The present invention relates generally and specifically to combining biological sensors with external machines using machine learning to form computerized representations capable of controlling effectors to deliver therapy or enhance performance. The invention integrates sensed signals from body systems and artificial devices with outputs from measureable body systems and artificial devices to create learned networks. Measurable body systems include the central and peripheral nervous systems, cardiovascular system, respiratory system, skeletal muscles and skin well as any other body systems that are capable of producing measurable signals. Artificial devices include diagnostic sensors, medical stimulating or prosthetic devices and/or non-medical systems. The invention has applications in sleep and wakefulness, sleep-disordered breathing, memory and cognition, monitoring and responding to obesity or heart failure and other conditions, or more generally in enhancing performance via external devices. This disclosure outlines several applications of this invention, using as an example methods and systems to enhance sleep-related bodily functions for use in normal individuals or patients with sleep-breathing disorders.
This application incorporates by reference the entire subject matter and application of attorney docket #2480-2 PCT (application PCT/US15/46819, filed Aug. 25, 2015).
The human body has long been interfaced with artificial devices or machines. Prosthetic limbs have for centuries been made of wood combined with metal and other materials. Through recent technological advances, devices often now have sophisticated materials, design and control for a specific purpose—such as a robotic limb (see for instance singularityhub.com/2013/07/24/dapas-brain-controlled-prosthetic-arm-and-a-bionic-hand-that-can-touch) or a glucose-sensing insulin infusion pump.
Many of these functions are mediated by the brain (central nervous system) and/or peripheral nervous system. These functions include classical “neurological” functions such as vision or hearing, but also nearly all activities of daily life, including learning, moving, or operating machinery.
In many situations, the body's ability to perform such functions is constrained. Constraint can take many forms and may be physical or functional. Physical constraints include an external obstacle preventing movement of a limb in an enclosed space such as may affect a warrior or scuba diver. A physical constraint may also be internal, such as loss of a limb from amputation. Functional constraints may include a classical disease, such as stroke that prevents an individual's ability to move the foot. However, functional constraints may also include underperformance on a task due to insufficient training, knowledge or acquisition of skills, or through disuse.
What is lacking is how devices can be used to automatically (“intelligently”) tailor therapy to restore lost physiological function or enhance an existing function in a specific individual. This inability for prior and current devices to automatically tailor therapy and restore or enhance a function is striking when examining how the human brain senses, integrates and controls bodily functions.
The prior art has extensively studied, yet imprecisely defined, which regions of the brain control bodily functions and how they interact with other physiological components in a network. Simple bodily functions, such as moving the biceps of the left arm or sensing from the right index finger, are well defined and often conserved between individuals. Functional mapping or “atlases” are debated even for some “simple sensations” such as visual recognition of a face. Moreover, other bodily functions including “higher cortical” functions are neither well defined nor conserved. This includes sleep, cognition, memory, mood, alertness, sensory-motor and many other activities.
Currently, machines that modulate bodily function are largely based on a precise detailed knowledge of physiology, which for the brain would include neuroimaging, brain mapping and peripheral nerve mapping of normal and abnormal functions. Unfortunately, such detailed knowledge is often incomplete. Mapping of functional locations often vary between individuals—and even in the same person at different times. Many functions are poorly mapped, such as memory, cognition and mental performance. Even for well mapped functions, studies that define physiological function often raise additional uncertainties in this functionality.
Mapping functional domains of a bodily function—the network of physiological systems associated with that function including sensed signals and biological effectors that control it—is difficult. Mapping of functional domains is particularly difficult for functions involving the brain. However, there is an urgent need to sense and modulate functional domains whose altered function may cause disease or suboptimal performance.
In traditional theory, sleep and wakefulness are modulated by brain regions including the posterior hypothalamus, while memory is encoded by the hippocampus and other regions of the limbic system. However, it is not clear what brain regions are responsible for controlling sleep, or for mediating abnormal breathing in central sleep apnea. Regions of the brainstem that control airway muscles are better characterized, such as nuclei in the medulla oblongata for the hypoglossal nerve (twelfth cranial nerve) that controls tongue movement. Yet, how nuclei are integrated into abnormal breathing to produce obstructive sleep apnea is not understood. As a result, it has been difficult to treat this condition even using novel systems that activate tongue motion to reduce obstruction.
Sleep is a bodily function that integrates the nervous system, skeletal muscle, cardiopulmonary, and other body systems. Sleep alternates with and enables subsequent wakefulness, and is required for normal functioning of most organ systems. Sleep is traditionally considered to be controlled by specific regions of the brainstem (primitive brain) that both regulate and are modulated by function of the higher brain (cerebral cortex), muscles controlling breathing, other involuntary muscles such as sphincters of the gastrointestinal and genitourinary tracts, voluntary muscles such as muscles of the legs or arms, sensory function, and other bodily functions.
Much work over several decades has strived to define which regions of the brain subtend bodily functions such as sleep. As outlined above, while functional mapping is well defined for “simple” functions such as controlling a defined muscle (e.g., the biceps of the upper arm) or sensation (e.g., the right index finger), it is far less clear for sleep. Interactions between the multiple organ systems impacted by sleep further complicate precise mapping.
An individual's ability to sleep may be compromised in many ways. Among the most important and common are sleep hypopnea (reduced breathing) and apnea (absence of breathing), in which impaired breathing in sleep interrupts sleep functioning, and primary sleep disorders such as insomnia, where the individual cannot sleep efficiently or sufficiently. All sleep disorders negatively impact wakefulness, producing daytime drowsiness that impairs daily activities. Sleep disorders can also lead to disorders from breathing such as low oxygen levels with metabolic effects including acidosis, disorders of the heart such as failure and abnormal rhythms, disorders of the immune system causing susceptibility to infection, psychological disorders such as stress, depression, anxiousness and psychosis, and several other states of poor functioning and disease.
Sleep apnea may be obstructive or central. Obstructive sleep apnea is increasingly recognized in individuals who snore, who are overweight and who may develop sequelae such as heart failure. Central sleep apnea is also common, yet is under-recognized and associated with comorbidities such as heart failure. It is likely that central sleep apnea also occurs alongside obstructive sleep apnea, since treatments that physically open the throat muscles and prevent obstruction may sometimes leave residual apnea.
Obstructive sleep apnea (OSA) results from complete or partial airway collapse in sleep. Conversely, central sleep apnea (CSA) results from reduced brain stimulation of the respiratory muscles in sleep. Both forms are typically diagnosed using overnight polysomnography (PSG), a test that measures at least eight (8) channels including the electroencephalogram (EEG), electrooculogram (EOG), electrocardiogram (ECG), chin electromyogram (EMG), airflow, respiratory “effort,” oxygen saturation (SaO2 or sat), and body position. However, this is a cumbersome test typically performed with an overnight hospital stay attended by physicians, is not well liked by patients, cannot easily be repeated to assess the impact of therapy and is difficult to perform at home. Recent studies have shown that commercial tests offered to circumvent traditional polysomnography are suboptimal at best.
From a polysomnogram, apnea is defined as absence of breathing (no tidal volume) for at least 10 seconds, while hypopnea is defined as decrease in tidal volume of 30% for at least 10 seconds accompanied by at least a 3% decrease in oxygen saturation or terminated by arousal from sleep. Apnea is defined as obstructive if accompanied by inspiratory effort against the occluded pharynx. Without such accompanying effort, apnea is defined as central. Similarly, hypopnea is obstructive if there are signs of upper airway flow limitation, and is otherwise considered central. The apnea-hypopnea index (AHI) is the ratio of apnea-to-hypopnea per hour of sleep, and is classified as no sleep apnea (AHI<5), mild sleep apnea (AHI of 5-15), moderate sleep apnea (AHI of 15-30) or severe sleep apnea (AHI>30).
Several treatments are available for obstructive sleep apnea, but these are often not well tolerated. The most commonly used treatment currently is continuous positive airway pressure to keep the airway open and reduce/eliminate obstruction. Other options include mechanical splints and even surgical procedures to reduce/eliminate obstruction. Some recent devices have applied stimulation to the muscles of the tongue or face to eliminate obstruction, but it is unclear how well they will work in the broad population.
Few strategies have been proposed to improve central sleep apnea—or more generally the central control of sleep. Since central sleep apnea may relate directly to sleep disorders, treatments for central sleep apnea may potentially also help other conditions. It is increasingly appreciated that central sleep apnea makes heart failure worse, and so treatment for central sleep apnea may improve symptoms of heart failure, and other cardiac and non-cardiac conditions such as insomnia and psychological sequelae.
Pharmacological drug therapy is often used to induce sleep, but these agents are not useful in sleep apnea. These drugs rarely mimic the natural stages of sleep, rarely induce rapid eye movement (REM) sleep that is essential for restfulness, and may paradoxically worsen sleep disorders and produce daytime drowsiness despite nighttime unconsciousness.
New therapeutic modalities are clearly needed to modulate the complex functions outlined above—often including a component of central or peripheral nervous system involvement. Emerging modalities involve electrical stimulation/modulation of brain or nervous system activity, typically at a specific target region. All these current modalities suffer from a significant common problem, as they attempt to perform therapy with no or minimal sensory input, feedback, or modulation of such therapy based upon the individual patient's neurological activity.
One example of electrical stimulation therapy is noninvasive or minimally invasive trigeminal nerve stimulation (e.g., NEUROSIGMA®) that is being evaluated to treat depression and seizures. Unfortunately, the true mechanism of action of such therapy is unclear. Whether this is due to the actual trigeminal nerve being stimulated, direct stimulation of the frontal lobe of the brain, indirect inhibition of cerebral blood flow or some other as yet unknown mechanism, still remains to be determined and will affect the ability of such therapy to be applied successfully. Additionally, this therapy is applied as a “one-size-fits-all” approach without any adaptation for individual patient responses.
Other non-invasive neuromodulation/stimulation approaches also being considered include stimulation of the vagus nerve for seizures (Carbomed, Inc.). Similar to trigeminal nerve stimulation, the mechanism is poorly understood, the actual stimulation of the vagus nerve is unclear via this noninvasive approach, and there is no individual patient adaptation. A number of technologies are attempting to treat depression via noninvasive transcranial application of an electrical and/or magnetic field (Neuronetics Inc., Neosync Inc., Brainsway Inc., Cervel Neurotech Inc., and Tal Medical Inc.). All of these approaches, even though they show interesting preliminary data, suffer from the same problems as above, namely, poor understanding of mechanism and lack of patient-tailored therapy due to a lack of feedback and adaptation for individual patients.
For apnea specifically, approaches that try to modulate obstructive sleep apnea, including stimulation of the hypoglossal nerve (Inspire Med Inc.) or other throat muscle (Apnex Medical Inc.)—are being evaluated but typically do not have individual patient-tailored therapies. In fact, whether direct management of the obstruction resolves the problem of apnea is also unclear due to commonality of a central sleep apnea component in most patients.
Invasive approaches to neuromodulation include vagal nerve stimulation to treat seizures and depression (Cyberonics), spinal cord stimulation to treat pain (such as Medtronic Inc., Boston Scientific Inc., Advanced Neuromed Inc.), direct deep brain stimulation to treat seizures (Medtronic Inc., Boston Scientific Inc.) or even cognitive disorders (Thync Inc.). However, these therapies target single components of the physiologic network for a bodily function, and are limited because they do not consider the remaining network. This may lead to suboptimal therapy, compensatory mechanisms that further diminish the efficacy of therapy, or unwanted effects. Moreover, these therapies are only as good as the accuracy of their specific targets, and brain/nerve regions are imprecisely defined for many bodily functions including sleep control, sleep-breathing conditions, cognition, alertness, memory, overall mental performance, or response to obesity.
Traditional therapies have also not typically been effective for managing central sleep apnea, other cognitive or performance functions, alertness, heart failure or obesity.
When devices are used for other functions, such as the increasing use of virtual environments, the goal is usually to create an illusionary or representative environment by feeding specific sensory inputs (primarily visual, tactile and/or auditory) to replicate existing real-world experiences. Unfortunately, such approaches may be limited in that normal pathways vary from individual to individual. Thus, simulating normal often may not accurately replicate that function for an individual nor represent normal for that individual.
In other situations, the use of devices to enhance or compensate for other functions such as motor tasks are limited or constrained. Constraint can take many forms and may be physical or functional. Physical constraints include an external obstacle preventing movement of a limb in an enclosed space such as may affect a warrior or scuba diver. A physical constraint may also be internal, such as loss of a limb from amputation. Functional constraints may include a classical disease, such as stroke that prevents an individual's ability to move the foot. However, functional constraints may also include underperformance on a task due to insufficient training, knowledge or acquisition of skills, or through disuse.
Many attempts have been made to address these constraints, using a familiar paradigm that body sensors (e.g., the eye), nervous function (e.g., the central and peripheral nervous system) and effector organs (e.g., a muscle group) can often be functionally mapped to specific anatomic locations. However, the precise locations of the brain or other physiological systems that control each task are not well defined. Such functional maps or “atlases” are often debated for complex functions. Much data has come from animal models that are not well suited to model or analyze complex human functions or mental functions.
It would be of great benefit to society to develop a device to enhance bodily functions by modulating its interfaced functional components. For instance, a device to restore sleep functionality, i.e., to prevent or treat central sleep disorders, would be of great value. Such devices may improve daytime performance in individuals without disease, or reduce symptoms in patients with disorders associated with central sleep apnea such as heart failure. It would also be of immense benefit to construct a device able to restore/enhance other functions such as motor activity or even some aspects of neural functioning without the need to define precise physiological, neural or other pathways to guide therapy. Currently, there are few methods in the prior art to achieve these goals.
The invention is designed to monitor and modulate a complex bodily function using a combined biological and machine approach. Unlike the prior art, the current invention uses machine learning to derive a robust relationship between sensed signatures of measurable body systems and bodily functions in animals and in particular human beings, but does not require presumed physiological relationships or mechanisms. The invention is then able to enhance performance of function or re-instate lost functions using this robust relationship or enciphered functional network.
For the purposes of this disclosure, the following definitions apply.
Associative learning is defined as the process of linking sensed signatures and other inputs with a body task. For this disclosure, body tasks are typically complex tasks rather than reflexive or other simple tasks. Associative learning may be iterative, such that associations are modified (“learned”) based upon patterns of change between these processes. For the purposes of this disclosure, this may be associating low impedance with abnormal breathing.
Bodily function is defined as the processes needed to perform a task, that may include physiologic or pathological processes. Examples include sleep, sleep apnea, mental performance, or the response to obesity. Bodily functions involve a network of several functional domains that often interact including the brain and central nervous system, peripheral nervous system, cardiovascular, pulmonary, gastrointestinal, genitourinary, immune, skin and other systems. A bodily function may result from biological activity/function, and may involve a non-biological or artificial component, e.g., reading with glasses, driving, using remote control unit, a patient moving a combined natural/cybernetic limb, etc.
Bodily signal means signals generated by and/or sensed from a human, animal, plant, bacterial or other single-cell-based body or multi-cell-based body. For purposes of this definition, viruses and prions are included. Bodily signals particularly include signals generated by and/or sensed from the human body. Bodily signals are generally associated with bodily functions. The term “non-bodily signal” indicates that it is generated from a source other than a single- or multi-cell-based body. Examples include an external “signal” from an external electrical source, machine, sensor, etc. When the term “signal” is used without the term “bodily” or “non-bodily”, the term “signal” indicates that it includes both “bodily signals” and “non-bodily” signals, i.e., it includes all signals.
Body means the physical structure of a single-celled organism, a multi-celled organism, viruses and prions. Organisms include animals (such as, but not limited to, humans), plants, bacteria, etc.
Effector is a means of performing a bodily task, and may include a physical effector such as for moving a limb or moving the diaphragm to enhance breathing during sleep, or an artificial effector such as a cybernetic limb or electrical stimulation to complete a task.
Effector response is the result of the effector, which may or may not complete a bodily task. For instance, if the effector is the triceps muscle in the arm, an effector response is to extend the arm by 30 degrees, while the entire task may be to fully straighten the arm.
Effector signal is the signal delivered by the invention to the effector to produce the effector response.
Functional domain is the aggregation of all the elements relating to a distinct bodily function, sometimes associated with a specific organ system or a combination of systems that results in the overall function, e.g., breathing. For a simple function, this may reduce to a sensed “dermatomal distribution”, for instance sensation at the shoulder is mediated by sensory nerves from the C435 distribution of the spinal cord. However, even such simple domains are more complex (and networked), in that shoulder sensation is mediated by nerves that also supply the heart. Functional domains include nerves, blood vessels, the lymphatic system, interstitial tissue planes and hormonal centers. Sensed signatures are measurable physiological parameters or indices, used individually or in combination from body systems above, that are linked with a body function and in aggregate describe that function.
Other definitions include a biological function, which means any function that is the direct result of natural biological activity such as breathing, heart beating, walking, running, sleeping, dreaming and so on.
A symbolic model herein is a mathematical representation of a function, linking measured sensed activity with a task even if complete physiological description for that task are lacking. It can also be termed a symbolic representation. This may include analog recorded physiological signals, digital coded ciphers, computer code, visual representations such as photographs or graphics, and so on, and can be used to aid in rapid, clear transformation to perform a specified method. Associative learning is an iterative process of linking processes, typically including sensed signatures and complex biological tasks, and modifying these associations (“learning”) based upon patterns of change between these processes (for instance, associating low impedance with abnormal breathing).
Enciphered network or enciphered functional network (EFN) is defined as a model associating measured parameters (sensed signatures) with aspects of the bodily task including effectors and other sensors. This enables monitoring and improved functionality of that body function. The enciphered network is designed to parameterize a functional domain, for example it links sensed activity with a task even if complete physiological or mechanistic description for that task is lacking. This departs from the traditional approach of meticulously mapping or recapitulating functions in each biological organ system. The network can be symbolic in the form of a symbolic representation such as symbolic code, in which case it may be a mathematical or other abstract representation. If applied to the nervous system, this can be termed an enciphered nervous system.
Encipher is defined as the process of coding information.
Enhanced performance or enhancement is defined as improvement to the normal healthy and non-diseased baseline function in an individual. Enhanced performance thus would not include therapies for disease such as pacing in an individual with abnormally slow heart rates or in a patient or an insulin pump in known diabetic patients.
External machine is defined as a mechanical, electrical, computational or other non-natural (native biological) device. This may be external to the body but can be in contact with or implanted within the body.
Extremity of the body is defined as limbs and associated structures of the body including arms, legs, hands, feet, fingers, toes, and subsegments thereof.
Functional domain is defined as the elements relating to a bodily task. This may include sensed elements, analysis elements and effector elements. Analysis elements may be “learned”, preprogrammed, reflexive, or passive. Each element may be biological, non-biological or artificial.
For example, a functional domain may sometimes be associated with a specific organ system or a combination of systems that results in the overall function, e.g., breathing. For a simple function, this may reduce to a sensed “dermatomal distribution”, for instance sensation at the shoulder is mediated by sensory nerves from the C435 distribution of the spinal cord. However, even simple domains may be more complex, in that shoulder sensation is mediated by nerves that also supply the heart. Functional domains thus include nerves, blood vessels, the lymphatic system, interstitial tissue planes and hormonal organs. Sensed signatures are measurable physiological parameters or indices from these organ systems, used individually or in combination that are associated with a body function and in aggregate describe that function.
Functionally associated is defined as sensed signals or functional domains that occur together when that function occurs. An example would be activity in portions of the brain controlling breathing with activity in muscles of breathing such as the intercostal muscles or diaphragm. Functional association does not need to follow biological pathways. For example, a functional association includes sensed activity in shoulder nerves with heart related problems such as angina, in which shoulder nerve activity is not part of the biological processes causing heart problems.
Machine learning is defined as a series of analytic methods and algorithms that can learn from and make predictions on data by building a model rather than following strictly static programming instructions. These machine learning approaches “learn” patterns and functions with at least some components that are not preprogrammed (i.e., instructed). In this sense, machine learning creates individualized solutions rather than generic ones. Machine learning can take many forms, including artificial neural networks, heuristics, deterministic rules and combined approaches.
Sensed signatures are defined as one or more signals from sensors related to a bodily task. Sensors may be biological, non-biological or artificial. Sensed signatures are inputs of the functional domain. Sensed signatures can be physiological parameters such as nerve firing rates and oxygenation level, that are associated with the function in question such as sleep disordered breathing.
Mental alertness is defined as an awake state that focuses on a specific desired task, that can be measured by performance at that task. Improved mental alertness is characterized by being awake and performing mental and other tasks well. Reduced mental alertness can include many states that include but are not limited to impaired performance of a task, “mental fatigue”, loss of focus, attention deficit, somnolescence, sleepiness, narcolepsy, sleep and disease processes that include the above as well as coma, “fugue” state and others.
“Task” means a piece of work, action or movement to be done, completed or undertaken. The term “bodily task” means a piece of work, action or movement to be done, completed or undertaken by a “body”, defined herein.
Therapeutically effective is defined as an effector function or dose of an intervention or therapy that produces measurable improvement in one or more patient outcomes. An example would be patterns of energy directed to the scalp to stimulate target regions controlling breathing, in order to treat central sleep apnea. Ideally, an intervention will minimize impact to other regions of the body, in this case the scalp which may be achieved by a small contact device rather than a cap that encompasses the entire scalp, or focusing energy from a non-contact device on the target region and not the entire head.
Other biological terms take their standard definitions, such as heart failure, tidal volume, sleep apnea, obesity and so on.
This invention creates an enciphered functional network. The potential number of uses of this invention are broad.
In one aspect, there is provided a method for interacting with the human body, the method including detecting bodily signals associated with one or more bodily functions at one or more sensors associated with the human body, processing the bodily signals to create one or more sensed signatures of the one of more bodily functions, processing the signatures using an enciphered functional network utilizing machine learning to determine one or more effector responses needed to control a bodily task, delivering via the enciphered functional network one or more effector signals (the effector signals based on the one or more effector responses), and controlling a bodily task.
In another aspect, there is provided a method to enhance performance of a bodily task, the method including detecting signals associated with the task at one or more sensors, processing the signals to create one or more sensed signatures, processing the signatures using an enciphered functional network to determine one or more effector responses needed to enhance performance of the bodily task, delivering via the enciphered functional network one or more effector signals (the effector signals based on the one or more effector responses), and enhancing performance of the task.
In another aspect, there is provided a method for treating a disease, the method including detecting signals associated with one or more bodily functions at one or more sensors associated with the human body, processing the signals to create one or more sensed signatures of the one of more bodily functions, processing the signatures using an enciphered functional network utilizing machine learning to determine one or more effector responses needed to treat a disease, delivering via the enciphered functional network one or more effector signals (the effector signals based on the one or more effector responses), and treating the disease.
In another aspect, there is provided a method for transforming nerve activity associated with one or more bodily functions, the method including detecting bodily signals of nerve activity associated with the one or more bodily functions at one or more sensors, processing the bodily signals to create one or more sensed signatures of the one or more bodily functions, processing the signatures using an enciphered functional network utilizing machine learning to determine one or more effector responses needed to transform nerve activity, delivering via the enciphered functional network one or more effector signals (the effector signals based on the one or more effector responses), and transforming nerve activity.
In another aspect, there is provided a method for controlling a device using an enciphered functional network, the method including detecting bodily signals from a body using one or more sensors, processing the bodily signals to create a sensed signature, processing the sensed signature using an enciphered functional network utilizing machine learning to determine one or more effector responses to control the device, delivering via the enciphered functional network one or more effector signals (the effector signals based on the one or more effector responses), and controlling the device.
In another aspect, there is provided a method to measure bodily function in an animal, the method including detecting bodily signals associated with sensory activation, processing the bodily signals to create one or more sensed signatures, and processing the sensed signatures using an enciphered functional network to determine one or more effector responses needed to enhance the bodily function of the animal.
In another aspect, there is provided a method of improving a specific human performance, the method including identifying one or more regions of a human body associated with parts of the brain that serve a specific function, placing low energy stimulating electrodes proximate to the one or more regions of the human body, applying stimulation through the electrodes to activate the parts of the brain, and measuring changes related to the parts of the brain to verify improvement of the specific human performance.
In another aspect, there is provided a method for treating a sleep disorder, the method including selecting one or more regions of a patient's central nervous system and/or peripheral nervous system associated with sleep functioning, and applying low energy stimulation through electrodes to activate the patient's one or more regions of central nervous system and/or peripheral nervous system to treat the sleep disorder.
In another aspect, there is provided a method of enhancing attention, the method including selecting one or more regions of a patient's central nervous system and/or peripheral nervous system associated with an attention disorder, and applying low energy stimulation through electrodes to activate parts of a patient's central nervous system and/or peripheral nervous system to treat the attention disorder.
In another aspect, there is provided a method of treating central sleep apnea, the method including identifying a target region from one or more local areas of the head and neck (the target region being functionally associated with one or parts of the brain that control sleep), and delivering a therapeutically effective amount of energy to stimulate the target region to treat the central sleep apnea, while minimizing stimulation of other regions of the body.
In another aspect, there is provided a method of modulating mental function, the method including identifying a target region selected from localized areas of the body (the target region being functionally associated with parts of the brain that govern the mental function), the mental function including one or more of alertness, cognition, memory, mood, attention and awareness, and delivering a therapeutically effective amount of energy to stimulate the target region to modulate the mental function, while minimizing stimulation of other regions of the body.
In another aspect, there is provided a system for interacting with the human body, the system including a processor and a memory storing instructions that, when executed by the processor, performs operations including detecting bodily signals associated with one or more bodily functions at one or more sensors associated with the human body, processing the bodily signals to create one or more sensed signatures of the one of more bodily functions, processing the signatures using an enciphered functional network utilizing machine learning to determine one or more effector responses needed to control a bodily task, delivering via the enciphered functional network one or more effector signals (the effector signals based on the one or more effector responses), and controlling a bodily task.
In another aspect, there is provided a system to enhance performance of one or more tasks, the system including a processor and a memory storing instructions that, when executed by the processor, performs operations including detecting signals associated with the task at one or more sensors, processing the signals to create one or more sensed signatures, processing the signatures using an enciphered functional network to determine one or more effector responses needed to enhance performance of the bodily task, delivering via the enciphered functional network one or more effector signals (the effector signals based on the one or more effector responses), and enhancing performance of the task.
In another aspect, there is provided a system to treat a disease, the system including a processor and a memory storing instructions that, when executed by the processor, performs operations including detecting bodily signals associated with one or more bodily functions at one or more sensors associated with the human body, processing the bodily signals to create one or more sensed signatures of the one of more bodily functions, processing the signatures using an enciphered functional network utilizing machine learning to determine one or more effector responses needed to treat a disease, delivering via the enciphered functional network one or more effector signals (the effector signals based on the one or more effector responses), and treating the disease.
In another aspect, there is provided a system to transform nerve activity associated with one or more biological functions, the system including a processor and a memory storing instructions that, when executed by the processor, performs operations including detecting bodily signals of nerve activity associated with the one or more bodily functions at one or more sensors, processing the bodily signals to create one or more sensed signatures of the one or more bodily functions, processing the signatures using an enciphered functional network utilizing machine learning to transform nerve activity, delivering via the enciphered functional network one or more effector signals (the effector signals based on the one or more effector responses), and transforming nerve activity.
In another aspect, there is provided a system to control a device using biological signals, the system including a processor and a memory storing instructions that, when executed by the processor, performs operations including detecting bodily signals from a body using one or more sensors, processing the bodily signals to create a sensed signature, processing the sensed signature using an enciphered functional network utilizing machine learning to determine one or more effector responses to control the device, delivering via the enciphered functional network one or more effector signals (the effector signals based on the one or more effector responses), and controlling the device.
In another aspect, there is provided a system to measure visual function in an animal, the system including a processor and a memory storing instructions that, when executed by the processor, performs operations including detecting bodily signals associated with sensory activation, processing the bodily signals to create one or more sensed signatures representing quantitative measures of sensation, and processing the sensed signatures using an enciphered functional network utilizing machine learning to determine one or more effector responses needed to enhance the bodily function of the animal.
In another aspect, there is provided a system for improving a specific human performance, the system including a processor and a memory storing instructions that, when executed by the processor, performs operations including identifying one or more regions of a human body associated with parts of the brain that serve a specific function, placing low energy stimulating electrodes proximate to the one or more regions of the human body, applying stimulation through the electrodes to activate the parts of the brain, and measuring changes related to the parts of the brain to verify improvement of the specific human performance.
In another aspect, there is provided a system for treating a sleep disorder, the system including a processor and a memory storing instructions that, when executed by the processor, performs operations including selecting one or more regions of a patient's central nervous system and/or peripheral nervous system associated with sleep functioning, and applying low energy stimulation through electrodes to activate the patient's one or more regions of central nervous system and/or peripheral nervous system to treat the sleep disorder.
In another aspect, there is provided a system to enhance attention, the system including a processor and a memory storing instructions that, when executed by the processor, performs operations including selecting one or more regions of a patient's central nervous system and/or peripheral nervous system associated with an attention disorder, and applying low energy stimulation through electrodes to activate parts of a patient's central nervous system and/or peripheral nervous system to treat the attention disorder.
In another aspect, there is provided a system to treat central sleep apnea, the system including a processor and a memory storing instructions that, when executed by the processor, performs operations including identifying a target region from one or more local areas of the head and neck (the target region being functionally associated with one or more parts of the brain that control sleep), and delivering a therapeutically effective amount of energy to stimulate the target region to treat the central sleep apnea, while minimizing stimulation of other regions of the body.
In another aspect, there is provided a system to modulate mental function, the system including a processor and a memory storing instructions that, when executed by the processor, performs operations including identifying a target region selected from localized areas of the body (the target region being functionally associated with parts of the brain that govern the mental function, including one or more of alertness, cognition, memory, mood, attention and awareness), and delivering a therapeutically effective amount of energy to stimulate the target region to modulate the mental function, while minimizing stimulation of other regions of the body.
One motivation for this invention is that detailed deterministic solutions for many complex bodily functions are inherently limited for therapy. This reflects several factors. First, there is inter-individual variation in regions of control—for instance, the biological neural network to talk in one person differs from the biological neural network to talk in another. For functions with a nervous component, this may represent the unique fashion in which higher cognitive functions and memories are shaped during growth and development in each person or potentially genetically established. Secondly, many brain functions are plastic—changes in the environment or disease can alter control regions. Changes can be gradual or abrupt, causing variations over years, months or even weeks that may reflect normal development, aging or dysfunction. This may explain why traditional therapies that are initially effective become ineffective over time. Thirdly, our conceptual knowledge of functional domains in the central and peripheral nervous system is in its infancy. It is thus a major challenge to understand a bodily function using a classical paradigm of observing then stimulating a specific target to map its region(s) of control.
Several innovations separate this invention from the prior art. First, the invention creates an enciphered network for the bodily function. This reconstructs the function as network of functional domains. This departs from the traditional approach of meticulously mapping or recapitulating each biological organ system. Instead, second, it identifies sensed signatures and effectors for each function. Signatures can be nervous or non-nervous system related. Third, the invention applies a feedback loop to apply measured and quantitatively determined therapy that may change over time even in the same individual. This is inherent because the enciphered network can be trained over time using an ongoing machine learning processes. Fourth, the core logic of the invention is patient-tailored, distinct from the majority of current devices that use “one-size-fits-all”, generic or stereotypic therapy. Fifth, therapy is adaptive through continued machine learning, such that a similar abnormality in the same individual may produce different signatures and/or require different effector responses at two or more distinct periods in time. Sixth, certain embodiments of the device combine biological and non-biologic devices, together or individually. The enciphered representation can accommodate novel signatures over time, that can be extrinsic artificial signals as well as intrinsic biological ones. Therapy can ultimately be delivered by an external device and/or by direct stimulation or inhibition of an effector. Embodiments include improvements of sleep apnea, the body's response to heart failure, obesity, alertness, memory and mental performance or cognition.
The concept of accessing functional domains for a task by measuring from or stimulating an interconnected region of the network, that may be neural, vascular or other, is novel at several levels and has not been addressed by devices in the prior art. One example way to better understand this concept is by considering the disease of central sleep apnea.
The functional domain for central sleep apnea in this invention includes sensed signatures of brain function (measurable on the EEG), reduced oxygenation levels and increased carbon monoxide levels in the blood (measurable from skin sensors), and in some individuals increased heart rate and altered patterns of heart rate and other less defined functions. Observed but unexplained signatures may include nocturnal rostral fluid shift from the legs (that may link sleep apnea with heart failure). Effector responses for central sleep apnea is to nerve function to the neck muscles, diaphragm, intercostal muscles and accessory muscles (measurable by nerve firing rates). The present invention will use these sensed signatures of brain or nerve activity, chest wall movement, bioimpedance at the skin (to assess for a rostral fall), or oxygenation for diagnosis and monitoring. In an embodiment for treatment, the invention may result in varying effector responses.
Chest wall impedance can be expressed in many forms. In this invention, the sensed signature of abnormal chest wall impedance includes a ratio of lower body impedance (e.g., leg, lower back) to higher body impedance (neck and chest)—i.e., higher impedance in lower body (less extracellular water), lower impedance in upper body (more extracellular water). This could also be expressed as upper-to-lower body conductance. This could also include measuring impedance to different forms, patterns, or waveforms of electrical energy.
In one preferred embodiment, machine learning is used to associate repeatedly measured signatures with normal breathing. If apnea arises during sleep, the invention will apply tailored therapy adaptively to alter domain activity and alleviate sleep apnea. The response to therapy (e.g., effector response) can also be assessed repeatedly via sensed signatures, and the therapy can be withdrawn or continued based upon these signatures. This differs from the prior art in which therapies such as continuous positive airway pressure or nerve stimulation are often delivered empirically, continuously or in predetermined fashions without adaptive algorithms to tailor therapy.
Other sensed signatures include altered nerve firing rates for mental performance or sleep, vasodilation during sleep, reduced skin galvanic resistance (from altered electrolytes or edema) in the body's response to heart failure or sleep-breathing disorders, altered skin absorption or emission of near-infrared or other components of the electromagnetic spectrum during sleep disorders, measured alterations to other forms of applied non-electrical energy including optical signals (altered reflectance), sound or ultrasound (different sonic reflectance and scattering), and potentially altered spectroscopic signals of body chemistry that can be sensed.
The network of functional domains is a unique approach for interfacing with bodily functions. For instance, a patient with heart pain (angina pectoris) or a heart attack (myocardial infarction) often experiences “radiated pain” to the left arm, shoulder or other regions. Some patients experience only arm pain from cardiac ischemia—i.e., arm pain is a sensed signature. This signature may not be relevant to other individuals a priori—but can be learned by the enciphered network for that individual. In this way, the invention can now detect nerve activity in the arm below the typical nerve firing rates for sensed “pain”, providing the device with an early warning sensor for heart pain (“angina”) to provide therapy or alert medical personnel.
In another example, patients with problems of the abdominal viscera (stomach, small intestine, large intestine) that may include normal “indigestion” as well as diseases often experience vague discomfort on the abdominal wall through imprecisely defined and variable visceral and somatic nerves. Massaging this region is an example of counter-stimulation that can alleviate the visceral organ pain. In one embodiment, the invention will thus provide algorithmically determined vibratory stimulation to appropriate skin regions within the “functional domain” of the bodily function to alleviate pain.
As yet another example, nerve firing in cutaneous or other accessible nerves (e.g., mucous membranes of mouth, anus, or skin of the external auditory meatus) may share neural control regions with other organs, such as heart pain or even abnormal heart rhythms. Effector signals can be delivered to specific regions of the functional domain to alleviate heart pain or other abnormalities. Other components of a functional domain may include blood vessel flow, vasomotor reactivity, skin electrical conductivity, heart rate or heart rate variations, breathing rate, cellular edema and other indices illustrated throughout the specification.
Therapy is individually tailored and not empirically delivered. Baseline signatures such as rates and patterns of nerve firing during a desired level of functioning are analyzed and learned in each individual and may be combined with other signatures within the enciphered functional network. In states such as sleep-disordered breathing, heart failure, fatigue and others, fluctuations outside this normal range are detected and can be used to monitor disease or performance. Therapy such as stimulating neck muscles for obstructive sleep apnea, stimulating accessory muscles or alertness centers for central apnea, or therapy for heart failure and other conditions can be monitored (e.g., by effector response) and tailored to machine learned signatures. Functionality can thus be modulated without direct knowledge or access to the primary physiological target and without detailed pathophysiological knowledge of that function.
Nerve signatures may be shared between many functions, e.g., based on dermatomal distributions of peripheral nerves. One example is sensation of the tip of shoulder blade at the “C234” region, control of deltoid muscle function by the “C56” region, and control of the diaphragm muscles and hence breathing at the “C345” region. Thus, sensation at the shoulder can indicate shoulder stimulation, or pain in portions of the heart adjacent to the diaphragm. Stimulation at these regions by direct electrical stimulation, vibratory stimuli, heat or other can produce a counter-irritant to the measured function.
Brain signatures can be assessed directly via the EEG or simplified EEG measured from the scalp by many types of electrical sensor. For instance, scalp activity in the alpha (7.5-12.5 Hz), beta (12.5-30 Hz) or gamma (25-40 Hz) bands indicate states of awakeness (wakefulness) or heightened or alertness; activity in the delta (0.1-3 Hz) or theta (4-7 Hz) bands indicate drowsy (or comatose) states. Depending on sensed activity, interventions can be applied to the scalp or other domains of the network while monitoring alpha, beta or gamma signatures to facilitate alertness. In each case, the invention is novel in that it derives patient-tailored signatures for a given function using machine learning, and will apply interventions algorithmically in a tailored feedback loop. In one preferred embodiment this will enhance sleep function.
Peripheral nerve signatures are numerous. For instance, increased nerve firing of the cervical sympathetic plexus in the head and neck may be associated with alertness or rapid eye movement (REM) sleep, and reduced activity may be associated with drowsiness or stages I-IV of sleep. Stimulation of those regions of the head and neck can be used to increase alertness. Increased firing of the accessory (XI), facial (VII) or other cranial nerves may indicate impending obstructive sleep apnea, and may provide targets for therapy.
There are several non-nerve domain signatures. For instance, deoxygenation of an oxygen sensor on the skin of a finger (via optical reflectance or plethysmography) can indicate hypopnea or apnea. Increased skin temperature or blood flow (absorption in red wavelengths on an optical sensor) may occur in stages I-IV sleep from parasympathetic activation. Novel skin sensors can detect changes in biomarkers such as glucose (to detect diabetic states, need to eat), INR (a test of blood thickness for some patients on blood thinners) and a new generation of sensors for drugs in the blood stream, chemical changes on the skin and so on, Interpretation of these signatures can be troublesome but is linked in this invention by machine learning to a specific function, e.g., fever increases skin temperature, but is accompanied by increased breathing rate and altered skin biochemistry/impedance (due to perspiration). By learning based on multiple signatures, temperature information can be used in this case to distinguish changes in breathing rates due to fever from that due to central sleep apnea.
This invention adapts to concepts of neural plasticity. Plasticity refers to alterations in the pathways of nerves and connections (synapses) from changes in behavior, environment, neural processes, thinking, and emotions, and also to changes resulting from injury. This concept has replaced prior teachings that the brain and nervous systema are static organs. New studies show that the brain changes in anatomy (structure) and physiology (functioning) over time. There are several examples, e.g., DARPA limb projects, stroke victims recovering function after months or years of physical or occupational therapy despite having infarcted the traditional brain areas for the target function. Plasticity is also observed in peripheral nerves, for instance the distribution of a functioning nerve (dermatome) can expand into an adjacent distribution of diseased nerve supply. In other words, a singular function can be assumed or subsumed by different regions of the central or peripheral nervous system, that will also have non-neural implications, e.g., on measured blood flow, galvanic skin resistance or other physiological parameters.
This invention uses the principle that continuous machine learning will enable its functionality to be retained even when plasticity occurs, and again without precise precise physiological mapping knowledge for that function. For instance, in classical Pavlovian training, dogs were trained to salivate when exposed to non-food-related stimuli that had previously been associated with food in training. In other words, a new trained stimulus—function interaction—was used without knowledge of detailed physiological linking for that function.
This invention also encompasses personalized learned feedback loops, to modulate a desired bodily function by algorithmic machine learning analogous to classical Pavlovian conditioning. In a training mode, stimulation is applied during normal periods—for instance, vibratory stimulation of the skin of the lower back on days of anticipated restful sleep. Subsequently, if sleep is interrupted, trained modes of stimulation are applied. This mode can be applied to various bodily functions including but not limited to alertness, memory, sleep and sleep-disordered breathing.
The present invention identifies functional domains empirically, and provides computational customized, individualized solutions. This differs from the prior art in which, for example, preferred embodiments for sleep disordered breathing stimulate cranial nerves (e.g., trigeminal or hypoglossal), but through unclear mechanisms that may in fact inadvertently work by training certain responses or stimulating other regions than those intended.
In another set of preferred embodiments, the enciphered network can be used to enhance body performance in non-disease states. One direction is to utilize unused body capacity. In daily western life, humans often underuse torso, leg and arm sensors and effectors yet frequently use eye and hand sensors and effectors. Stimulation of underused regions by a device can extend the sensory capacity (bandwidth) of an individual. When combined with artificial sensors, these underused regions can also be used to provide a “sixth sense” (see drawings) to extend sensation to biologically unsensed stimuli (e.g., a carbon monoxide sensor can provide vibratory stimuli to unused portions of the body), to train the body (e.g., improve alertness) or other function.
Enhancement of performance may require specific stimulation patterns that vary based on frequencies, amplitudes and sites of stimulation. This information can be derived by machine learning of sensed signatures or patterns in each individual. Another approach is to use patterns from individuals who are highly functioning in that desired modality—from a de-identified database, by crowd sourced data collection from wearable devices or by other means. These patterns can then serve as inputs for machine learning algorithms in the enciphered network, that will interface them to the symbolic representation for an individual to tailor them appropriately.
Effector stimulation should avoid inadvertent recruitment of existing bodily functions by applying non-physiological or atypical physiological stimuli. This can be achieved by using neural frequencies or patterns that are not part of normal processing or pathways, such as outside the normal sensed frequency, or with a different pattern, or at a different (lower) amplitude. Using other examples in this disclosure, the invention may detect subclinical nerve firing in the functional domain for cardiac ischemia as an early warning for angina, or application of subclinical amplitudes of nerve stimulation to the accessory muscles to stimulate breathing (for central sleep apnea) or neck (to improve alertness). These safeguards will avoid invoking behavioral change, sensation by the brain and/or changing memory of an event (Redondo et al., Nature 2014).
The invention can work with several types of sensors individually or in combination. Examples include solid physical sensors such as FINE (singularityhub.com/2013/07/24/dapas-brain-controlled-prosthetic-arm-and-a-bionic-hand-that-can-touch/), traditional ECG- or EEG- electrical sensors, non-solid sensors such as electrostatic creams, sensors for bioimpedance, piezoelectric film sensors, printed circuit sensors, photosensitive film, thermosensitive film, and external-oriented sensors not in contact with the body such as video, IR, temperature, gas sensors, as well as other sensors. These sensors detect stimuli and transduce the information through a constructed/created (non standard or non-somatotopic) path to active nerves.
Processing elements include a digital signal processor to interface with output elements that can stimulate different parts/nerves of the body, or cause mechanical action in an external machine. Such elements could include traditional computing machines with integrated circuits in isolation or networked (e.g., cloud computing), biological computing, integrated biological/artificial devices (cybernetic) or utilizing unused biological capacity to perform specific, directed tasks. One potential embodiment is to use unused computational capacity of the central nervous system to perform pattern recognition in lieu of programming an artificial computer for this purpose. This can be accomplished by training an individual to recognize a visual/auditory/olfactory or other sensation and then sensing the sensed signature of that evoked response when that stimulus is subsequently encountered.
Effector elements can include direct electrical outputs, piezoelectrical devices, visual/infrared or other stimulatory systems, nerve stimulating electrodes or servo motors to control a limb, digitized electronic signals such as radiofrequency or infrared transmissions, or even virtualized data such as avatars in a virtual world interface or elements in a large database that can be queried, as well as other effector elements now existing or yet to be developed.
Applications of effector elements can be for diagnostic purposes such as detecting stimuli or body functions (e.g., visual function, visual disease progression, mood, alertness, detecting injury such as traumatic brain injury, cardiac electrical and/or mechanical function, subclinical seizure detection), detecting external world situations or environments without subjecting the human body to discomfort (e.g., sensing heat in a fire, detecting oxygen or toxic gas content in the external environment such as a mine).
Effectors can be applied for medically related therapy such as brain related function (e.g., brain stimulation for patients with sleep disorders or central apnea, biofeedback for stroke rehabilitation, deep brain stimulation for motion or seizure disorders), other neurological diseases (e.g., substituting artificial sensor data in patients with peripheral neuropathy, biofeedback stimulation of muscles), cardiac disease (e.g., arrhythmias treated with implanted devices, cardiac function improved with mechanical or electrical devices), response to obesity, or other organ disease modified with directed electrical or mechanical elements.
Applications of machine learned therapy using this invention can be for training, learning and performing of physiological activities or mechanical, non-physiologic functions. Unlike the prior art that applies non-specific stimulation, e.g., transcranial direct current stimulation (ref: www.scientificamerican.com/article/amping-up-brain-function), the present invention can sense, machine learn, optimize, and then deliver specific therapy modulated via a feedback loop. This will provide tailored therapy to modify many complex functions.
Other applications for this invention include improving athletic performance after injury (e.g., direct stimulation to muscles to regain lost function, biofeedback to maintain heart rate within desired range during controlled exercise, brain stimulation), enhancing sensory perceptions (e.g., artificial visual sensors for facial recognition, artificial auditory sensors to detect previously inaudible information), performing tasks in non-typical ways by overcoming constraints or developing more efficient solutions (e.g., driving a car with small finger movements or eye motion amplified by artificial device, controlling an external device biologically, e.g., small eye or limb movements to control a computer interface). Examples of mechanical functions include biological operation of a mechanical exoskeleton for soldiers, performing tasks too difficult or dangerous for humans such as deep sea exploration, armed combat, or basic tasks such as controlling a computer, video games or remote controls.
In summary, the invention incorporates a combined biological-artificial network, referred to as enciphered network (or representation), to modulate specific tasks (such as complex bodily functions often requiring brain or nerve involvement, or higher cortical functions). Sensors (biological or artificial) sense the activity of the measured task. This sensed activity is enciphered, and then machine learning algorithms and specific hardware modulate the network using biological, artificial or hybrid effectors (e.g., stimulating electrodes). The network can directly augment a function (e.g., sleep), or form a new function via existing elements (“retasking” a function, e.g., associating lower back stimulation with sleep).
The enciphered network can operate internally using symbolic internal representation specific to each task. Specific representations of each task may be important because the pattern, frequency, and amplitude of stimulation differ considerably between tasks—e.g., modulating electrical activity on the scalp versus the neck or other parts of the body, or stimulating neural elements versus blood vessels.
Some embodiments are illustrated by way of example and not limitation in the figures of the accompanying drawings, in which:
A system and method for enhancing and modifying complex functions of the body are disclosed herein. In the following description, for the purposes of explanation, numerous specific details are set forth in order to provide a thorough understanding of example embodiments. It will be evident, however, to one skilled in the art, that an example embodiment may be practiced without all of the disclosed specific details.
The invention modulates and enhances complex and higher bodily functions by modulating a series of functional domains. Typically, the complex function will include a component of brain or nervous activity. One innovation is the creation of an enciphered (symbolic) representation to model the complex function. Such a representation model may also be called a network, and is learned in this invention. This is created by, then used to interpret sensed signals from functional domains that comprise the function. The enciphered network is then used to effect change. In one preferred embodiment, this is applied to improve sleep apnea, but other embodiments modulate heart failure, obesity, alertness, mood, memory and mental performance or cognition.
The biological sensors 104 can sense biological signals, from an individual, from another individual, or from a database of signals 118. The biological sensors 104 can be wearable.
External sensors 110 can sense biological signals, from an individual, from another individual or from a database of signals 118. In turn, signals may arise from the central nervous system, peripheral nervous system, cardiovascular system, pulmonary system, gastrointestinal system, genitourinary system, skin or other systems.
External sensors 110 can provide many types of information including, but not limited to, those normally sensed including pressure/physical movement (tactile, touch sensation), temperature (thermal information, infrared sensing), chemical (galvanic skin resistance, impedance, detection of specific ions from the skin), sound (auditory sensation), electromagnetic radiation in the visible spectrum (visual sensation), movement (a measure of muscle function and balance).
External sensors 110 can also provide information related to normal sensation but that is not normally sensed including, but not limited to, the invisible electromagnetic spectrum (such as gamma radiation, X-rays, radiowaves), sound waves outside the normal physiological range for humans (roughly 20 Hz to 20 kHz) but including the range sensed by animals (for instance, dogs can sense higher frequencies).
External sensors 110 can provide information outside normal sensory modalities including, but not limited to, toxins such as carbon monoxide or excessive carbon dioxide, forms of radiation (such as alpha and beta radiation), biotoxins such as toxins of Escherichia coli bacteria associated with food poisoning (type 0157), anthrax or other agents. Clearly, such information would be of value for military and security applications.
In
In
Machine learning algorithms of the enciphered functional network are enabled using artificial intelligence (autobot, fuzzy logic circuits). This can be done via neural networks (e.g., 3 layer back-propagation networks or other designs), techniques of deep learning, heuristics, linear classifiers or other forms of fuzzy logic. An important feature of such systems is that they do not need to know much about the specifics of human pathophysiology, but need to learn information about factors influencing behavior that is provided by the sensed signatures. They are thus well suited to the problem of complex bodily functions that are often incompletely defined or mapped pathophysiologically. This is not structured by theoretical “textbook” classification schemes. Certain elements of the system can be layered as rule-based, using for instance deterministic solutions from a database such as the dermatomal distribution of a nerve in the shoulder or the fact that some fluctuations in skin oxygenation reflect heart rate.
Thus, the symbolic model of simple and complex functions is akin to representing something that is visualized by an “impressionist” painter rather than a detailed physiological representation—by one trained in the “realist” school. Again, this approach is based largely on the premise that in addition to the primary physiological systems required for a task, that is difficult to precisely define, secondary networked regions become involved.
Machine learning nominally links signatures with normal function 250 in order to create a patient specific range to detect abnormal function 290 as outliers. In practice, the best results are obtained when the machine learning algorithms perform repeated pattern classification interactions 255 between sensed signatures for normal 250 and abnormal 290 functions. This interconnectivity is necessary, but its complexity makes the system ideally suited for a computational machine learning paradigm to modify and treat the networks 235.
In
Numerous other nerve-related parameters are possible, e.g., nuclear scans of neuro-tissue function, e.g., MIBG scanning for autonomic ganglia, metabolic quantification using positron emission tomography based sensor information, serum levels of norepinephrine and other nerve-related signatures familiar to one skilled in the art.
Non-nerve signatures 335 represent other modalities 340 that are not primarily in the nervous system. Represented modalities have one or more defined signatures, e.g., hypervolemia is detectable by reduced electrical impedance of tissue, sympathetic activation via “clammy skin”—reduced galvanic skin resistance and altered ionic composition, apnea via reduced oxygenation measurable as reduced skin absorption in the near-infrared end of the electromagnetic spectrum. These signatures also possess information on location 345, rate 350 and temporal patterns over time 355. Numerous other parameters are currently possible and may develop over time and be incorporated into this invention, e.g., tissue concentrations of neurohormones such as B-type natriuretic peptide or prolactin from a pharmacological sensor, signal intensity from a photodetector to detect drug concentrations in skin or cutaneous blood vessels, drug or alcohol levels in exhaled breath from an oropharyngeal sensor, drug or alcohol levels in urine from a penile sensor, and other sensors relevant to the functional domain under consideration.
The network of sensed signatures exemplified in
Note that not all possible measured signatures are needed for the invention to work—in simple clinical practice, heart failure can be monitored quite well from the simple measure of weight gain alone; this invention uses machine learning to mathematically weight the most important signature but also to use information from whatever is currently available.
Modulation of nerve-related domains 410 can be linked to modulation of non-nervous domains by modulation connection 415. Moreover, the central and peripheral nervous network 420 can be linked to the non-nervous system physiologic network by network connection 425.
The invention can also use external sensors (
More specifically,
In measuring the actual physiology of sleep-disordered breathing in an individual 1200, biological signals are sensed 1205. These include biological signals of control regions 1210 including activation of the amydala and other parts of the limbic system that control alertness, wakefulness and relate to sleep. These signals have scalp representations that can be detected by skin nerve sensors, but can also be detected by medical devices such as the BOLD signal from functional magnetic resonance imaging, or metabolic images from positron emission tomography in medical applications. Physiologically, sleep is also triggered from intrinsic but natural signals such as darkness, sound (e.g., soothing music or the sound of waves), tactile sense (e.g., massage of parts of the body). The intrinsic sleep control regions of the brain 1210 then integrate these inputs with sensors related to breathing including low oxygenation, measureable in the fingertips 1225, that stimulates breathing, and stimulation of the diaphragm 1220 to enable ventilation of the lungs.
The schematic shown in the left panel of
In the right panel of
The computational element 1255 forms a symbolic relationship between sensed signals and biological function (e.g., elements 250-290 in
The symbolic relationship of the enciphered network in
In the simplest case, the enciphered symbolic relationship is a matrix in which a signal X causes a function Y; for instance, a noxious stimulus such as pain sensed by a sensor/sensory nerve in the leg (X) causes activity in a motor nerve causing withdrawal of that leg (Y). This function is not represented in the device based upon a detailed neurophysiological representation of leg sensation (in the primary somatosensory cortex, in the post-central gyrus), or the precise nerves that control the leg. Instead, this function is mapped empirically—sensation on any nerve associated with the painful stimulus can result in actions leading to leg withdrawal.
The advantage of this approach is that it can analyze the multiple effects of a particular stimulus. For instance, an acute painful stimulus often produces activation on nerves remote from the original site of stimulation. Hence, pain in the leg, that may be inaccessible, may be detected from nerve activity quite distant from the sensation such as the chest wall, that may be more accessible.
In
Moreover,
Item 1310 applies the symbolic model of the enciphered network, as identified in
As described above, the symbolic model uses machine learning to map sensor input to normal and abnormal function of that bodily functionality. This comprises training sets of different patterns for that individual, that are both personalized and continuously adaptive.
In
In
Furthermore, the invention can 1325 artificially generate signals needed to stimulate the muscle. Since the frequency and amplitude of nerve activity that controls a muscle lies within a range for each individual, the enciphered network can simulate the nerve activity controlling the quadriceps femoris muscle and deliver it programmatically to regions of the skin associated with contraction and relaxation of that muscle for that individual (part of the functional domain). This can be used when the nerve is degenerated or anesthetized (for instance, to prevent pressure ulcers in patients on prolonged ventilation). It can also be used for performance enhancement—for instance, to perform isometric exercises during rest or sleep to prevent or reverse muscle atrophy, or to improve muscle function or increase metabolic rate to lose weight.
In
In
Another embodiment of performance improvement (step 1345) is to increase alertness. Stimulation of the scalp in the temporal region and other function-specific zones can increase brain activity in these regions. The invention tailors stimulation to the enciphered representation of awakeness (i.e., alertness). As a corollary, drowsiness can be detected by the enciphered network and used in a feedback loop to trigger low intensity stimulation by a cutaneous device elsewhere on the body. This has several applications, including detecting and trying to prevent drowsiness while driving, in the intensive care unit during pre-comatose states or during drug-overdoses, as a monitor for excessive alcohol or medication ingestion, or during excessive fatigue states, e.g., in the military.
Sensors can detect alertness versus drowsiness from large groups of neurons using electroencephalography (EEG) over a wide range of frequencies. EEG signals have a broad spectral content but exhibit specific oscillatory frequencies. The alpha activity band (8-13 Hz) can be detected from the occipital lobe (or from electrodes placed over the occipital region of the scalp) during relaxed wakefulness and increase when the eyes close. The delta band is 1-4 Hz, theta from 4-8 Hz, beta from 13-30 Hz and gamma from 30-70 Hz. Faster EEG frequencies are linked to thought (cognitive processing) and alertness, and EEG signals slow during sleep and during drowsiness states such as coma and intoxication.
In
Step 1450 in
Several embodiments exist. In step 1460, the invention uses a biological signal to control an external device (e.g., motor nerve control of a prosthetic limb), or an external signal to control a biological function (e.g., external signal stimulation of a skeletal muscle). As described, skeletal muscle is typically stimulated by nerve activity at a frequency of 7-15 Hz (varying with precise nerve distribution, see Dorfman et al. Electroencephalography and Clinical Neurophysiology, 1989; 73: 215-224). Such external stimulation can improve muscle strength by stimulating it, and would enable performance improvement of, e.g., programmable improvement in leg muscle function. Another example is to treat central sleep apnea, using an external sensor of oxygen desaturation (“desat”) to activate a device that stimulates the phrenic nerve and hence the diaphragm. This may have substantial clinical implications.
In
In
In
In step 1525, a signal multiplexor links the intrinsic or extrinsic signals in order to control the desired programmed function. In step 1530, this is achieved to enhance biological leg function (e.g., via cutaneous/direct electrical stimulation as described). In step 1535, this is performed to control a prosthetic limb.
The invention summarized in
The invention outlined in
The current invention extends known interventions based upon cortical plasticity. For instance, it is known that the dermatomal distribution of a functioning peripheral nerve expands when an adjacent distribution is served by a diseased nerve. In other words, the same function can now be served by different regions of the central or peripheral nervous system.
The invention also substantially extends normal plasticity—by programming desired and directed regions of the body to sense and effect functions normally reserved for other regions of the body that are currently inaccessible (e.g., in military combat) or unavailable (e.g., due to disease).
The invention also substantially advances normal plasticity by integrating external sensors (e.g., for normally inaudible sound frequencies or sensations) or devices (e.g., prosthetic limbs, other electronic devices) into the ENS.
In
Nomograms of the detrimental impact of sensed signals are used to calibrate sensing and delivery of therapy functions from the enciphered nervous system. For carbon monoxide, exposures at 100 ppm (0.01%) or greater can be dangerous to human health. Accordingly, in the United States, Federal agencies such as OSHA put a highest limit on long-term workplace exposure levels of 50 ppm, but individuals should not be exposed to an upper limit (“ceiling”) of 100 ppm. Exposures of 800 ppm (0.08%) lead to dizziness, nausea, and convulsions within 45 min, with the individual becoming insensible within 2 hours. Clearly, detecting this toxin early would have extremely practical implications in industrial environments, for instance. Other nomograms can be developed to identify thresholds for “safe” versus “actionable” exposure to various stimuli including but not limited to chemicals, biological toxins, radiation, electrical stimuli, visual stimuli and auditory stimuli.
Steps 1910 and 1915 link the pattern (e.g., a face) to the biological sensed response—for instance, activity of nerves in the scalp over the parietal lobes of the brain, or over the forehead indicating “recognition”. This is used to create the elements of enciphered nervous system for this task (step 1920). This will be personalized, but can also take inputs from a multi-person (population, crowd-sourced) enciphered nervous system. Once this link has been made, then presentation of the pattern will result in a “sensed” biological pattern, which is used by the multiplexer or control logic in step 1925 to deliver a “1” (recognized) or “0” (not recognized) to control a device (step 1930) (e.g., external computer classifier) or stimulate the individual via a surrogate sensation (step 1935) (e.g., vibration at the left upper arm if a recognized pattern is detected). Uses for this invention include pure biocomputing (pattern recognition of familiar or abstract shapes/codes), formally encoding and enhancing memory of faces for a particular person, and security such that only a hostile pattern/face elicits a specific surrogate sensation or activates a device. One other advantage of this approach over waiting for a cognitive recognition of the pattern is that this can function as a “background process” and/or provide faster pattern recognition.
Thus, this invention can improve and enhance function of traditional senses, if a device is used that integrates sensors that sense outside the normal physiological range can be used to enhance the range of normal physiological sensation. For instance, sensing signals in the “inaudible to humans” part of the frequency spectrum, transducing the signal to the audible range, and transmitting it via bony conduction using a device could be used for private communication, encryption, recreational or other purposes. Medically, this invention could be used to compensate for hearing loss. This same invention with sensors of vibration could be used to compensate for loss of this sensation in certain neurological diseases such as peripheral neuropathy, by transmitting this sensation to an intact sensation in a different part of the body.
Important safety issues must be raised at this stage. While no untoward, dangerous or otherwise undesired functionality has been observed with this invention, certain limits must be imposed. First, no stimulation intensity provided by the device can reach painful or dangerous levels. Second, no sensory input can be allowed to reach disturbing or undesired levels. Third, any sensor or device (effector) should have acceptable and tested safety profiles.
In operation as described in
The computer system 2000 may be implemented as or incorporated into various devices, such as a personal computer (PC), a tablet PC, a personal digital assistant (PDA), a mobile device, a palmtop computer, a laptop computer, a desktop computer, a communications device, a control system, a web appliance, or any other machine capable of executing a set of instructions (sequentially or otherwise) that specify actions to be taken by that machine. Further, while a single computer system 2000 is illustrated, the term “system” shall also be taken to include any collection of systems or sub-systems that individually or jointly execute a set, or multiple sets, of instructions to perform one or more computer functions.
As illustrated in
In a particular embodiment, as depicted in
In an alternative embodiment, dedicated hardware implementations, such as application specific integrated circuits, programmable logic arrays and other hardware devices, can be constructed to implement one or more of the methods described herein. Applications that may include the apparatus and systems of various embodiments can broadly include a variety of electronic and computer systems. One or more embodiments described herein may implement functions using two or more specific interconnected hardware modules or devices with related control and data signals that can be communicated between and through the modules, or as portions of an application-specific integrated circuit. Accordingly, the present system encompasses software, firmware, and hardware implementations.
In accordance with various embodiments, the methods described herein may be implemented by software programs tangibly embodied in a processor-readable medium and may be executed by a processor. Further, in an exemplary, non-limited embodiment, implementations can include distributed processing, component/object distributed processing, and parallel processing. Alternatively, virtual computer system processing can be constructed to implement one or more of the methods or functionality as described herein.
It is also contemplated that a computer-readable medium includes instructions or receives and executes instructions 2020 responsive to a propagated signal, so that a device connected to a network 2024 can communicate voice, video or data over the network 2024. Further, the instructions 2020 may be transmitted or received over the network 2024 via the network interface device 2008.
While the computer-readable medium is shown to be a single medium, the term “computer-readable medium” includes a single medium or multiple media, such as a centralized or distributed database, and/or associated caches and servers that store one or more sets of instructions. The term “computer-readable medium” shall also include any medium that is capable of storing, encoding or carrying a set of instructions for execution by a processor or that cause a computer system to perform any one or more of the methods or operations disclosed herein.
In a particular non-limiting, example embodiment, the computer-readable medium can include a solid-state memory, such as a memory card or other package, which houses one or more non-volatile read-only memories. Further, the computer-readable medium can be a random access memory or other volatile re-writable memory. Additionally, the computer-readable medium can include a magneto-optical or optical medium, such as a disk or tapes or other storage device to capture carrier wave signals, such as a signal communicated over a transmission medium. A digital file attachment to an e-mail or other self-contained information archive or set of archives may be considered a distribution medium that is equivalent to a tangible storage medium. Accordingly, any one or more of a computer-readable medium or a distribution medium and other equivalents and successor media, in which data or instructions may be stored, are included herein.
In accordance with various embodiments, the methods described herein may be implemented as one or more software programs running on a computer processor. Dedicated hardware implementations including, but not limited to, application specific integrated circuits, programmable logic arrays, and other hardware devices can likewise be constructed to implement the methods described herein. Furthermore, alternative software implementations including, but not limited to, distributed processing or component/object distributed processing, parallel processing, or virtual machine processing can also be constructed to implement the methods described herein.
It should also be noted that software that implements the disclosed methods may optionally be stored on a tangible storage medium, such as: a magnetic medium, such as a disk or tape; a magneto-optical or optical medium, such as a disk; or a solid state medium, such as a memory card or other package that houses one or more read-only (non-volatile) memories, random access memories, or other re-writable (volatile) memories. The software may also utilize a signal containing computer instructions. A digital file attachment to e-mail or other self-contained information archive or set of archives is considered a distribution medium equivalent to a tangible storage medium. Accordingly, a tangible storage medium or distribution medium as listed herein, and other equivalents and successor media, in which the software implementations herein may be stored, are included herein.
Thus, a system and method of identifying a source of a heart rhythm disorder, by identification of rotational or focal activation in relation to one or more spatial elements associated with the source of the heart rhythm disorder, have been described. Although specific example embodiments have been described, it will be evident that various modifications and changes may be made to these embodiments without departing from the broader scope of the invention. Accordingly, the specification and drawings are to be regarded in an illustrative rather than a restrictive sense. The accompanying drawings that form a part hereof, show by way of illustration, and not of limitation, specific embodiments in which the subject matter may be practiced. The embodiments illustrated are described in sufficient detail to enable those skilled in the art to practice the teachings disclosed herein. Other embodiments may be utilized and derived, such that structural and logical substitutions and changes may be made without departing from the scope of this disclosure. This Detailed Description, therefore, is not to be taken in a limiting sense, and the scope of various embodiments is defined only by the appended claims, along with the full range of equivalents to which such claims are entitled.
Such embodiments of the inventive subject matter may be referred to herein, individually and/or collectively, by the term “invention” merely for convenience and without intending to voluntarily limit the scope of this application to any single invention or inventive concept if more than one is in fact disclosed. Thus, although specific embodiments have been illustrated and described herein, it should be appreciated that any arrangement calculated to achieve the same purpose may be substituted for the specific embodiments shown. This disclosure is intended to cover any and all adaptations or variations of various embodiments. Combinations of any of the above-described embodiments, and other embodiments not specifically described herein, may be used and are fully contemplated herein.
The Abstract is provided to comply with 37 CFR § 1.72(b) and will allow the reader to quickly ascertain the nature and gist of the technical disclosure. It is submitted with the understanding that it will not be used to interpret or limit the scope or meaning of the claims.
In the foregoing description of the embodiments, various features are grouped together in a single embodiment for the purpose of streamlining the disclosure. This method of disclosure is not to be interpreted as reflecting that the claimed embodiments have more features than are expressly recited in each claim. Rather, as the following claims reflect, inventive subject matter lies in less than all features of a single disclosed embodiment. Thus the following claims are hereby incorporated into the Description of the Embodiments, with each claim standing on its own as a separate example embodiment.
This application is a continuation of U.S. patent application Ser. No. 15/443,956, filed on Feb. 27, 2017, which is a continuation of International Application No. PCT/US2015/047820, filed Aug. 31, 2015, which claims priority to U.S. Provisional Application No. 62/043,760, filed Aug. 29, 2014; International Application No. PCT/US2015/047820 is also a continuation-in-part of International Application No. PCT/US2015/046819, filed Aug. 25, 2015, which in turn claims priority to U.S. Provisional Application No. 62/043,760, filed Aug. 29, 2014, the entire contents of which are incorporated by reference in their entirety. This application is also a continuation-in-part of International Application No. PCT/US2015/046819, filed Aug. 25, 2015, which in turn claims priority to U.S. Provisional Application No. 62/043,760, filed Aug. 29, 2014, the entire contents of which are incorporated by reference in their entirety.
Number | Date | Country | |
---|---|---|---|
62043760 | Aug 2014 | US | |
62043760 | Aug 2014 | US | |
62043760 | Aug 2014 | US |
Number | Date | Country | |
---|---|---|---|
Parent | 15443956 | Feb 2017 | US |
Child | 16895861 | US | |
Parent | PCT/US2015/047820 | Aug 2015 | US |
Child | 15443956 | US |
Number | Date | Country | |
---|---|---|---|
Parent | PCT/US2015/046819 | Aug 2015 | US |
Child | PCT/US2015/047820 | US | |
Parent | PCT/US2015/046819 | Aug 2015 | US |
Child | 15443956 | US |