Systems and methods for pacing based on pulse pressure

Abstract

Disclosed are methods and systems for outputting a pacing rate, including: receiving a heart rate and a pulse pressure of a patient; determining a first pacing rate to modify the heart rate based on the heart rate and the pulse pressure; outputting the first pacing rate; receiving a subjective input subsequent to outputting the first pacing rate; determining a second pacing rate based on the subjective input and first pacing rate; and outputting the second pacing rate.

Description

SYSTEMS AND METHODS FOR PACING BASED ON PULSE PRESSURECROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims benefit of US Provisional Patent Application No. 63 / 701,123 filed on September 30, 2024. The entirety of this application is hereby incorporated herein by reference.TECHNICAL FIELD

[0002] Various examples of the present disclosure relate generally to cardiac pacing based on at least one input, and more particularly, to systems and methods for outputting cardiac pacing rates based on pulse pressure (PP).BACKGROUND

[0003] Many factors, such as Hypertension (HTN), may contribute to cardiovascular mortality. Patients with heart conditions, such as Hypertension, may require heart pacing in order to help the different heart chambers beat in sync. Heart chambers that have a synchronized beat are able to pump blood more efficiently throughout the body.

[0004] Generally, a pacing rate for a pacemaker may be determined by an internal algorithm of the pacemaker. The base rate of the pacemaker, or the lowest heart rate allowed, may be set by a cardiologist taking into account the output of rhythm studies and other tests or by a machine learning algorithm. Current approaches to treatment rely on a patient’s presence at a provider’s facility to allow for pacing rate adjustment.

[0005] Unless otherwise indicated herein, the materials described in this section are not prior art to the claims in this application and are not admitted to be prior art, or suggestions of the prior art, by inclusion in this section.SUMMARY

[0006] A cardiac pacing system for treating heart failure with preserved ejection fraction (“HFpEF”) including an implantable pulse generator configured to deliver atrial pacing signals, a sensor to measure systolic blood pressure (“SBP”) and diastolic blood pressure (“DBP”) in real time, a controller to calculate a pulse pressure (“PP”) value based on a difference between the SBP and DBP, and a pacing algorithm executed by the controller that may determine an optimal atrial pacing rate based on the calculated PP value. The pacing algorithm may adjust the atrial rate to one or more of improve patient exercise tolerance and provide symptomatic relief.

[0007] The pacing algorithm may determine changes in PP over time and may adjust atrial pacing rate proportionally to a detected percent change. The pacing algorithm may calculate an average PP over a predetermined number of cardiac cycles and may use the averaged PP to determine the atrial pacing rate. The pacing algorithm may compute a moving average or weighted average of PP values across n intervals to smooth transient fluctuations. The pacing algorithm may adjust atrial pacing rate upward when PP decreases below a patient-specific threshold and downward when PP increases above the threshold. The controller may apply a derivative of PP with respect to time (“dPP / dt”) to determinedynamic changes in vascular compliance and adjusts atrial pacing accordingly. The controller may calculate a normalized PP index by dividing PP by mean arterial pressure (“MAP”) and use the PP index to determine pacing rate. The controller may compute one or more of a spectral and a frequency-domain analysis of PP variability and adjust atrial pacing rate to maintain PP variability within a therapeutic range.

[0008] A method of controlling atrial pacing in a patient with HFpEF including measuring SBP and DBP on a beat-to-beat basis, calculating PP as a difference between SBP and DBP, determining a pacing rate based on PP or a mathematical derivative thereof, and delivering atrial pacing at the determined rate to improve exercise capacity and symptomatic relief.

[0009] The determining the pacing rate may include calculating a percent change in PP relative to a baseline PP value. The determining the pacing rate may include averaging PP values across a plurality of cardiac cycles. The method may further include adjusting the pacing rate based on both PP and an additional hemodynamic parameter selected from stroke volume, cardiac output, or heart rate variability.

[0010] A non-transitory computer-readable medium storing instructions which, when executed by a controller of a cardiac pacemaker, cause the controller to acquire SBP and DBP values, compute PP values based on SBP-DBP, determine changes in PP according to one or more of percent change, averaged values, or variability analysis, adjust atrial pacing rate to optimize patient functional capacity.BRIEF DESCRIPTION OF THE DRAWINGS

[0011] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate various examples and, together with the description, serve to explain the principles of the disclosed examples.

[0012] Aspects of the disclosure may be implemented in connection with examples illustrated in the attached drawings. These drawings show different aspects of the present disclosure and, where appropriate, reference numerals illustrating like structures, components, materials, and / or elements in different figures are labeled similarly. It is understood that various combinations of the structures, components, and / or elements, other than those specifically shown, are contemplated and are within the scope of the present disclosure. Moreover, there are many examples described and illustrated herein.

[0013] FIG. 1 depicts an exemplary environment for determining a pacing rate, according to one or more examples.

[0014] FIG. 2A depicts a flow diagram of an exemplary method for determining a pacing rate, according to one or more examples.

[0015] FIG. 2B depicts a flow diagram of another exemplary method for determining a pacing rate, according to one or more examples.

[0016] FIG. 3 depicts a flowchart of an exemplary method for determining a pacing rate, according to one or more examples.

[0017] FIG. 4 depicts a further flow diagram of an exemplary method for determining a pacing rate via a pseudo-subjective machine learning model, according to one or more examples.

[0018] FIG. 5 depicts an example of training a machine learning model, according to one or more examples.

[0019] FIG. 6 depicts an example of a computing device, according to one or more examples.

[0020] Notably, for simplicity and clarity of illustration, certain aspects of the figures depict the general structure and / or manner of construction of the various examples. Descriptions and details of well-known features and techniques may be omitted to avoid unnecessarily obscuring other features. Elements in the figures are not necessarily drawn to scale; the dimensions of some features may be exaggerated relative to other elements to improve understanding of the examples. For example, one of ordinary skill in the art appreciates that the side views are not drawn to scale and should not be viewed as representing proportional relationships between different components. The side views are provided to help illustrate the various components of the depicted assembly, and to show their relative positioning to one another.DETAILED DESCRIPTION

[0021] Various examples of the present disclosure relate generally to methods and systems for cardiac pacing rate programming.

[0022] As will be discussed in more detail below, in various examples, systems and methods are described for using machine learning to determine cardiac pacing rate programming. By training a machine-learning model, e.g., via supervised or semi-supervised learning, to learn associations between training data (e.g., physiological data) and ground truth data (e.g., cardiac pacing programs), the trained machine-learning model may be usable to output a pacing program for a patient.

[0023] For example, a closed loop system that is driven by one or more physiologic endpoints such as systolic blood pressure (“SBP”) may be used to change one or more types of pacing rate (e.g., atrial, ventricular, biventricular, or some other combination) and / or deliver a subthreshold pacing stimulus that isn’t strong enough to cause the heart to beat but nevertheless stimulates the autonomic nervous system via neuromodulation. Blood pressure regulated atrial pacing (“BRT”) may result in a narrowing of pulse pressure, which may be beneficial in known causes heart failure, such as preserved ejection fraction (“HFpEF”) and heart failure with reduced ejection fraction (“HfirEF”) .

[0024] Blood pressure may be measured from some source (e.g., a cuff, a wearable, an invasive method such as an arterial line) and sent (e.g., wirelessly) to a computing device (e.g., a tablet, computer, smartwatch). The machine learning models may be used to compute a best pacing rate of some type (atrial, ventricular, etc.), which would then be sent (e.g. wirelessly) to a pacemaker. Pulse pressure (“PP”) may be post processed into a derivative of pulse pressure with respect to time and sent with the blood pressure to the computing device, such that it may be either the simplest derivation of SBP-DBP or the change in PP for one or more intervals such as P1-P2, or Pl-Pn. It may be a percentage change, a change in slope, or any other indicator of change.

[0025] Reference will now be made in detail to examples of the present disclosure. The present disclosure is not limited to any single aspect or example thereof, nor is it limited to any combinations and / or permutations of such aspects and / or examples. Moreover, each of the aspects of the present disclosure, and / or examples thereof, may be employed alone or in combination with one or more of the other aspects of the present disclosure and / or examples thereof. For the sake of brevity, certain permutations and combinations are not discussed and / or illustrated separately herein.

[0026] Notably, an example or implementation described herein as “exemplary” is not to be construed as preferred or advantageous, for example, over other examples or implementations. The term “exemplary” is used in the sense of “example” rather than “ideal.”

[0027] Wherever possible, the same reference numbers will be used throughout the drawings to refer to the same or like parts. In the discussion that follows, relative terms such as “about,” “substantially,” “approximately,” etc. are used to indicate a possible variation of ±10% in a stated numeric value.

[0028] In this disclosure, the term “based on” means “based at least in part on.” The singular forms “a,” “an,” and “the” include plural referents unless the context dictates otherwise.

[0029] As used herein, the terms “comprises,” “comprising,” “includes,” “including,” or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements, but may include other elements not expressly listed or inherent to such process, method, article, or apparatus.

[0030] The term “or” is used disjunctively, such that “at least one of A or B” includes, (A), (B), (A and A), (A and B), etc. Relative terms, such as, “substantially,” “approximately,” and “generally,” are used to indicate a possible variation of ±10% of a stated or understood value.

[0031] It will also be understood that, although the terms first, second, third, etc. are, in some instances, used herein to describe various elements, these elements should not be limited by these terms. These terms are only used to distinguish one element from another. For example, a first contact could be termed a second contact, and, similarly, a second contact could be termed a first contact, without departing from the scope of the various described examples. The first contact and the second contact are both contacts, but they are not the same contact. In addition, the terms “first,” “second,” and the like, herein do not denote any order, quantity, or importance, but rather are used to distinguish an element or a structure from another. Moreover, the terms “a” and “an” herein do not denote a limitation of quantity, but rather denote the presence of one or more of the referenced items.

[0032] As used herein, the term “if’ is, optionally, construed to mean “when” or “upon” or “in response to determining” or “in response to detecting,” depending on the context. Similarly, the phrase “if it is determined” or “if [a stated condition or event] is detected” is, optionally, construed to mean “upon determining” or “in response to determining” or “upon detecting [the stated condition or event]” or “in response to detecting [the stated condition or event],” depending on the context.

[0033] Terms like “provider,” “medical provider,” or the like generally encompass an entity, person, or organization that may seek information, resolution of an issue, or engage in any other type of interaction with a user, e.g., to provide medical care, medical intervention or advice, or the like. Terms like “user,” “patient,” or the like generally encompass any person (e.g., an individual, a medical provider, etc.) or entity who is using a device, programming a device, obtaining information, seeking resolution of an issue, or the like.

[0034] Terms like “pacemaker” or the like generally encompass a device that may help control a user’s heartbeat, for example, to prevent the user’s heart from beating too fast or too slow. A pacemaker may include one or more sensors and / or one or more defibrillators, which may determine heart rate and provide electrical impulses, respectively. Terms like “pacing rate” or the like generally encompass the electrical pulses generated by the pacemaker and provided to one or more chambers of the heart to maintain an adequate heart rate. Terms like “base rate” or the like generally encompass the lowest heart rate allowed by a pacemaker. Terms like “lookup table” or the like generally encompass data that may determine the pacing rate and / or the base rate. A lookup table may be stored, e.g., in a database, in the form of a data table.

[0035] According to implementations of the disclosed subject matter, changes to a pacing rate and / or a base rate may be based on a lookup table. A lookup table may be a table with the patient attributes (e.g., height, weight, gender, medical condition, physiologic input, current state, etc.), objective inputs, subjective inputs, and / or the like and may be referenced to determine one or more pacing rates. Such a lookup table may be accessed by a pacemaker or a processing device in communication with a pacemaker.

[0036] According to implementations of the disclosed subject matter, cardiac pacing may be determined based on physiological inputs such as, but not limited to, blood pressure and / or heart rate, as further discussed herein. Such physiological input based cardiac pacing may be used to treat conditions such as, but not limited to, drug resistant hypertension (DRH), DRH with diastolic congestive heart failure (“DCHF”), HFpEF, HfirEF, etc.

[0037] Blood pressure may be detected using a blood pressure measuring device (a “device” or a “blood pressure device”). A blood pressure may be a sensed value, a blood pressure, a sensed value converted into one or more other formats (e.g., by a processor), or the like. A blood pressure may indicate how much pressure a user’s blood exerts against the user’s artery walls when the user’s heart beats (e.g., a systolic blood pressure). A blood pressure may indicate how much pressure a user’s blood exerts against the user’s artery walls when the user’s heart is resting between beats (e.g., diastolic blood pressure).

[0038] According to implementations of the disclosed subject matter, blood pressure may refer to systolic and diastolic blood pressure. In addition, blood pressure may include or may be used to determine pulse pressure (PP) which may be the difference between systolic and diastolic blood pressure measured in millimeters of mercury (mmHg) or any other applicable unit, differential, ratio, and / or thelike. Pulse pressure (PP) may represent the force that the heart generates each time it contracts. For reference purposes, a healthy pulse pressure (PP) may be 40 mmHg which may be the difference between systolic and diastolic blood pressure.

[0039] A widened pulse pressure (PP) may be associated with poor outcomes in many forms of heart failure, and particularly in HFpEF. As will be discussed in greater detail below, a widened pulse pressure (PP) may be managed by advancing the reduction of diastolic relaxation properties of the left ventricle as the disease progresses.

[0040] A blood pressure measuring device may include any type of blood pressure monitor or cuff such as, for example, a pneumatic cuff relying on mechanical compression of a peripheral artery cuff (e.g., to be attached to brachial artery, ankle, wrist, etc.), a non-pneumatic cuff (e.g., which analyzes an arterial waveform and function anywhere on the body where the arterial pulse contour can be sensed such as at a wrist), or an implantable sensor within a blood vessel or heart chamber. A blood pressure measuring device may be a light-based device such as a photoplethysmography (PPG) device. A blood pressure measuring device may output blood pressure in a first format which may be converted to a second format such that a processing component receiving blood pressure information may be configured to utilize such information in the second format and may not be configured to utilize such information in the first format.

[0041] Physiological inputs, as discussed herein, include, but are not limited to, a pulse pressure, a blood-pressure, heart rate, biomarker level (e.g., cortisol, atrial natriuretic peptide (ANP), B-type natriuretic peptide (BNP), N-terminal pro b-type natriuretic peptide (NT-proBNP), etc.), blood oxygen level, glucose level, blood electrolytes level, an accelerometer value, respiratory rate sensor value (e.g., via diaphragmatic movement), thoracic impedance, impedance (e.g., as a correlate of right ventricular function), environmental parameter, ambient oxygen concentration (e.g., SP02), humidity, portions of cardiac rate such as atrial rate, ventricular rate, atrioventricular conduction, the presence of rhythm irregularities, autonomic nervous system (ANS) function, glucose, skin electrolytes, galvanic skin response, PPG values, Electroencephalogram (EEG) wave, urination parameters, etc. Such physiological inputs may be provided by one or more sensors, devices, or the like. Systolic blood pressure (SBP) and diastolic blood pressure (DBP) may be sensed by one or more blood pressure sensing devices.

[0042] According to implementations of the disclosed subject matter, cardiac pacing may be determined based on environmental parameters. Such environmental parameters include one or more of the patient’s diet, a time of day, an ambient temperature, the patient’s location, an ambient oxygen concentration, and / or a humidity.

[0043] Additionally, cardiac pacing may be determined based on subjective input from the patient. This subjective input can include emotional parameters, such as the patient’s (or a patient’s provider’s) reporting of a state of emotional well-being, physical well-being, comfort level, etc. Alternatively,cardiac pacing may be determined by a medical provider or by a machine learning algorithm trained to begin, maintain, modify, and / or end cardiac pacing. In instances where cardiac pacing is determined by a machine learning algorithm, a medical provider or other professional may manually indicate the use of the machine learning algorithm for pacing, or the pacing by the machine learning algorithm may be initiated, performed, or terminated automatically.

[0044] As discussed herein, cardiac pacing may be based on physiological parameters, environmental parameters, and / or subjective / emotional parameters. It will be understood that such parameters include changes to such parameters. For example, cardiac pacing may be based one or more of a change such as a change in clinical status, a change in medication, a change in other physiologic parameters, a change in other diagnostic testing such as in vitro diagnostics (e.g., blood tests and the like), changes based on procedures during and / or after surgery, endoscopy, cardiac ablation, renal denervation, etc.

[0045] As used herein, a “machine-learning model” generally encompasses instructions, data, or a model configured to receive input, and apply one or more of a weight, bias, classification, or analysis on the input to generate an output. Accordingly, techniques disclosed herein may be implemented to determine or modify pacing (e.g., by a pacing device such as a pacemaker) based on physiological, environmental, and / or subjective inputs. A pacing output may be modified in accordance with an algorithm or machine learning output. The output may include, for example, a classification of the input, an analysis based on the input, a design, process, prediction, or recommendation associated with the input, or any other suitable type of output. A machine-learning model is generally trained using training data, e.g., experiential data or samples of input data, which are fed into the model in order to establish, tune, or modify one or more aspects of the model, e.g., the weights, biases, criteria for forming classifications or clusters, or the like. Aspects of a machine-learning model may operate on an input linearly, in parallel, via a network (e.g., a neural network), or via any suitable configuration. By virtue of such training, a machine-learning model is converted from an un-trained and un-specific model to a model that is unique to and specifically configured for the particular purpose for which it is trained. In an example, training of a machine-learning model is analogous to a method of production in which the article produced is the trained model having unique characteristics by virtue of its particular training. Moreover, the result of training a machine-learning model using particular training data and for a particular purpose results in a technical solution to an inherently technical problem.

[0046] The execution of the machine-learning model may include deployment of one or more machine learning techniques, such as linear regression, logistical regression, random forest, gradient boosted machine (GBM), deep learning, or a deep neural network. Supervised or unsupervised training may be employed. For example, supervised learning may include providing training data and labels corresponding to the training data, e.g., as ground truth. Unsupervised approaches may include clustering, classification or the like. K-means clustering or K-Nearest Neighbors may also be used, which may be supervised or unsupervised. Combinations of K-Nearest Neighbors and an unsupervisedcluster technique may also be used. Any suitable type of training may be used, e.g., stochastic, gradient boosted, random seeded, recursive, epoch or batch-based, etc.

[0047] For example, pacing may be modified to improve a blood pressure related condition by increasing or decreasing blood pressure, via determined cardiac pacing outputs, based on observed biomarker levels. The modification may result in an increase in a cardiac pacing rate or amplitude, a decrease in a cardiac pacing rate or amplitude, an acceleration of a cardiac pacing rate, a deceleration of a cardiac pacing rate, and / or the like. Such modified pacing may, at least in part, improve a given medical condition for a patient (e.g., a blood pressure condition). Such conditions may include, for example, hypertension, hypotension, DRH, DRH with diastolic congestive heart failure (DCHF), HFpEF, and / or the like.

[0048] Presented below are various aspects of machine learning techniques that may be adapted to determine cardiac pacing instructions. As will be discussed in more detail below, machine learning techniques may be adapted to output a cardiac pacing rate or program, may include one or more aspects according to this disclosure, e.g., a particular selection of training data, a particular training process for the machine-learning model, operation of a particular device suitable for use with the trained machinelearning model, operation of the machine-learning model in conjunction with particular data, modification of such particular data by the machine-learning model, etc., or other aspects that may be apparent to one of ordinary skill in the art based on this disclosure.

[0049] FIG. 1 shows an exemplary environment 100 for a closed- loop system for determining an adjusted physiologic pacing rate, according to one or more examples. In some examples, the systems and devices of the environment 100 may communicate in any arrangement. As will be discussed herein, systems or devices of the environment 100 may communicate in order to one or more of generate, train, or use a machine-learning model to determine a physiologic pacing rate, among other activities.

[0050] Such a pacing rate may be determined, for example, upon detection of a medical condition such as Hypertensive Heart Disease and / or may be determined independent of a medical condition (e.g., based on one or more physiological inputs, subjective inputs, environmental inputs, and / or the like). Physiological pacing rates may include any applicable properties for cardiac pacing such as, but not limited to, pulse pressure (PP), frequency of pacing, amplitude of pacing, duration of pacing, acceleration of pacing, deceleration of pacing, etc.

[0051] In an exemplary example, a pacing rate to pace an atrium of a heart may utilize the input of pulse pressure to regulate the programmed right atrial pacing rate of a dual chamber pacemaker. In addition, a pacing rate may also incorporate other physiologic inputs such as diastolic blood pressure, mean arterial blood pressure, and non-blood pressure associated variables such as pulse oximetry, body temperature, medication type, etc. alone or in any combination.

[0052] Environment 100 of FIG. 1 depicts at least one physiological input measuring device 110, a component 120 to receive the input (e.g., an electronic device, a patient device, a network device, acloud device, etc.), and pacing system 130. Some or all of these components may be connected via a network 140.

[0053] Physiological input measuring device 110 may include one or more sensors such as, but not limited to, a blood-pressure sensors, heart rate sensors, biomarker level sensors, blood oxygen level sensors, glucose level sensors, blood electrolytes level sensors, an accelerometer, motion sensors, position sensors, respiratory rate sensors, impedance sensors, environmental sensors, ambient oxygen condition sensors, humidity sensors, ANS sensors, glucose sensors, skin electrolytes sensors, galvanic skin sensors, PPG sensors, EEG sensors, EKG sensors, fluid sensors, volume sensors, light sensors, cameras, and / or the like. For example, a physiological input measuring device 110 may be used to determine a pulse pressure for a given patient. As further discussed herein, a pacing rate (e.g., an atrial pacing rate) may be determined based on the pulse pressure. The pacing rate may be determined such that the pacing rate is used to pace a patient’s heart such that the pulse pressure reaches a target pulse pressure. Accordingly, one or more pulse pressures may be used as inputs (e.g., to a machine learning model) and corresponding outputs (e.g., from the machine learning model) may one or more pulse rates that cause a current pulse pressure to reach a target pulse pressure over a period of time.

[0054] It will be understood that the process discussed herein may be implemented iteratively such that a current pulse pressure is used to determine first pacing rate that is predicted to cause current pulse pressure to reach a target pulse pressure as a result of implementing the first pacing rate using a pacing device (e.g., a pacemaker). The first pacing rate may be provided to a pacing device which outputs the pacing rate using one or more electrical signals. An updated pulse pressure may be received (e.g., using physiological input measuring device 110) after a period of time post implementing the first pacing rate. Based on the updated pulse pressure, a second pacing rate may be determined such that the second pacing rate is predicted to cause the updated pulse pressure to reach the target pulse pressure as a result of implementing the second pacing rate using a pacing device (e.g., a pacemaker). Such an iterative process may be performed one or a plurality of times (e.g., until the target pulse pressure is reached or a pulse pressure within a threshold value of the target pulse pressure is reached).

[0055] The machine learning model may be trained using supervised, unsupervised, or semi-supervised learning based on training data that includes historical or simulated pulse pressures, historical or simulated pacing rates, historical or simulated changes in pulse pressure based on changes in pacing rates, and / or the like. The training data may be based on historical or simulated patients and / or based on a cohort of historical or simulated patients. The cohort of historical or simulated patients may be selected by a training machine learning model based on patient data such that the training machine learning model receives a superset of potential patients and their respective patient properties (e.g., demographic information, medical information, medical history, medication information, lifestyle information, etc.). The training machine learning model may output a subset of patients that overlap or correlate with the patient for whom the pacing rate is to be determined. The training machine learning model may outputthe subset of patients and the training data to train the machine learning model may be determined to include only historical or simulated data for this subset (or cohort) of patients.

[0056] One or more of the components of environment 100 of FIG. 1 may communicate with each other and / or other systems, e.g., via network 140. In some examples, network 140 may connect one or more components of environment 100 via a wired connection. In some examples, network 140 may connect one or more aspects of environment 100 via an electronic network connection, for example a wide area network (WAN), a local area network (LAN), personal area network (PAN), or the like. In some examples, the electronic network connection includes the internet, and information and data provided between various systems occurs online. “Online” may mean connecting to or accessing source data or information from a location remote from other devices or networks coupled to the Internet. Alternatively, “online” may refer to connecting or accessing an electronic network (wired or wireless) via a mobile communications network or device. The Internet is a worldwide system of computer networks — a network of networks in which a party at one computer or other device connected to the network may obtain information from any other computer and communicate with parties of other computers or devices. The most widely used part of the Internet is the World Wide Web (often- abbreviated “WWW” or called “the Web”). A “website page,” a “portal,” or the like generally encompasses a location, data store, or the like that is, for example, hosted and / or operated by a computer system so as to be accessible online, and that may include data configured to cause a program such as a web browser to perform operations such as send, receive, or process data, generate a visual display and / or an interactive interface, or the like. In any case, the connections within the environment 100 may be network, wired, any other suitable connection, or any combination thereof.

[0057] In an example, pacing system 130 may be used to generate or train a machine-learning model. For example, such a system may include instructions for generating the machine-learning model, the training data and ground truth, or instructions for training the machine-learning model. A resulting trained-machine-leaming model may then be provided to the pacing system 130.

[0058] Generally, a machine -learning model includes a set of variables, e.g., nodes, neurons, filters, etc., that are tuned, e.g., weighted or biased, to different values via the application of training data. In supervised learning, e.g., where a ground truth is known for the training data provided, training may proceed by feeding a sample of training data into a model with variables set at initialized values, e.g., at random, based on Gaussian noise, a pre-trained model, or the like. The output may be compared with the ground truth to determine an error, which may then be back-propagated through the model to adjust the values of the variable. In unsupervised learning, patterns, correlations, or clusters of input samples may be used to determine one or more metrics or features of the samples usable to differentiate between related subsets of the samples. In semi-supervised learning, unsupervised and supervised approaches may be combined.

[0059] Training may be conducted in any suitable manner, e.g., in batches, and may include any suitable training methodology, e.g., stochastic or non-stochastic gradient descent, gradient boosting, random forest, etc. In some examples, a portion of the training data may be withheld during training or used to validate the trained machine-learning model, e.g., compare the output of the trained model with the ground truth for that portion of the training data to evaluate an accuracy of the trained model. The training of the machine-learning model may be configured to cause the machine-learning model to learn associations between [training] data and [ground truth] data, such that the trained machine-learning model is configured to determine an output [claim] in response to the input [claim] data based on the learned associations. Particular selection or application of training data, such as discussed in various examples of this disclosure, may inhibit or reduce impact of concerns such as biasing (e.g., via selection, truncation, or the like), overfitting, under-fitting, etc.

[0060] In some instances, training using one set or type of data may be used or adapted to another set of data. For example, a modal initially trained on one data set may require less samples or time to train on a second data set. In another example, initial training may result in a base model that may be tuned with an additional data set so as to form a particularized model specific to circumstances of the additional data set.

[0061] In various examples, the variables of a machine-learning model may be interrelated in any suitable arrangement in order to generate the output. For example, the machine-learning model may include one or more convolutional neural network (“CNN”) configured to identify features in the physiological data, and may include further architecture, e.g., a connected layer, neural network, etc., configured to determine a relationship between the identified features in order to determine a location in the physiological data.

[0062] In some instances, different samples of training data or input data may not be independent. For example, training data from different testing, trials, programs, and the like may include samples of training data obtained by different entities and for different purposes. Thus, in some examples, the machine-learning model may be configured to account for or determine relationships between multiple samples.

[0063] For example, in some examples, the machine-learning model of the pacing system 130 may include a Recurrent Neural Network (“RNN”). Generally, RNNs are a class of feed-forward neural networks that may be well adapted to processing a sequence of inputs. In some examples, the machinelearning model may include a Long Shor Term Memory (“LSTM”) model or Sequence to Sequence (“Seq2Seq”) model. An LSTM model may be configured to generate an output from a sample that takes at least some previous samples or outputs into account. A Seq2Seq model may be configured to, for example, receive a sequence of non-optical in vivo images as input, and generate a sequence of locations, e.g., a path, in the medical imaging data as output.

[0064] In a reinforcement learning model, environmental data (e.g., data describing a current state of a system) is evaluated using a policy in order to determine a next action of an agent. A scoring function or metric is usable to objectively quantify a state of the environment, e.g., to evaluate whether the action of the agent was desirable or not. The policy may include, for example, one or more tunable metrics or any suitable machine learning architecture such as a neural network in which the output nodes correspond to possible actions of the agent. In some instances, the policy includes a recurrent network structure or the like that obtains or retains data on previous actions of the agent or states of the environment. Training the reinforcement model may include, for example, a random forest of policy perturbations (e.g., to control for a final score outcome), weighting of policy parameters via back propagation based on a score for a state of the environment after performance of a particular action, etc.

[0065] In an adversarial network, a network may be trained using the output of a different model, rather than or in addition to using ground truth. For example, output of a mathematical (formula-based) model or a different machine learning model may be compared with output from a model to be trained. In an example, the difference in the results may be back propagated. In another example, the comparison may include a scoring, and instances in which the model to be trained outperformed the adversary may cause reinforcement of the model.

[0066] Any suitable type of machine learning model or combination of machine learning models may be used. Operations conducted by one model in some examples may be distributed amongst a plurality of models in other examples, or vice versa.

[0067] Although depicted as separate components in FIG. 1, it should be understood that a component or portion of a component in the environment 100 may, in some examples, be integrated with or incorporated into one or more other components. For example, a portion of the physiological input measuring device 110 may be integrated into the component 120 or the like. In another example, the pacing system 130 may be integrated with a data storage system (not shown). In some examples, operations or aspects of one or more of the components discussed above may be distributed amongst one or more other components. Any suitable arrangement or integration of the various systems and devices of the environment 100 may be used.

[0068] According to implementations of the disclosed subject matter, one or more systems or methods disclosed herein may be utilized for pacing rate programming. FIG. 2A depicts a flow diagram 200 determine (e.g., program) one or more cardiac pacing rates. It will be understood that the steps described in reference to FIG. 2A are an example only. Steps shown in FIG. 2A may be performed in the order described herein or in any applicable order. Further, for simplicity, flow diagram 200 is provided in reference to blood pressure data. However, it will be understood that any physiologic input (e.g., as discussed herein) may be substituted for blood pressure in accordance with the techniques discussed in reference to flow diagram 200.

[0069] As shown in FIG. 2A, the process of flow diagram 200 may start at step 202. Physiologic property data (e.g., pulse pressure data) of a subject may be received at step 204. For example, pulse pressure data may be received via physiological input measuring device 110 based on one or more sensor inputs detected at physiological input measuring device 110. Physiological property data, at step 204, may be received at, for example, component 120 via network 140. As discussed herein, physiological property data received at step 204 may be determined in a first format associated with physiological input measuring device 110 and may be converted into a second format to be utilized by component 120. Physiological property data received at step 204 may be converted from the first format to the second format via physiological input measuring device 110 and / or via component 120. According to an implementation, component 120 may be part of, associated with, or in communication with pacing system 130.

[0070] At step 206, subject information of a given subject may be received. Such subject information may include, for example, patient attributes (e.g., demographic attributes, height, weight, ethnicity, medical conditions, medication information, etc.). Subject information may be provided via physiological input measuring device 110 (which may be the same as or different than the physiological input measuring device that provides pulse pressure information at step 204). Alternatively, or in addition, subject information may be provided via component 120 (e.g., via user input or storage associated with component 120) and / or may be provided via a separate component (e.g., a remote component, database, server, electronic medical record program, etc.).

[0071] At step 212, a determination may be made regarding whether a physiologic property (e.g., PP) associated with the input received at step 204 meets or exceeds a threshold physiologic value (e.g., if PP is greater than approximately 60 mmHg). The threshold physiologic value may be a lower bound of an unacceptable physiologic value range (e.g., an unacceptable PP range). If the physiologic property does not meet or exceed the threshold physiologic value, then step 210 may be performed. At step 210, a determination may be made whether the physiologic property is within a second threshold physiologic value (e.g., if PP is equal to or less than 40 mmHg). The second threshold physiologic value may be an upper bound of an acceptable physiologic value range (e.g., an acceptable PP range). If the physiologic property is below the second threshold physiologic value, then step 208 may be performed and the process of flow diagram 200 may be terminated. Accordingly, step 208 may be performed to terminate the process if the physiologic property (e.g., pulse pressure) received at step 204 is within an acceptable physiologic value range. For example, the physiologic property data being in such an acceptable range may not require an adjusted pacing rate.

[0072] Returning to step 210, if the physiologic property is above the second threshold physiologic value (e.g., higher than the acceptable physiologic value range), step 214 may be performed. For example, as shown in flow diagram 200, if pulse pressure is greater than approximately 40 mmHg and less than approximately 60 mmHg, an existing pacing rate may be adjusted to be a predetermined ordynamically determined amount (e.g., approximately 5% of a patient heart rate or physiologic property such as pulse pressure) lower. The pacing rate may be adjusted to be lowered by the amount in comparison to the existing pacing rate and / or may be adjusted to be lowered in comparison to a heart rate (e.g., a heart rate detected by an ECG sensor).

[0073] According to an example, the pacing rate may be adjusted based on an output of a pacing machine learning model. The pacing machine learning model may be trained in accordance with the techniques disclosed herein. For example, the pacing machine learning model may be trained to adjust one or more weights, layers, biases, synapses, nodes, and / or the like based on training data that may include one or more of historical or simulated physiologic input values, subject information, acceptable physiologic value ranges, pacing rates, pacing rate changes, subjective inputs, and / or the like. The pacing machine learning model may receive, as inputs, one or more of the physiologic input value received at step 204, the subject information received at step 206, the acceptable physiologic value range applied at step 212 and step 210 (e.g. , as output by a threshold machine learning model), and / or the like. The pacing machine learning model may output a pacing rate (e.g., an actual rate, a percentage change, a ratio, etc.) based on the inputs.

[0074] Next, at step 218A, a subjective input may be received from the subject. For example, a patient may be provided a prompt (e.g., via a graphical user interface) requesting the patient to provide an assessment of his or her own wellbeing. According to an implementation of the disclosed subject matter, the subject may be provided a prompt, via a graphical interface, comprising graphical components generated based on one or more of the physiologic property received or determined based on the input received at step 204, based on the amount of deviation of the physiological property from the second threshold physiologic value, and / or the like. For example, the prompt may include larger icons for receiving the subjective input if the amount of deviation of the physiological property from the second threshold physiologic value is above a threshold amount. An order of the requested input may be determined based on the one or more of the physiologic property received or determined based on the input received at step 204, based on the amount of deviation of the physiological property from the second threshold physiologic value, and / or the like. For example, a graphical component corresponding to the subject feeling ill may be ordered above a graphical component corresponding to the subject feeling healthy if the amount of deviation of the physiological property from the second threshold physiologic value is above a threshold amount.

[0075] If, at step 218A, a positive indication is received indicating that the subject is feeling the same or better than prior to the pacing rate adjustment at step 214, a waiting period of a predetermined or dynamically determined time period (e.g., 3 minutes) may be implemented at step 220A. Upon expiration of the time period, another subjective input may be received from the subject at step 218B. According to implementations of the disclosed subject matter, a dynamically determined time period may be determined using an algorithm or time-based machine learning model. The time-based machinelearning model may be trained in accordance with the techniques disclosed herein. For example, the time-based machine learning model may be trained to adjust one or more weights, layers, biases, synapses, nodes, and / or the like based on training data that may include one or more of historical or simulated physiologic input values, subject information, acceptable physiologic value ranges, and / or the like. The time-based machine learning model may receive, as inputs, one or more of the physiologic input value received at step 204, the subject information received at step 206, the acceptable physiologic value range applied at step 212 and step 210, and / or the like. The time-based machine learning model may output a dynamically determined time period such that, for example, the dynamically determined time period provides sufficient time for the subject to notice an effect of the pacing rate change implemented at step 214. Further, the output dynamically determined time period may not exceed a time period such that the likelihood of the subject being effected by external factors other than the pacing rate change is below an acceptable threshold likelihood. Accordingly, the time-based machine learning model may, at least in part, determine a likelihood of the subject being effected by external factors.

[0076] If, at step 218B, a positive indication is received indicating that the subject is feeling better or the same as prior to step 220A, the process may be considered successful and may terminate at step 224. It will be understood that indications (e.g., via subjective inputs received at steps 218A, 218B, 218C, 218D, 218E, and / or 218F) may be provided via an interface in a manner similar to that discussed in reference to step 218A.

[0077] If at step 218 A, a negative indication is received indicating that the subject is feeling worse, then, at step 226A, the pacing rate will revert to the same pacing rate at step 202, and the appointment will end 208A. Alternatively, at step 226A, the pacing rate may be adjusted to an intermediate rate (e.g., a rate higher than the pacing rate at step 202 but lower than the pacing rate at step 222).

[0078] Similarly, if at step 218B a negative indication is received indicating that the subject is feeling worse, then the pacing rate will revert to the same pacing rate at step 202, and the appointment will end 208A. Alternatively, at step 226B, the pacing rate may be adjusted to an intermediate rate (e.g., a rate higher than the pacing rate at step 202 but lower than the pacing rate at step 222

[0079] Still referring to FIG. 2A, if the physiological property value (e.g., pulse pressure) at step 212 is greater than the threshold amount (e.g., approximately 60 mmHg), the pacing rate may be adjusted to be raised by a predetermined or dynamically determined amount (e.g., 10%) at step 216. The pacing rate may be adjusted by the amount in comparison to the existing pacing rate and / or may be adjusted to be lowered in comparison to a heart rate (e.g., a heart rate detected by an ECG sensor).

[0080] Next, at step 218C, a subjective input may be received from the subject. If, at step 218C, a positive indication is received indicating that the subject is feeling the same or better than prior to the pacing rate adjustment at step 216, a waiting period of a predetermined or dynamically determined time period (e.g., approximately 3 minutes) may be implemented at step 220B. Upon expiration of the time period, another subjective input may be received from the subject at step 218D.

[0081] If, at step 218D, a positive indication is received indicating that the subject is feeling better or the same as prior to step 220B, the process may be considered successful and may terminate at step 224. It will be understood that indications (e.g., via subjective inputs received at steps 218A, 218B, 218C, 218D, 218E, and / or 218F) may be provided via an interface in a manner similar to that discussed in reference to step 218A.

[0082] If at step 218C, a negative indication is received indicating that the subject is feeling worse, then, the pacing rate may be adjusted to be raised by a predetermined or dynamically determined amount (e.g., 5%) at step 222, as discussed herein. Similarly, if at step 218D a negative indication is received indicating that the subject is feeling worse, then the pacing rate may be adjusted to be raised by a predetermined or dynamically determined amount (e.g., 5%) at step 222, as discussed herein.

[0083] Next, at step 218E, a subjective input may be received from the subject. If, at step 218E, a positive indication is received indicating that the subject is feeling the same or better than prior to the pacing rate adjustment at step 222, a waiting period of a predetermined or dynamically determined time period (e.g., approximately 3 minutes) may be implemented at step 220C. Upon expiration of the time period, another subjective input may be received from the subject at step 218F. If, at step 218F, a positive indication is received indicating that the subject is feeling better or the same as prior to step 222, the process may be considered successful and may terminate at step 224.

[0084] If at step 218E, a negative indication is received indicating that the subject is feeling worse, then, at step 226B, the pacing rate will revert to the same pacing rate at step 202, and the process may terminate at step 208B. Alternatively, at step 226B, the pacing rate may be adjusted to be higher than the pacing rate at step 202 but lower than the pacing rate at step 222. Similarly, if at step 218F a negative indication is received indicating that the subject is feeling worse, then the pacing rate will revert to the same pacing rate at step 202, and the process may terminate at step 208B. Alternatively, at step 226B, the pacing rate may be adjusted to be higher than the pacing rate at step 202 but lower than the pacing rate at step 222.

[0085] As discussed herein, a dose-response relationship to modifying a pacing rate (e.g., an atrial pacing rate) is provided herein. Accordingly, a pacing rate output to a pacing system may be based on a current or prior pacing rate. The output pacing rate may be a percentage or ratio of the current or prior pacing rate. The percentage or ratio may be predetermined (e.g., approximately 5%, approximately 10%, etc.) and / or may be determined based on or more factors discussed herein (e.g., via an algorithm or a pacing machine learning algorithm). Accordingly, for a first patient, the output pacing rate may be a percentage or ratio that is different for a second patient. As another example, for a first patient, the output pacing rate may be a percentage or ratio at a first time that is different than the output pacing rate for the first patient at a second time (e.g., based on a change in the patient information, physiological factors, etc.) In some examples, the adjustment of the base pacing rate or the atrial pacing rate may be adjusted based on a patient’s tolerance and / or may not exceed the pacing maximum allowed by physiologic orsafety guardrails (e.g., approximately 88 bpm). Such physiologic or safety guardrails may be determined based on patient information, device characteristics, safety regulations, and / or the like.

[0086] According to examples of the disclosed subject matter, the threshold and / or the second threshold (e.g., as described in reference to step 212 and / or step 210) may be output by a threshold machine learning model. The threshold machine learning model may be trained in accordance with the techniques disclosed herein. For example, the threshold machine learning model may be trained to adjust one or more weights, layers, biases, synapses, nodes, and / or the like based on training data that may include one or more of historical or simulated physiologic input values, subject information, acceptable physiologic value ranges, pacing rates, pacing rate changes, changes in physiological values, and / or the like. The threshold machine learning model may receive, as inputs, one or more of the physiologic input value received at step 204, the subject information received at step 206, and / or the like. The threshold machine learning model may output the threshold and / or second threshold, either of which may be patient specific. According to an example, an output of the threshold machine learning model may be provided as an input into the time-based machine learning model which may provide an output, as discussed herein, at least in part of the threshold machine learning model output.

[0087] According to an example of the disclosed subject matter, a complex lookup table may be accessed to determine a pacing rate. A primary lookup variable for the complex lookup table may be one or more physiological parameter values (e.g., blood pressure values or pulse pressure values) which may include reference systolic / diastolic numbers associated with respective pulse pressure readings. Each pair of systolic / diastolic numbers may correspond to an applicable pacing rate. Hence, in response to an input physiological parameter value (e.g., pulse pressure reading), the device and / or algorithm implementing the change of pacing rate may perform a look up of the new pacing rate to be programmed. An implementation of the complex lookup table (lookup table) may include a function that retrieves data from a pre-programmed and / or pre-populated array of data. For example, such pre-programmed and / or pre-populated array of data may be populated using the threshold machine learning model discussed herein. According to this example, the threshold machine learning model may be provided physiological parameter values as an input, and may generate corresponding pacing rates to populate the lookup table. For example, pulse pressure (PP) may be looked up and a pacing rate may be determined using a device, based on a populated lookup table.

[0088] In some examples, the lookup table may be populated based on and / or may include a patient’s height, weight, demographics, disease quantification (e.g. biomarkers such as one or more QT interval and / or one or more QTc interval), etc. The lookup table can be generated, for example, either using an empirical series of measurements in the clinic and / or a series of past records of the patient. Accordingly, a lookup table may be generated, for example, empirically or based on past historical information.

[0089] In some examples, the lookup table may not be static. In accordance with the current disclosure, a feedback system may be maintained wherein, for each new pacing rate that is programmed, theresultant physiological parameter values may be measured and / or monitored. If the desired control is achieved, changes to the look up table may not be needed. If the desired control is not achieved, a new target pacing rate may be tried / tested (e.g., in accordance with flow diagram 200 of FIG. 2A). If this rate results in the desired control, this experience may be recorded. After a certain number of such experiences, which may be a parameterized number, the lookup table entry may be modified to reflect the experiences. As such, a lookup table may be developed for each patient. Accordingly, a lookup table may be personalized (e.g., over time) based on feedback and learning, as discussed herein.

[0090] In certain clinical situations, it may be necessary to review modifications to the lookup table, even if such changes are algorithmically implemented autonomously. Hence, implementations of the current disclosure may include a built-in alert and / or communication mechanism that may be configured to transmit the modifications to a provider and may affect the changes, once approved. A user (e.g., a physician) or automated system may select, for example, either a review-gated mode or autonomous mode for the lookup table modifications or may set a threshold beyond which a change needs to be reviewed. Accordingly, a review (e.g., a manual review) may be implemented and may be triggered based on one or more thresholds.

[0091] At a macro level, correlating the personalized tables over similar patients (e.g., similar in height, weight, demographics, disease conditions, etc.) may yield improved starting points for the lookup table of each patient and / or also lead to the modifications of the lookup tables across a cohort. Such cross communication and group-evolution of lookup tables may be accomplished across multiple patient devices through a set of cloud services, anonymously. Such cross communication may be done at any desired or determined periodic frequency. Accordingly, cohort level data may be improved by aggregating personal tables and / or personal tables may be improved using cohort level data.

[0092] In some instances, modifications to a lookup table may inadvertently result in the patient feeling worse. In this case, a previous lookup table may be restored. Accordingly, previous versions of lookup tables may be stored and any may be restored, e.g., as a roll back feature. Accordingly, previous versions of a lookup table may be restored or rolled back.

[0093] In some instances, the lookup table may be a substitute implementation used in compute- constrained and / or battery-constrained platforms. For such a platform, the outputs of an algorithm may be pre-computed and stored, e.g., as a cache. This configuration may reduce the need for computation, reduce latency, and / or improve both the thermal management as well as battery life of certain implementations. Accordingly, a complex algorithm (e.g., running in the background) may be used to process one or more scenarios (e.g., a series of scenarios) and corresponding outputs may be prerecorded.

[0094] While the disclosed methods, devices, and systems are described with exemplary reference to pacing rate programming via a pacemaker, it should be appreciated that the disclosed examples may be applicable to any environment, such as a desktop or laptop computer, an automobile entertainmentsystem, a home entertainment system, etc. Also, the disclosed examples may be applicable to any type of Internet protocol.

[0095] According to an example, as shown in FIG. 2B, the process of flow diagram 270 describes an example of a pacing algorithm. The example may be an improvement to pacing programs by serving as an adaptive mechanism to calculate a desired right atrial pacing rate based on multiple variables (e.g., an existing pacing rate, pulse pressure (PP), current blood pressure, one or more physiological parameters, current heart-rate, patient’s current activity level, etc.). According to this exemplary example, the algorithm may use a three step approach to calculate an optimal right atrial pacing rate and make calculations more efficient by using patient specific values, from the configuration selected by a clinician, or derived from a machine learning algorithm that has “learned” an individual patient’s physiologic (e.g., pulse pressure (PP) dynamics and other correlated data that includes other parameters of interest, such as age, medications, etc. For simplicity, blood pressure and / or pulse pressure are used as an example physiologic value hereafter. However, it will be understood that blood pressure may be replaced with any applicable physiological value discussed herein. Similarly, for simplicity, example pulse pressure values are provided herein. It will be understood that values corresponding to one or more other physiologic attributes may be substituted for the example blood pressure values provided herein. This example may provide a personalized, patient-specific real-time closed loop adjustment to right atrial pacing for treating a variety of cardiovascular conditions including, but not limited to, hypertension and heart failure.

[0096] This example may be an extension of the examples disclosed herein, and may be based on extending the benefits of prior examples to patients with a high pulse pressure (PP) (e.g., greater than approximately 40 mmHg). For example, prior pacing programs may be designed to only consider systolic and / or diastolic blood pressure. This example may apply the treatment range based on pulse pressure in combination with physiologic and / or environmental variables for which pacing may be applied. This example may include an algorithm that is configurable and capable of being enhanced by Al methods, such as ML. Thus, this example may make pacing programs even more effective as a realtime closed-loop system for a wide range of disease states, beginning with hypertension and heart failure.

[0097] As shown in FIG. 2B, this example includes a pulse pacing range. Multiple stages of pacing programs may be defined. Multiple stages may include a Stage-I pacing (Stage-I Pacing where PP is, for example, between approximately 40 mmHg to approximately 60 mmHg) and a Stage-II pacing (Stage-II Pacing where PP is greater than 60 mmHg). There may also be a no pacing stage (no Pacing where SBP is less than 40 mmHg). Each stage may refer to a different level of therapy.

[0098] In this example, Stage-I patients may receive lower doses of pacing therapy and the automatic or configurable increments of treatment may also be smaller. For Stage-II patients, the dosage and increments of pacing therapy may be higher than Stage-I, making Stage-II pacing may be calculated to determine more aggressive pacing values. For example, Stage-I pacing may be configurable to reduceboth the duration and magnitude of right atrial pacing changes, which may result in Stage-I dosing that is less intense (lower duration, and / or lower rate increase or decrease) than Stage-II.

[0099] In this example, flow diagram 270 illustrates a high level decision making flowchart. Each pulse pressure value may be assessed against the 3-bands approach from table 271 (No Pacing, Stage-I, or Stage-II). A decision may be made to classify a given blood pressure value into one of the three bands. If the value falls into the “No Pacing” band, no therapy will be delivered.

[0100] If the value is classified into the Stage-I band, a Stage-I lookup table may be used to calculate the applicable pacing value to be sent as a programming command to the pacemaker. As disclosed herein, a lookup table may be output by a machine learning model based on one more inputs and / or applicable training data (e.g., supervised or unsupervised training data).

[0101] According to an example, a patient physiologic value may be detected and / or received (e.g., a pulse pressure value detected using a blood pressure device). The patient physiologic value may be analyzed and a determination may be made whether the physiologic value corresponds to Stage-I, Stage- II, or Stage-Ill. Based on the determination of the applicable stage, a corresponding stage specific program may be activated. Similarly, based on the determination of the applicable stage, a corresponding look-up table may be generated, received, and / or loaded. By activating a given stage specific program, the pacing based technology may be improved as a single program may be selected from a plurality of programs such that targeted pacing may be provided to a patient.

[0102] In this example, a preliminary calculation may be based on a table and / or a pacing algorithm may be designed to incorporate various configurable parameters. These parameters may be divided into categories such patient historical values, patient physiological parameters, and patient environmental and geographical parameters. A non-exhaustive list may include the following:

[0103] Patient historical values may include (but are not limited to): Which Pacing value resulted in better qualify of life; Success of a pacing program; Individual successful treatment dynamics developed for each patient; Recent food / drink intake; Amount of sleep last night; Physical activities (e.g., recent or ongoing); Any emotional events; Any ongoing sickness; Current medications; etc.

[0104] Patient physiological parameters may include (but are not limited to): pulse pressure, Demographics (e.g., male / female, age, body morphology, race); Medications; Pulse oximetry; Hemodynamics (e.g., cardiac output and systemic vascular resistance); Height; Weight; Systolic Blood Pressure; Diastolic Blood Pressure; Mean Arterial Pressure; Heart-rate; Oxygen saturation; Body temperature; Fluid status; Bioimpedance; Systemic vascular resistance; etc.

[0105] Patient environmental and geographical parameters may include (but are not limited to): Date & time of the day; Location (e.g., latitude & longitude and altitude); Outside temperature; Humidify; Motion sensor input; etc.

[0106] Some of these values may be fed through a patient feedback module, which may be an optional module for the entire algorithm. Other variables may be derived through sensors and other automationor calculations. Consideration of these configurable values may be through traditional mathematics, a machine learning algorithm, or a hybrid approach of the two. Machine learning scenarios may calculate patient specific recommendations by learning from a patient’s history and may also consider cohortbased learnings, available from a server, to apply towards calculating an optimal or recommended pacing value for a given dosage delivery.

[0107] Flow diagram 270 may start at step 272. Step 273 may include checking a parameter, such as pulse pressure. Based on the parameter, a determination is made at step 274, to determine if a patient falls within a Stage-I pacing program. If the patient is determined to be on a Stage-I pacing program (e.g., a PP between 40 mmHg and 60 mmHg), step 274a may include activating a Stage-I program using a Stage-I pacing algorithm to determine and adjust a pacing parameter for a patient. After a period of time (e.g., 30 minutes, 3 hours, or 3 days) step 274b may include receiving a patient’s feedback.

[0108] Receiving a patient’s feedback may include receiving a subjective indication from the patient related to how the patient is feeling based on the implementation of the current pacing rate. In one example, receiving a status update may include a subjective input received from the subject. For example, a patient may be provided a prompt (e.g., via a graphical user interface) requesting the patient to provide an assessment of his or her own wellbeing. According to an implementation of the disclosed subject matter, the subject may be provided a prompt, via a graphical interface, comprising graphical components that are generated and displayed based on the pre-determined status update schedule. For example, a prompt may provide a selectable graphical component corresponding to the subject feeling ill and selectable graphical component corresponding to the subject feeling well after the patient’s baseline pacing rate was set to 5%. Alternatively, or in addition, receiving patient feedback may include receiving a pseudo-subjective input, as discussed herein.

[0109] The patient feedback (or pseudo-subjective input) may be combined, in step 276, with configurable, patent specific parameters (physiological and historical). This information may then be included, at step 274a, and utilized as part of Stage-I pacing algorithm to determine an updated pacing parameter. The loop may continue in such a manner where after each adjustment, patient feedback is received, and the new information may be used to update the pacing parameter determined using the Stage-I pacing algorithm.

[0110] Alternatively, if it is determined at step 274 that the patient is not in a Stage-I pacing (e.g., the PP of the patient is greater than 60 mmHg) a determination at step 275 will determine if the patient falls within a Stage-II pacing program. If the patient is determined to fall into a Stage-II pacing program, step 275a may include activating a Stage-II program using a Stage-II pacing algorithm to determine a pacing parameter for a patient. After the pacing parameter is adjusted, step 275b may include receiving a patient’s feedback, such as by using methods described above. The patient feedback may be combined in step 276 with configurable, patient specific parameters (physiological and historical). Thisinformation may then be included again at step 275 a and used as part of Stage-II pacing algorithm to determine an updated pacing parameter.

[0111] Generally, a Stage-II pacing algorithm may follow a similar strategy as a Stage-I processing. In the example, when comparing a Stage-I dosage with a Stage-II dosage, the Stage-II pacing program will likely have a higher dosage, larger increments, and more steps. After each adjustment, a patient’s feedback is received, and that new information may be used to update pacing parameters using the Stage- II pacing algorithm.

[0112] If it is determined at step 275 that the patient is not in a Stage-II pacing, the process will revert back to step 273 where patient parameters (such as pulse pressure) may be monitored.

[0113] Step 274a and step 275a of Stage-I and Stage-II respectively, may be configurable to include therapeutic parameters such as: Duration between two consecutive Pacing therapies (e.g., settling time or refractory period); Priority level with respect to other pacemaker’s built-in algorithms; Number of increments (stair steps model) and different % for each increments; Criteria for step back and different % for each step back values; Total number of steps before max out or reaching minimum; etc.

[0114] Additionally, step 274a and 275a may include a pacing algorithm that allows clinicians to configure various rules based on which patient can be remotely monitored better. For example, such rules may include (but are not limited to) the following: If the patient is not feeling well after 3 consecutive attempts of pacing therapy, alert the clinician; If the patient remains at a max possible pacing value for 5 days, alert the physician; If a pacemaker or ICD rejects a prescribed pacing value 3 consecutive times, alert the physician; If a patient is jogging, but a sensory input is at a level where it needs additional confirmation, alert the patient through smartphone app notification, and ask for confirmation, etc

[0115] Overall, this example may be designed to have a maximum configurability to customize pain therapies for every patient, prescribe a pain therapy that considers various external parameters, leverage machine learning to determine appropriate pacing programs and parameters, incorporate advanced monitoring and alerting mechanisms made available to a patient, caretaker, and clinician, and harvest relevant data available in a cloud for machine learning purposes.

[0116] FIG. 3 shows a flowchart 300 for determining pacing rate, according to one or more examples. For simplicity, flowchart 300 is provided in reference to pulse pressure (PP). However, it will be understood that any applicable physiological parameter value, such as those discussed herein, may be used in reference to flowchart 300. At step 302, the beginning of a session, a patient’s pulse pressure (PP) and heart rate may be received. The pulse pressure (PP) may be detected using a blood pressure cuff or a continuous blood pressure device. The pulse pressure (PP) data may be provided to a processor that is local to a pacing device or a remote component external to the pacing device (e.g., a user device, a cloud component, an external processor, etc.), such as component 120 of FIG. 1.

[0117] At step 304, a first pacing rate may be determined to modify the heart rate received at step 302, based on the pulse pressure (PP) input received at step 302. For example, if a patient’s initial pulse pressure (PP) was 60 mmHg and heart rate was 60 bpm, the first pacing rate may be set at 66 bpm. In another example, if a patient’s initial blood pressure was 45 mmHg, and heart rate was 60 bpm, the first pacing rate may be set at 58 bpm. The first pacing rate may be determined based on the techniques disclosed herein in reference to FIG. 2A. For example, the first pacing rate may be determined based on a threshold value and / or second threshold value output by a threshold machine learning model and / or an output generated by a pacing machine learning model, as discussed herein. Alternatively, or in addition, the first pacing rate may be determined in accordance with a complex lookup table. As discussed herein, a complex lookup table may be populated using outputs output by a threshold machine learning model and / or pacing machine learning model.

[0118] According to examples disclosed herein, a pacing rate may be determined on-demand using the threshold machine learning model and / or pacing machine learning machine learning model. For example, component 120 may receive the pulse pressure and / or heart rate at step 302 and provide the same to the threshold machine learning model and / or pacing machine learning machine to receive a pacing rate. Alternatively, pacing system 130 may be provided a complex lookup table which may be populated based on the threshold machine learning model and / or pacing machine learning machine. Pacing system 130 may maintain a static copy of a complex lookup table which may be updated from time to time. According to this example, battery efficiency may be optimized at the pacing system 130 by using the static complex lookup table to receive, determine, or otherwise identify pacing rates.

[0119] At step 306, the first pacing rate may be output and the output may be received at a cardiac pacing device (e.g., pacing system 130). The cardiac pacing device may be configured to pace based on the first pacing rate. For example, the first pacing rate may be output by a pacing machine learning algorithm and / or by a complex lookup table populated at least in part based on a pacing machine learning model.

[0120] After a defined period of time (e.g., as determined based on a time-based machine learning model), a patient may be provided an interface to rate their wellbeing (e.g., as discussed in reference to steps 218A, 218B, 218C, 218D, 218E, and / or 218E of FIG. 2A). The interface may be populated in accordance with the techniques discussed herein.

[0121] At step 308, a patient’s input regarding their wellbeing may be received following the output of the first pacing rate at step 306. For example, a patient may indicate that they are feeling better, worse, or the same. If the patient is feeling better or same (e.g., based on a positive indication provided via the provided interface), the session may terminate as there may be no need for further adjustment of the pacing rate. However, if the patient indicates that they are feeling worse (e.g., a negative indication), the session may continue to step 310.

[0122] At step 310, a second pacing rate is determined based on input received from the patient at step 308. For example, if the first pacing rate was set at 66 bpm, the second pacing rate may be set at 63 bpm, based on input received from the patient at step 308. In another example, if the first pacing rate was set at 58 bpm, and the patient input a negative indication at step 308, the second pacing rate may be adjusted in accordance with the techniques disclosed herein (e.g., the second pacing rate may be set to the initial rate at the beginning of the session, such as 60 bpm). At step 312, the second pacing may be output and the output may be received at a cardiac pacing device. The cardiac pacing device may be configured to pace based on the second pacing rate.

[0123] FIG. 4 depicts a flow diagram 400 of an exemplary method for determining a pacing rate via a pseudo-subjective machine learning model, according to one or more examples. As shown in FIG. 4, the pseudo-subjective input 408 may be output by a pseudo-subjective machine learning model 406. The pseudo-subjective machine learning model 406 may be trained to output a pseudo-subjective input 408 based on one or more objective inputs. The pseudo-subjective machine learning model 406 may be trained by modifying one or more layers, weights, biases, synapsis, and / or the like of the model based on the training data. The machine learning model may be trained to output a pseudo-subjective input 408 based on one or more sensor inputs which may correspond to physiological inputs (e.g., pulse pressure (PP), blood pressure, heart-rate, temperature, sympathetic nerve activity (SNA), etc.). These sensor inputs may be sensed by one or more sensors, such as those shown in FIG. 4 as sensor input A 402 and / or sensor input B 404.

[0124] The pseudo-subjective machine learning model 406 may be trained using historical patient information 410, such as, patient subjective inputs, patient medications (e.g., patient provided, received by a system or component, medication compliance information, etc.), health history, and / or the like. For example, pseudo-subjective machine learning model 406 may be trained using historical subjective inputs provided by a patient (e.g., in accordance with flow diagram 200 of FIG. 2A). The pseudo- subjective machine learning model 406 may correlate or otherwise associate such subjective inputs with corresponding physiological inputs captured at or about the time the subjective inputs are provided. Accordingly, the pseudo-subjective machine learning model 406 may be trained to output pseudo- subjective inputs that correspond to the subjective inputs a patient is likely to provide, based on one or more current physiological states the patient is experiencing. As a simplified example, a given patient may typically provide a negative indication as a subjective input when the patient’s heart rate is above a given amount. Accordingly, the pseudo-subjective machine learning model 406 may generate a negative indication pseudo-subjective input when sensor input A 402 and / or sensor input B 404 senses a heart rate above the given heart rate. It will be understood that although this simplified example is based on a single objective input (e.g., heart rate), the pseudo-subjective machine learning model 406 may be trained to output a pseudo-subjective input based on one or more objective inputs, trends related to such objective inputs, changes in such objective inputs, and / or the like.

[0125] Alternatively, or in addition, cohort population information 412 may be used to train pseudo- subjective machine learning model 406. For example, pseudo-subjective machine learning model 406 may be trained using historical or simulated subjective inputs provided by a cohort of patients (e.g., in accordance with flow diagram 200 of FIG. 2A). The pseudo-subjective machine learning model 406 may correlate or otherwise associate such cohort based subjective inputs with corresponding cohort physiological inputs captured at or about the time the subjective inputs are provided. Accordingly, the pseudo-subjective machine learning model 406 may be trained to output pseudo-subjective inputs that correspond to the subjective inputs a patient is likely to provide, based on one or more current physiological states the patient is experiencing. As a simplified example, a cohort of patients may typically provide a negative indication as a subjective input when the patients’ temperature is above a given amount. Accordingly, the pseudo-subjective machine learning model 406 may generate a negative indication pseudo-subjective input when sensor input A 402 and / or sensor input B 404 senses that a given patient’s temperature is above the given temperature.

[0126] The following is an example that corresponds with the subject matter discussed herein. Drug resistant hypertension (DRH) is defined as blood pressure (BP) that remains above goal despite concomitant use of >3 different classes of antihypertensive drugs, administered at maximally tolerated doses, including a diuretic. Patients with DRH are at high risk for having major cardiovascular events. The prevalence and incidence of DRH is expected to increase as the global population continues to age, with an overall increase in the number of affected individuals as the general population grows.

[0127] Recent efforts to address the problem of DRH have included the development and investigation of device-based therapies. Because many patients have high resting sympathetic activity, one focus in the past decade has been the development of several implantable devices, including those intended to target the autonomic nervous system, regulate left ventricular preload or alter mechanical arterial properties. Non-pharmacological neuromodulation devices that modulate sympathetic activity using electrical activation of the carotid baroreflex, catheter-based renal nerve ablation, and new algorithms for permanent ventricular pacing are supported by experimental studies and early clinical trials. However, the relationship between hypertension (HTN) and conventional clinical permanent cardiac pacing is not well established. Many older patients requiring permanent pacing also have persistent HTN with systolic blood pressures (SBP) above recommended levels. A significant reduction in SBP and diastolic (DBP) has been observed among such patients. In this example, permanent cardiac pacing, particularly with elderly patients having DRH, may result in pacing-mediated BP lowering.

[0128] One or more implementations disclosed herein may be applied by using a machine learning model, another Al system such as a neural network, or a non-AI rules-based system. For example, a machine learning model may be used to determine a state machine and / or a next state. As shown in flow diagram 500 of FIG. 5, training data 512 may include one or more of stage inputs 514 and known outcomes 518 related to a machine learning model to be trained. The stage inputs 514 may be from anyapplicable source including an input or system discussed herein (e.g., an output from a step or aspect of FIGs. 1-4). The known outcomes 518 may be included for machine learning models generated based on supervised or semi-supervised training. An unsupervised machine learning model might not be trained using known outcomes 518. Known outcomes 518 may include known or desired outputs for future inputs similar to or in the same category as stage inputs 514 that do not have corresponding known outputs.

[0129] Known outcomes 518 may include a recent Phase I study including a human clinical trial investigating the safety and efficacy of pacing therapy in HFpEF. From this study, Relieve HFpEF-II, a data analysis revealed statistically significant correlations between the efficacy of pacing therapy therapy in HFpEF associated with hypertension through the Minnesota Living with Heart Failure Questionnaire and an NYHA score. Observations included the ability to exercise assessed with a low level treadmill exercise and the six-minute walk test. While the correlations were expected, further analysis of blood pressure data revealed the following additional unexpected findings:Pacing therapy lowered blood pressure in hypertensive patients (expected result)Pacing therapy prevented spontaneous drops in diastolic blood pressure that occurs in HFpEF and is a poor prognostic sign, (new finding)Pacing therapy prevents the drop in pulse pressure seen in HFpEF and narrows pulse pressure in patients who already exhibit widened pulse pressure, suggesting improved prognosis (new finding)The narrowing of pulse pressure associated with pacing therapy is associated with improved MLWHF score, improved NYHA score, and improved exercise tolerance (treadmill test and six- minute walk test)The measurement of pulse pressure is an important physiologic variable in HFpEF that can be employed by pacing therapy, alone or with “standard” systolic pressure endpoints to regulate Right Atrial Pacing for pacing therapies and further improve the treatment and outcomes in HFpEF- Not only systolic blood pressure, but also diastolic blood pressure, is important as an input to pacing therapy, alone or in combination with systolic pressure and other variables

[0130] Results from a 24-hour pacing therapy, where pulse pressure was used as source blood pressure data include the following:More patients showed a reduction in pulse pressure when compared to standard pacing therapy patients (Fisher’s exact test p-value of 0. 1086 or a Chi-squared p-value of 0.0596)When treating week 3 as the baseline, pulse pressure was reduced more in comparison to standard pacing therapy (t-test p-value of 0.0254)The linear regression, with pulse pressure as the covariate and change from baseline as the response, shows that subjects with narrower pulse pressure tend to walk longer distances in thesix-minute walk test (p-value=O. 1154) and have better MLHFQ scores (p-value=0.0468) in comparison to subjects with wider pulse pressure

[0131] Results from twice-a-day pulse pressure determinations using blood pressure cuffs (not 27 times in 24 hours as above)Pacing therapy has more patients showing a reduction in pulse pressure (Fisher’s exact test p- value of 0.0768 or a Chi-squared p-value of 0.0291)When treating week 3 as the baseline, pulse pressure was reduced more in comparison to standard pacing therapy (t-test p-value of 0.0448)The linear regression, with pulse pressure as the covariate and change from baseline as the response, shows that subjects with narrower pulse pressure tend to walk for a greater amount of time on a treadmill (p-value = 0.0762) and for greater distances in the six-minute walk test (p- value=0.0374) in comparison to subjects with wider pulse pressure.

[0132] A key is that two different blood pressure measurement techniques produced the same results. Thus, although the total number of subjects in the trial was only 16, and most trials using blood pressure as an endpoint involved hundreds or thousands of subjects, the concordance of the two methods, particularly as the twice a day blood pressure measurements totaled 1540 for the trial as opposed to 48 measurements for the 24-hour blood pressure monitors, give added confidence that the findings are real and reproducible.

[0133] The training data 512 and a training algorithm 520 may be provided to a training component 530 that may apply the training data 512 to the training algorithm 520 to generate a machine learning model. According to an implementation, the training component 530 may be provided comparison results 516 that compare a previous output of the corresponding machine learning model to apply the previous result to re-train the machine learning model. The comparison results 516 may be used by the training component 530 to update the corresponding machine learning model. The training algorithm 520 may utilize machine learning networks and / or models including, but not limited to a deep learning network such as Deep Neural Networks (DNN), Convolutional Neural Networks (CNN), Fully Convolutional Networks (FCN) and Recurrent Neural Networks (RCN), probabilistic models such as Bayesian Networks and Graphical Models, and / or discriminative models such as Decision Forests and maximum margin methods, or the like.

[0134] In general, any process or operation discussed in this disclosure that is understood to be computer-implementable, such as the flows and / or process discussed herein (e.g., in FIGS. 1-5), etc., may be performed by one or more processors of a computer system, such any systems or devices used to implement the techniques disclosed herein. A process or process step performed by one or more processors may also be referred to as an operation. The one or more processors may be configured to perform such processes by having access to instructions (e.g., software or computer-readable code) that, when executed by the one or more processors, cause the one or more processors to perform the 1processes. The instructions may be stored in a memory of the computer system. A processor may be a central processing unit (CPU), a graphics processing unit (GPU), or any suitable types of processing unit.

[0135] FIG. 6 depicts an example system 600 that may execute techniques presented herein. FIG. 6 is a simplified functional block diagram of a computer that may be configured to execute techniques described herein, according to examples of the present disclosure. Specifically, the computer (or “platform” as it may not be a single physical computer infrastructure) may include a data communication interface 660 for packet data communication. The platform may also include a central processing unit 620 (“CPU”), in the form of one or more processors, for executing program instructions. The platform may include an internal communication bus 610, and the platform may also include a program storage and / or a data storage for various data files to be processed and / or communicated by the platform such as ROM 630 and RAM 640, although the system 600 may receive programming and data via network communications. The system 600 also may include input and output ports 650 to connect with input and output devices such as keyboards, mice, touchscreens, monitors, displays, etc. Of course, the various system functions may be implemented in a distributed fashion on a number of similar platforms, to distribute the processing load. Alternatively, the systems may be implemented by appropriate programming of one computer hardware platform.

[0136] The general discussion of this disclosure provides a brief, general description of a suitable computing environment in which the present disclosure may be implemented. In one example, any of the disclosed systems, methods, and / or graphical user interfaces may be executed by or implemented by a computing system consistent with or similar to that depicted and / or explained in this disclosure. Although not required, aspects of the present disclosure are described in the context of computerexecutable instructions, such as routines executed by a data processing device, e.g., a server computer, wireless device, and / or personal computer. Those skilled in the relevant art will appreciate that aspects of the present disclosure can be practiced with other communications, data processing, or computer system configurations, including: Internet appliances, hand-held devices (including personal digital assistants (“PDAs”)), wearable computers, all manner of cellular or mobile phones (including Voice over IP (“VoIP”) phones), dumb terminals, media players, gaming devices, virtual reality devices, multiprocessor systems, microprocessor-based or programmable consumer electronics, set-top boxes, network PCs, mini-computers, mainframe computers, and the like. Indeed, the terms “computer,” “server,” and the like, are generally used interchangeably herein, and refer to any of the above devices and systems, as well as any data processor.

[0137] Aspects of the present disclosure may be embodied in a special purpose computer and / or data processor that is specifically programmed, configured, and / or constructed to perform one or more of the computer-executable instructions explained in detail herein. While aspects of the present disclosure, such as certain functions, are described as being performed exclusively on a single device, the presentdisclosure may also be practiced in distributed environments where functions or modules are shared among disparate processing devices, which are linked through a communications network, such as a Local Area Network (“LAN”), Wide Area Network (“WAN”), and / or the Internet. Similarly, techniques presented herein as involving multiple devices may be implemented in a single device. In a distributed computing environment, program modules may be located in both local and / or remote memory storage devices.

[0138] As discussed herein, a memory may include a device or system that is used to store information for immediate use in a computer or related computer hardware and digital electronic devices. Contents of memory can be transferred to storage (e.g., via virtual memory). Memory may be implemented as semiconductor memory, where data is stored within memory cells built from MOS transistors on an integrated circuit. Semiconductor memory may include volatile and / or non-volatile memory. Examples of non-volatile memory include flash memory and read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM) memory, and the like. Examples of volatile memory include primary memory such as dynamic random-access memory (DRAM) and fast CPU cache memory such as static random-access memory (SRAM).

[0139] Aspects of the present disclosure may be stored and / or distributed on non-transitory computer- readable media, including magnetically or optically readable computer discs, hard-wired or preprogrammed chips (e.g., EEPROM semiconductor chips), nanotechnology memory, biological memory, or other data storage media. Alternatively, computer implemented instructions, data structures, screen displays, and other data under aspects of the present disclosure may be distributed over the Internet and / or over other networks (including wireless networks), on a propagated signal on a propagation medium (e.g., an electromagnetic wave(s), a sound wave, etc.) over a period of time, and / or they may be provided on any analog or digital network (packet switched, circuit switched, or other scheme).

[0140] Program aspects of the technology may be thought of as “products” or “articles of manufacture” typically in the form of executable code and / or associated data that is carried on or embodied in a type of machine-readable medium. “Storage” type media include any or all of the tangible memory of the computers, processors or the like, or associated modules thereof, such as various semiconductor memories, tape drives, disk drives and the like, which may provide non-transitory storage at any time for the software programming. All or portions of the software may at times be communicated through the Internet or various other telecommunication networks. Such communications, for example, may enable loading of the software from one computer or processor into another, for example, from a management server or host computer of the mobile communication network into the computer platform of a server and / or from a server to the mobile device. Thus, another type of media that may bear the software elements includes optical, electrical and electromagnetic waves, such as used across physicalinterfaces between local devices, through wired and optical landline networks and over various air-links. The physical elements that carry such waves, such as wired or wireless links, optical links, or the like, also may be considered as media bearing the software. As used herein, unless restricted to non- transitory, tangible “storage” media, terms such as computer or machine “readable medium” refer to any medium that participates in providing instructions to a processor for execution.

[0141] The terminology used above may be interpreted in its broadest reasonable manner, even though it is being used in conjunction with a detailed description of certain specific examples of the present disclosure. Indeed, certain terms may even be emphasized above; however, any terminology intended to be interpreted in any restricted manner will be overtly and specifically defined as such in this Detailed Description section. Both the foregoing general description and the detailed description are exemplary and explanatory only and are not restrictive of the features, as claimed.

[0142] Other examples of the disclosure will be apparent to those skilled in the art from consideration of the specification and practice of the invention disclosed herein. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the invention being indicated by the following claims.

[0143] It should be understood that examples in this disclosure are exemplary only, and that other examples may include various combinations of features from other examples, as well as additional or fewer features. It should be appreciated that in the above description of examples of the invention, various features of the invention are sometimes grouped together in a single example, figure, or description thereof for the purpose of streamlining the disclosure and aiding in the understanding of one or more of the various inventive aspects. This method of disclosure, however, is not to be interpreted as reflecting an intention that the claimed invention requires more features than are expressly recited in each claim. Rather, as the following claims reflect, inventive aspects lie in less than all features of a single foregoing disclosed example. Thus, the claims following the Detailed Description are hereby expressly incorporated into this Detailed Description, with each claim standing on its own as a separate example of this invention.

[0144] Furthermore, while some examples described herein include some but not other features included in other examples, combinations of features of different examples are meant to be within the scope of the invention, and form different examples, as would be understood by those skilled in the art. For example, in the following claims, any of the claimed examples can be used in any combination.

[0145] Thus, while certain examples have been described, those skilled in the art will recognize that other and further modifications may be made thereto without departing from the spirit of the invention, and it is intended to claim all such changes and modifications as falling within the scope of the invention. For example, functionality may be added or deleted from the block diagrams and operations may be interchanged among functional blocks. Steps may be added or deleted to methods described within the scope of the present invention.

[0146] The above disclosed subject matter is to be considered illustrative, and not restrictive, and the appended claims are intended to cover all such modifications, enhancements, and other implementations, which fall within the true spirit and scope of the present disclosure. Thus, to the maximum extent allowed by law, the scope of the present disclosure is to be determined by the broadest permissible interpretation of the following claims and their equivalents, and shall not be restricted or limited by the foregoing detailed description. While various implementations of the disclosure have been described, it will be apparent to those of ordinary skill in the art that many more implementations are possible within the scope of the disclosure. Accordingly, the disclosure is not to be restricted except in light of the attached claims and their equivalents.

Claims

CLAIMSWhat is claimed is:

1. A cardiac pacing system for treating heart failure with preserved ejection fraction (“HFpEF”), comprising: an implantable pulse generator configured to deliver atrial pacing signals; a sensor configured to measure systolic blood pressure (“SBP”) and diastolic blood pressure (“DBP”) in real time; a controller configured to calculate a pulse pressure (“PP”) value based on a difference between the SBP and DBP; and a pacing algorithm executed by the controller configured to determine an optimal atrial pacing rate based on the calculated PP value, wherein the pacing algorithm is configured to adjust the atrial rate to one or more of improve patient exercise tolerance and provide symptomatic relief.

2. The system of claim 1, wherein the pacing algorithm is configured to determine changes in PP over time and adjust atrial pacing rate proportionally to a detected percent change.

3. The system of claim 1, wherein the pacing algorithm is configured to calculate an average PP over a predetermined number of cardiac cycles and use the averaged PP to determine the atrial pacing rate.

4. The system of claim 1, wherein the pacing algorithm is configured to compute a moving average or weighted average of PP values across n intervals to smooth transient fluctuations.

5. The system of claim 1, wherein the pacing algorithm is configure to adjust atrial pacing rate upward when PP decreases below a patient-specific threshold and downward when PP increases above the threshold.

6. The system of claim 1, wherein the controller is configured to apply a derivative of PP with respect to time (“dPP / df ’) to determine dynamic changes in vascular compliance and adjusts atrial pacing accordingly.

7. The system of claim 1, wherein the controller is configured to calculate a normalized PP index by dividing PP by mean arterial pressure (“MAP”) and use the PP index to determine pacing rate.

8. The system of claim 1, wherein the controller is configured to compute one or more of a spectral and frequency-domain analysis of PP variability and adjusts atrial pacing rate to maintain PP variability within a therapeutic range.

9. A method of controlling atrial pacing in a patient with HFpEF, comprising: measuring SBP and DBP on a beat-to-beat basis; calculating PP as a difference between SBP and DBP; determining a pacing rate based on PP or a mathematical derivative thereof; and delivering atrial pacing at the determined rate to improve exercise capacity and symptomatic relief.

10. The method of claim 9, wherein determining the pacing rate comprises calculating a percent change in PP relative to a baseline PP value.

11. The method of claim 9, wherein determining the pacing rate comprises averaging PP values across a plurality of cardiac cycles.

12. The method of claim 9, further comprising adjusting the pacing rate based on both PP and an additional hemodynamic parameter selected from stroke volume, cardiac output, or heart rate variability.

13. A non-transitory computer-readable medium storing instructions which, when executed by a controller of a cardiac pacemaker, cause the controller to: acquire SBP and DBP values; compute PP values based on SBP-DBP; determine changes in PP according to one or more of percent change, averaged values, or variability analysis; and adjust atrial pacing rate to optimize patient functional capacity.

Images