find patterns of the complex numbers in the s-domain to differentiate between the disease states (e.g., HCM and LVH).

Description

Attorney Docket No.10046-631WO1 8448 CHE CHARACTERIZATION OF DISEASE STATES USING SYSTEM DYNAMIC CONTROL OF ECG SIGNALS Cross-Reference to Related Applications

[0001] This application claims the benefit of priority to U.S Provisional Application No. 63 / 676,692, filed July 29, 2024, entitled “DIAGNOSIS OF HYPERTROPHIC CARDIOMYOPATHY (HCM) AND LEFT VENTRICULAR HYPERTROPHY (LVH) USING DYNAMIC CONTROL ANALYSIS OF ECG SIGNALS,” which is expressly incorporated by reference herein in its entirety. Background

[0002] An electrocardiogram (ECG) is a recording of the heart's electrical activity across multiple cardiac cycles. ECG are typically presented as a graph of voltage versus time of the electrical activity of the heart using electrodes placed on the skin. ECG can also be acquired via wearable devices and implanted devices. ECG-based diagnosis is very economical and much less expensive as compared to other medical modalities.

[0003] An electrocardiogram (ECG or EKG) is a quick test that is typically used to check the heartbeat and record the electrical signals in the heart to diagnose heart attacks, arrhythmias, chest pain, and heart problems.

[0004] There is a benefit to using ECG to diagnose or prescreen for disease. Summary

[0005] An exemplary system and method are disclosed that can differentiate between Hypertrophic Cardiomyopathy (HCM) and Left Ventricular Hypertrophy using only ECG data analyzed via dynamic control analysis. In some embodiments, the exemplary system and method is configured to transform ECG data in the time domain into complex numbers in the frequency domain (s-domain). A pattern recognition or AI classifier can then find patterns of the complex numbers in the s-domain to differentiate between HCM and LVH.

[0006] The ECG data of LVH and HCM are similar to each other due to similar symptoms in the heart. However, the treatments for these conditions are different from one another, and LVH and HCM can cause severe outcomes like heart failure if left alone. HCM is the most common cause of SCD in people under the age of 35, especially, young athletes. The current standard of care to differentiate between LCH and HCM includes echocardiography and genetic testing, but both echocardiography and genetic testing are expensive and require more equipment compared to an ECG.Attorney Docket No.10046-631WO1 8448 CHE

[0007] The exemplary system and method allow for ECG, as a more cost effective / lower cost and accessible method of clinical measurement, to be used to diagnose HCM. The exemplary system and method can be implemented in an analysis system, a standalone device, or integrated with existing ECG machines used in clinical offices or hospitals.

[0008] In an aspect, a method is disclosed of pre-screening or identifying misdiagnosis of hypertrophic cardiomyopathy patient as having left ventricular hypertrophy, the method comprising: providing, by a processor, a data set consisting of electrocardiographic (ECG) signal data acquired via an ECG device of a patient, wherein the patient has been diagnose or scored as having a high or moderate likelihood of left ventricular hypertrophy (LVH); generating, by the processor, via a frequency domain transform operation (e.g., s-domain transform operation, e.g., Laplace transform operator), complex number values (e.g., frequency domain poles, eigenvalues, or complex number expressions), or a number derived therefrom, of the ECG signal data set; and determining, by the processor, (i) executing a trained AI model using the generated complex number values, (ii) executing a rule-based algorithm using the generated complex number values, or (iii) analyzing an S-domain plot generated from the generated complex number values, an indicator of a presence or non-presence of hypertrophic cardiomyopathy, wherein the indicator is outputted and used by a clinician to subsequently (i) diagnose the presence of hypertrophic cardiomyopathy or (ii) revise the diagnosis of left ventricular hypertrophy to the presence of hypertrophic cardiomyopathy.

[0009] In some embodiments, the output indicator is used to order an echocardiogram or a genetic test for validation of the presence of hypertrophic cardiomyopathy.

[0010] In some embodiments, the ECG device is a standard 12-lead ECG equipment. In some embodiments, the ECG signal data includes one or more signals obtained from at least Lead I, Lead II, Lead III, V1 through V6 leads, and / or aVR, aVL and aVF for a standard 12- lead ECG equipment. In some embodiments, the ECG signal data includes at least one signal acquired from lead V5 and lead V6 of the ECG device. In some embodiments, the ECG signal data includes a recording of at least 1 second (e.g., at least 5 seconds, at least 10 seconds, at least 15 seconds, at least 20 seconds, at least 25 seconds, at least 30 seconds, or at least 1 minute).

[0011] In some embodiments, the patient has been diagnosed or scored as having a high or moderate likelihood of left ventricular hypertrophy (LVH) based on Cornell criteria or Sokolow-Lyon Criteria.

[0012] In some embodiments, the patient has been diagnosed or scored based on the Cornell criteria or the Sokolow-Lyon Criteria using the data set used to determine the presenceAttorney Docket No.10046-631WO1 8448 CHE or non-presence of hypertrophic cardiomyopathy, wherein the data set is first used to pre-screen to identify (i) potential presence or (ii) presence of left ventricular hypertrophy, and wherein the data set is then used to screen or determine for (i) potential presence or (ii) presence of hypertrophic cardiomyopathy.

[0013] In some embodiments, the patient has been diagnosed or scored based on the Cornell criteria or the Sokolow-Lyon Criteria using a first data set acquired via an ECG device to determine the presence or non-presence of left ventricular hypertrophy, wherein the data set used to screen or determine for (i) potential presence or (ii) presence of hypertrophic cardiomyopathy is acquired after the first data set.

[0014] In some embodiments, the trained AI model is trained using an influence function (e.g., to approximate a change in model parameters and predictions upon upweighting or removing a training data point).

[0015] In some embodiments, the method further includes preprocessing, by the processor (e.g., using analog and digital filters), the provided data set to remove the DC and high- frequency noise of the electrocardiographic (ECG) signal data prior to generating the complex number values.

[0016] In some embodiments, the method further includes preprocessing, by the processor (e.g., using a magnitude filter or a pulse detector), the provided data set or a cleaned version of the provided data set to extract ECG signatures prior to generating the complex number values, wherein the extracted ECG signatures are used for the frequency domain transform operation.

[0017] In some embodiments, the method further includes at least one of: detecting peaks in the provided data set (e.g., using peak detector, e.g., Pan Tompkin-based filter); normalizing the detected peaks; removing peaks having a peak height outside a predefined range; removing peaks having a peak distance outside a predefined range; or a combination thereof.

[0018] In some embodiments, the method further includes determining a location of a QRS complex in each ECG cycle of the ECG signal data; and separating the QRS complexes around a respective peak of each QRS interval with a tuned trunked window, where the QRS complexes of each interval are transformed from the time domain to the s-domain.

[0019] In some embodiments, the complex number values comprise frequency domain poles having natural frequency and damping values, wherein the indication of hypertrophic cardiomyopathy is determined by the damping values, natural frequency values, or a combination of either one, having a value above a pre-defined threshold.

[0020] In some embodiments, the ECG signal data includes one or more signals obtained from at least Lead I, Lead II, Lead III, V1through V6leads, and / or aVR, aVL and aVF for aAttorney Docket No.10046-631WO1 8448 CHE standard 12-lead ECG equipment. In some embodiments, the ECG signal data includes at least one signal acquired from lead V5 and lead V6 of the ECG device. In some embodiments, the ECG signal data includes a recording of at least 1 second (e.g., at least 5 seconds, at least 10 seconds, at least 15 seconds, at least 20 seconds, at least 25 seconds, at least 30 seconds, or at least 1 minute).

[0021] In another aspect, a method is disclosed comprising: providing, by a processor, a data set consisting of electrocardiographic (ECG) signal data acquired via an ECG device of a patient (e.g., wherein the patient has only been screened with the ECG device); generating, by the processor (e.g., via a frequency domain transform operation, e.g., s-domain transform operation, e.g., Laplace transform operator, linear operators that converts ordinary differential equations to algebraic state space matrices), complex number values (e.g., frequency domain poles or complex number expressions), or a number derived therefrom, of the ECG signal data set, including a damping value; and determining, by processor, (i) executing a trained AI model using the generated complex number values, (ii) executing rule-based algorithm using the generated complex number values, (iii) analysis a plotted s-domain plot generated from the generated complex number values, an indicator of a presence or non-presence of cardiac condition, wherein the indicator is outputted and used by a clinician to subsequently (i) diagnose the presence of the cardiac condition or (ii) revised a prior diagnose made using the ECG signal data.

[0022] In some embodiments, damping value and natural frequency are obtained from time domain ordinary differential equations and eigenvalue decomposition of a linear state space system matrix.

[0023] In some embodiments, the indicator of the presence or non-presence of the cardiac condition includes an indicator for bradycardia (among others, e.g., heart diseases, neurological diseases (e.g., epilepsy, Parkinson, Alzheimer's), muscle function problems).

[0024] In another aspect, a method is disclosed of pre-screening or identifying misdiagnosis of a patient as having a first disease, the method comprising: providing, by a processor, a data set consisting of electrocardiographic (ECG) signal data acquired via an ECG device of a patient, wherein the patient has been diagnose or scored as having a high or moderate likelihood of a first disease (e.g., left ventricular hypertrophy (LVH)); generating, by the processor, dynamic control domain poles, complex number expressions, or damping ratio and natural frequency values, or a number derived therefrom, of the ECG signal data set; and determining, by the processor, (i) executing a trained AI model using the generated complex number values, (ii) executing a rule-based algorithm using the generated complex numberAttorney Docket No.10046-631WO1 8448 CHE values, an indicator of a presence or non-presence of a second disease (e.g., hypertrophic cardiomyopathy), wherein the indicator is outputted and used by a clinician to subsequently (i) diagnose the presence of the second disease or (ii) revise the misdiagnosis of the first disease to the second disease.

[0025] In another aspect, a system is disclosed that can perform any one of the above- discussed methods. The system includes a processor and memory having instructions stored thereon to perform the method.

[0026] In another aspect, a non-transitory computer-readable medium is disclosed, having instructions stored thereon, wherein execution of the instructions by a processor causes the processor to perform any one of the above-discussed methods. Brief Description of the Drawings

[0027] Fig. 1A shows an example diagram of a computing device that can be used to determine indicators of a disease or disorder using only ECG data.

[0028] Fig. 1B shows an example system diagram for determining indicators of a disease or disorder using only ECG data.

[0029] Figs. 2A-2C show embodiments of a method for determining indicators of hypertrophic cardiomyopathy using a trained AI model (Fig.2A), s-domain plot analysis (Fig. 2B), and a rule-based algorithm (Fig.2C).

[0030] Fig.3 shows a representation of the complex number values.

[0031] Fig. 4 shows the data for four patients in the s-domain with the natural frequency and damping values illustrated.

[0032] Figs. 5A-5B show an example of no overlapping ECG lines (Fig. 5A) and overlapping ECG lines (Fig.5B).

[0033] Fig.6 shows an exemplary system overview for the determination of indicators for hypertrophic cardiomyopathy.

[0034] Fig. 7 shows the result of a preliminary digitization process to convert paper ECG waveforms to digitized signals.

[0035] Fig.8 shows a fully processed ECG waveform following an exemplary digitization methodology. Detailed Description

[0036] DefinitionsAttorney Docket No.10046-631WO1 8448 CHE

[0037] As used herein, the term “subject” refers to any animal (e.g., a mammal), including, but not limited to, humans, non-human primates, rodents, and the like. Typically, the terms “subject” and “patient” are used interchangeably herein in reference to a human subject.

[0038] Some references, which may include various patents, patent applications, and publications, are cited in a reference list and discussed in the disclosure provided herein. The citation and / or discussion of such references is provided merely to clarify the description of the disclosed technology and is not an admission that any such reference is “prior art” to any aspects of the disclosed technology described herein. In terms of notation, “[n]” corresponds to the nth reference in the list. For example, [1] refers to the first reference in the list. All references cited and discussed in this specification are incorporated herein by reference in their entireties and to the same extent as if each reference was individually incorporated by reference.

[0039] Example Method

[0040] An exemplary method is based on the premise that the cardiovascular system can be treated as a dynamic system and that under HCM, the dynamic system reacts abnormally due to temporal and spatial destabilization. Beat signatures can thus be derived from the ECG signal, e.g., using peak detection techniques in which the signatures can be used to estimate the pole of the cardiovascular system in the s-domain. A pole is defined as a complex number that represents the natural frequency and shape of a beat. The location of the poles on the complex plane is representative of the state of the cardiovascular system.

[0041] Because the exemplary method can be used to diagnose HCM using ECG testing, it can be less expensive and more broadly available compared to the other methods. The exemplary method can use s-domain poles, or equivalents in the time domain, to readily quantify and compare the ECG shape information for diagnosis. The exemplary method can be performed in a standalone device, a cloud device, or edge device configured to receive electrocardiographic signal data. In some aspects, the method is implemented in a clinical work flow, as shown in Fig.6.

[0042] Referring specifically to Figs. 2A-2C, exemplary methods 200 (shown as 200a, 200b, 200c) are illustrated for determining an indicator of a disease or disorder (e.g., hypertrophic cardiomyopathy). Method 200 includes providing 210, by a processor, a data set consisting of electrocardiographic (ECG) signal data acquired via an ECG device of a patient. In some embodiments, the patient has been diagnosed or scored as having a high or moderate likelihood of a particular disease or disorder (e.g., left ventricular hypertrophy (LVH)). In some embodiments, the ECG signal data includes one or more signals obtained from at least Lead I,Attorney Docket No.10046-631WO1 8448 CHE Lead II, Lead III, V1through V6leads, and / or aVR, aVL and aVF for a standard 12-lead ECG equipment. In some embodiments, the ECG signal data includes at least one signal acquired from lead V5 and lead V6 of the ECG device. In some embodiments, the ECG signal data includes a recording of at least 1 second (e.g., at least 5 seconds, at least 10 seconds, at least 15 seconds, at least 20 seconds, at least 25 seconds, at least 30 seconds, or at least 1 minute).

[0043] The method 200 then includes generating 220, by the processor, via a frequency domain transform operation (e.g., s-domain transform operation, e.g., Laplace transform operator), complex number values (e.g., frequency domain poles or complex number expressions), damping values and natural frequency and their equivalent expressions (e.g., generated using time-domain), any use of the angle obtained by using the ratio of the real part of the complex number over the imaginary part of the complex number, or a derived form of this ratio, or a number derived therefrom, of the ECG signal data set. Fig. 4 shows example data for four patients in the s-domain with the natural frequency and damping values illustrated.

[0044] Referring specifically to Fig.2A, the method 200a further includes executing 230a a trained artificial intelligence (AI) model using the generated complex number values via the processor. In some embodiments, the trained AI model includes a neural network that has been trained with training data, including complex number values corresponding to a particular disease state. The training operation after the fully connected layer, in the provided example, is configured to employ conventional operations, e.g., employing gradient descent and various normalization operations, and thus are not further described herein. Other training operations may also be employed. Once the trained AI model is obtained, complex number values (e.g., frequency domain poles or complex number expressions), damping values and natural frequencies (either via frequency domain transform operation or time-domain ordinary differential equations or Eigenvalue decomposition), or a number derived therefrom, of the ECG signal data set derived from ECG signal data from a patient can be supplied to the AI model to determine 240 an indicator of a presence or non-presence of the disease or disorder.

[0045] In some embodiments, the trained AI model is trained using an influence function (e.g., to approximate a change in model parameters and predictions upon upweighting or removing a training data point).

[0046] Fig.2B illustrates another method 200b for determining an indicator of a disease or disorder (e.g., hypertrophic cardiomyopathy) according to the present disclosure. In method 200b, the generated complex number values (e.g., frequency domain poles or complex number expressions), damping values and natural frequency and their equivalent expressions (e.g., generated using time-domain), any use of the angle obtained by using the ratio of the real partAttorney Docket No.10046-631WO1 8448 CHE of the complex number over the imaginary part of the complex number, or a derived form of this ratio, or a number derived therefrom, of the ECG signal data set obtained from step 220 are translated to an s-domain plot. The resulting plot may be analyzed 230b (e.g., via the processor or a look-up table (LUT)) to determine 240 the indicator of a presence or non- presence of the disease or disorder.

[0047] Fig.2C shows another method 200c according to the present disclosure. In method 200c, the s-domain generated from 220 is provided to a rule-based algorithm 230c using the generated complex number values. The rule-based algorithm may include any model having a set of predetermined rules and corresponding classification groupings based on a set of predetermined ranges of parameters.

[0048] The determined indicator 240 (e.g., HCM index or HCM likelihood score) can be outputted 250 and used by a clinician to subsequently (i) diagnose the presence of a disease or disorder (e.g., hypertrophic cardiomyopathy) or (ii) revise the diagnosis of another disease or disorder (e.g., left ventricular hypertrophy).

[0049] In some embodiments, the indicator (e.g., HCM index or HCM likelihood score) is outputted as a report. As used herein, the term “report” is intended to describe a presentation of multidimensional data. A report can include both image data and textual data and may be considered as an analysis tool that can be used to view, manipulate, and print data. Reports can further include medical reports related to the treatment and care of a patient.

[0050] In some embodiments, the method is employed to prescreen or identify misdiagnosis of a hypertrophic cardiomyopathy patient as having left ventricular hypertrophy.

[0051] In some embodiments, the report is displayed on a graphical user interface for clinical or informative review. The term “graphical user interface,” or GUI, may be used in the singular or the plural to describe one or more graphical user interfaces and each of the displays of a particular graphical user interface. Therefore, a GUI may represent any graphical user interface, including but not limited to, a web browser, a touch screen, or a command line interface (CLI) that processes information and efficiently presents the information results to the user. In general, a GUI may include a plurality of user interface (UI) elements, some or all associated with a web browser, such as interactive fields, pull-down lists, and buttons operable by the business suite user. These and other UI elements may be related to or represent the functions of the web browser.

[0052] In some embodiments, the output indicator is used to order an echocardiogram or a genetic test for validation of the presence of hypertrophic cardiomyopathy (e.g., if the output indicator is shown to fall within a predetermined range or exceed a predetermined threshold).Attorney Docket No.10046-631WO1 8448 CHE

[0053] In some embodiments, the ECG device includes a standard 12-lead ECG equipment.

[0054] LVH Pre-screening. There are established criteria that guide the diagnosis of LVH on an ECG, the most well-known being the Cornell criteria and Sokolow-Lyon criteria. ECGs map the electrical activity of the heart. Both LVH and HCM can be seen on an ECG due to pathologic structural changes that manifest as characteristic electrical changes. It can be challenging distinguishing HCM from LVH by relying on the time domain representation of an ECG due to overlap in the electrical profiles of both conditions resulting from similar underlying structural changes. However, while minute differences in how HCM and LVH thicken the heart muscle may appear similar in the time domain, the data transformed into the s-domain can illustrate subtle differences between HCM and LVH.

[0055] In some embodiments, the patient has been diagnosed or scored as having a high or moderate likelihood of left ventricular hypertrophy (LVH) based on Cornell criteria or Sokolow-Lyon Criteria.

[0056] HCM Screening from LVH analysis. In some embodiments, the patient has been diagnosed or scored based on the Cornell criteria or the Sokolow-Lyon Criteria using the data set to determine the presence or non-presence of hypertrophic cardiomyopathy, and the data set is first used to pre-screen to identify (i) potential presence or (ii) presence of left ventricular hypertrophy, and then later used to screen or determine for (i) potential presence or (ii) presence of hypertrophic cardiomyopathy.

[0057] In some embodiments, the patient has been diagnosed or scored based on the Cornell criteria or the Sokolow-Lyon Criteria using a first data set acquired via an ECG device to determine the presence or non-presence of left ventricular hypertrophy and the data set is used to screen or determine (i) potential presence or (ii) presence of hypertrophic cardiomyopathy acquired after the first data set. Left ventricular hypertrophy (LVH) generally refers to a thickening of the left ventricle, which is typically caused by an increased left ventricular load. While almost all patients with hypertrophic cardiomyopathy (HCM) have LVH, not all patients with LVH have HCM. Modern computers and humans lack the ability to determine whether a patient has HCM just by seeing markers for LVH on a raw ECG plot. That LVH could be HCM, but it also could be from other causes such as high blood pressure, aortic stenosis, or even vigorous exercise. In clinical applications, HCM is normally diagnosed following expensive procedures, such as when a patient has unexplained wall thickness >15 mm on an echo or MRI, or >13 mm with some additional criteria i.e. family history of HCM, etc. The vast majority of HCM is genetic, however, the standard of care is not usually to identifyAttorney Docket No.10046-631WO1 8448 CHE the specific genetic variant, but rather to screen first degree relatives at regular intervals with echos to see if they are developing signs of HCM. Early diagnosis can be critical because HCM greatly increases the risk of sudden cardiac death, a risk that can be largely mitigated by promptly administering treatment (e.g., using defibrillators) in patients who are considered high risk.

[0058] Example Data Processing. In some embodiments, the method further includes preprocessing, by the processor (e.g., using analog and digital filters), the provided data set to remove the DC and high-frequency noise of the electrocardiographic (ECG) signal data prior to generating the complex number values. For example, the preprocessing can include digitization of paper ECG waveforms and / or other filtering processes.

[0059] In some embodiments, the method further includes preprocessing, by the processor (e.g., using a magnitude filter or a pulse detector), the provided data set or a cleaned version of the provided data set to extract ECG signatures prior to generating the complex number values, wherein the extracted ECG signatures are used for the frequency domain transform operation.

[0060] In some embodiments, the method includes at least one of: detecting peaks in the provided data set (e.g., using peak detector, e.g., Pan Tompkin-based filter); normalizing the detected peaks; removing peaks having a peak height outside a predefined range; and removing peaks having a peak distance outside a predefined range; or a combination thereof.

[0061] In some embodiments, the method includes determining the location of a QRS complex in each ECG cycle of the ECG signal data; and separating the QRS complexes around a respective peak of each QRS interval with a tuned trunked window, where the QRS complexes of each interval are transformed from the time domain to the s-domain (also referred to as the “complex frequency domain”).

[0062] In some embodiments, the complex number values include frequency domain poles having natural frequency and damping values, wherein the indication of hypertrophic cardiomyopathy is determined by the damping values, natural frequency values, or a combination of either one, having a value above or below a pre-defined threshold.

[0063] In some embodiments, the ECG data set includes at least one signal acquired from lead V5 and lead V6 of the ECG device (e.g., for a standard 12-lead ECG equipment). The ECG data set can be data from other leads in a 12 lead ECG, e.g., V1 to V4, Lead I, Lead II, Lead III, aVR, aVL, aVF.

[0064] The exemplary method can be used in cardiac clinics to evaluate heart conditions. The exemplary method can be implemented into current medical processes with no extra cost and can be readily accessible to economically disadvantaged populations.Attorney Docket No.10046-631WO1 8448 CHE

[0065] Because the exemplary method is based on ECG signals that are routinely collected for HCM and LVH patients, if the algorithm cannot be used as the only diagnosis tool due to accuracy issues, it still can be added as a supplementary function as long as it gives a diagnostic indication.

[0066] From experimental results, it is observed that the pole locations of HCM are different from those of the LVH in the complex plane. The difference is used to distinguish HCM from LVH.

[0067] From experimental results, it is also observed that the damping value locations of HCM are different from those of the LVH in the complex plane. The difference can be used to distinguish HCM from LVH.

[0068] The methodology may be employed for all physiological signal-based diagnoses and prognoses of other diseases or disorders, including but not limited to heart diseases, neurological diseases (epilepsy, Parkinson's, Alzheimer's), muscle function problems, etc. In some embodiments, the disease or disorder includes: hypertrophic, premature ventricular contraction-mediated, arrhythmogenic right ventricular, peripartum, stress, LV non- compaction, ischemic, tachycardia-mediated, alcohol-induced, drug-induced, medication- induced, chemo-induced, hypertensive tachyarrhythmias - sinus tachycardia, atrial fibrillation, atrial flutter, atrial tachycardia, atrioventricular re-entry tachycardia, atrioventricular nodal re- entry tachycardia, ventricular tachycardia, ventricular fibrillation, conduction abnormalities e.g., heart block, bundle branch block, sinus bradycardia, genetic conditions e.g., Brugada Syndrome, long QT syndrome valvular, aortic stenosis, aortic regurgitation, mitral stenosis, mitral regurgitation, tricuspid stenosis, tricuspid regurgitation, pulmonic stenosis, pulmonic regurgitation, LV dysfunction, electrolyte abnormalities, amyloidosis, sarcoidosis, myocarditis, pericarditis, congenital HIV hemochromatosis, and Chagas Disease.

[0069] Obtaining ^^, ^ using Laplace transform

[0070] The Laplace transform is defined as: ^^^^ = ℒ^^^^^^ ≡ ^ ^^^^^^^^^^ = ^ ^^^^^^^^^^^^^^^ ^1^

[0071] where ^ = & + !^, a complex number with &, ^ ∈ *.

[0072] After applying Laplace transform, the input and output relationship of a system or signal, also defined as a transfer function, can be described as:Attorney Docket No.10046-631WO1 8448 CHE ^= ,^^^ . ^ ^ ^ ^+^ ^ / 0 ^ − 12 … ^ − 14-^^^ =^^ − 5 ^ ^ ^ ^3^2 … ^ − 5^

[0073] Where 7^^^ is the input and 8^^^ is the output. Term . / 0is the DC gain of the system. The numerator binomial terms 91:;:∈92…4;are called zeros and the denominator binomial terms 95<;<∈92…^;are called

[0074] For the present example,=considered.

[0075] Once the poles are obtained, they can be used for feature characterization. The polescould be real numbers or complex conjugate pairs. If the pole is a real pole, we only have ^ =&. Details of complex poles are explained as below.

[0076] Figure 3 shows two poles, represented by a pair of complex conjugates, s= σ±jω_d. The natural frequency (ω_n) and the damping ratio (ζ) of the system can be obtained from s. The relationship between these parameters is calculated using the following equations. ^^ = ?&@ + ^A@ , ^ = ^^B =^ ?^C^^DC(4)

[0077] Onederived forms can be used to diagnose HCM from LVH, and this can apply toother heart abnormalities. By way of non-limiting example, the methodology can be extended to a number of other diseases or disorders including cardiomyopathies e.g., hypertrophic, premature ventricular contraction-mediated, arrhythmogenic right ventricular, peripartum, stress, LV non-compaction, ischemic, tachycardia-mediated, alcohol-induced, drug-induced, medication-induced, chemo-induced, hypertensive tachyarrhythmias - sinus tachycardia, atrial fibrillation, atrial flutter, atrial tachycardia, atrioventricular re-entry tachycardia, atrioventricular nodal re-entry tachycardia, ventricular tachycardia, ventricular fibrillation, conduction abnormalities e.g., heart block, bundle branch block, sinus bradycardia, genetic conditions e.g., Brugada Syndrome, long QT syndrome valvular, aortic stenosis, aortic regurgitation, mitral stenosis, mitral regurgitation, tricuspid stenosis, tricuspid regurgitation, pulmonic stenosis, pulmonic regurgitation, LV dysfunction, electrolyte abnormalities, amyloidosis, sarcoidosis, myocarditis, pericarditis, congenital HIV hemochromatosis, and Chagas Disease.

[0078] Obtaining ωn, ζ using time domain methods

[0079] The input output relationship can also be represented by a series of 1storder and2nd order forms in the time domain as below:Attorney Docket No.10046-631WO1 8448 CHE 1st order E AFA^ + 8 = 7^^^C 2ndorderA F (5)

[0080] E, ^, and ^diagnose HCM from LVH,

[0081] to Frequency domain or s-domain analysis, the exemplary method may employ other analyses to determine natural frequency and damping ratio or other dynamic control parameters in the time domain (such as using ordinary differential equations or eigenvalue decomposition methods). The damping ratios and natural frequency can be similarly used for the analysis to estimate the HCM index or likelihood of HCM.

[0082] Example Analysis System

[0083] Analysis system. The exemplary system is preferably an analysis system configured to interface to a database having ECG data to provide an analysis of the medical data. In some embodiments, the system can analyze data from a digital signal or a standard paper printout of a 12-lead ECG machine. In some embodiments, the analysis is performed on an ECG machine or a wearable device or implanted device configured to measure or acquire ECG from a user or patient.

[0084] Fig.1A shows an exemplary computing device 150 for determining indicators of a disease or disorder using only ECG data of a subject.

[0085] Computing device 150 is configured to receive raw electrocardiogram (ECG) data 152 from an ECG device. The raw ECG data 152 is provided to a pre-processing unit 154 to first condition the raw ECG data 152 to a computer-readable form for subsequent processing and analysis. Depending on the form of the raw ECG data (e.g., ECG printouts, analog / digital ECG waveform recordings), the pre-processing unit 154 can include various subcomponents for signal conditioning and noise removal, including, for example, a digitization module, a segmentation module, a filtration module, and / or a normalization module.

[0086] The processed signal 155 is then subjected to a frequency transform operation (e.g., s-domain transform operation, e.g., Laplace transform operator) and feature extraction 158 to determine a set of complex number values (e.g., frequency domain poles or complex number expressions), or a set of number derived therefrom from the processed signal data 155. By way of illustrative example, the resulting complex number values 157 can include frequency domain poles or their equivalent forms (e.g. complex number) having natural frequency and damping values. In some embodiments, damping value and natural frequency are obtained from time domain ordinary differential equations and eigenvalue decomposition of a linear state spaceAttorney Docket No.10046-631WO1 8448 CHE system matrix. In this regard, the indication of the disease or disorder (e.g., HCM) can be determined, via the disease or disorder evaluation module, 160 by the damping values, natural frequency values, or a combination of either one, having a value above or below a pre-defined threshold.

[0087] In some embodiments, the disease or disorder evaluation module is configured to execute a trained AI model using the generated complex number values to determine an indicator of a presence or non-presence of a disease or disorder. In some embodiments, the disease or disorder evaluation module is configured to execute a rule-based algorithm using the generated complex number values to determine an indicator of a presence or non-presence of a disease or disorder. In some embodiments, the disease or disorder evaluation module is configured to analyze s-domain information generated from the generated complex number values to determine an indicator of a presence or non-presence of a disease or disorder.

[0088] As shown in Fig.1A, the disease or disorder evaluation module 160 may further be supplemented with patient specific information stored on memory 162 to further enhance the determination of indicators of the disease or disorder using only ECG data of a subject. Although the computing device 150 shown in Fig. 1A includes a local memory 162 storing patient specific information, other aspects may retrieve patient specific identifiers from a user input or a remote database. In other aspects, patient specific information is not used in the determination of an indicator of a presence or non-presence of a disease or disorder.

[0089] Finally, the determined indicator 164 may be outputted and used by a clinician to subsequently (i) diagnose the presence of a disease or disorder (e.g., hypertrophic cardiomyopathy) or (ii) revise the misdiagnosis of the disease or disorder (e.g., from left ventricular hypertrophy to the presence of hypertrophic cardiomyopathy).

[0090] Fig. 1B shows a diagram of an example closed-loop system 100 determining indicators of a disease or disorder using only ECG data of a subject.

[0091] System 100 includes an ECG device 110 which is configured with at least two electrodes 114 for obtaining ECG signals from a user’s skin. The electrodes 114 are electrically coupled to a power source 112, such as a battery, for providing power. The ECG device 110 may further include other sensors 116, such as an accelerometer and / or an optical sensor. The electrodes 114 of the ECG device 110 are configured to record clinically relevant ECG signal data. The term ECG signal data, as used herein, refers to data that represents an ECG signal that is either sensed or otherwise reconstructed. In various embodiments, the additional sensors 116 are utilized in the preprocessing of ECG signal data to reduce noise and / or enhanceAttorney Docket No.10046-631WO1 8448 CHE detection. The ECG device includes other electronic components, such as an amplifier and analog-to-digital converter (ADC) 120 for signal conditioning.

[0092] The ECG signal data is received by computing device 122 where operations 124 are implemented to reduce noise such that clinically relevant features may be properly obtained from the resulting filtered ECG shape 126. The filtered ECG shape is then subjected to a frequency domain transform operation 128 (e.g., s-domain transform operation, e.g., Laplace transform operator) to determine complex number values 130 (e.g., frequency domain poles or complex number expressions), or a number derived therefrom, of the ECG signal data set by the computing device 122. From there, the computing device 122 is configured to (i) execute a trained AI model using the generated complex number values 130, (ii) execute a rule-based algorithm using the generated complex number values 130, or (iii) analyze an s-domain plot generated from the generated complex number values 130 to determine an indicator 132 of a presence or non-presence of a disease or disorder (e.g., hypertrophic cardiomyopathy). The indicator 132 is outputted (e.g., to the display 118 or via a report) and used by a clinician to subsequently (i) diagnose the presence of the disease or disorder or (ii) revise a diagnosis.

[0093] In its most basic configuration, the computing device 122 includes at least one processing unit and system memory. Depending on the exact configuration and type of computing device, system memory 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.

[0094] The processing unit may be a standard programmable processor that performs arithmetic and logic operations necessary for the operation of the computing device. While only one processing unit is shown, multiple processors may be present. As used herein, processing unit and processor refers to a physical hardware device that executes encoded instructions for performing functions on inputs and creating outputs, including, for example, but not limited to, microprocessors (MCUs), microcontrollers, graphical processing units (GPUs), and application-specific circuits (ASICs). Thus, while instructions may be discussed as executed by a processor, the instructions may be executed simultaneously, serially, or otherwise executed by one or multiple processors. The computing device may also include a bus or other communication mechanism for communicating information among various components of the computing device.

[0095] Computing devices may have additional features / functionality. For example, the computing device may include additional storage such as removable storage and non- removable storage including, but not limited to, magnetic or optical disks or tapes. ComputingAttorney Docket No.10046-631WO1 8448 CHE devices may also contain network connection(s) that allow the device to communicate with other devices, such as over the communication pathways described herein. The network connection(s) may take the form of modems, modem banks, Ethernet cards, universal serial bus (USB) interface cards, serial interfaces, token ring cards, fiber distributed data interface (FDDI) cards, wireless local area network (WLAN) cards, radio transceiver cards such as code division multiple access (CDMA), global system for mobile communications (GSM), long- term evolution (LTE), worldwide interoperability for microwave access (WiMAX), and / or other air interface protocol radio transceiver cards, and other well-known network devices. Computing devices may also have input device(s) such as keyboards, keypads, switches, dials, mice, trackballs, touch screens, voice recognizers, card readers, paper tape readers, or other well-known input devices. Output device(s) such as printers, video monitors, liquid crystal displays (LCDs), touch screen displays, displays, speakers, etc., may also be included. The additional devices may be connected to the bus in order to facilitate the communication of data among the components of the computing device. All these devices are well-known in the art and need not be discussed at length here.

[0096] The processing unit may be configured to execute program code encoded in tangible, computer-readable media. Tangible, computer-readable media refers to any media that is capable of providing data that causes the computing device (i.e., a machine) to operate in a particular fashion. Various computer-readable media may be utilized to provide instructions to the processing unit for execution. Example tangible, computer-readable media may include but is not limited to volatile media, non-volatile media, removable media, 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. System memory, removable storage, and non-removable storage are all examples of tangible computer storage media. Example tangible, computer-readable recording media include, but are not limited to, an integrated circuit (e.g., field-programmable gate array or application- specific IC), a hard disk, an optical disk, a magneto-optical disk, a floppy disk, a magnetic tape, a holographic storage medium, a solid-state device, RAM, ROM, electrically erasable program read-only memory (EEPROM), flash memory or other memory technology, CD-ROM, digital versatile disks (DVD) or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices.

[0097] In light of the above, it should be appreciated that many types of physical transformations take place in the computer architecture to store and execute the software components presented herein. It also should be appreciated that the computer architecture mayAttorney Docket No.10046-631WO1 8448 CHE include other types of computing devices, including hand-held computers, embedded computer systems, personal digital assistants, and other types of computing devices known to those skilled in the art.

[0098] In an example implementation, the processing unit may execute program code stored in the system memory. For example, the bus may carry data to the system memory, from which the processing unit receives and executes instructions. The data received by the system memory may optionally be stored on the removable storage or the non-removable storage before or after execution by the processing unit.

[0099] The exemplary system and method may be implemented (1) as a sequence of computer-implemented acts or program modules running on a computing system and / or (2) as interconnected machine logic circuits or circuit modules within the computing system. The implementation is a matter of choice dependent on the performance and other requirements of the computing system. Accordingly, the logical operations described herein are referred to variously as state operations, acts, or modules. These operations, acts, and / or modules can be implemented in software, in firmware, in special purpose digital logic, in hardware, and any combination thereof. It should also be appreciated that more or fewer operations can be performed than shown in the figures and described herein. These operations can also be performed in a different order than those described herein.

[0100] Although the system 100 shown in Fig. 1B contains a local computing device for signal interpretation and processing, other embodiments may utilize a network interface to transmit the ECG signal obtained from the ECG device to another computing device. As used herein, the term “network interface” refers to any signal, data, and / or software interface with a component, network, and / or process. By way of non-limiting example, a network interface may include one or more of FireWire (e.g., FW400, FW110, and / or other variation.), USB (e.g., USB2), Ethernet (e.g., 10 / 100, 10 / 100 / 1000 (Gigabit Ethernet), 10-Gig-E, and / or other Ethernet implementations), MoCA, Coaxsys (e.g., TVnet™), radio frequency tuner (e.g., in- band or OOB, cable modem, and / or other protocol), Wi-Fi (802.11), WiMAX (802.16), PAN (e.g., 802.15), cellular (e.g., 3G, LTE / LTE-A / TD-LTE, GSM, and / or other cellular technology), IrDA families, and / or other network interfaces. As used herein, the term “Wi-Fi” includes one or more of IEEE-Std.802.11, variants of IEEE-Std. 802.11, standards related to IEEE-Std. 802.11 (e.g., 802.11 a / b / g / n / s / v), and / or other wireless standards. As used herein, the term “wireless” means any wireless signal, data, communication, and / or other wireless interface. By way of non-limiting example, a wireless interface may include one or more of Wi-Fi, Bluetooth, 3G (3GPP / 3GPP2), HSDPA / HSUPA, TDMA, CDMA (e.g., IS-95A,Attorney Docket No.10046-631WO1 8448 CHE WCDMA, and / or other wireless technology), FHSS, DSSS, GSM, PAN / 802.15, WiMAX (802.16), 802.20, narrowband / FDMA, OFDM, PCS / DCS, LTE / LTE-A / TD-LTE, analog cellular, CDPD, satellite systems, millimeter wave or microwave systems, acoustic, infrared (i.e., IrDA), and / or other wireless interfaces.

[0101] Cloud system. The computer system is capable of executing the software components described herein for the exemplary method or systems. In an embodiment, the computing device may comprise two or more computers in communication with each other that collaborate to perform a task. For example, but not by way of limitation, an application may be partitioned in such a way as to permit concurrent and / or parallel processing of the instructions of the application. Alternatively, the data processed by the application may be partitioned in such a way as to permit concurrent and / or parallel processing of different portions of a data set by the two or more computers. In an embodiment, virtualization software may be employed by the computing device to provide the functionality of a number of servers that are not directly bound to the number of computers in the computing device. For example, virtualization software may provide twenty virtual servers on four physical computers. In an embodiment, the functionality disclosed above may be provided by executing the application and / or applications in a cloud computing environment. Cloud computing may comprise providing computing services via a network connection using dynamically scalable computing resources. Cloud computing may be supported, at least in part, by virtualization software. A cloud computing environment may be established by an enterprise and / or can be hired on an as-needed basis from a third-party provider. Some cloud computing environments may comprise cloud computing resources owned and operated by the enterprise as well as cloud computing resources hired and / or leased from a third-party provider.

[0102] Example Machine Learning Analysis

[0103] Machine Learning. In addition to the machine learning features described above, the various analysis system can be implemented using one or more artificial intelligence and machine learning operations. The term “artificial intelligence” can include any technique that enables one or more computing devices or comping systems (i.e., a machine) to mimic human intelligence. Artificial intelligence (AI) includes but is not limited to knowledge bases, machine learning, representation learning, and deep learning. The term “machine learning” is defined herein to be a subset of AI that enables a machine to acquire knowledge by extracting patterns from raw data. Machine learning techniques include, but are not limited to, logistic regression, support vector machines (SVMs), decision trees, Naïve Bayes classifiers, and artificial neural networks. The term “representation learning” is defined herein to be a subset of machineAttorney Docket No.10046-631WO1 8448 CHE learning that enables a machine to automatically discover representations needed for feature detection, prediction, or classification from raw data. Representation learning techniques include, but are not limited to, autoencoders and embeddings. The term “deep learning” is defined herein to be a subset of machine learning that enables a machine to automatically discover representations needed for feature detection, prediction, classification, etc., using layers of processing. Deep learning techniques include but are not limited to artificial neural networks or multilayer perceptron (MLP).

[0104] Machine learning models include supervised, semi-supervised, and unsupervised learning models. In a supervised learning model, the model learns a function that maps an input (also known as feature or features) to an output (also known as target) during training with a labeled data set (or dataset). In an unsupervised learning model, the algorithm discovers patterns among data. In a semi-supervised model, the model learns a function that maps an input (also known as a feature or features) to an output (also known as a target) during training with both labeled and unlabeled data.

[0105] Neural Networks. An artificial neural network (ANN) is a computing system including a plurality of interconnected neurons (e.g., also referred to as “nodes”). This disclosure contemplates that the nodes can be implemented using a computing device (e.g., a processing unit and memory as described herein). The nodes can be arranged in a plurality of layers such as an input layer, an output layer, and optionally one or more hidden layers with different activation functions. An ANN having hidden layers can be referred to as a deep neural network or multilayer perceptron (MLP). Each node is connected to one or more other nodes in the ANN. For example, each layer is made of a plurality of nodes, where each node is connected to all nodes in the previous layer. The nodes in a given layer are not interconnected with one another, i.e., the nodes in a given layer function independently of one another. As used herein, nodes in the input layer receive data from outside of the ANN, nodes in the hidden layer(s) modify the data between the input and output layers, and nodes in the output layer provide the results. Each node is configured to receive an input, implement an activation function (e.g., binary step, linear, sigmoid, tanh, or rectified linear unit (ReLU), and provide an output in accordance with the activation function. Additionally, each node is associated with a respective weight. ANNs are trained with a dataset to maximize or minimize an objective function. In some implementations, the objective function is a cost function, which is a measure of the ANN’s performance (e.g., error such as L1 or L2 loss) during training, and the training algorithm tunes the node weights and / or bias to minimize the cost function. This disclosure contemplates that any algorithm that finds the maximum or minimum of the objective functionAttorney Docket No.10046-631WO1 8448 CHE can be used for training the ANN. Training algorithms for ANNs include but are not limited to backpropagation. It should be understood that an ANN is provided only as an example machine learning model. This disclosure contemplates that the machine learning model can be any supervised learning model, semi-supervised learning model, or unsupervised learning model. Optionally, the machine learning model is a deep learning model. Machine learning models are known in the art and are therefore not described in further detail herein.

[0106] A convolutional neural network (CNN) is a type of deep neural network that has been applied, for example, to image analysis applications. Unlike traditional neural networks, each layer in a CNN has a plurality of nodes arranged in three dimensions (width, height, depth). CNNs can include different types of layers, e.g., convolutional, pooling, and fully- connected (also referred to herein as “dense”) layers. A convolutional layer includes a set of filters and performs the bulk of the computations. A pooling layer is optionally inserted between convolutional layers to reduce the computational power and / or control overfitting (e.g., by downsampling). A fully-connected layer includes neurons, where each neuron is connected to all of the neurons in the previous layer. The layers are stacked similar to traditional neural networks. GCNNs are CNNs that have been adapted to work on structured datasets such as graphs.

[0107] Other Supervised Learning Models. A logistic regression (LR) classifier is a supervised classification model that uses the logistic function to predict the probability of a target, which can be used for classification. LR classifiers are trained with a data set (also referred to herein as a “dataset”) to maximize or minimize an objective function, for example, a measure of the LR classifier’s performance (e.g., error such as L1 or L2 loss), during training. This disclosure contemplates that any algorithm that finds the minimum of the cost function can be used. LR classifiers are known in the art and are therefore not described in further detail herein.

[0108] A Naïve Bayes’ (NB) classifier is a supervised classification model that is based on Bayes’ Theorem, which assumes independence among features (i.e., the presence of one feature in a class is unrelated to the presence of any other features). NB classifiers are trained with a data set by computing the conditional probability distribution of each feature given a label and applying Bayes’ Theorem to compute the conditional probability distribution of a label given an observation. NB classifiers are known in the art and are therefore not described in further detail herein.

[0109] A k-NN classifier is an unsupervised classification model that classifies new data points based on similarity measures (e.g., distance functions). The k-NN classifiers are trainedAttorney Docket No.10046-631WO1 8448 CHE with a data set (also referred to herein as a “dataset”) to maximize or minimize a measure of the k-NN classifier’s performance during training. This disclosure contemplates any algorithm that finds the maximum or minimum. The k-NN classifiers are known in the art and are therefore not described in further detail herein.

[0110] Other classifiers may be used, e.g., Random Forest classifier, Genetic Algorithm classifier, among others. Experimental Results and Additional Examples

[0111] The following examples are set forth below to illustrate the methods and results according to the disclosed subject matter. These examples are not intended to be inclusive of all aspects of the subject matter disclosed herein, but rather to illustrate representative methods and results. These examples are not intended to exclude equivalents and variations of the present invention which are apparent to one skilled in the art.

[0112] Example 1:

[0113] An initial study was conducted that validated the hypothesis for several ECG recordings from PhysioNet and collaborating physician. In the study, currently, data from patients with HCM have been evaluated. By plotting the data in terms of natural frequency and damping with respect to the time of the lead V6, the study observed that the HCM patients had a higher damping term compared to the LVH patients. This is consistent across all patients with HCM. The leads V5 and V6 measure the left side of the heart.

[0114] The ECGs of the heartbeats are plotted in the s-domain with natural frequency and damping values. The HCM data used to plot this is from 12 lead ECG graphs, while the LVH data was taken from the “A large scale 12-lead electrocardiogram database for arrhythmia study” database from Physionet.

[0115] The first two HCM plots were taken from ECG graphs that had no overlapping ECG lines on any of the leads, while the last two HCM plots used ECG graphs that had overlapping ECG lines on most of the V leads besides V6. This is not relevant to the plots above because these plots used the V6 lead. Below is an example of no overlapping ECG lines (Fig.5A) and overlapping ECG lines (Fig.5B).

[0116] In the study, the indication of HCM is mainly determined by the damping value. However, it was observed that the natural frequency also shows a small change from LVH to HCM based on limited testing sets. The analysis can, for example, be a function of both damping and natural frequency values (e.g., 90% zeta and 10% omega).

[0117] DiscussionAttorney Docket No.10046-631WO1 8448 CHE

[0118] The exemplary study sought to find a cost-effective method to distinguish between Hypertrophic Cardiomyopathy (HCM) and Left Ventricular Hypertrophy (LVH) using an ECG signal as the step toward HCM diagnosis, bypassing expensive alternatives like echocardiograms and genetic testing that can cost thousands of dollars. HCM is a rare heart disease that is found in around 1 in 500 adults. LVH is found in around 10%-20% of adults. The diagnoses of these diseases are difficult to differentiate using only ECGs. The symptoms for these conditions are similar, but the treatments are different. If these heart conditions are not properly treated, they can lead to many debilitating conditions, including heart failure and cardiac arrest. Therefore, it is vital to differentiate between HCM and LVH. Our innovation aims to find an inexpensive and reliable way to early distinguish between LVH and HCM using an ECG that is widely available and costs less than $100. According to Growth Plus Reports, the global market for HCM is worth $1.12 billion and is projected to be $1.35 billion by 2031.

[0119] ECG-based diagnosis is much less expensive and more widely available compared to echocardiograms and genetic testing. Our innovation is an algorithm used to analyze ECG data. The developed algorithm can convert the ECG signal from the time domain to data points in the frequency domain. It then can identify and distinguish HCM from LVH based on the frequency data points. With our innovation, there is no need to incur additional costs associated with echocardiograms and genetic testing. Our innovation also enables an early diagnosis that leads to early and accurate treatments. This makes our innovation widely available to a broader public and highly competitive in the marketplace.

[0120] The exemplary method and system can be unique in two aspects. First, existing methods need to look at ECG data, echocardiograms and genetic testing. It is costly and not widely available. On the other hand, our method only relies on standard ECG data. Second, existing algorithms that try to analyze ECG data either only look at the time domain data or look at frequencies themself. None of them analyzes the complex number expression or its equivalent variation form (such as natural frequency plus the damping value). By analyzing the complex form of the ECG data in the s-domain, the exemplary system and method can find the differences using ECG data only.

[0121] When employed using only ECG data, the exemplary system and method would be less costly compared to diagnosing methods like echocardiograms or genetic testing. It is also more available to communities that may not have access to places that provide services like echocardiograms.

[0122] Example 2:Attorney Docket No.10046-631WO1 8448 CHE

[0123] Initially, ECG images were digitized into signals using the package ecg-digitize [1]. However, the algorithm typically performs well only on high-quality images, whereas much of the practical data is relatively low quality. As a result, the study further aimed to perform preliminary digitization-mainly focusing on background removal and noise reduction. An initial processed image, as in Fig.7 is obtained.

[0124] Image Preprocessing and Region of Interest Selection. In the exemplary study, the methodology begins with interactive region of interest (ROI) selection to isolate the ECG waveform from ancillary chart elements: ROI = 9^^, 8^ ∣ ^2 ≤ ^ ≤ ^@, 82 ≤ 8 ≤ 8@;

[0125] where ^^2, coordinates defining thebounding rectangle

[0126] Color Space Analysis and Signal Isolation. The core signal isolation employs unsupervised K-means clustering in the RGB color space to identify distinct color regions, a technique successfully applied to the digitization of paper ECGs [2].

[0127] K-means Color Clustering. The algorithm identifies distinct color regions corresponding to the background, grid lines, and the ECG trace itself. An example algorithm for K-means Color Segmentation is shown below: Input: RGB image N^ℎ × Q × 3^, number of clusters R = 3Output: Clustered labels S^ℎ × Q^:1. Reshape N → *^ℎ ⋅ Q × 3^2. Apply K-means: * → 9V2, V@, … , VW;3. Compute cluster centers: X: = mean^V:^4. Generate labels: S = arg m:in  ‖* − X:‖@

[0128] Target Color Identification. ECG traces are typically rendered in black or dark ink, corresponding to RGB values near ^0,0,0^. The algorithm identifies the cluster center closest to the target color using Euclidean distance: c `∗ = ^

[0129] Binary Maskwhere pixels belonging to the selected cluster are marked as foreground: 1if S ∗e^^, 8^ = f ^^, 8^ = `0 otherwiseAttorney Docket No.10046-631WO1 8448 CHE

[0130] Morphological Image Processing and Edge-based Text Removal. To eliminate textual annotations that may interfere with signal extraction, edges are identified first.

[0131] Canny Edge Detection: An edge detector was used similar to that of J. Canny, "A Computational Approach to Edge Detection," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. PAMI-8, no. 6, pp. 679-698, Nov. 1986, which is hereby incorporated by reference in its entirety. g= Canny^Ngray , & = 0.85$

[0132] Connected Component Analysis: Components with a small area, likely corresponding to text, are removed from the mask.

[0133] For each connected component VV:in g : 9If Area ^VV:^ < Etext , e^VV:^ = 0 where Etext = 1000 pixels

[0134] Noise operations are usedto clean the binary pepper non-signal artifacts [4].

[0135] Median Filtering:

[0136] em = median_filter ^e, kernel_size = 3 × 3^

[0137] Morphological Opening:

[0138] emm = remove_small_objects ^em, min_size = Enoise ^

[0139] where Enoiserepresents the minimum pixel threshold for noise removal.

[0140] Signal Extraction Algorithm and Column-wise Scanning.The processed binary image emmundergoes systematic column-wise analysis to extract the vertical position of the signal at each horizontal point [2].

[0141] For ` = 0 to width −1: ,column ^`^ = 98 ∣ emm^8, `^ = 1, 8min ≤ 8 ≤ 8max ;

[0142] Dual Signal Generation. Raw Signal: =raw = 9^`, 8^ ∣ 8 ∈ ,column ^`^, ∀`;

[0143] Smoothed=smooth ^`^ = 9mean^,column ^`^^ if ,column ^`^ ≠ ∅

[0144] Coordinate Transformation and Scaling and Pixel-to-Physical Coordinate Mapping. The final step is to map the extracted pixel coordinates to physical units of time (seconds) and amplitude (millivolts).

[0145] Horizontal (Time) Scaling: ^pixe − ^^ = l rst⋅ − +

[0146] VerticalAttorney Docket No.10046-631WO1 8448 CHE 8w − 88wxyzs{u| = }1 − sv~| rst8 − 8 ^ ⋅ ^^ruv − ^rst^ + ^rstruv rst

[0147] ^ruv^ = ^0,100^(amplitude.

[0148] Signal Centering 8centered = 8physical − mean^8physical $

[0149] Temporal^^ = lin samplesnormalized space^0,2.5, ^samples $, ^^ =2.5

[0150] The

[0151] the resulting image can be further processed as described in Example 1.

[0152] Example 3:

[0153] Using the disclosed s-domain analysis method, it is possible to convert the signal from the time-domain to the frequency domain using mathematical tools such as Laplace transform or Fourier transform. The data noise and known uncertainties can also be converted to the frequency domain and used for analysis. For instance, the noise can be expressed as a finite sum of sinusoids: ^^^^ = ∑^Wd2 ^Wsin ^^W^ + ФW^ (1)

[0154] Then, using Laplace transform: ℒ9^^^^; = ∑^Wd2 ^Wℒ9sin^^W^ + ФW^ ; (2)

[0155] The Eq (2) becomes: ^ℒ9^^^^; = ∑^Wd2 ^ ^ Ф^^^W ^^C^^C^^(3)

[0156] a probability with confidence level diagnosis can be determined.

[0157] Analysis of the unknown noise and uncertainty of ECG signals with AI. Data driven approaches are used to model the unknown noise and uncertainties. Data driven approaches generally are computationally expensive. In order to speed up the online diagnosis process, an AI tool including an influence function has been developed.

[0158] Briefly, an influence function was derived to estimate how the removal of a trainingtrajectory impacts the predictive accuracy of a learned linear dynamics model ^^^^^, 7^^.Attorney Docket No.10046-631WO1 8448 CHE Considering a discrete-time dynamical system where state ^^ ∈ ℝ^^ at time t evolves to ^^^2based on the current state and control input 7^ ∈ ℝ^^:^ = ^^^ , 7 ^ +^^2 ^ ^ Q^

[0159] where ^^. , . ^ represents the true underlying dynamics, which may be linear orunknown, and Q^ ∽ ^^0, ∑^ ^ is assumed to be i.i.d. Gaussian process noise. The influencefunction (shown as IF1 below) provides a computationally efficient estimate of how much the predictive loss would change upon removal of trajectory (τk), by leveraging gradients and the Hessian ^^^^ ^ evaluated only at the original parameters. The model parameters θ are estimated from a dataset ^ = 9E2, … , E^;, which consists of N trajectories. Each trajectory EW =f^^^W^^ , 7^W^^ ^^W^ ^^^2^^2 ^^ ^d^ is a time-ordered sequence of state-action-nextstate transitions, with Hkof trajectory τk. The optical parameters ^^are obtained byminimizing an empiral risk function, such as the total sum of squared prediction errors over all transitions observed in ^:

[0160] S^^, ^^ = ∑^2 ℒ ^^^ = ∑^Wd2 ∑ ^ −@ Wd W ^^ ^^2,^ Φ^^^,^, 7^,^^^^^^∈^ @, ^the datasetand ℒW^^^ signifies the cumulative loss contribution from all transitions within trajectory τk. The learned parameters are thus ^^ = ¡¢£ ¤`^^S^^, ^^.

[0162] IF1 provides an approximation of the influence of removing trajectory (τk) on the predictive loss Lpred (θ): N^1^EW, S¥¦^A$ ≔ ^∇^ℒW ^^^$^©^^^^^^2∇^S¥¦^A^^^^.

[0163] If LpredS^^ ª ª @¥¦^A^ $ = ∑ ^^^^^2,^ − Φ^ ^^^@ ,

[0164] Its∇^S¥¦^A^^^$ = −2 ∑ Φ©ªª^ (^^^2,^ − Φ^ª^^)^

[0165] IF1model’s generalization performance, with correlations typically exceeding 0.68.

[0166] Additional data and discussion regarding the derivation of the influence function can be found in Li, Jiachen, et al. "Influence Functions for Data Attribution in Linear System Identification and LQR Control." arXiv:2506.11293 (2025), which is expressly incorporated by reference in its entirety.Attorney Docket No.10046-631WO1 8448 CHE

[0167] This AI tool can predict the importance of each data point in the modeling process by using gradients and the Hessian matrix, thereby reducing the amount of retraining of the model with each data point.

[0168] It must also be noted that, as used in the specification and the appended claims, the singular forms “a,” “an,” and “the” include plural referents unless the context clearly dictates otherwise. Ranges may be expressed herein as from “about” or “approximately” one particular value and / or to “about” or “approximately” another particular value. When such a range is expressed, other exemplary embodiments include one particular value and / or the other particular value. As used herein, the term “is associated with”, as in A is associated with B, means that A refers to B, is B, identifies a feature of B, or indicates that B exists.

[0169] By “comprising” or “containing” or “including,” is meant that at least the name compound, element, particle, or method step is present in the composition or article or method but does not exclude the presence of other compounds, materials, particles, method steps, even if the other such compounds, material, particles, method steps have the same function as what is named.

[0170] In describing example embodiments, terminology will be resorted to for the sake of clarity. It is intended that each term contemplates its broadest meaning as understood by those skilled in the art and includes all technical equivalents that operate in a similar manner to accomplish a similar purpose. It is also to be understood that the mention of one or more steps of a method does not preclude the presence of additional method steps or intervening method steps between those steps expressly identified. Steps of a method may be performed in a different order than those described herein without departing from the scope of the present disclosure. Similarly, it is also to be understood that the mention of one or more components in a device or system does not preclude the presence of additional components or intervening components between those components expressly identified.

[0171] Further, while certain embodiments have been described using a particular combination of hardware and software, it should be recognized that other combinations of hardware and software are also possible. Certain embodiments may be implemented only in hardware, or only in software, or using combinations thereof. In one example, the software may be implemented with a computer program product containing computer program code or instructions executable by one or more processors for performing any or all of the steps, operations, or processes described in this disclosure, where the computer program may be stored on a non-transitory computer-readable medium. The various processes described herein can be implemented on the same processor or different processors in any combination.Attorney Docket No.10046-631WO1 8448 CHE

[0172] Where devices, systems, components or modules are described as being configured to perform certain operations or functions, such configuration can be accomplished, for example, by designing electronic circuits to perform the operation, by programming programmable electronic circuits (such as microprocessors) to perform the operation such as by executing computer instructions or code, or processors or cores programmed to execute code or instructions stored on a non-transitory memory medium, or any combination thereof. Processes can communicate using a variety of techniques, including, but not limited to, conventional techniques for inter-process communications, and different pairs of processes may use different techniques, or the same pair of processes may use different techniques at different times.

[0173] The following patents, applications and publications as listed below and throughout this document are hereby incorporated by reference in their entirety herein. References [1] D. Chen, Z. Ma, B. C. Li, Z. Yan, and W. Li, "Drowsiness detection with electrooculography signal using a system dynamics approach," Journal of Dynamic Systems, Measurement, and Control, vol.139, no.8, p.081003, 2017. [2] S. Bakshi, T. Feng, D. Chen, and W. Li, "Real-Time Bradycardia Prediction in Preterm Infants Using a Dynamic System Identification Approach," Journal of Engineering and Science in Medical Diagnostics and Therapy, vol.3, no.1, p.011006, 2020. [3] Li, Jiachen, et al. "Influence Functions for Data Attribution in Linear System Identification and LQR Control." arXiv:2506.11293 (2025), which is incorporated by reference in its entirety. [4] J. D. Fortune, N. E. Coppa, K. T. Haq, H. Patel, and L. G. Tereshchenko, "Digitizing ECG image: A new method and open-source software code," Computer Methods and Programs in Biomedicine, vol.221, p.106890, 2022. [5] P. M. W. M. Bandara and A. G. B. P. Jayasekara, "A Novel Method for Digitization of Paper ECGs," 2017 IEEE International Conference on Industrial and Information Systems (ICIIS), Peradeniya, Sri Lanka, 2017, pp.1-6. [6] J. Canny, "A Computational Approach to Edge Detection," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. PAMI-8, no.6, pp.679-698, Nov.1986. [7] R. C. Gonzalez and R. E. Woods, Digital Image Processing, 3rd ed. PrenticeHall, Inc., 2008.

Claims

Attorney Docket No.10046-631WO1 8448 CHE Claims What is claimed is:

1. A method of pre-screening or identifying misdiagnosis of hypertrophic cardiomyopathy patient as having left ventricular hypertrophy, the method comprising: providing, by a processor, a data set consisting of electrocardiographic (ECG) signal data acquired via an ECG device of a patient, wherein the patient has been diagnosed or scored as having a high or moderate likelihood of left ventricular hypertrophy (LVH); generating, by the processor, via a frequency domain transform operation (e.g., s- domain transform operation, e.g., Laplace transform operator, linear operator), complex number values (e.g., frequency domain poles or complex number expressions), or damping values and natural frequency and their equivalent expressions (e.g., generated using time- domain), any use of an angle obtained by using a ratio of a real part of a complex number value over an imaginary part of the complex number value, or a derived form of the ratio, or a number derived therefrom, of the ECG signal data; and determining, by the processor, by (i) executing a trained artificial intelligence (AI) model using the generated complex number values, (ii) executing a rule-based algorithm using the generated complex number values, or (iii) analyzing an s-domain information generated from the generated complex number values, an indicator of a presence or non- presence of hypertrophic cardiomyopathy, wherein the indicator is outputted and used by a clinician to subsequently (i) diagnose the presence of hypertrophic cardiomyopathy or (ii) revise the misdiagnosis of left ventricular hypertrophy to the presence of hypertrophic cardiomyopathy.

2. The method of claim 1, wherein the indicator is used to order an echocardiogram or a genetic test for validation of the presence of hypertrophic cardiomyopathy.

3. The method of claim 1 or 2, wherein the ECG device is a standard 12-lead ECG, or 3- lead ECG, or an ECG with any number of leads that is between 3 to 12 equipment.

4. The method of any one of claims 1-3, wherein the patient has been diagnosed or scored as having a high or moderate likelihood of left ventricular hypertrophy (LVH) based on Cornell criteria or Sokolow-Lyon Criteria.Attorney Docket No.10046-631WO1 8448 CHE 5. The method of claim 4, wherein the patient has been diagnosed or scored based on the Cornell criteria or the Sokolow-Lyon Criteria using the data set used to determine the presence or non-presence of hypertrophic cardiomyopathy, wherein the data set is first used to pre-screen to identify (i) potential presence or (ii) presence of left ventricular hypertrophy; and wherein the data set is then used to screen or determine for (i) potential presence or (ii) presence of hypertrophic cardiomyopathy.

6. The method of claim 4, wherein the patient has been diagnosed or scored based on the Cornell criteria or the Sokolow-Lyon Criteria using a first data set acquired via an ECG device to determine the presence or non-presence of left ventricular hypertrophy; and wherein the data set used to screen or determine for (i) potential presence or (ii) presence of hypertrophic cardiomyopathy is acquired after the first data set.

7. The method of any one of claims 1-6 further comprising: preprocessing, by the processor (e.g., using analog and digital filters), the provided data set to remove DC and high-frequency noise of the electrocardiographic (ECG) signal data prior to generating the complex number values.

8. The method of any one of claims 1-7 further comprising: preprocessing, by the processor (e.g., using a magnitude filter or a pulse detector), the provided data set or a cleaned version of the provided data set to extract ECG signatures prior to generating the complex number values, wherein the extracted ECG signatures are used for the frequency domain transform operation.

9. The method of any one of claims 1-8 further comprising: at least one of: detecting peaks in the provided data set (e.g., using peak detector, e.g., Pan Tompkin-based filter); normalizing the detected peaks; removing peaks having a peak height outside a predefined range; removing peaks having a peak distance outside a predefined range; or a combination thereof.Attorney Docket No.10046-631WO1 8448 CHE 10. The method of any one of claims 1-9 further comprising: determining a location of a QRS complex in each ECG cycle of the ECG signal data; and separating the QRS complexes around a respective peak of each QRS interval with a tuned trunked window, wherein the QRS complexes of each interval are transformed from a time-domain representation to an s-domain representation.

11. The method of any one of claims 1-10, wherein the complex number values comprise frequency domain poles or their equivalent forms (e.g. complex number) having natural frequency and damping values, wherein the indication of hypertrophic cardiomyopathy is determined by the damping values, natural frequency values, or a combination of either one, having a value above or below a pre-defined threshold.

12. The method of any one of claims 1-11, wherein the ECG data set includes at least one signal acquired from lead V5 and lead V6 of the ECG device (e.g., for a standard 12-lead ECG equipment).

13. The method of any one of claims 1-12, wherein the trained AI model is trained using an influence function (e.g., to approximate a change in model parameters and predictions upon upweighting or removing a training data point).

14. A method comprising: providing, by a processor, a data set consisting of electrocardiographic (ECG) signal data acquired via an ECG device of a patient (e.g., wherein the patient has only been screened with the ECG device); generating, by the processor (e.g., via a frequency domain transform operation, e.g., s- domain transform operation, e.g., Laplace transform operator, linear operator), complex number values (e.g., frequency domain poles or complex number expressions), damping values and natural frequency (e.g., generated using time-domain), or a number derived therefrom, of the ECG signal data set; and determining, by a processor, (i) executing a trained artificial intelligence (AI) model using the generated complex number values, (ii) executing a rule-based algorithm using the generated complex number values, (iii) analysis of a plotted S-domain plot generated fromAttorney Docket No.10046-631WO1 8448 CHE the generated complex number values, an indicator of a presence or non-presence of cardiac condition, wherein the indicator is outputted and used by a clinician to subsequently (i) diagnose the presence of the cardiac condition or (ii) revised a prior diagnose made using the ECG signal data.

15. The method of claim 14, wherein the indicator of a presence or non-presence of the cardiac condition includes an indicator for bradycardia (among others, e.g., heart diseases, neurological diseases (e.g., epilepsy, Parkinson's, Alzheimer's), muscle function problems).

16. A method of pre-screening or identifying misdiagnosis of a patient as having a first disease, the method comprising: providing, by a processor, a data set consisting of electrocardiographic (ECG) signal data acquired via an ECG device of a patient, wherein the patient has been diagnosed or scored as having a high or moderate likelihood of a first disease (e.g., left ventricular hypertrophy (LVH)); generating, by the processor, via a frequency domain transform operation (e.g., s- domain transform operation, e.g., Laplace transform operator, linear operator), complex number values (e.g., frequency domain poles or complex number expressions), damping values and natural frequencies (either via frequency domain transform operation or time- domain ordinary differential equations or Eigenvalue decomposition), or a number derived therefrom, of the ECG signal data set; and determining, by the processor, (i) executing a trained artificial intelligence (AI) model using the generated complex number values, (ii) executing a rule-based algorithm using the generated complex number values, and / or (iii) analyzing an S-domain plot generated from the generated complex number values, an indicator of a presence or non-presence of a second disease (e.g., hypertrophic cardiomyopathy), wherein the indicator is outputted and used by a clinician to subsequently (i) diagnose the presence of the second disease or (ii) revise the misdiagnosis of the first disease to the second disease.

17. The method of claim 16, wherein the output indicator is used to order a second test for validation of the presence of the second disease.Attorney Docket No.10046-631WO1 8448 CHE 18. The method of claims 16 or 17, wherein the first disease is selected from the group consisting of: hypertrophic, premature ventricular contraction-mediated, arrhythmogenic right ventricular, peripartum, stress, LV non-compaction, ischemic, tachycardia-mediated, alcohol-induced, drug-induced, medication-induced, chemo-induced, hypertensive tachyarrhythmias - sinus tachycardia, atrial fibrillation, atrial flutter, atrial tachycardia, atrioventricular re-entry tachycardia, atrioventricular nodal re-entry tachycardia, ventricular tachycardia, ventricular fibrillation, conduction abnormalities e.g., heart block, bundle branch block, sinus bradycardia, genetic conditions e.g., Brugada Syndrome, long QT syndrome valvular, aortic stenosis, aortic regurgitation, mitral stenosis, mitral regurgitation, tricuspid stenosis, tricuspid regurgitation, pulmonic stenosis, pulmonic regurgitation, lv dysfunction, electrolyte abnormalities, amyloidosis, sarcoidosis, myocarditis, pericarditis, congenital HIV hemochromatosis, and Chagas Disease.

19. The method of claim 16-18, wherein the ECG device is a standard 12-lead ECG equipment or an ECG device with 3 to 12 leads.

20. The method of any one of claims 16-19, wherein the patient has been diagnosed or scored based on a clinical criteria used for determining the presence or non-presence of the first disease, wherein the data set is first used to pre-screen to identify (i) potential presence or (ii) presence of the first disease; and wherein the data set is then used to screen or determine for (i) potential presence or (ii) presence of the second disease.

21. The method of any one of claims 16-19, wherein the patient has been diagnosed or scored based on a clinical criteria used for determining the presence or non-presence of the first disease; and wherein the data set used to screen or determine for (i) potential presence or (ii) presence of the second disease is acquired after the first data set.

22. A method of pre-screening or identifying misdiagnosis of a patient as having a first disease, the method comprising: providing, by a processor, a data set consisting of electrocardiographic (ECG) signal data acquired via an ECG device of a patient, wherein the patient has been diagnosed orAttorney Docket No.10046-631WO1 8448 CHE scored as having a high or moderate likelihood of a first disease (e.g., left ventricular hypertrophy (LVH)); generating, by the processor, dynamic control domain poles or complex number expressions (e.g., damping ratio or natural frequency values), or a number derived therefrom, of the ECG signal data set; and determining, by the processor, (i) executing a trained artificial intelligence (AI) model using the generated complex number values, (ii) executing a rule-based algorithm using the generated complex number values, an indicator of a presence or non-presence of a second disease (e.g., hypertrophic cardiomyopathy), wherein the indicator is outputted and used by a clinician to subsequently (i) diagnose the presence of the second disease or (ii) revise the misdiagnosis of the first disease to the second disease.

23. A system comprising: a processor; and a memory having instruction stored thereon, wherein execution of the instructions by the processor causes the processor to perform any one of the methods of claims 1-22.

24. The system of claim 23, wherein the system is an analysis system (e.g., remote or cloud system).

25. The system of claim 23, wherein the system is a wearable device or implanted device.

26. The system of claim 23, wherein the system is an ECG machine.

27. A non-transitory computer-readable medium having instruction stored thereon, wherein execution of the instructions by a processor causes the processor to perform any one of the methods of claims 1-22 or any one of systems of claims 23 – 26.

Images