Patent Application
20230190164
The present disclosure relates to medical devices, systems, and methods and in particular, to a small form-factor device for providing electrocardiogram (ECG) monitoring.
Cardiovascular diseases are the leading cause of death in the world. In 2008, 30% of all global death can be attributed to cardiovascular diseases. It is also estimated that by 2030, over 23 million people will die from cardiovascular diseases annually. Cardiovascular diseases are prevalent across populations of first and third world countries alike, and affect people regardless of socioeconomic status.
Arrhythmia is a cardiac condition in which the electrical activity of the heart is irregular or is faster (tachycardia) or slower (bradycardia) than normal. Although many arrhythmias are not life-threatening, some can cause cardiac arrest and even sudden cardiac death. Indeed, cardiac arrhythmias are one of the most common causes of death when traveling to a hospital. Atrial fibrillation (A-fib) is the most common cardiac arrhythmia. In A-fib, electrical conduction through the ventricles of heart is irregular and disorganized. While A-fib may cause no symptoms, it is often associated with palpitations, shortness of breath, fainting, chest pain or congestive heart failure and also increases the risk of stroke. A-fib is usually diagnosed by taking an electrocardiogram (ECG) of a subject. To treat A-fib, a patient may take medications to slow heart rate or modify the rhythm of the heart. Patients may also take anticoagulants to prevent stroke or may even undergo surgical intervention including cardiac ablation to treat A-fib. In another example, an ECG may provide decision support for Acute Coronary Syndromes (ACS) by interpreting various rhythm and morphology conditions, including Myocardial Infarction (MI) and Ischemia.
Often, a patient with A-fib (or other type of arrhythmia) is monitored for extended periods of time to manage the disease. For example, a patient may be provided with a Holter monitor or other ambulatory electrocardiography device to continuously monitor the electrical activity of the cardiovascular system for e.g., at least 24 hours. Such monitoring can be critical in detecting conditions such as acute coronary syndrome (ACS), among others.
The American Heart Association and the European Society of Cardiology recommends that a 12-lead ECG should be acquired as early as possible for patients with possible ACS when symptoms present. Prehospital ECG has been found to significantly reduce time-to-treatment and shows better survival rates. The time-to-first-ECG is so vital that it is a quality and performance metric monitored by several regulatory bodies. According to the national health statistics for 2015, over 7 million people visited the emergency department (ED) in the United States (U.S.) with the primary complaint of chest pain or related symptoms of ACS. In the US, ED visits are increasing at a rate of or 3.2% annually and outside the U.S. ED visits are increasing at 3% to 7%, annually.
The novel features of the disclosure are set forth with particularity in the appended claims. A better understanding of the features and advantages of the present disclosure will be obtained by reference to the following detailed description that sets forth illustrative embodiments, in which the principles of the invention are utilized, and the accompanying drawings of which:
It is to be understood that the present disclosure is not limited in its application to the details of construction, experiments, exemplary data, and/or the arrangement of the components set forth in the following description. The embodiments of the present disclosure are capable of other embodiments or of being practiced or carried out in various ways. Also, it is to be understood that the terminology employed herein is for purpose of description and should not be regarded as limiting.
In the following detailed description of embodiments of the present disclosure, numerous specific details are set forth in order to provide a more thorough understanding of the disclosure. However, it will be apparent to one of ordinary skill in the art that the concepts within the disclosure can be practiced without these specific details. In other instances, well-known features have not been described in detail to avoid unnecessarily complicating the description.
An electrocardiogram (ECG) provides a number of ECG waveforms that represent the electrical activity of a person's heart. An ECG monitoring device may comprise a set of electrodes for recording these ECG waveforms (also referred to herein as “taking an ECG”) of the patient's heart. The set of electrodes may be placed on the skin of the patient in multiple locations and the electrical signal (ECG waveform) recorded between each electrode pair in the set of electrodes may be referred to as a lead. Varying numbers of leads can be used to take an ECG, and different numbers and combinations of electrodes can be used to form the various leads. Example numbers of leads used for taking ECGs are 1, 2, 6, and 12 leads.
The ECG waveforms (each one corresponding to a lead of the ECG) recorded by the ECG monitoring device may comprise data corresponding to the electrical activity of the person's heart. A typical heartbeat may include several variations of electrical potential, which may be classified into waves and complexes, including a P wave, a QRS complex, a T wave, and a U wave among others, as is known in the art. Stated differently, each ECG waveform may include a P wave, a QRS complex, a T wave, and a U wave among others, as is known in the art. The shape and duration of these waves may be related to various characteristics of the person's heart such as the size of the person's atrium (e.g., indicating atrial enlargement) and can be a first source of heartbeat characteristics unique to a person. The ECG waveforms may be analyzed (typically after standard filtering and “cleaning” of the signals) for various indicators that are useful in detecting cardiac events or status, such as cardiac arrhythmia detection and characterization. Such indicators may include ECG waveform amplitude and morphology (e.g., QRS complex amplitude and morphology), R wave-ST segment and T wave amplitude analysis, and heart rate variability (HRV), for example.
As noted above, ECG waveforms are generated from measuring multiple leads (each lead formed by a different electrode pair), and the ECG waveform obtained from each different electrode pair/lead may be different/unique (e.g., may have different morphologies/amplitudes). This is because although the various leads may analyze the same electrical events, each one may do so from a different angle.
There are different “standard” configurations for electrode placement that can be used to place electrodes on the patient. For example, an electrode placed on the right arm can be referred to as RA. The electrode placed on the left arm can be referred to as LA. The RA and LA electrodes may be placed at the same location on the left and right arms, preferably near the wrist in some embodiments. The leg electrodes can be referred to as RL for the right leg and LL for the left leg. The RL and LL electrodes may be placed on the same location for the left and right legs, preferably near the ankle in some embodiments. Lead I is typically the voltage between the left arm (LA) and right arm (RA), e.g. I=LA−RA. Lead II is typically the voltage between the left leg (LL) and right arm (RA), e.g. II=LL−RA. Lead III is the typically voltage between the left leg (LL) and left arm (LA), e.g. III=LL−LA. Augmented limb leads can also be determined from RA, RL, LL, and LA. The augmented vector right (aVR) lead is equal to RA−(LA+LL)/2 or −(I+II)/2. The augmented vector left (aVL) lead is equal to LA−(RA+LL)/2 or I−II/2. The augmented vector foot (aVF) lead is equal to LL−(RA+LA)/2 or II−I/2.
Lead I+Lead II/3.
It should be noted that a set of two or more leads may be transformed to generate a full, 12-lead ECG. Such transformation may be performed using a machine learning model (e.g., a neural network, deep-learning techniques, etc.). The machine learning model may be trained using 12-lead ECG data corresponding to a population of individuals. The data, before being input into the machine learning model, may be pre-processed to filter the data in a manner suitable for the application. For example, data may be categorized according to height, gender, weight, nationality, etc. before being used to train one or more machine learning models, such that the resulting one or models are finely-tuned the specific types of individuals. In a further embodiment, the machine learning model may be further trained based on a user's own ECG data, to fine-tune and personalize the model even further to decrease any residual synthesis error.
As discussed herein, a 12-lead ECG should be acquired as early as possible for patients with possible ACS when symptoms present as prehospital ECG has been found to significantly reduce time-to-treatment and shows better survival rates. In addition, current ambulatory ECG devices such as Holter monitors, are typically bulky and difficult for subjects to administer without the aid of a medical professional. For example, the use of Holter monitors requires a patient to wear a bulky device on their chest and precisely place a plurality of electrode leads on precise locations on their chest. These requirements can impede the activities of the subject, including their natural movement such as bathing and showering. Once an ECG is taken by such devices, the ECG is sent to the subject's physician who then analyzes the ECG waveforms and provides a diagnosis and other recommendations. Currently, this process often must be performed through hospital administrators and health management organizations and many patients do not receive feedback in an expedient manner.
A number of handheld ECG measurement devices are known, including devices that may adapt existing mobile telecommunications device (e.g., smartphones) so that they can be used to record ECS. However, such devices either require the use of external (e.g., plug-in) electrodes, or include electrodes in a housing that are difficult to properly hold and apply to the body. Many ECG monitors are also limited to acquiring limb leads (e.g., due to size and other constraints). However, as people age, their QRS and T-wave vector may gradually move from the frontal plane to the horizontal plane, thus increasing the importance of acquiring data from a horizontal plane lead.
As discussed herein, for patients potentially suffering from ACS, including Myocardial Infarction (MI) and Ischemia, a 12 lead ECG should be taken as early as possible to reduce the time to diagnosis and the time to treatment. The ECG monitoring device in accordance with embodiments of the present disclosure may provide decision support to physicians for ACS from the home of a patient itself, and provides a convenient way for doctors to order 12-lead ECG tests and view reports as often as is necessary for them to manage the health of their patients, especially if they suspect ACS. In addition, an ECG monitoring device in accordance with embodiments of the present disclosure may prevent a patient from undergoing the inconvenience and disruption of an office visit and may save the cost and time of utilizing an ECG technician in the physician's office.
The computing device 305 may comprise hardware to perform the functions described herein.
The positioning of the electrodes 310 allows the user to connect the ECG monitoring device 300 to themselves such that each electrode 310 contacts the appropriate location (e.g., as indicated in and discussed in further detail with respect to
The computing device 305 may further comprise a transceiver 308, which in some embodiments may implement any appropriate protocol for transmitting the ECG data to a computing device 405 (which may represent e.g., a cloud service) or a mobile computing device 450 (which may be remote from the ECG monitoring device 300 as illustrated in
Removing the EMG and motion artifacts allows for a cleaner signal and greater diagnostic quality. The user may position the electrodes such that electrodes are in contact with the RA, V2, LA, and LL positions respectively. In some examples, a mobile computing device 405 alerts the user to the appropriate location to place the electrodes on the body of the user. In another example, the chest and arm locations are marked and shown by a healthcare professional who prescribes the ECG monitoring device 300 to the patient. In some embodiments, one or more of the electrodes 310 of the ECG monitoring device 300 may comprise dry electrodes, which are easier for at home use because they stay in place more easily and have a longer lifetime use. Dry electrodes tend to give lower signal quality due to a lower level of conductive contact. In other embodiments, one or more of the electrodes 310 may be wet electrodes, e.g., when the ECG monitoring device 300 is being operated by a healthcare professional or the user has been trained to use wet electrodes. The wet electrodes may provide better conductive contact and produce a higher quality signal. The length of the housing cable 320 may accommodate users of various sizes. The housing cable 320 may have a length that corresponds to (e.g., fits with no slack between electrodes 310) a user with a particular height (e.g., six foot 2 inches), users that are smaller in height than the height the housing cable 320 corresponds to may simply have additional slack between electrodes once the ECG monitoring device 300 has been coupled to them.
In some embodiments, the housing cable 320 may have elastic properties and may have a length corresponding to a user with a smaller particular height (e.g., 4 foot 7 inches). The signal cables 325 may each be implemented using a flexible printed circuit (not shown) with an over mold (not shown) comprising e.g., a thermoplastic polyurethane material or any other appropriate material to provide a conductive channel through which analog signals generated by an electrode 310 may be routed to the computing device 305. In some embodiments, all of the electrodes 310 may be implemented using a single flexible printed circuit that runs the length of the housing cable 320. In these embodiments, the closer the individual is to the height the housing cable 320 corresponds to, the less tension will be applied the housing cable 320.
In the example
Referring back to
Vpred=f(W, Vx)
Where Vpred are the predicted V leads, f( ) is a transfer function with input leads Vx (I, II, and V2 in this case), and coefficients (W). Appropriate coefficients W that will minimize the error between Vpred and actual sampled V lead signals (Vreal) (Min err=E|(Vreal−Vpred)|{circumflex over ( )}2) must be found.
Locating such appropriate coefficients W is a problem that may be solved using any appropriate method such as a supervised learning task or a curve fitting problem. In some embodiments, the lead synthesis software 410 may utilize linear optimization and the least square method (LS):
Vpred=f(W, Vx)-->Vpred=Vx W
By stacking many paired samples of Vx to form a matrix X and Y (given as Y=X·W), linear optimization and the LS method can be used to solve for W. More specifically, processing device 206 may prepare the matrix X and Y for Y=X·W, using the LS method to obtain a conversion coefficient:
X′=X{circumflex over ( )}t
The covariance of X may then be given as:
CovX=X′X
W may then be calculated as follows:
Xinv=inv(CovX)
W=CovX_inv*X′*Y, (here we assume CovX is a full rank matrix).
Training data comprising e.g., 100,000 ECGs with average beats may be used for the training of the lead conversion ML model. The final conversion model is a matrix W having a 3 by 5 shape. The leads V1, V3, V4, V5, and V6 are predicted using the input leads I, II, V2.
In some embodiments, the computing device 405 may quantify the quality of the lead conversion ML model and determine whether a different lead conversion ML model (e.g., a more individualized model, or a multiple-dipole model) should be used. Techniques for quantifying the quality of a lead conversion ML model may be complicated since most statistical similarity methods are more closely related to amplitudes of every point, like the popular R-square method. However, the overall ECG morphology patterns which are used for interpretations are not based on amplitudes alone. For example, a Q wave is important for myocardial infarction detection, but its amplitude is generally much smaller than an R wave. One way to measure the quality of conversion is to compare the important ECG features used for ECG interpretations. Any appropriate algorithm (such as GE's EK12/12SL algorithm) can be used for this purpose. The data shown in table 1 below was obtained using the R-square (R2) algorithm (shown directly below), however, any appropriate 12-lead measurement algorithm may be used.
As can be seen, R-V1 is higher than the other V leads. Theoretically, V1 is the most difficult to predict from leads I, II, V2, since it represents more right ventricular activity, while the input leads are more reflective of the left ventricular electric field. The V3 and V4 signals usually have higher amplitudes than other leads due to their proximity to the heart.
However, the computing device 405 may also determine that the performance provided by the single dipole global model is not sufficient. For example,
A single pole cardiac source model may cover all phases of cardiac signal progression, including both atrial and ventricular depolarization and repolarization, while being relatively simple. However, such a model may be oversimplified in certain circumstances since a multiple-dipole model generally provides better accuracy than a single dipole model. Thus, in some embodiments, the computing device 405 may utilize a lead conversion ML model based on a multiple-dipole conversion model. The below table illustrates R-square statistics when a multiple-dipole model is used. As can be seen, both the QRS and ST-T segments show improved accuracy (relative to the single-dipole model in table 1 above), while the P-wave results are not improved. As a result, the lead conversion ML model used by computing device 405 may consider depolarization and repolarization separately.
Referring to the basic optimization equation (Vpred=f(W, Vx)), it is clear that both a linear function f( ) and a non-linear function f( ) can be used for finding the W. In some embodiments, the computing device 405 may utilize a nonlinear lead conversion model in situations where the increased computational burden required for the use of a nonlinear model is justified by significantly superior performance.
A number of deep learning methods may also be used to synthesize a full 12 lead set from the set of leads measured by the ECG monitoring device 300. For example, the lead conversion ML model may utilize artificial neural networks (ANNs) for supervised classification, where the outcome of the model represents the probability of the input sample to be in a specific class of data or exhibits some peculiar characteristics. In another example, a data driven approach based on convolutional neural networks (CNNs) is used. By using convolution operations, the lead conversion ML model may take into account the correlation among temporally closed input samples to infer a single output data point. More specifically, a single output sample (each precordial lead) at a generic time t is affected by all the input samples (all limb leads) from t−τ to t+τ. The value of τ, which represents the receptive field of the network, highly depends on the model architecture and typically increases with its depth, i.e., the number of consecutive layers. The ability to generalize on unseen data, and avoid over-fitting issues, is of primary importance for all data driven approaches. Complex models, along with small datasets, may lead to excellent performance on the training set, but may perform poorly on unseen data. Any appropriate regularization method may be used to optimize the model, such as inter and intra-layer normalization (e.g., batch normalization and layer normalization), and data augmentation techniques. Finally, to improve the effectiveness and efficiency of the model, the use of residual connections, i.e., an identity mapping that allow gradients to flow through a layer during the backpropagation of gradient-based optimization algorithms may be utilized.
The computing device 405 may execute an ECG waveform interpretation software 415 in order to perform interpretation based on the synthesized full 12-lead set of ECG waveforms using an interpretation ML model. The ECG waveform interpretation software 415 may comprise an interpretation ML model which may function to determine (based on the full 12 lead set measured/generated by the computing device 405) interpretations indicating myocardial ischemia (anterior, lateral, ischemia), myocardial infarction (anterior, lateral mi), left and right bundle branch block and right ventricular hypertrophy, among others. The interpretation ML model may be trained to perform well on morphology-based abnormalities using the converted 12-leads. More specifically, the interpretation ML model may comprise a deep neural network (DNN) model that is trained with converted lead signals, so that it can identify new ECG feature patterns, even if they are not identical with the original ones, thus enabling the interpretation ML model to differentiate among different abnormalities. The interpretation ML model may be a convolutional DNN with 6 residual blocks and 3 fully connected layers. The ML interpretation model may also have dropout and batch normalization layers to improve the generalization.
The interpretation ML model may be trained using a 12-lead ECG database (not shown), which has ECG data for a large number of 12-lead ECGs, each with e.g., 10 seconds of data. The 12-lead ECG database may include ECG data with various types of ECG abnormalities. In the training data, morphology-based ECGs may be clustered into 6 categories: ischemia, infarction, left bundle branch block (LBBB), right bundle branch block (RBBB), left ventricle hypertrophy (LVH), and others. Of those ECGs, a majority may be used for training, while the remainder are used for testing. Experimental data has shown that training the interpretation ML model on a converted lead set optimizes the interpretation performance. Thus, in some embodiments the interpretation ML model may be further trained on a converted lead set to optimize its interpretation performance. More specifically, the interpretation ML model is first trained and tested with the originally sampled 12-lead data and the interpretation performance is recorded. The interpretation ML model is then reinitialized and retrained/retested with converted 12-lead data, and the interpretation performance is recorded.
The exemplary computer system 800 includes a processing device 802, a main memory 804 (e.g., read-only memory (ROM), flash memory, dynamic random access memory (DRAM), a static memory 806 (e.g., flash memory, static random access memory (SRAM), etc.), and a data storage device 818, which communicate with each other via a bus 830. Any of the signals provided over various buses described herein may be time multiplexed with other signals and provided over one or more common buses. Additionally, the interconnection between circuit components or blocks may be shown as buses or as single signal lines. Each of the buses may alternatively be one or more single signal lines and each of the single signal lines may alternatively be buses.
Computing device 800 may further include a network interface device 908 which may communicate with a network 820. Processing device 802 represents one or more general-purpose processing devices such as a microprocessor, central processing unit, or the like. More particularly, the processing device may be complex instruction set computing (CISC) microprocessor, reduced instruction set computer (RISC) microprocessor, very long instruction word (VLIW) microprocessor, or processor implementing other instruction sets, or processors implementing a combination of instruction sets. Processing device 802 may also be one or more special-purpose processing devices such as an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), a digital signal processor (DSP), network processor, or the like. The processing device 802 is configured to execute lead synthesis and interpretation generation instructions 825, for performing the operations and steps discussed herein.
The data storage device 815 may include a machine-readable storage medium 828, on which is stored one or more sets of lead synthesis and interpretation generation instructions 825 (e.g., software) embodying any one or more of the methodologies of functions described herein. The lead synthesis and interpretation generation instructions 825 may also reside, completely or at least partially, within the main memory 804 or within the processing device 802 during execution thereof by the computer system 800; the main memory 804 and the processing device 802 also constituting machine-readable storage media. The lead synthesis and interpretation generation instructions 825 may further be transmitted or received over a network 820 via the network interface device 808.
Terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. For example, as used herein, the singular forms “a”, “an” and “the” are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms “comprises” and/or “comprising,” when used in this specification, specify the presence of stated features, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, steps, operations, elements, components, and/or groups thereof. As used herein, the term “and/or” includes any and all combinations of one or more of the associated listed items and may be abbreviated as “/”.
Spatially relative terms, such as “under”, “below”, “lower”, “over”, “upper” and the like, may be used herein for ease of description to describe one element or feature's relationship to another element(s) or feature(s) as illustrated in the figures. It will be understood that the spatially relative terms are intended to encompass different orientations of the device in use or operation in addition to the orientation depicted in the figures. For example, if a device in the figures is inverted, elements described as “under” or “beneath” other elements or features would then be oriented “over” the other elements or features. Thus, the exemplary term “under” can encompass both an orientation of over and under. The device may be otherwise oriented (rotated 90 degrees or at other orientations) and the spatially relative descriptors used herein interpreted accordingly. Similarly, the terms “upwardly”, “downwardly”, “vertical”, “horizontal” and the like are used herein for the purpose of explanation only unless specifically indicated otherwise.
Although the terms “first” and “second” may be used herein to describe various features/elements, these features/elements should not be limited by these terms, unless the context indicates otherwise. These terms may be used to distinguish one feature/element from another feature/element. Thus, a first feature/element discussed below could be termed a second feature/element, and similarly, a second feature/element discussed below could be termed a first feature/element without departing from the teachings of the present invention.
As used herein in the specification and claims, including as used in the examples and unless otherwise expressly specified, all numbers may be read as if prefaced by the word “about” or “approximately,” even if the term does not expressly appear. The phrase “about” or “approximately” may be used when describing magnitude and/or position to indicate that the value and/or position described is within a reasonable expected range of values and/or positions. For example, a numeric value may have a value that is +/−0.1% of the stated value (or range of values), +/−1% of the stated value (or range of values), +/−2% of the stated value (or range of values), +/−5% of the stated value (or range of values), +/−10% of the stated value (or range of values), etc. Any numerical range recited herein is intended to include all sub-ranges subsumed therein.
While preferred embodiments of the present disclosure have been shown and described herein, it will be obvious to those skilled in the art that such embodiments are provided by way of example only. Numerous variations, changes, and substitutions will now occur to those skilled in the art without departing from the invention. It should be understood that various alternatives to the embodiments of the invention described herein may be employed in practicing the invention. It is intended that the following claims define the scope of the invention and that methods and structures within the scope of these claims and their equivalents be covered thereby.
