Systems and methods for generating information indicative of cardiac abnormalities

Abstract

A system for generating information indicative of a heart abnormality (e.g., aortic valve stenosis) includes a signal interface (101) for receiving a signal indicative of heart motion and generated by a motion sensor system (103) having mechanical contact with a chest of an individual. The system includes a processing system (102) configured to extract samples from the motion signal each having a time length of one or more heartbeat cycles such that a time phase of a heartbeat-related feature is the same within each sample relative to a start of the sample, an integral transform for each sample is calculated to map each sample to a basis function of the integral transform, and the samples are classified to correspond to a health case or a cardiac anomaly based on coefficients resulting in the integral transform for each sample.

Description

TECHNICAL FIELD

[0001] The present invention relates generally to generating information indicative of a heart abnormality, such as aortic stenosis "AS", heart valve disease, heart failure, atrial fibrillation and / or other heart conditions. More specifically, the present invention relates to a system for generating information indicative of a heart abnormality. Furthermore, the present invention relates to a method and a computer program for generating information indicative of a heart abnormality. BACKGROUND

[0002] Abnormalities that can occur in the cardiovascular system, if not diagnosed and properly treated and / or remedied, can gradually reduce the ability of the cardiovascular system to maintain blood flow that meets the needs of the individual's body, especially when the individual encounters physical stress. For example, aortic stenosis "AS" occurs when the aortic valve narrows, preventing blood from flowing normally. Aortic stenosis is a common valvular heart disease affecting up to about 4% of older adults. If left untreated, aortic stenosis can lead to various complications, including heart failure and increased mortality. During the latent and early stages of the disease, many patients are asymptomatic. Auscultation of aortic murmurs can still be detected, which prompts further investigation, cardiology referral, and echocardiography. When aortic stenosis is not yet severe, observation is recommended for most asymptomatic patients. Patients with severe aortic stenosis symptoms are typically referred for aortic valve replacement.

[0003] Currently, the primary method for diagnosing aortic stenosis is echocardiography. The use of echocardiography is technically demanding and can require years of training. Furthermore, echocardiography machines are expensive. These factors limit its availability for screening large populations, such as in a clinic setting. Therefore, there is a need for a technique to generate information indicative of a heart abnormality, such as aortic stenosis, without the need for expensive equipment and specialized surgical personnel. SUMMARY

[0004] The following presents a simplified summary in order to provide a basic understanding of some aspects of various application embodiments. The summary is not an extensive overview of the application. It is neither

[0005] According to the present invention, a system for generating information indicative of a heart abnormality, including but not limited to aortic stenosis "AS", heart valve disease, heart failure, and atrial fibrillation, is provided. The system comprises:

[0006] - a signal interface for receiving a motion signal indicative of heart motion and generated by a motion sensor system (e.g. an accelerometer and / or a gyroscope) having mechanical contact with the chest of the individual, and

[0007] - a processing system coupled to the signal interface and configured to:

[0008] - extract from the motion signal each sample having a time length of one or more heart beat periods such that the time phase of a heart beat related feature, such as the closing of the mitral and tricuspid "S1" and the closing of the aortic and pulmonary "S2" valves, is identical within each sample having that feature relative to the start of the sample,

[0009] - compute an integral transform of each sample to map each sample to basis functions of the integral transform, and

[0010] - classify the samples based on coefficients resulting in the integral transform of each sample into one of predefined classes, such as for example a healthy case, a heart abnormality case, and an uncertain (i.e. undefined) case, set an indicator signal to represent the result of the classification, and output the indicator signal.

[0011] The indicator signal can be sent to the individual and can include instructions to seek treatment and / or advice from a healthcare provider. Alternatively or in addition, the indicator signal can be sent to a healthcare provider and can indicate whether the individual has a heart abnormality. The indicator signal can include the heart measurement data, e.g. the motion signal, transmitted to the healthcare provider.

[0012] The above classification can be performed by a machine learning classification algorithm, such as a gradient boosting classifier "GBC" that is trained using a training database comprising first data collected from healthy people with motion sensors (e.g., accelerometers and / or gyroscopes), and second data collected from people with a cardiac abnormality, such as, for example, aortic stenosis. Gradient boosting classifiers are based on a function gradient algorithm that repeatedly selects a function in a direction pointing to a weak hypothesis or negative gradient that minimizes a loss function. Gradient boosting classifiers combine several weak learning models to produce a strong predictive model. Other classification algorithms can also be used that classify samples based on coefficients resulting from an integral transform of each sample to correspond to a healthy case or a cardiac abnormality. Thus, the present invention is not limited to any particular classification method. For example, the integral transform can be a continuous wavelet transform "CWT" in which the basis functions are different time shifts and time stretches or contractions of a basis wavelet, or a Fourier transform in which the basis functions are sinusoidal functions, such as exponential functions with imaginary arguments having different frequencies, or a Laplace transform in which the basis functions are exponential functions with complex-valued arguments. Thus, the present invention is not limited to any particular integral transform. The coefficients resulting from the integral transform are typically scalar products between the sample and the basis functions of the integral transform. The classification can be performed by a classification system that includes a signal processing model, a signal matching model, and a classification model. The signal processing model can be configured to generate a time length value for each of one or more signal samples. The signal matching model can receive input from the signal processing model including the one or more signal samples. The signal matching model can predict at least one corresponding point on each signal sample. The signal matching model can map the one or more corresponding points of each signal sample to create a signal sample stack. The signal matching model can output data including the signal sample stack to the classification model. The classification model can generate a prediction of an appropriate classification category for the signal sample stack from one or more features of each signal sample. The classification model can output an indication of the classification category, which can be a cardiac abnormality case, a healthy case, or an indeterminate case.

[0013] In some embodiments, the signal matching model can generate a feedback output including data indicating that one or more signal samples are not in compliance with the sample. The signal processing model can receive the feedback data from the classification model and adjust one or more time length values. The classification model can provide classification feedback output data to the signal matching model. The signal matching model can adjust the mapping of one or more signal samples based on the classification feedback output data.

[0014] A system for generating information indicating cardiac abnormalities may include a sensor system comprising one or more accelerometers for generating motion signals indicative of cardiac motion and / or one or more gyroscopes for generating motion signals representing cardiac motion. The signal interface may also be configured to receive motion signals from an external device including a suitable sensor system; that is, it is emphasized that the system does not necessarily include means for generating motion signals indicative of cardiac motion. The system may be, for example, a smartphone or another handheld device including accelerometers and / or gyroscopes. The smartphone may be an Apple iPhone, an Android phone, a Google Pixel phone, a Motorola phone, or other types of smartphones. The smartphone or other handheld device may be placed on an individual's chest to generate the aforementioned motion signals caused by cardiac motion. The system may include, for example, a patch or wearable sensor capable of contacting the individual's chest when the individual is prone. The system may include a medical provider device or other handheld medical device.

[0015] In this document, the term "accelerometer" encompasses various sensors used to measure linear lateral acceleration. For example, an accelerometer can be a microelectromechanical system (MEMS) based on the law of inertia. Similarly, the term "gyroscope" encompasses various sensors used to measure angular rotation. For example, a gyroscope can be a MEMS based on the effect of the Coriolis force acting on a rotating object within the MEMS gyroscope.

[0016] According to the present invention, a method for generating information indicating cardiac abnormalities, such as aortic stenosis, valvular heart disease, heart failure, and / or atrial fibrillation, is also provided. The method includes:

[0017] - Receives motion signals indicative of heart movement and generated by a motion sensor system that has mechanical contact with the individual's chest.

[0018] - Extract samples from the motion signal for each time length having one or more heartbeat cycles, such that the time phase of the heartbeat-related feature is the same relative to the start of the sample within each sample having that feature.

[0019] - Calculate the integral transform for each sample to map each sample to the basis functions of the integral transform, and

[0020] - Based on the coefficients that cause the integral transformation of each sample, the samples are classified into one of the predefined categories, such as, for example, healthy cases, cardiac abnormality cases, and uncertain cases. An indicator signal is set to represent the classification result, and the indicator signal is output.

[0021] According to the present invention, a computer program for generating information indicating cardiac abnormalities, such as aortic stenosis, valvular heart disease, heart failure, and / or atrial fibrillation, is also provided. The computer program includes computer-executable instructions for controlling a programmable processing system to:

[0022] - Receives motion signals indicative of heart movement and generated by a motion sensor system that has mechanical contact with the individual's chest.

[0023] - Extract samples from the motion signal for each time length having one or more heartbeat cycles, such that the time phase of the heartbeat-related feature is the same relative to the start of the sample within each sample having that feature.

[0024] - Calculate the integral transform for each sample to map each sample to the basis functions of the integral transform, and

[0025] - Based on the coefficients that cause the integral transformation of each sample, the samples are classified into one of the predefined categories, such as, for example, healthy cases, cardiac abnormality cases, and uncertain cases. An indicator signal is set to represent the classification result, and the indicator signal is output.

[0026] In some embodiments, the present invention may include a system for generating an individual cardiac condition indicator, the system comprising:

[0027] - A signal interface configured to receive inputs including motion signals indicative of heart movement generated by a motion sensor system having mechanical contact with the individual's chest, and

[0028] - The processing system is configured as follows:

[0029] i) Identify one or more characteristics of each sample of the motion signal, including heart rate-related features, over the duration of the motion signal.

[0030] ii) Apply an integral transform to each sample to map each sample to the basis functions of the integral transform.

[0031] iii) Classify samples into one of the categories based on the coefficients that cause the integral transformation of each sample, and

[0032] iv) Output an indicator signal including the sample classification results.

[0033] In some embodiments, the processing system is further configured to select a subset of samples from the motion signals, wherein the subset of samples has corresponding characteristic heartbeat-related features.

[0034] In some embodiments, the motion sensor system may include an accelerometer, a gyroscope, or both. In some embodiments, one or more heart-related characteristics may include one or more of amplitude characteristics, frequency characteristics, power characteristics, intensity characteristics, energy characteristics, or any combination thereof. In some embodiments, categories may include healthy cases, cardiac abnormality cases, and indeterminate cases.

[0035] In some embodiments, the indicator signal may also include one or more of the following: a notification predicting classification to a category, an instruction to contact a healthcare professional based on the predicted classification, data output regarding a motion signal indicating heart movement, a display of a motion signal indicating heart movement, or any combination thereof.

[0036] In some embodiments, the present invention may include a method for generating an individual cardiac condition indicator, the method comprising:

[0037] - A signal interface configured to receive input, including motion signals indicative of heart movement generated by a motion sensor system that has mechanical contact with the individual's chest.

[0038] - Identify one or more characteristics of each sample of a motion signal over its time duration, including heart rate-related features.

[0039] - Apply an integral transform to each sample to map each sample to the basis functions of the integral transform.

[0040] - Based on the coefficients that cause the integral transformation of each sample, the samples are classified into one of the categories representing different cardiac conditions.

[0041] - Output includes an indicator signal for the sample classification results.

[0042] In some embodiments, the method further includes selecting a subset of samples from the motion signal, wherein the subset of samples has corresponding characteristic heartbeat-related features.

[0043] In some embodiments, the motion sensor system may receive input from an accelerometer, a gyroscope, or both. In some embodiments, identifying one or more characteristic features may further include identifying one or more of amplitude features, frequency features, power features, intensity features, energy features, or any combination thereof. In some embodiments, categories may include healthy cases, cardiac abnormality cases, and indeterminate cases. In some embodiments, the output indicator signal may further include, based on the classification, an instruction to contact a healthcare professional, output data related to a signal indicating cardiac motion, or an output of a motion signal indicating cardiac motion, or any combination thereof.

[0044] In some embodiments, the present invention may include a system for indicating cardiac condition, the system comprising:

[0045] - An input model, configured to receive input including motion signals from a motion sensor system that has mechanical contact with the individual's chest.

[0046] - Sample recognition model, which is configured as follows:

[0047] a) Identify one or more characteristics of the motion signal received from the input model, including heart rate-related features, over the duration of the signal.

[0048] b) Generate one or more samples of motion signals by mapping one or more characteristic features of one or more samples received from the input model.

[0049] - A classification model configured to generate a prediction that classifies the one or more samples into one of the categories based on the integral transform coefficients of each of the one or more samples.

[0050] - Output model, which is configured to output a classification indication received from the classification model of the one or more samples.

[0051] In some embodiments, input received from the motion sensor system may include input from an accelerometer, a gyroscope, or both. In some embodiments, one or more heartbeat-related characteristics may include one or more of amplitude characteristics, frequency characteristics, power characteristics, intensity characteristics, energy characteristics, or any combination thereof. In some embodiments, categories may include healthy cases, cardiac abnormality cases, and indeterminate cases. In some embodiments, the signal may also include one or more of the following: instructions to contact a healthcare professional based on classification, data output related to a signal indicating heart movement, a display of a signal indicating heart movement, or any combination thereof.

[0052] In some embodiments, the sample recognition model is further configured to adjust the recognition of one or more feature characteristics based on feedback data from the classification model. In some cases, the feedback data may include indications that one or more feature characteristics do not match. In some cases, the feedback data may include indications that one or more feature characteristics are irrelevant.

[0053] In some embodiments, the sample identification model can also be configured to generate one or more modified samples based on feedback data from the classification model. In some cases, the feedback data may include data indicating that one or more samples do not conform. In some cases, the feedback data may include data indicating that one or more samples are noise samples.

[0054] In some embodiments, one or more modified samples can be generated from modifications to the mapping of one or more feature characteristics of one or more samples. In some cases, the mapping can be modified by excluding one or more samples from the mapping. In some embodiments, the mapping can be modified by using different feature characteristics for the mapping.

[0055] In some embodiments, the classification model may also be configured to generate one or more modified classifications based on feedback data from the sample recognition model. In some embodiments, the output model is further configured to generate update indications for the modified classifications.

[0056] According to the present invention, a computer program product is also provided. The computer program product includes a non-volatile computer-readable medium, such as an optical disc "CD" encoded with the computer program according to the present invention, a flash drive encoded with computer software according to the present invention, or a digital download of the computer program according to the present invention.

[0057] Computer-readable media, such as computer-executable code, can take many forms, including but not limited to tangible storage media, carrier media, or physical transmission media. Non-volatile storage media can include, for example, optical discs or disks, such as any storage device in any one or more computers, which can be used to implement databases. Volatile storage media include dynamic memory, such as the main memory of such a computer platform. Tangible transmission media include coaxial cables, copper wires, and optical fibers, including the wires that form the bus within a computer system. Carrier transmission media can take the form of electrical or electromagnetic signals, or sound or light waves, such as those generated during radio frequency (RF) and infrared (IR) data communications. Therefore, common forms of computer-readable media include, for example: floppy disks, hard disks, magnetic tapes, any other magnetic media, optical disc read-only memory "CD-ROM", digital video disc "DVD" or DVD-ROM, any other optical media, punched card tape, any other physical storage medium with a perforated pattern, random access memory "RAM", read-only memory "ROM", programmable read-only memory "PROM" and erasable programmable read-only memory "EPROM", flash memory EPROM, any other memory chip or cartridge, a carrier wave for transmitting data or instructions, a cable or link for transmitting such a carrier wave, or any other medium from which a computer can read program code and / or data. Many of these forms of computer-readable media may involve carrying one or more sequences of one or more instructions to a processor for execution.

[0058] Exemplary and non-limiting embodiments are described in the appended dependent claims.

[0059] When read in conjunction with the accompanying drawings, the exemplary and non-limiting embodiments relating to the construction and operation methods, as well as their additional objects and advantages, will be best understood from the following description of specific exemplary embodiments.

[0060] The verbs “contain” and “include” are used in this document as open restrictions, neither excluding nor requiring the existence of unreferenced features.

[0061] Unless otherwise expressly stated, the features described in the appended dependent claims may be freely combined with each other.

[0062] Furthermore, it should be understood that the use of “a” or “an” in this document, i.e., the singular form, does not exclude the plural form. As used herein, unless the context clearly indicates otherwise, the singular forms “a,” “an,” and “the” include plural references. Unless otherwise stated, references to “or” in this document are intended to include “and / or.”

[0063] As used herein, the term “approximately” in some cases means a quantity approximately equal to the quantity, in some cases it is close to 10%, 5%, or 1% of the quantity, including increments, and in some cases, in terms of percentages it means a quantity that is 10%, 5%, or 1% greater or less than the percentage, including increments.

[0064] As used in this article, the phrases “at least one,” “one or more,” and “and / or” are open-ended expressions that are operationally both conjunctions and disjunctions. For example, each of “at least one of A, B, and C,” “at least one of A, B, or C,” “one or more of A, B, and C,” and “A, C, and / or B” implies a single A, a single B, a single C, A and B together, A and C together, B and C together, or A, B, and C together.

[0065] The terms "some embodiments," "other embodiments," or "specific embodiments" used in this specification mean that a particular feature, structure, or characteristic associated with that embodiment is included in at least one embodiment. Therefore, the phrases "in some embodiments," "in other embodiments," or "in specific embodiments" appearing in various places in this specification do not necessarily refer to the same embodiment. Furthermore, in one or more embodiments, specific features, structures, or characteristics can be combined in any suitable manner. Attached Figure Description

[0066] The exemplary and non-limiting embodiments and their advantages are explained in more detail below with reference to the accompanying drawings, in which:

[0067] Figure 1a The diagram illustrates a system for generating information indicative of the heart, according to an exemplary and non-limiting embodiment.

[0068] Figure 1b Show Figure 1a The example signal generated in the system shown.

[0069] Figure 1c Show Figure 1a The functional block diagram of the system shown, and

[0070] Figure 2 This is a flowchart of a method for generating information indicating cardiac abnormalities according to exemplary and non-limiting embodiments. Detailed Implementation

[0071] The specific examples provided in the following description should not be construed as limiting the scope of the invention. Unless otherwise expressly stated, the lists and sets of examples provided in the specification are not exhaustive.

[0072] Figure 1a A schematic diagram of a system 100 for generating information indicative of cardiac abnormalities, such as aortic stenosis, valvular heart disease, heart failure, and / or atrial fibrillation, according to an exemplary and non-limiting embodiment, is shown. System 100 includes a signal interface 101 for receiving motion signals indicative of cardiac motion and generated by a motion sensor system 103 having mechanical contact with the chest of an individual 107. System 100 includes a processing system 102 coupled to the signal interface 101. The processing system 102 is configured to:

[0073] - Extract samples from the motion signal for each time length having one or more heartbeat cycles, such that the temporal phase of heartbeat-related features (e.g., the closure of the mitral and tricuspid valves "S1" and the closure of the aortic and pulmonary valves "S2") is the same relative to the start of the sample within each sample having that feature.

[0074] - Calculate the integral transform for each sample to map each sample to the basis functions of the integral transform, and

[0075] - Based on the coefficients that cause the integral transformation of each sample, the samples are classified into one of the predefined categories, such as, for example, healthy cases, cardiac abnormality cases, and uncertain cases, and an indicator signal is set to represent the classification result.

[0076] The sensor system 103 described above includes at least one accelerometer and / or at least one gyroscope. The sensor system 103 may also include, for example, one or more inertial measurement units (IMUs), each IMU including an accelerometer and a gyroscope, and each gyroscope may be, for example, a microelectromechanical system (MEMS). The duration of the motion signal may, for example, range from tens of seconds to several days. In some cases, the duration of the signal can be approximately 10 seconds, approximately 20 seconds, approximately 30 seconds, approximately 40 seconds, approximately 50 seconds, approximately 1 minute, approximately 2 minutes, approximately 3 minutes, approximately 4 minutes, approximately 5 minutes, approximately 6 minutes, approximately 7 minutes, approximately 8 minutes, approximately 9 minutes, approximately 10 minutes, approximately 15 minutes, approximately 20 minutes, approximately 25 minutes, approximately 30 minutes, approximately 35 minutes, approximately 40 minutes, approximately 45 minutes, approximately 50 minutes, approximately 55 minutes, approximately 1 hour, approximately 2 hours, approximately 3 hours, approximately 4 hours, approximately 5 hours, approximately 6 hours, approximately 7 hours, approximately 8 hours, approximately 9 hours, approximately 10 hours, 12 hours, approximately 14 hours, approximately 16 hours, approximately 18 hours, approximately 20 hours, approximately 22 hours, approximately 24 hours, approximately 2 days, approximately 3 days, approximately 4 days, approximately 5 days, approximately 6 days, approximately 7 days, or more than approximately 7 days.

[0077] The indicator signal output by system 100 may be, for example, a message displayed on the screen of user interface 104. The indicator signal may include instructions to seek treatment and / or advice from a healthcare provider. System 100 may be configured to transmit the indicator signal to the healthcare provider. Furthermore, cardiac measurement data, such as recorded waveforms of motion signals indicating heart movement, may be sent to the healthcare provider.

[0078] exist Figure 1a In the exemplary embodiment shown, sensor system 103 is connected to signal interface 101 via transmitter 110 and one or more data transmission links, each data transmission link being, for example, a radio link or a wired link. Data transmission from transmitter 110 to signal interface 101 can be performed directly or via data transmission network 105 (such as, for example, a telecommunications network). Figure 1aIn the exemplary case shown, transmitter 110 is a radio transmitter. The system including processing system 102 may also be integrated with a sensor system; that is, system 100 includes a sensor system. In this exemplary case, the system is configured to generate motion signals, and the signal interface may be simple wiring from the sensor system to the processing system. The system including the integrated sensor system may be, for example, a smartphone or another handheld device that can be placed on the individual's chest during the measurement phase. The smartphone may be an Apple iPhone, an Android phone, a Google Pixel phone, a Motorola phone, or other types of smartphone. The handheld device may be, for example, a patch or wearable sensor that can contact the individual's chest when the individual is prone. The handheld device may be a medical provider device or another handheld medical device.

[0079] The system according to exemplary and non-limiting embodiments is configured to record motion signals indicative of cardiac motion. The recorded motion signals may be generated within a time window having a fixed start and end time, or within a sliding time window having a fixed length and moving over time. The system may include an internal memory 106 for recording signals and / or the system may include a data port for connecting to an external memory. The system may include a wireless transceiver for wirelessly transmitting data to and receiving data from the external memory.

[0080] In the system according to the exemplary and non-limiting embodiments, the processing system 102 is configured to compute the continuous wavelet transform "CWT" for each sample. The basis functions of the CWT are different time-shifted and time-scaled versions of the basis wavelet (i.e., the mother wavelet), and the CWT coefficients are scalar products between the sample and the CWT transform basis functions. The time scale of the basis wavelet is represented by a scale parameter, while the time shift of the basis wavelet is represented by a translation value. Therefore, each CWT coefficient is associated with a specific scale parameter value and a specific translation value.

[0081] In the system according to the exemplary and non-limiting embodiments, the processing system 102 is configured to compute the Fourier transform of each sample. The basis functions of the Fourier transform are exponential functions of imaginary variables with different frequencies, and the Fourier coefficients are scalar products between the sample and the Fourier transform basis functions.

[0082] In the system according to the exemplary and non-limiting embodiments, the processing system 102 is configured to implement a gradient boosting classifier "GBC" to classify samples to correspond to healthy cases, cases of cardiac abnormalities (such as, for example, aortic stenosis), or uncertain cases. The gradient boosting classifier "GBC" is a machine learning classification algorithm that is trained using a training database comprising first data collected from healthy individuals and second data collected from individuals with cardiac abnormalities (e.g., aortic stenosis).

[0083] In the system according to the exemplary and non-limiting embodiments, the processing system 102 is configured to directly apply a classification algorithm (e.g., GBC described above) to the coefficients that cause the integral transform of the samples, i.e., without preprocessing the coefficients to classify the samples as corresponding to healthy cases or cardiac abnormality cases (such as, for example, aortic stenosis), or indeterminate cases. In the system according to another exemplary and non-limiting embodiment, the processing system 102 is configured to preprocess the coefficients that produce the integral transform and apply the classification algorithm to the result of the preprocessing. An exemplary preprocessing is given below when the integral transform is a continuous wavelet transform "CWT".

[0084] The application of CWT to samples of motion signals results in a coefficient matrix representing information about the samples in the time-frequency domain. This matrix captures the intensity and distribution of frequencies in the samples over time, such that the first dimension of the matrix corresponds to different scale parameter values ​​of the CWT, i.e., different frequency bands, and the second dimension corresponds to different shift values ​​of the CWT. By extracting specific features from these coefficients, insights into the underlying hemodynamics can be gained, and potential abnormalities can be detected.

[0085] For each scale parameter value of CWT and each sample corresponding to a different time window on the motion signal, the following three example features can be extracted: 1)-3).

[0086] 1) Maximum Amplitude: This feature identifies the maximum amplitude of the CWT coefficients used for each scale parameter value. It can help capture the dominant frequency component in each sample.

[0087] 2) Average value: Calculate the average value of the CWT coefficients for each scale parameter value, giving a measurement of the average power or intensity of the frequency component over time.

[0088] 3) Total Energy: This is the sum of squares of the CWT coefficients at each scale parameter value. It represents the total energy or power in the motion signal for a specific scale parameter.

[0089] The following example array illustrates the calculation of the three features 1)-3) above. This array can be a set of coefficients obtained from the CWT of a given sample, so i) the CWT has a given scaling parameter value, ii) the translation value of the CWT changes to have ten different values, that is, the array is a single row of the above coefficient matrix:

[0090] [1, 2, 3, 4, 5, 6, 7, 8, 9, 10].

[0091] In this exemplary case, the maximum amplitude is the maximum absolute value in the array = Max{abs(1, 2, 3, 4, 5, 6, 7, 8, 9, 10)} = 10.

[0092] In this exemplary case, the average is the average of all numbers in the array = Mean{1, 2, 3, 4, 5, 6, 7, 8, 9, 10} = 5.5.

[0093] In this example, the total energy is the sum of the squares of the numbers in the array = 1. 2 +2 2 +3 2 +4 2 +5 2 +6 2 +7 2 +8 2 +9 2 +10 2 = 385.

[0094] Therefore, for the example array above, the calculated three features 1)-3) are:

[0095] Maximum amplitude: 10

[0096] Average: 5.5

[0097] Total energy: 385.

[0098] These features can be computed for each value of the CWT's scale parameter and for each sample of the motion signal. The computed features can then be fed into a classification algorithm, such as a gradient boosting classifier.

[0099] The same preprocessing (such as the example above) is advantageously applied to the training data for training classification algorithms and the data based on which to detect possible cardiac abnormalities (such as aortic stenosis).

[0100] In a system according to an exemplary and non-limiting embodiment, processing system 102 is configured to compute the integral transform of the samples without preprocessing them. In a system according to another exemplary and non-limiting embodiment, processing system 102 is configured to preprocess the samples before computing the integral transform. Preprocessing may include, for example, filtering the samples with a filter that suppresses frequency components outside the normal frequency band of the samples, and thus represents noise.

[0101] In the system according to the exemplary and non-limiting embodiments, the processing system 102 is configured to perform dynamic time warp (DTW) to measure the similarity between samples and reject samples that differ most significantly from the majority of samples. Thus, the samples most severely corrupted by noise can be discarded from further processing, i.e., integral transformation and classification. However, other methods can also be used to remove noise-contaminated samples.

[0102] In the system according to exemplary and non-limiting embodiments, signal interface 101 is configured to receive electrocardiogram (ECG) signals, and processing system 102 is configured to find R-peaks from the ECG signals and determine the duration of a sample based on the time difference between consecutive R-peaks. Heart rate can also be detected from motion signals based on aortic opening (AO) and / or aortic closure (AC) peaks and / or other heartbeat-related features in the motion signal.

[0103] exist Figure 1a In the exemplary embodiment shown, a single-lead ECG system is present, which includes ECG electrodes 108a and 108b and a differential amplifier 109 configured to generate an ECG signal. Transmitter 110 is configured to transmit the ECG signal to signal interface 101 in addition to the motion signal. The ECG system may also be part of system 100, thus system 100 includes ECG electrodes 108a and 108b, differential amplifier 109, and lines connecting the ECG electrodes to the differential amplifier.

[0104] exist Figure 1a In the exemplary embodiment shown, the motion signal includes a first component and a second component. The first component of the motion signal is generated by a first motion sensor 103a, which is part of the motion sensor system 103 and has mechanical contact with a first location on the chest of the individual 107. The second component of the motion signal is generated by a second motion sensor 103b, which is also part of the motion sensor system 103 and has mechanical contact with a second location on the chest of the individual 107. Figure 1a In the exemplary case shown, the first motion sensor 103a is located on the upper part of the sternum of individual 107, and the second motion sensor 103b is located on the lower part of the sternum of individual 107. The first motion sensor 103a may include an accelerometer and / or a gyroscope. Accordingly, the second motion sensor 103b may include an accelerometer and / or a gyroscope. Measurements from two different locations on the chest of individual 107 improve the reliability of detecting cardiac abnormalities such as aortic stenosis. In the exemplary case where each of the first motion sensor 103a and the second motion sensor 103b includes an accelerometer and a gyroscope, the motion signal has four signal components. If the corresponding accelerometer or gyroscope is a triaxial accelerometer or gyroscope, each of these signal components may have three sub-components. The integral transform for each sub-component can be calculated individually, or the sub-components can be combined with, for example, the following formula, and then the integral transform can be calculated on the result r obtained by the formula:

[0105] r i = (1)

[0106] Where i is the index that increases over time, q xi It is the i-th value of the x-component of cardiac motion, such as acceleration in the x-direction of coordinate system 199 or rotation about a geometric axis parallel to the x-axis of coordinate system 199, q yi qzi is the i-th value of the y-component of the heart's motion in coordinate system 199, and qzi is the i-th value of the z-component of the heart's motion in coordinate system 199.

[0107] Figure 1b An example waveform of the motion signal 150 generated by the triaxial accelerometer is shown, and using Equation 1 above, q xi q yi and q zi These represent the accelerations at time i in the x, y, and z directions of coordinate system 199, respectively. Furthermore, Figure 1b An electrocardiogram (ECG) signal 151 is shown. In this exemplary case, the processing system 102 is configured to find R-peaks from the ECG signal, and the sample of motion signal 151 is a portion of motion signal 150 between consecutive R-peaks. Figure 1b In the example samples, Sp1, Sp2, Sp3, Sp4, Sp5, and Sp6 are represented.

[0108] Figure 1c Show Figure 1a The system functional block diagram shown is as follows: Function block 161 is a signal interface for receiving motion signals indicating heart movement; function block 162 is a sample extractor for extracting samples from the motion signals; function block 163 is an integral transform calculator for calculating the integral transform of each sample; function block 164 is a classifier for classifying samples into one of predefined categories based on the coefficients that cause the integral transform of the samples; and function block 165 is an output unit for setting an indicator signal to represent the classification result and outputting the indicator signal. Function blocks 161-165 may represent corresponding parts of software modules and / or hardware processors.

[0109] Processing system 102 and Figure 1c The functional blocks shown can be implemented, for example, using one or more processor circuits, each of which can be a programmable processor circuit equipped with appropriate software, a special-purpose hardware processor (such as, for example, an application-specific integrated circuit "ASIC"), or a configurable hardware processor (such as, for example, a field-programmable gate array "FPGA"). Memory 106 can be implemented, for example, using one or more memory circuits, each of which can be, for example, a random access memory "RAM" device.

[0110] Figure 2A flowchart illustrating a method according to an exemplary and non-limiting embodiment is shown, the method being used to generate information indicative of cardiac abnormalities, such as aortic stenosis, valvular heart disease, heart failure, and / or atrial fibrillation. The method includes the following actions:

[0111] - Action 201: Receive motion signals indicative of cardiac motion and generated by a motion sensor system that has mechanical contact with the individual's chest.

[0112] - Action 202: Extract samples from the motion signal for each time length having one or more heartbeat cycles, such that the time phase of the heartbeat-related feature is the same relative to the start of the sample within each sample having that feature.

[0113] - Action 203: Compute the integral transform for each sample to map each sample to the basis functions of the integral transform, and

[0114] - Action 204: Based on the coefficients that cause the integral transformation of each sample, classify the sample into one of the predefined categories, such as, for example, healthy cases, cardiac abnormality cases, and uncertain cases, and set an indicator signal to indicate the classification result.

[0115] The method according to exemplary and non-limiting embodiments includes generating the aforementioned motion signal via a motion sensor system having mechanical contact with an individual's chest. The method according to another exemplary and non-limiting embodiment includes reading the motion signal from a memory, in which case the motion signal has been generated in advance and recorded in the memory. The method according to exemplary and non-limiting embodiments includes receiving the motion signal from an external data transmission system. Therefore, the generation of the aforementioned motion signal, i.e., the measurement step, is not a necessary or essential step in the method according to embodiments of the present invention.

[0116] The method according to exemplary and non-limiting embodiments includes calculating the integral transform of the samples without preprocessing them.

[0117] A method according to another exemplary and non-limiting embodiment includes preprocessing the samples before computing the integral transform of the samples. Preprocessing may include, for example, filtering the samples.

[0118] In the method according to the exemplary and non-limiting embodiments, the integral transform of each sample is the continuous wavelet transform "CWT" of the sample.

[0119] In the method according to the exemplary and non-limiting embodiments, the integral transform of each sample is the Fourier transform of the sample.

[0120] The method according to exemplary and non-limiting embodiments includes applying a classification algorithm (e.g., GBC) directly to the coefficients that cause the integral transformation of the samples, i.e., without preprocessing the coefficients, in order to classify the samples as corresponding to healthy cases or cardiac abnormality cases (such as, for example, aortic stenosis) or uncertain cases.

[0121] The method according to another exemplary and non-limiting embodiment includes preprocessing the coefficients that generate the integral transform and applying a classification algorithm to the result of the preprocessing.

[0122] The method according to exemplary and non-limiting embodiments includes performing dynamic time warp “DTW” to measure the similarity between samples and rejecting samples that differ most from the majority of samples.

[0123] In the method according to an exemplary and non-limiting embodiment, the motion signal includes: a first component generated by a first motion sensor, which is part of a motion sensor system and has mechanical contact with a first location on the chest of an individual (e.g., on the upper part of the sternum); and a second component generated by a second motion sensor, which is part of a motion sensor system and has mechanical contact with a second location on the chest of an individual (e.g., on the upper part of the sternum).

[0124] In the method according to exemplary and non-limiting embodiments, at least a portion of the motion signal is generated by an accelerometer having mechanical contact with the individual's chest.

[0125] In the method according to exemplary and non-limiting embodiments, at least a portion of the motion signal is generated by a gyroscope having mechanical contact with the individual's chest.

[0126] The method according to exemplary and non-limiting embodiments includes receiving an electrocardiogram (ECG) signal, identifying R-peaks from the ECG signal, and determining the duration of a sample based on the time difference between consecutive R-peaks within the R-peaks.

[0127] In the method according to the exemplary and non-limiting embodiments, the cardiac abnormality is aortic stenosis.

[0128] A computer program according to exemplary and non-limiting embodiments includes computer-executable instructions for controlling a programmable processing system to perform a method according to any of the exemplary and non-limiting embodiments described above.

[0129] A computer program according to exemplary and non-limiting embodiments includes a software module for generating information indicative of cardiac abnormalities, such as aortic stenosis, valvular heart disease, heart failure, atrial fibrillation, and / or other cardiac abnormalities. The software module includes computer-executable instructions for controlling a programmable processing system to:

[0130] - Receives motion signals indicative of heart movement, generated by a motion sensor system that has mechanical contact with the individual's chest.

[0131] - Extract samples from the motion signal for each time length having one or more heartbeat cycles, such that the time phase of the heartbeat-related feature is the same relative to the start of the sample within each sample having that feature.

[0132] - Calculate the integral transform for each sample to map each sample to the basis functions of the integral transform, and

[0133] - Based on the coefficients that cause the integral transformation of each sample, the samples are classified into one of the predefined categories, such as, for example, healthy cases, cardiac abnormality cases, and uncertain cases.

[0134] A software module can be, for example, a subroutine or function implemented in a suitable programming language and a compiler suitable for that programming language and the programmable processing system. It is worth noting that source code corresponding to a suitable programming language represents a computer-executable software module, because the source code contains the information needed to control the programmable processing system to perform the aforementioned actions, and compilation only changes the format of that information. Furthermore, an interpreter can be provided for the programmable processing system so that the source code implemented in a suitable programming language does not need to be compiled before execution.

[0135] Computer program products according to exemplary and non-limiting embodiments include: computer-readable media, such as an optical disc "CD" encoded with a computer program according to embodiments of the present invention, a flash drive encoded with a computer program according to the present invention, or a digital download of a computer program according to the present invention.

[0136] Computer-readable media according to exemplary and non-limiting embodiments are encoded with a computing program according to embodiments of the present invention.

[0137] Computer-readable media, such as computer-executable code, can take many forms, including but not limited to tangible storage media, carrier media, or physical transmission media. Non-volatile storage media include, for example, optical discs or disks, any storage device such as any one or more computers, such as those used to implement the database shown in the accompanying drawings. Volatile storage media include dynamic memory, such as the main memory of such a computer platform. Tangible transmission media include coaxial cables; including lines that constitute a bus within a computer system. Carrier transmission media can take the form of electrical or electromagnetic signals, or sound or light waves, such as those generated during radio frequency (RF) and infrared (IR) data communications. Therefore, common forms of computer-readable media include: floppy disks, hard disks, magnetic tapes, any other magnetic media, optical disc read-only memory (CD-ROM), digital video disc (DVD) or DVD-ROM, any other optical media, punched card tape, any other physical storage medium with a perforated pattern, random access memory (RAM), read-only memory (ROM), programmable read-only memory (PROM) and erasable programmable read-only memory (EPROM), flash memory (EPROM), any other memory chip or cartridge, a carrier wave for transmitting data or instructions, a cable or link for transmitting such a carrier wave, or any other medium from which a computer can read programming code and / or data. Many of these forms of computer-readable media may involve carrying one or more sequences of one or more instructions to a processor for execution.

[0138] The signals according to exemplary and non-limiting embodiments are encoded to carry information defining a computer program according to an embodiment of the invention.

[0139] In some embodiments, the system for indicating cardiac condition may include an input model configured to receive input including motion signals from a motion sensor system having mechanical contact with the individual's chest. In some embodiments, the motion signals may be read from memory, in which case the motion signals have been measured and recorded in memory beforehand. In some embodiments, the motion signals may be received from an external data transmission system. Therefore, according to embodiments of the invention, measurement is not a necessary and essential step. In some embodiments, input may be received from an accelerometer, a gyroscope, or both.

[0140] In some embodiments, the system may further include a sample recognition model, which is configured to:

[0141] - Identify one or more characteristics of the motion signal received from the input model over the signal's time length, including heartbeat-related features, and

[0142] - Generate one or more samples for each motion signal by mapping one or more characteristic features of one or more samples received from the input model.

[0143] In some embodiments, one or more characteristic features include one or more of amplitude features, frequency features, power features, intensity features, energy features, or any combination thereof. In some embodiments, the characteristic feature may be a heartbeat-related feature, such as the closure of the mitral and tricuspid valves "S1" and the closure of the aortic and pulmonary valves "S2". In some embodiments, the feature may include maximum amplitude: this feature identifies the maximum amplitude of the CWT coefficients for each scale parameter value. It can help capture the dominant frequency component within each sample. In some embodiments, the feature may include mean: calculating the mean of the CWT coefficients for each scale parameter value yields a measure of the average power or intensity of that frequency component over time. In some embodiments, the feature may include total energy: this is the sum of squares of the CWT coefficients at each scale parameter value. It represents the total energy or power in the motion signal for a particular scale parameter.

[0144] In some embodiments, the system may further include a classification model configured to generate a prediction of the sample’s classification into one of the categories based on the integral transform coefficients of the sample.

[0145] In some embodiments, the classification model may apply an integral transform to the samples without preprocessing them. In other embodiments, the classification model may preprocess the samples and then apply an integral transform to the preprocessed samples. Preprocessing may include, for example, filtering the samples. In some embodiments, the integral transform for each sample may be a continuous wavelet transform (CWT) of the sample. In other embodiments, the integral transform for each sample may be a Fourier transform of the sample.

[0146] In some embodiments, the classification model can directly apply the classification algorithm to the coefficients, resulting in an integral transformation of the samples. In some embodiments, the coefficients can be preprocessed, and the classification algorithm can be applied to the preprocessed results. In some embodiments, the Dynamic Time Warp (DTW) function can be applied to measure the similarity between samples and reject samples that differ most significantly from the majority of samples.

[0147] In some embodiments, the categories may include healthy cases, cardiac abnormality cases, and uncertain cases.

[0148] In some embodiments, the system may further include an output model configured to output an indicator signal representing a classification result received from the classification model. In some embodiments, the indicator signal may further include one or more of the following: instructions to contact healthcare professionals based on the classification result, data related to motion signals indicating cardiac motion, a display of motion information indicating cardiac motion, or any combination thereof.

[0149] In some embodiments, the signal may be output as a visual notification, an audio notification, or a haptic notification, or any combination thereof.

[0150] In some embodiments, the sample identification model can also be configured to adjust the identification of one or more features based on feedback data from the predictive classification model. In some embodiments, the feedback data may include non-conforming data. In some embodiments, the feedback data may include classifying the sample as an uncertain case, a healthy case, or a case of cardiac abnormality.

[0151] In some embodiments, the sample identification model can also be configured to predictively generate one or more modified samples based on feedback data from the predictive classification model. In some embodiments, the feedback data may include non-conforming data. In some embodiments, the feedback data may include classifying samples as uncertain cases, healthy cases, or cardiac abnormality cases.

[0152] In some embodiments, one or more modified samples can be generated from a mapping of one or more characteristics of one or more samples. In some embodiments, one or more samples can be excluded from the mapping. In some embodiments, one or more samples can be copied from the mapping.

[0153] In some embodiments, the classification model can also be configured to generate one or more modified classifications based on feedback data from the sample recognition model. In some embodiments, the feedback data may also include one or more non-compliant samples or data points.

[0154] In some embodiments, the output model can also be configured to generate an updated indication of the modified classification. In some embodiments, the updated indication may include an indication that classifies the sample into one of three cases: cardiac abnormality, healthy, or indeterminate. In some embodiments, the indication may be an error message. In some embodiments, the indication may be a request to generate additional motion signals.

[0155] In some embodiments, a system for generating an indication of an individual's cardiac condition includes:

[0156] - A signal interface configured to receive inputs including signals indicative of heart movement generated by a motion sensor system having mechanical contact with the individual's chest; and

[0157] - The processor is configured as follows:

[0158] i) Identify one or more characteristic features of each motion signal over its time length.

[0159] ii) Apply an integral transform to each sample to map each sample to the basis functions of the integral transform.

[0160] iii) Classify samples into one of the predefined categories based on the integral transform coefficients of each sample, and

[0161] iv) Output a signal that includes the results of the sample classification.

[0162] In some embodiments, the processor is further configured to select a subset of samples from the motion signal, wherein the subset of samples has corresponding characteristic features.

[0163] In some embodiments, the motion sensor system includes an accelerometer, a gyroscope, or both.

[0164] In some embodiments, one or more characteristic features include one or more of amplitude features, frequency features, power features, intensity features, energy features, or any combination thereof.

[0165] In some embodiments, one or more categories include healthy cases, cardiac abnormality cases, and uncertain cases.

[0166] In some embodiments, the signal may further include one or more of the following: instructions to contact healthcare professionals based on classification, data output related to signals indicating heart movement, display of signals indicating heart movement, or any combination thereof.

[0167] In some embodiments, a method for generating an indication of an individual's cardiac condition includes:

[0168] - Receive input, including signals indicating heart movement generated by a motion sensor system that has mechanical contact with the individual's chest.

[0169] - Identify one or more characteristic features of each motion signal over its time length.

[0170] - Apply an integral transform to each sample to map each sample to the basis functions of the integral transform.

[0171] - Based on the integral transform coefficients of each sample, classify the samples into one of the predefined categories, and

[0172] - The output includes a signal that represents the sample classification results.

[0173] In some embodiments, the method further includes selecting a subset of samples from the motion signal, wherein the subset of samples has corresponding characteristic features.

[0174] In some embodiments, receiving input from a motion sensor system includes receiving input from an accelerometer, a gyroscope, or both.

[0175] In some embodiments, identifying one or more characteristic features also includes identifying one or more of amplitude features, frequency features, power features, intensity features, energy features, or any combination thereof.

[0176] In some embodiments, the categories include healthy cases, cardiac abnormality cases, and uncertain cases.

[0177] In some embodiments, the output signal may further include one or more of the following: outputting instructions to contact healthcare professionals based on predictive classification, outputting data about signals indicating heart movement, or outputting a display of signals indicating heart movement, or any combination thereof.

[0178] In some embodiments, a system for indicating cardiac condition includes:

[0179] i) An input model configured to receive input, including motion signals, from a motion sensor system that has mechanical contact with the individual's chest.

[0180] ii) Sample recognition model, which is configured as follows:

[0181] a) Identify one or more characteristic features of each motion signal received from the input model over the duration of the signal, and

[0182] b) Generate one or more samples for each motion signal by mapping one or more feature characteristics of one or more samples received from the input model.

[0183] iii) A classification model configured to classify the samples into one of predefined categories based on the coefficients of the integral transform of each of the one or more samples, and

[0184] iv) Output model, which is configured to output the classification results received from the predictive classification model.

[0185] In some embodiments, the input received from the motion sensor system includes input from an accelerometer, a gyroscope, or both.

[0186] In some embodiments, one or more characteristic features include one or more of amplitude features, frequency features, power features, intensity features, energy features, or any combination thereof.

[0187] In some embodiments, the categories include healthy cases, cardiac abnormality cases, and uncertain cases.

[0188] In some embodiments, the signal may further include one or more of the following: instructions to contact healthcare professionals based on predictive classification, data output related to a signal indicating heart movement, a display of a signal indicating heart movement, or any combination thereof.

[0189] In some embodiments, the sample recognition model is also configured to adjust the recognition of one or more feature characteristics based on feedback data from the predictive classification model.

[0190] In some embodiments, the sample identification model is further configured to predictively generate one or more modified samples based on feedback data from the predictive classification model.

[0191] In some embodiments, one or more modified samples are generated from modifications to the mapping of one or more characteristic features of one or more samples.

[0192] In some embodiments, the predictive classification model is also configured to generate one or more modified classifications based on feedback data from the sample recognition model.

[0193] In some embodiments, the output model is further configured to generate an update indication of the modified classification.

[0194] Example Case

[0195] In this example, a gradient boosting classifier "GBC" was trained using a training dataset collected from 106 individuals. Of these individuals, 57 were diagnosed with aortic stenosis "AS," and 49 had a history of heart disease, excluding valvular heart disease and heart failure. The motion signals associated with the training dataset were generated by accelerometers on the individuals' chests. The integral transform used during the training phase was the continuous wavelet transform "CWT," and as described earlier in this paper, the feature maximum amplitude, mean, and total energy were calculated and fed into the gradient boosting classifier.

[0196] A test dataset was collected from an additional 106 age- and sex-matched individuals. Of these 106 individuals, 7 were excluded due to issues during data collection. These issues included poor signal quality during recording, such as a poor signal-to-noise ratio (SNR), or a lack of information including the individual's true state at the time of recording. This left 99 individuals. The motion signals associated with the test dataset were generated by accelerometers on the individuals' chests.

[0197] Table 1 shows the results achieved using the method according to the embodiment on the test dataset, where the integral transform is a continuous wavelet transform "CWT", the maximum amplitude, average value, and total energy of the aforementioned features are calculated, and fed into a gradient boosting classifier "GBC" trained with the aforementioned training dataset. The predictions passed by this method were examined by a cardiologist. In Table 1, true positives are individuals with aortic stenosis "AS" who are correctly diagnosed with AS, and true negatives are individuals without AS who are accurately diagnosed without AS:

[0198] Table 1.

[0199] Predicted positive Predicted negative True positive 46 4 True negative 6 43

[0200] Remark

[0201] The specific examples provided in the description above should not be construed as limiting the scope of the invention. Unless otherwise expressly stated, the group of lists and examples provided in the description above is not exhaustive.

Claims

1. A system (100) comprising: - a signal interface (101) for receiving a motion signal, the motion signal being indicative of heart motion and being generated by a motion sensor system (103) having mechanical contact with a chest of an individual, and - a processing system (102) coupled to the signal interface, characterized in that the processing system is configured to: - extract from the motion signal each sample having a time length of one or more heart beat periods such that a time phase of a heart beat related feature is identical within each sample having the feature with respect to a beginning of the sample, - compute an integral transform of each sample to map each sample to basis functions of the integral transform, and - classify a sample to one of predefined classes based on coefficients resulting in the integral transform of each sample, set an indicator signal to represent a classification result, and output the indicator signal.

2. The system of claim 1, wherein, the processing system is configured to compute a continuous wavelet transform of each sample, the continuous wavelet transform being the integral transform.

3. The system of claim 1, wherein, the processing system is configured to compute a Fourier transform of each sample, the Fourier transform being the integral transform.

4. The system of any one of claims 1-3, wherein, the processing system is configured to perform dynamic time warping to measure similarity between samples and to reject a sample that is most different from most samples.

5. The system of any one of claims 1-4, wherein, the motion signal comprises a first signal component produced by a first motion sensor (103a) being part of the motion sensor system and having mechanical contact with a first position on the chest of the individual and a second signal component produced by a second motion sensor (103b) being part of the motion sensor system and having mechanical contact with a second position on the chest of the individual.

6. The system of any one of claims 1-5, wherein, at least a part of the motion signal is produced by an accelerometer having mechanical contact with the chest of the individual.

7. The system of any one of claims 1-6, wherein, at least a part of the motion signal is produced by a gyroscope having mechanical contact with the chest of the individual.

8. The system of any one of claims 1-7, wherein, the device comprises the motion sensor system (103).

9. The system of claim 8, wherein, the motion sensor system comprises one or more accelerometers.

10. The system of claim 8 or 9, wherein, the motion sensor system comprises one or more gyroscopes.

11. The system of any one of claims 1-10, wherein, the signal interface is configured to receive an electrocardiogram signal and the processing system is configured to find R-peaks from the electrocardiogram signal and to determine the time length of the samples based on a time length of a time period between successive ones of the R-peaks.

12. The system of claim 11, wherein, the device comprises electrocardiogram electrodes (108a, 108b) and electrical wires connecting the electrocardiogram electrodes to a differential amplifier (109) configured to produce the electrocardiogram signal.

13. The system of any one of claims 1-12, wherein, the predefined classes are: a healthy case, a heart abnormality case, and an uncertain case.

14. The system of claim 13, wherein, the heart abnormality is aortic stenosis.

15. A method for indicating a heart abnormality, the method comprising: - receiving (201) a motion signal, the motion signal being indicative of heart motion and being generated by a motion sensor system having mechanical contact with a chest of an individual, characterized in that the method comprises: - extracting (202) from the motion signal each sample having a time length of one or more heart beat periods such that the time phase of a heart beat related feature is identical within each sample having this feature with respect to the beginning of the sample, - computing (203) an integral transform of each sample to map each sample to basis functions of the integral transform, and - classifying (204) each sample to one of predefined classes based on coefficients resulting in the integral transform of the sample, setting an indicator signal to represent the classification result, and outputting the indicator signal.

16. A computer program comprising computer executable instructions for controlling a programmable processing system: - receiving a motion signal, the motion signal being indicative of a heart motion and being generated by a motion sensor system having mechanical contact to the chest of an individual, characterized in that the computer program comprising computer executable instructions for controlling the programmable processing system: - extracting from the motion signal each sample having a time length of one or more heart beat periods such that the time phase of a heart beat related feature is identical within each sample having this feature with respect to the beginning of the sample, - computing an integral transform of each sample to map each sample to basis functions of the integral transform, and - classifying each sample to one of predefined classes based on coefficients resulting in the integral transform of the sample, setting an indicator signal to represent the classification result, and outputting the indicator signal.

17. A method for determining a cardiac abnormality, the method comprising: (a) receiving a motion signal related to a motion of a heart of a subject; (b) generating a plurality of samples of the motion signal, wherein at least a portion of the plurality of samples has a first time length; (c) matching one or more corresponding points of each of the at least a portion of the plurality of samples having the first time length to generate a stack of signal samples; and (d) determining a cardiac abnormality based at least in part on comparing one or more features of the matched samples of the stack of signal samples to one or more reference features. generating an indication of the determination of the cardiac abnormality.

18. The method of claim 17, further comprising: The indication comprises an indication of a cardiac abnormality, an indication of a healthy case, or an indication of an inconclusive case.

19. The method of claim 18, wherein, The motion signal is received from one or more sensors.

20. The method of any one of claims 17-19, further comprising: The one or more sensors are configured to generate the motion signal based at least in part on measuring a motion of a heart.

21. The method of claim 20, wherein, The one or more sensors comprise an accelerometer, a gyroscope, or both.

22. The method of claim 20, wherein, A time length value is generated for each of the plurality of samples of the motion signal.

23. The method of any of claims 17-22, further comprising: The at least a portion of the plurality of samples of the motion signal is matched based at least in part on the time length value of the respective one or more samples.

24. The method of claim 23, further comprising: The one or more points of the at least a portion of the plurality of samples are matched based on a relative distance of the one or more points to the time length value.

25. The method of claim 24, further comprising: Generating a sample of the plurality of samples is an indication of a non-compliant sample.

26. The method of claim 24, further comprising: ​ 27. The method of claim 26, further comprising: generating the indication of the non-conforming sample based at least in part on an indication that one or more points of the non-conforming sample do not match at least a portion of the one or more samples.

28. The method of claim 27, further comprising: modifying the time length value of the non-conforming sample based at least in part on feedback comprising the indication of the non-conforming sample.

29. The method of claim 27, further comprising: modifying one or more points of the non-conforming sample based at least in part on feedback comprising the indication of the non-conforming sample.

30. The method of claim 28, further comprising: modifying one or more points of the non-conforming sample based at least in part on a modified time length value of the non-conforming sample.

31. The method of claim 29 or 30, further comprising: matching the modified one or more points of the non-conforming sample to at least a portion of the plurality of samples of the motion signal.

32. The method of claim 31, further comprising: removing the indication of the non-conforming sample.

33. The method of claim 32, further comprising: adding the non-conforming sample with the removed indication to the signal sample stack.

34. The method of any one of claims 17-33, wherein, the time length comprises one or more heart beat cycles.

35. The method of any one of claims 17-34, wherein, generating the signal sample stack further comprises determining an integral transform of each of the plurality of samples of the motion signal.

36. The method of claim 35, wherein, determining the cardiac abnormality further comprises grouping the plurality of samples of the motion signal based at least in part on coefficients related to the integral transform of each of the one or more samples.

37. The method of any one of claims 17-36, wherein, the one or more features of the signal sample stack comprise one or more of: an amplitude feature, a frequency feature, a power feature, an intensity feature, an energy feature, or any combination thereof.

38. The method of claim 37, further comprising: determining the cardiac abnormality based at least in part on the grouping of the one or more features of the signal sample stack.

39. The method of claim 38, further comprising: determining the cardiac abnormality based at least in part on an association of the grouped one or more features of the signal stack with an indication of a cardiac abnormality.

40. The method of any one of claims 17-39, wherein, the cardiac abnormality comprises aortic stenosis (AS), valvular heart disease, heart failure, or atrial fibrillation, or any combination thereof.

41. A system for determining a cardiac abnormality, the system comprising: a memory; and the one or more processors comprise: (a) a signal interface configured to receive a motion signal, wherein the motion signal relates to motion of a heart of a subject; (b) a signal processing model configured to generate (i) a plurality of signal samples extracted from the motion signal, and (ii) a time length value corresponding to each of the plurality of signal samples; (c) a signal matching model configured to match one or more corresponding points of at least a portion of the plurality of signal samples to generate a signal sample stack, wherein the signal sample stack comprises matched signal samples; (d) a classification model configured to: (i) determine one or more features of the signal sample stack, and (ii) determine a cardiac abnormality based at least in part on a comparison of the one or more features of the signal sample stack to one or more reference features.

42. The system of claim 41, further comprising: a sensor system.

43. The system of claim 42, wherein, the sensor system comprises a gyroscope, an accelerometer, or both.

44. The system of any one of claims 41-43, wherein, the signal matching model is further configured to match the at least a portion of the plurality of samples based at least in part on the time length value of each of the at least a portion of the plurality of signal samples.

45. The system of claim 44, wherein, The signal matching model is further configured to match one or more points of each signal sample of the at least a portion of the plurality of signal samples based at least in part on a relative distance of the one or more points to a beginning or an end of each respective signal sample.

46. The system of claim 44, wherein, The signal matching model is further configured to generate an indication that a signal sample in the plurality of samples is an out-of-sample.

47. The system of claim 46, wherein, The signal matching model is further configured to generate the indication of the out-of-sample based at least in part on an indication that one or more points of the out-of-sample do not match one or more signal samples in the plurality of signal samples.

48. The system of claim 47, wherein, The signal processing model is further configured to modify the time length value of the out-of-sample based at least in part on feedback comprising the indication of the out-of-sample.

49. The system of claim 47, wherein, The signal processing model is further configured to modify one or more points of the out-of-sample based at least in part on feedback comprising the indication of the out-of-sample.

50. The system of claim 48, wherein, The signal processing model is further configured to modify one or more points of the out-of-sample based at least in part on the modified time length value of the out-of-sample.

51. The system of claim 49 or 50, wherein, The signal matching model is further configured to match the modified one or more points of the out-of-sample to at least a portion of the plurality of samples of the motion signal.

52. The system of claim 51, wherein, The signal processing model is further configured to remove the indication of the out-of-sample.

53. The system of claim 52, wherein, The signal matching model is further configured to add the out-of-sample with the removed out-of-sample indication to the signal sample stack.

54. The system of any one of claims 41-53, wherein, The time length comprises one or more heart beat cycles.

55. The system of any one of claims 41-54, wherein, The signal matching model is further configured to generate the signal sample stack by determining an integral transform of each of the plurality of signal samples of the motion signal.

56. The system of claim 55, wherein, The signal matching model is further configured to generate the signal sample stack by matching the integral transform of at least a subset of the plurality of signal samples.

57. The system of claim 56, wherein, The signal matching model is further configured to generate the signal sample stack by matching coefficients related to the integral transform of at least a subset of the plurality of signal samples.

58. The system of any one of claims 41 to 57, wherein, The one or more features of the signal sample stack comprise one or more of: an amplitude feature, a frequency feature, a power feature, an intensity feature, an energy feature, or any combination thereof.

59. The system of claim 58, wherein, The classification model is further configured to determine the cardiac abnormality based at least in part on a comparison of the one or more features of the signal sample stack to reference features of one or more signal samples not in the signal sample stack.

60. The system of claim 58, wherein, The classification model is further configured to determine the cardiac abnormality based at least in part on an association of the one or more features of the signal sample stack to an indication of a cardiac abnormality.

61. The system of claim 41, wherein, The cardiac abnormality comprises aortic stenosis (AS), a heart valve disease, heart failure, or atrial fibrillation, or any combination thereof.

62. A system for cardiac abnormality classification, the system comprising a memory and one or more processors, the processors comprising: (a) a signal processing model configured to generate a time length value for each of one or more signal samples; (b) a signal matching model configured to: (i) receive input from a signal processing model comprising the one or more signal samples, (ii) predict at least one corresponding point on each signal sample, (iii) map one or more corresponding points of each signal sample to generate a signal sample stack; (c) a classification model configured to: (i) receive output data comprising the signal sample stack from the signal matching model, and (ii) generate a prediction of an appropriate classification category of a cardiac abnormality for the signal sample stack from one or more features of each signal sample in the signal sample stack.

63. The system of claim 62, wherein, The classification model is further configured to output an indication of the classification category, the indication being a cardiac abnormality case, a healthy case, or an indeterminate case.

64. The system of claim 62 or 63, wherein, The signal matching model is further configured to generate feedback output comprising data indicating that one or more signal samples are non-compliant samples.

65. The system of any one of claims 62-64, wherein, The signal processing model is further configured to receive the feedback data from the classification model and adjust one or more time length values.

66. The system of any one of claims 62-65, wherein, The classification model is further configured to provide classification feedback output data to the signal matching model.

67. The system of any one of claims 62 to 66, wherein, The signal matching model is further configured to adjust the mapping of the one or more signal samples based on the classification feedback output data.

68. The system of any one of claims 62 to 67, further comprising: An input model configured to receive input comprising motion signals from a motion sensor system in mechanical contact with a chest of an individual.

69. The system of claim 68, wherein, The motion signals are read from the memory.

70. The system of claim 68, wherein, The motion signals are received from an external data transmission system.

71. The system of claim 68, wherein, The motion signals are received from an accelerometer, a gyroscope, or both.

72. The system of any one of claims 62-71, wherein, The system further comprises a sample identification model configured to: (i) identify one or more characteristic heartbeat-related features of the motion signals received from the input model over a time length of a signal, and (ii) generate one or more samples of each motion signal by mapping the one or more characteristic features of one or more samples received from the input model.

73. The system of claim 72, wherein, The one or more characteristic features comprise one or more of: an amplitude feature, a frequency feature, a power feature, an intensity feature, an energy feature, or any combination thereof.

74. The system of claim 72, wherein, The characteristic features include heartbeat-related features such as the closure of the mitral and tricuspid valves value "S1" and the closure of the aortic and pulmonary valves "S2".

75. The system of claim 73 or 74, wherein, The features further comprise a maximum amplitude.

76. The system of claim 73 or 74, wherein, The features further comprise an average power or average intensity of frequency components over time.

77. The system of claim 73 or 74, wherein the features further comprise: A sum of squares of CWT coefficients at each scale parameter value.

78. The system of claim 73 or 74, wherein, The features further comprise a total energy or power in the motion signal for a particular scale parameter.

79. The system of any one of claims 62-78, wherein, The classification model is further configured to generate a prediction of a classification of the sample into one of the categories based on the integral transform coefficients of the sample.

80. The system of claim 79, wherein, The classification model is further configured to apply the integral transform to the sample without pre-processing the sample.

81. The system of claim 79, wherein, The classification model is further configured to: pre-process the samples, and subsequently apply the integral transform to the pre-processed samples.

82. The system of claim 81, wherein, The pre-processing of the samples includes filtering the samples.

83. The system of claim 81, wherein, The integral transform of each sample is a continuous wavelet transform (CWT) of the sample.

84. The system of claim 81, wherein, The integral transform of each sample is a Fourier transform of the sample.

85. The system of claim 81, wherein, The classification model is further configured to apply a classification algorithm to generate the integral transform of the samples.

86. The system of claim 81, further comprising: Pre-processing coefficients related to the samples.

87. The system of claim 81, wherein, The classification model is further configured to apply a dynamic time warping (DTW) function to determine similarity between the samples.

88. The system of claim 87, wherein, The classification model is further configured to reject one or more samples that are most different from the majority of the samples.

89. The system of claim 87, wherein, The classification model is further configured to classify the sample data as a healthy case, a cardiac abnormality case, or an uncertain case.

90. The system of any one of claims 62-89, wherein, The system further comprises an output model configured to output an indicator signal representing a classification result received from the classification model.

91. The system of claim 90, wherein, The indicator signal includes one or more of: an instruction to contact a healthcare professional based on the classification result, data regarding the motion signal indicative of cardiac motion, or a display of the motion signal indicative of cardiac motion, or any combination thereof.

92. The system of claim 91, wherein, The signal is generated as a visual notification, an audio notification, or a haptic notification, or any combination thereof.

93. The system of any one of claims 62-92, wherein, The sample identification model is further configured to adjust identification of the one or more characteristic features based on feedback data from the predictive classification model.

94. The system of claim 93, wherein, The feedback data includes non-conforming data.

95. The system of claim 93, wherein, The feedback data is used to classify the sample as an uncertain case, a healthy case, or a cardiac abnormality case.

96. The system of any one of claims 62-95, wherein, The sample identification model is further configured to predictively generate one or more modified samples based on feedback data from the predictive classification model.

97. The system of claim 96, wherein, The feedback data is further configured to include non-conforming data.

98. The system of claim 97, wherein, The feedback data is used to classify the sample as an uncertain case, a healthy case, or a cardiac abnormality case.

99. The system of any one of claims 96 to 98, wherein, The one or more modified samples are generated by modifying a mapping of the one or more characteristic features of the one or more samples.

100. The system of claim 99, wherein, One or more samples are excluded from the mapping.

101. The system of claim 99, wherein, One or more samples are duplicated in the mapping.

102. The system of any one of claims 62-101, wherein, The classification model is further configured to generate one or more modified classifications based on feedback data from the sample identification model.

103. The system of claim 102, wherein, The feedback data includes one or more non-conforming samples or data points.

104. The system of claim 102 or 103, further comprising an output model configured to generate an updated indication of the modified classification.

105. The system of claim 104, wherein, The updated indication includes an indication of classifying the sample as one of a cardiac abnormality case, a healthy case, and an uncertain case.

106. The system of claim 105, wherein, The indication is an error message.

107. The system of claim 106, wherein, The indication includes a request to generate additional motion signals.

108. A system for generating an indication of a cardiac condition of an individual, comprising: (a) a signal interface configured to receive input, the input including a signal indicative of cardiac motion generated by a motion sensor system in mechanical contact with a chest of an individual; and (b) a processor configured to: i) identify one or more characteristic features of each motion signal over a time length of the signal, ii) apply an integral transform to each sample to map each sample to a basis function of the integral transform, iii) classify the sample to one of a predefined category based on a coefficient of the integral transform of each sample, and iv) output a signal including a result of the sample classification.

109. The system of claim 108, wherein, The processor is further configured to select a subset of samples from the motion signals, wherein the subset of samples have respective characteristic features.

110. The system of claim 108 or 109, wherein, The motion sensor system includes an accelerometer, a gyroscope, or both.

111. The system of any one of claims 108-110, wherein, The one or more characteristic features include one or more of an amplitude feature, a frequency feature, a power feature, an intensity feature, an energy feature, or any combination thereof.

112. The system of any one of claims 108-111, wherein, The one or more categories include a healthy case, a cardiac abnormality case, and an indeterminate case.

113. The system of any one of claims 108-112, wherein, The signal further includes one or more of an indication to contact a healthcare professional based on the classification, a data output related to the signal indicative of cardiac motion, or a display of the signal indicative of cardiac motion, or any combination thereof.

114. A method of generating an indication of a cardiac condition of an individual, comprising: (a) receiving input including a signal indicative of cardiac motion generated by a motion sensor system in mechanical contact with a chest of the individual; (b) identifying one or more characteristic features of each motion signal over a time length of the signal; (c) applying an integral transform to each sample to map each sample to a basis function of the integral transform; (d) classifying the sample to one of a predefined category based on a coefficient of the integral transform of each sample; and (e) outputting a signal including a result of the classification of the sample.

115. The method of claim 114, further comprising: The processor is further configured to select a subset of samples from the motion signals, wherein the subset of samples have respective characteristic features.

116. The method of claim 114 or 115, wherein, Receiving the input from the motion sensor system includes receiving input from an accelerometer, a gyroscope, or both.

117. The method of any one of claims 114-116, wherein, Identifying the one or more characteristic features further includes identifying one or more of an amplitude feature, a frequency feature, a power feature, an intensity feature, an energy feature, or any combination thereof.

118. The method of any one of claims 114-117, wherein, The category includes a healthy case, a cardiac abnormality case, and an indeterminate case.

119. The method of any one of claims 114-118, wherein, Outputting the signal further includes one or more of outputting an indication to contact a healthcare professional based on the predicted classification, outputting data related to the signal indicative of cardiac motion, or outputting a display of the signal indicative of cardiac motion, or any combination thereof.

120. A system for indicating a cardiac condition, comprising: (a) an input model configured to receive input including motion signals from a motion sensor system in mechanical contact with a chest of an individual; (b) a sample identification model configured to: (i) identify one or more characteristic features of each motion signal received from the input model over a time length of the signal, and (ii) generate one or more samples of each motion signal by mapping the one or more characteristic features of one or more samples received from the input model. (c) a classification model configured to classify each of the one or more samples into one of a predefined class based on coefficients of the integral transform of the sample; and (d) an output model configured to output a result of the classification received from the predictive classification model.

121. The system of claim 120, wherein, The input received from the motion sensor system includes input from an accelerometer, a gyroscope, or both.

122. The system of claim 120 or 121, wherein, The one or more characteristic features include one or more of: an amplitude feature, a frequency feature, a power feature, an intensity feature, an energy feature, or any combination thereof.

123. The system of any one of claims 120-122, wherein, The class includes a healthy case, a cardiac abnormality case, and an indeterminate case.

124. The system of any one of claims 120-123, wherein, The signal further includes one or more of: an indication to contact a healthcare professional based on the predictive classification, a data output related to the signal indicative of cardiac motion, or a display of the signal indicative of cardiac motion, or any combination thereof.

125. The system of any one of claims 120-124, wherein, The sample identification model is further configured to adjust identification of the one or more characteristic features based on feedback data from the predictive classification model.

126. The system of any one of claims 120-125, wherein, The sample identification model is further configured to predictively generate one or more modified samples based on feedback data from the predictive classification model.

127. The system of claim 126, wherein, The one or more modified samples are generated from a modification of the mapping of the one or more characteristic features of the one or more samples.

128. The system of claim 127, wherein, The predictive classification model is further configured to generate one or more modified classifications based on feedback data from the sample identification model.

129. The system of claim 128, wherein, The output model is further configured to generate an updated indication of the modified classification.

130. A system for generating an indication of a cardiac condition of an individual, the system comprising: (a) a signal interface configured to receive input including a motion signal indicative of cardiac motion generated by a motion sensor system in mechanical contact with a chest of an individual; and (b) a processing system configured to: (i) identify one or more characteristic heartbeat-related features of each sample of the motion signal over a length of time of the motion signal, (ii) apply an integral transform to each sample to map each sample to a basis function of the integral transform, (iii) classify a sample into one of a class based on coefficients of the integral transform resulting in each sample, and (iv) output an indicator signal including a result of the classification of the sample.

131. The system of claim 130, wherein, The processing system is further configured to select a subset of the samples from the motion signal, wherein the subset of the samples have respective characteristic heartbeat-related features.

132. The system of claim 130 or 131, wherein, The motion sensor system is further configured to include an accelerometer, a gyroscope, or both.

133. The system of any one of claims 130-132, wherein, The one or more characteristic heartbeat-related features are further configured to include one or more of: an amplitude feature, a frequency feature, a power feature, an intensity feature, an energy feature, or any combination thereof.

134. The system of any one of claims 130-133, wherein, The class is further configured to include a healthy case, a cardiac abnormality case, and an indeterminate case.

135. The system of any one of claims 130-134, wherein, The indicator signal is further configured to further include one or more of a notification of a predicted classification to a category, an indication to contact a healthcare professional based on the predicted classification, a data output related to the motion signal indicative of heart motion, or a display indicative of heart motion, or any combination thereof.

136. A method for generating an indication of a cardiac condition of an individual, the method comprising: (a) a signal interface configured to receive an input comprising a motion signal indicative of heart motion generated by a motion sensor system in mechanical contact with a chest of an individual; (b) identifying one or more characteristic heartbeat-related features of each sample of the motion signal over a temporal length of the motion signal; (c) applying an integral transform to each sample to map each sample to a basis function of the integral transform; (d) classifying each sample to one of categories representing different cardiac conditions based on coefficients of the integral transform resulting in each sample; and (e) outputting an indicator signal comprising a result of the classification of the samples.

137. The method of claim 136, further comprising: selecting a subset of the samples from the motion signal, wherein the subset of the samples have corresponding characteristic heartbeat-related features.

138. The method of claim 136 or 137, wherein, receiving the input from the motion sensor system is further configured to include receiving the input from an accelerometer, a gyroscope, or both.

139. The method of any one of claims 136-138, further comprising: identifying one or more of an amplitude feature, a frequency feature, a power feature, an intensity feature, an energy feature, or any combination thereof.

140. The method of any one of claims 136-139, wherein, the categories include a healthy case, a cardiac abnormality case, and an indeterminate case.

141. The system of any one of claims 136-140, further comprising: outputting an indication to contact a healthcare professional based on the classification, outputting data related to the signal indicative of heart motion, or outputting a display of the motion signal indicative of heart motion, or any combination thereof.

142. A system for indicating a cardiac condition, the system comprising: (a) an input model configured to receive an input comprising a motion signal from a motion sensor system in mechanical contact with a chest of an individual; (b) a sample identification model configured to: (i) identify one or more characteristic heartbeat-related features of the motion signal received from the input model over a temporal length of the signal, and (ii) generate one or more samples of the motion signal by mapping the one or more characteristic features of one or more samples received from the input model; (c) a classification model configured to generate a prediction of a classification of each of the one or more samples to one of categories based on integral transform coefficients of each of the one or more samples; and (d) an output model configured to output an indication of the classification received from the classification model of the one or more samples. the input received from the motion sensor system is further configured to include input from an accelerometer, a gyroscope, or both.

143. The system of claim 142, wherein, the one or more characteristic heartbeat-related features are further configured to include one or more of an amplitude feature, a frequency feature, a power feature, an intensity feature, an energy feature, or any combination thereof.

144. The system of claim 142 or 143, wherein, the categories are further configured to include a healthy case, a cardiac abnormality case, and an indeterminate case.

145. The system of any one of claims 142-144, wherein, ​ 146. The system of any one of claims 142-145, wherein, The signal is further configured to further include one or more of an indication to contact a healthcare professional based on the classification, a data output regarding the signal indicative of heart motion, or a display of the signal indicative of heart motion, or any combination thereof.

147. The system of any one of claims 142-146, wherein, The sample identification model is further configured to adjust identification of the one or more characteristic features based on feedback data from the classification model.

148. The system of claim 147, wherein, The feedback data includes an indication that the one or more characteristic features are not compliant.

149. The system of claim 148, wherein, The feedback data further includes an indication that the one or more characteristic features are not relevant.

150. The system of any one of claims 142-149, wherein, The sample identification model is further configured to generate one or more modified samples based on feedback data from the classification model.

151. The system of claim 150, wherein, The feedback data includes data indicating that one or more samples are not compliant.

152. The system of claim 151, wherein, The feedback data further includes data indicating that one or more samples are noise samples.

153. The system of claim 152, wherein, The one or more modified samples are generated from a modification of the mapping of the one or more characteristic features of the one or more samples.

154. The system of claim 153, wherein, The mapping is modified by excluding one or more samples from the mapping.

155. The system of claim 154, wherein, The mapping is modified by using different characteristic features for the mapping.

156. The system of any one of claims 142-156, wherein, The classification model is further configured to generate one or more modified classifications based on feedback data from the sample identification model.

157. The system of any one of claims 142-156, wherein, The output model is further configured to generate an updated indication of the modified classification.

Images