Augmented human intelligence using artificial intelligence with human in the loop

Abstract

Augmented human intelligence using artificial intelligence (AI) with a human in the loop is disclosed. The AI is trained with data sets with brain-muscle patterns. Brain and muscle signals are obtained for a subject by EEG and EMG sensors to input into the AI. The AI augments human intelligence by overlapping natural human responses with real-time, goal-directed adjustments to instantaneous brain-muscle reactions. The AI enables the human to repeat such adjustments in order to re-wire the brain to adopt new reactions.

Description

Attorney Docket No.: SYNP2024PCT07WO0AUGMENTED HUMAN INTELLIGENCE USING ARTIFICIAL INTELLIGENCE WITH HUMAN IN THE LOOPCROSS-REFERENCE TO RELATED APPLICATION

[0001] This application claims priority to US Provisional Patent Application Ser. No. 63 / 715,651 filed on November 4, 2024. This application also cross-references to co-pending US Patent Application Ser. No. 18 / 741,732 filed on June 12, 2024. All disclosures are herein incorporated by reference for all purposes.FIELD OF THE INVENTION

[0002] The present disclosure generally relates to the enhancement of human performance through augmented intelligence development in humans using artificial intelligence (Al) trained with instantaneous brain-muscle reaction datasets. The augmented Al can be implemented to treat, rehabilitate and raise performance in those with cognitive and / or mobility deficits to re-learn the correct use of muscles and the ability to isolate muscles for activity.BACKGROUND

[0003] Treating or rehabilitating a patient with neurological damage, dysfunction or performance deficits typically involves repetitively performing tasks related to the functional impairment. For example, for a patient suffering from partial paralysis of a limb, such as an arm, rehabilitation includes repetitively moving the arm. This also includes, for example, rehabilitating patients suffering emotional paralysis from stress disorders.

[0004] Conventional rehabilitation models, as discussed, do not evaluate feedback based on body (muscle) and mind (brain) signals. As discussed, feedback for conventional rehabilitation models only considers physical motion at a gross or obvious level, including position, trajectory and velocity.Attorney Docket No.: SYNP2024PCT07WO0

[0005] The present disclosure relates to the use of an artificial intelligence (Al) with a human in the loop to augment human intelligence, taking into consideration of the brain-muscle signal patterns from rehabilitation by augmenting and accelerating conscious learning.SUMMARY

[0006] Effective interventions for accelerated recovery of patients with neurological damage or dysfunction and those with performance deficits are disclosed. In one embodiment, an artificial intelligence (Al) for augmenting human performance is disclosed. The Al includes a neural network configured with a human in the loop. The neural network is trained with brainmuscle data sets of different activities and from subjects of different demographics. The Al includes biosensors for sensing brain and muscle signals from a subject. Inputs of the neural network are from the biosensors. The inputs include external and internal environmental inputs to a central nervous system (CNS) and a peripheral nervous system (PNS) of a subject, and real-time audio-visual feedback inputs of the reactions to the CNS and PNS. The neural network generates outputs based on the inputs. The outputs include external environment outputs through motor and cognitive processes (MCPs) and through a musculo-skeletal system (MSK), internal environmental outputs from the CNS and PNS, and modulated outputs from the real-time audio-visual feedback of the CNS and PNS, wherein the modulated output serves as the real-time audio-visual feedback inputs of the reactions to the CNS and PNS to form the loop. The neural network is configured to perform a gap analysis of the muscle-brain signal combinations of the subject when performing an activity with ideal muscle-brain signal combinations to establish thresholds for the subject to be achieved for rehabilitation.

[0007] Another embodiment relates to a method of improving performance of a subject. The method includes providing a neural network trained with brain-muscle data sets of different activities from subjects of different demographics. The subject is fitted with biosensors to collect brain and muscle signals when performing a prescribed activity. BrainAtorney Docket No.: SYNP2024PCT07WO0 and muscle signals are collected as inputs to the neural network. The neural network performs a gap analysis of the muscle-brain signal combinations of the subject with ideal muscle-brain signal combinations to establish thresholds for the subject to achieve to improve performance.

[0008] These and other advantages and features of the embodiments herein disclosed, will become apparent through reference to the following description and the accompanying drawings. Furthermore, it is to be understood that the features of the various embodiments described herein are not mutually exclusive and can exist in various combinations and permutations.BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In the drawings, like reference characters generally refer to the same parts throughout the different views. Also, the drawings are not necessarily drawn to scale, with emphasis instead generally being placed upon illustrating the principles of the invention. In the following description, various embodiments of the present invention are described with reference to the following drawings, in which:

[0010] Fig. 1 shows a simplified embodiment of an Al architecture with a human 110 in the loop;

[0011] Figs. 2a-2b illustrate the difference with and without the use of an Al with a human in the loop;

[0012] Fig. 3 shows a process 300 flow for training the Al with a human in the loop; and|0013| Fig. 4 shows a simplified block diagram of a neural network.Attorney Docket No.: SYNP2024PCT07WO0DETAILED DESCRIPTION

[0014] Embodiments relate to augmented human intelligence using an Al with a human in the loop to enhance human performance. Instantaneous brain-muscle-endocrine responses are internal human responses to interactions with the external environment. These responses are used to determine how one adapts, survives and interacts with the world. This, in turn, determines many aspects of health, quality of life and longevity. The responses, as such, can be used as datasets to train the Al to determine human adaptation and maladaptation to various situations, time periods and activities / actions.

[0015] The Al with human in the loop is an Al architecture that can assess human intelligence by comparing natural human responses versus the machine learned adaptive human responses. This can be used to predict markers of disease onset and progression based on a progressive gap analysis between natural human responses in an individual or a group of individuals and the machine learned reaction patterns.

[0016] The Al can be used to rewire the brain to adopt new reactions. For example, by overlapping natural human responses with real-time, goal-directed adjustments to instantaneous brain-muscle reactions, this can be used to enhance human performance or to rehabilitate subjects (patients) with cognitive and / or mobility deficits to re-learn the correct use of muscles and the ability to isolate muscles for activity.

[0017] In addition, the Al can be implemented as a preventative health modality. For example, the Al can be used to re-educate the sensory-motor neuromuscular system to take conscious adaptation beyond the automatic flight-fight response. This frees human progress and performance from the reptilian brain, enabling one to naturally respond with the relaxation response in real-time.

[0018] Fig. 1 shows an embodiment of an Al architecture 100 with a human 110 in the loop. The Al, for example, is a neural network. In one embodiment, the Al receives inputs on theAtorney Docket No.: SYNP2024PCT07WO0 input side 140 and generates outputs on the output side 150. Input signals on the input side include signals from the central nervous system (CNS) and peripheral nervous system (PNS). In one embodiment, input signals include” i) input signals from the external environment to the CNS; ii) input signals from the external environment to the PNS; iii) input signals from the internal environment to the CNS; iv) input signals from the internal environment to the PNS; v) inputs from the real-time audio-visual feedback of the reactions to the CNS; and vi) intputs from the real-time audio- visual feedback of the reactions to the PNS.100191 Input signals from the external environment to the CN S include, for example, inputs to the eyes, ears, nose and tongue while the body is static or in motion, such as performing a task. The inputs may also include inputs from instruments, such as SPO2 sensors, heart rate variability sensors, brain stimulators as well as other similar types of instruments. As for inputs from the external environment to the PNS, they may include touch, temperature, impact, and slippage while the body is static or in motion, such as performing a task. These input signals may also include inputs from instruments such as galvanic skin conductance sensors, infra-red or electric heating systems, cold packs, muscle stimulators or other similar types of instruments.

[0020] Input signals from the internal environment to the PNS may self-regulated / non- regulated brain reactions, such as, for example, from processing vision, audition, smell and taste while the body is static or in motion, such as performing a task. Inputs from the internal environment to the PNS include self-regulated / non-regulated muscle reactions, such as, for example, pain, trauma, weight bearing on the floor through the feet on a static or moving surface, loss of or regaining balance while the body is static or in motion, such as performing a task.Attorney Docket No.: SYNP2024PCT07WO0

[0021] As for the input signals from the real-time audio-visual feedback of the reactions to the CNS and PNS, these are the modulated output signals from the real-time audio-visual feedback of the CNS and PNS outputs, as will be subsequently described. For example, the output signals are fed back as inputs to the Al, forming a continuous loop. In other words, the CNS and PNS feedforward input signals are the feedback output signals.

[0022] The feedforward input signals and feedback output signals are generated continuously. A change may take place within 50-5000 milliseconds (ms). The recording of changes, however, depends on the sampling rate of the sensors, such as brain (EEG) and muscle (EMG) sensors. In the case where the sampling rate is 1000 / sec, then sampling is performed every 10 ms. The running data window may 100 ms. The running window may be adjusted based on performance needs.

[0023] Reactions to the CNS is the perception of change achieved with the biofeedback signals, for example, within 50-500 ms, along with Afferent nerve feedback experienced by the same CNS due to the adaptive changes achieved by the person when responding to the biofeedback and feedforward. A reaction, as far as PNS is concerned, is the sensation of the position of the body in space that is adopted to achieve the changes in the biofeedback signals. These are of the nature of proprioceptive sensations. The other reaction input is the knowledge of those adaptive reactions that can now be repeated volitionally. Volition may be defined as any repetition that is equal to or greater than 3 out of five attempts, which is as close to realtime as possible in the same 50-500 ms as a target. It is understood that the starting time frames may be as long as 2-5 seconds to respond with a deliberate and accurate muscle isolation.

[0024] Afferent nerves provide feedback (sensory information) from the body to the CNS. Afferemt nerves, known as sensory neurons, transmit impulses from the sensory receptors to the periphery, such as the skin, muscles and organs, to the CNS, such as the brain and spinal cord. On the other hand, Efferent nerves carry motor commands from the CNS to the musclesAttorney Docket No.: SYNP2024PCT07WO0 and glands. Efferent nerves, known as motor neurons, transmit impulses from the CNS to effector organs, such as muscles and glands, informing them how to respond. The afferent nerves forms the feedback pathway while the efferent nerves form the response pathway.

[0025] Variations in the input datasets may be classified and tagged as follows: i) at rest calibration; ii) calibration post relaxation; iii) based on tasks performed; iv) physical situations - (relaxed, static, mobile, comfortable, uncomfortable, challenging, extreme, and painful); v) mental situations - (relaxed, static, mobile, comfortable, uncomfortable, challenging, extreme, and painful); vi) emotional situations - (relaxed, static, mobile, comfortable, uncomfortable, challenging, extreme, and painful), vii) different time periods within the above situations; and viii) different activities within the above situations and time periods.

[0026] Output signals on the output side include signals from the CNS and PNS. In one embodiment, output signals include: i) output signals from the external environment through motor and cognitive processes (MCPs); ii) output signals from the external environment through the musculo-skeletal system (MSK); iii) output signals from the CNS to the internal environment (sensory recognition and processing ability); iv) output signals from the PNS to the internal environment (proprioceptive, motor and sense of body in 3D space);Atorney Docket No.: SYNP2024PCT07WO0 v) modulated outputs from the real-time audio-visual feedback of CNS outputs above; and vi) modulated outputs from the real-time audio-visual feedback of the PNS outputs above.

[0027] Output signals from the external environment through MCPs may include, for example, facial expression, sweating, body language, eye movement and other similar types of indicators. Outputs signals from the external environment through the MSK may include, for example, posture adjustment, arm and leg movement, tongue movement, balance adjustment through the foot and head and truck movement as well as other types of movements.

[0028] As for output signals from the CNS to the internal environment, they may include a change of heart rate, a change in body temperature as well as other similar types of changes to the body. Output signals from the PNS to the internal environment based on proprioceptive, motor and sense of body in 3D space may include sensory input for the brain to plan and execute a task. The modulated outputs from the real-time audio-visual feedback of CNS and PNS outputs serve as feedforward inputs to the Al.

[0029] Variations in the output datasets may be classified and tagged similarly to the variations in the input datasets.

[0030] As described, the feedforward signals from the external and internal environments can come from multiple sources or sensors However, the self-regulation outputs (and hence inputs) will come only from EEG and EMG sensors.[0031| The neural network includes n hidden layers 130i-n. Hidden layers are the intermediate layers in the neural network that process data from inputs on the input side 140 data and perform computations to produce outputs on the output side 150. The hidden layers not directly visible to the user. They are the core of Al learning, extracting complex patterns and performing non-linear transformations on data to enable tasks like image recognition andAtorney Docket No.: SYNP2024PCT07WO0 natural language processing. The more hidden layers a network has, the deeper the learning and the more complex the patterns it can represent.

[0032] A hidden layer in deep learning is a layer of artificial neurons between the input and output layers of a neural network. It transforms inputs through weighted connections and activation functions to help the network learn patterns. The more numbers of hidden layers, the deeper the learning.

[0033] The number of hidden layers will differ (and is configurable) based on the use-case. This could be manual or automated by the use of a rule engine where the use cases are defined. As shown, the Al includes 4 hidden layers. Providing an Al, as discussed, will more layers may also be useful.

[0034] In one embodiment, the Al includes a combination of a convolutional neural network (CNN) providing feedforward and a recurrent neural network (RNN) providing feedback. The inputs to the CNN include signals from the environment and the outputs of the RNN; inputs to the RNN include the instantaneous brain-muscle reactions and the outputs of the CNN. Nonmodulated and / or modulated brain-muscle reactions form the outputs to the environment as motor process outputs. Other configurations of the Al may also be useful.

[0035] As discussed, the Al can be implemented to treat, rehabilitate and raise performance in those with cognitive and / or mobility deficits to re-leam the correct use of muscles and the ability to isolate muscles for activity. The ability to retrain or re-leam the correct use of muscles and the ability to isolate muscles for the activity facilitates accelerated recovery of patients with neurological damage or dysfunction as well as those performance deficits. This is achieved by helping patients to re-wire their brains and to adopt to new reactions. For example, a patient can be retrained to make conscious choices instead of relying on the instantaneous brain-muscle-endocrine reactions.Atorney Docket No.: SYNP2024PCT07WO0

[0036] Figs. 2a-2b illustrate the difference with and without the use of an Al with a human 110 in the loop, respectively. As shown in Fig. 2a, the Al has retrained a patient’ s brain to, for example, help the patient to correctly use the muscles. This results in the patient making a conscious choice, resulting in an upward spiral 262, synonymous with learning and improvement. On the other hand, without the Al, as shown in Fig. 2b, the patient has an unconscious reaction because the brain has not been retrained to overcome the instantaneous brain-muscle-endocrine reactions This results in a downward spiral 264, synonymous with compensation and no improvement. As such, the Al augments human intelligence through the training and expansion of conscious choice over instantaneous brain-muscle-endocrine reactions.

[0037] Muscle-use profiles vary widely in amplitude and frequency of EMG signals, depending on the tasks required. For example, depending on the task or activity, different muscles are used, while exerting strength as well as control simultaneously. In addition, performing any task requires the processing of sensory and cognitive inputs to make the execution of a motor task possible. The Al trained with such signal combinations from muscles and brain facilitates gap analysis from a patient’s current status and establishes thresholds to be achieved for rehabilitation.

[0038] The Al can be used to help a patient re-leam the ideal muscle-use profile and the ideal cognitive performance profile. The AT assistance is dynamic For example, depending on which direction or mix of muscle combinations a patient is trending towards learning, the Al determines if this direction is beneficial or if a course correction is required.

[0039] As an example, a patient or subj ect is being trained to perform the activity of pouring water into a cup from a bottle using the Al. Other activities may also be performed by the subject using the Al. Pouring water requires a multi-muscle strategy closely coupled with thresholds of attention span and the ability to process the sequence and balance whileAtorney Docket No.: SYNP2024PCT07WO0 maintaining the working memory. The interaction between cognition and muscle strategy here is different and is known to be different even for the same person from one repetition of the action to the next.

[0040] This “black box” of the interaction between cognition and movement can be opened up and made visible through the training of the Al engine. The training may use, for example, input signals and patterns from a large number of patients in their recovery journey. Using machine learning, functional thresholds for the brain and muscle signals, including their distribution, can be established. These thresholds may then be used by a patient to set goals towards regaining the function, for example, of pouring water into a cup from a bottle.100411 Although the dataset relates to the activity of pouring water into a cup from a bottle, other datasets of other activities can also be collected. The various datasets, as discussed, can be classified or tagged based on the activities as well as other variables. In the case where a patient needs more frequent tracking and close monitoring, an embedded sensor with Al on the edge embedded in microelectronic hardware (e.g., chips with FPGA capabilities or processors) is implantable subcutaneously at the location of a specific muscle with interaction with a mobile APP, which can track the activity of a particular muscle and compare with Al driven “normative” data patterns could be utilized.

[0042] Fig 3 shows a process 300 flow for training the Al with a human in the loop. As shown, the Al receives brain metrics and muscle metrics inputs 303 and 307 The brain and muscle metrics inputs are obtained from brain (e.g., EEG) and muscle (e.g., EMG) sensors.[0043| The brain metrics 303 are the CNS outputs and can be obtained from the total brain or specific brain area(s). The brain metrics, as discussed, can be read or obtained with EEG sensors. The brain metrics may or may not include readings from a fNIRS device. The brain metrics are used as a separate layer of inputs to the Al. The inputs, in one embodiment, are the real-time audio-visual feedback or the reactions to the CNS.Attorney Docket No.: SYNP2024PCT07WO0

[0044] As for the muscle metrics, they include the PNS outputs which can be sensed from individual muscles or muscle group combinations by EMG sensors. As such, the PNS may be undefined combinations (e.g, n channels of inputs 307i-n), from individual muscles such as pronator / supinator or then groups such as distal muscles (finger and wrist muscles) or proximal muscles (wrist and pronator / supinator). The number of channels depends on, for example, the use case or activity of interest.

[0045] The PNS outputs are the modulated outputs based on the real-time audio-visual feedback. The PNS outputs are then used as inputs to the AT from the real-time audio-visual feedback of the reactions to the PNS, forming the loop.

[0046] The brain metrics are processed by a hidden layer of the Al. For example, brain asymmetry 305, delta to alpha ratio 323, relative alpha 333 and relaxed focus % time 343 are determined Al based on the brain metrics. Brain asymmetry is the ratio of the total power of the four EEG frequency bands (alpha, beta, delta, theta) between the two brain hemispheres as a proportion of total power across both hemispheres. Delta to Alpha ratio is the ratio of the total power of the delta and alpha EEG frequency bands. Relative alpha is the ratio of the total alpha power and the total power across all four frequency bands (alpha, beta, delta, theta. Delta is the frequency band between 0.5-4 Hz, Theta is 4.5-7.5 Hz, Alpha is 8-12 Hz and Beta is 12.5-35 Hz.

[0047] The muscle metrics are processed by n hidden layers, each analyzing one of the muscle metrics 307i.n, depending on the number of layers needed, according to the activity to be analyzed. The hidden layers determine the agonist EMG amplitudes, the antagonist EMG amplitudes 327i-n and the agonist-antagonist balances 327i-n. The agonist-antagonist balance is the ratio of the difference in EMG amplitudes of the agonist and antagonist muscles as a proportion of the agonist EMG amplitude at any time point t, where t is, for example, captured every 10 ms if the running window is 100 ms and the sampling rate is 1000 / sec.Atorney Docket No.: SYNP2024PCT07WO0

[0048] The outputs of the Al 353 and 357 are then combined. For example, the EEG and EMG are separate patterns. The separate EEG and EMG are combined based on the same time stamp in the time series data, resulting in combined EEG-EMG metrics. For example, at the point of firing a muscle, does attention remain stable, or at the point of doing a mental task, does the muscle experience tightening? There are many such combinations whose significance with respect to the different use cases that the Al learning will throw up in order to personalize therapy and inform the scientific model

[0049] In the case where a patient performs 24 sessions of performing an activity, such as pouring water into a cup from a bottle, data from all 24 sessions and all repetitions within the data set are available from creating the training data set. Each set of data is tagged to the task type and the demographics of the patient. Features could be extracted and categorized for comparison with another session. For example, every Xthsession, such as every 6thsession, is compared with the previous Xthsession. Other values of x may also be useful. This presents a dynamic target threshold to be achieved through training for individuals and different use cases, rather than some standard, inaccurate normative data. In other words, the target thresholds are personalized to each individual and is based on the individual’s current status or performance. Based on the target thresholds, the Al can prescribe a specific activity to perform. For example, the specific activity may be displayed on a display or projected on a surface. As the patient improves, target thresholds are adjusted and appropriate activities are prescribed to continue to improve rehabilitation or performance.

[0050] Such features could be correlated to standardised clinical scales to provide “global” thresholds in brain and muscle metrics to be achieved for the more gross movements or tasks, such as reaching out with the hand. Data from the individual patients could then help personalize their therapy using their individual reaction data and the patterns seen therein forAttorney Docket No.: SYNP2024PCT07WO0 more subtle and fine tasks, such as threading a needle or cutting with a knife. In addition, the fine tasks may also involve the gross tasks, such as reaching out.

[0051] Brain (EEG) and muscle sensors (EMG) are positioned to obtain inputs for the Al. Regarding EEG sensors, they may be positioned at F3, F4, C2, C4, P2, P4 01 and 02 As for the EMG sensors, they may be positioned at the finger flexor, thumb flexor, wrist extensor, pronator and supinator. The stages of the activity (pouring water into a cup from a bottle) to be analyzed may include: i) reaching out with the dominant hand opened to grasp the bottle; ii) grasping the bottle with a cylindrical grasp using the fingers iii) lifting bottle up by engaging the elbow; iv) pouring by maintaining grasp and pronating the arm; v) stop pouring by supinating until the bottle is upright; vi) placing the bottle on the table and releasing the grasp; and vii) retracting and relaxing the arm.

[0052] Brain and muscle activity and relaxation thresholds, as learned by the Al, can be used to provide feedback for comparison to patient performance at any time. The thresholds may be in the form of bands based on the severity, tolerance and type of medical condition, as determined from the training data set. The Al may determine areas of good and poor performance in real-time for an individual patient when performing the activity. This will help evolve the personalized learning pattern of the patient before progressing to optimize it to as close to the natural learning pattern as possible. The focus is to reduce compensation as a pathway to accelerate re-learning.

[0053] Another example of an activity may be writing with a pen. Writing requires a distal, multi-muscle strategy closely coupled with thresholds of attention spans and the ability to process sequences and word construction while maintaining the working memory. TheAttorney Docket No.: SYNP2024PCT07WO0 interaction between cognition and muscle strategy is different and is known to be different even for the same person from one repetition of a word or sentence to the next. The interaction between cognition and muscle strategy (movement) is also different when one is supposed to simply write an alphabet, word, sentence or create a paragraph.

[0054] This “black box” of the interaction between cognition and movement can be opened up and made visible through the training of an Al, for example, using such input signals and patterns from a large number of patients in their recovery journey. The Al, through learning, can establish functional thresholds in the brain and muscle signals as well as their distribution. These thresholds may then be used by a patient to set goals towards regaining the function of writing functionally and creatively for self-expression.

[0055] For those who need more frequent tracking and close monitoring, an embedded sensor with Al on the edge embedded in microelectronic hardware (e.g., chips with FPGA capabilities or processors) that is implantable subcutaneously at the location of a specific muscle with interaction with a mobile APP, which can track the activity of a particular muscle and compare with Al driven “normative” data patterns could be utilised.

[0056] In the case of writing with a pen, EEG sensors are located on the head, similar to the activity of pouring water into a cup from a bottle, [correct?] As for the EMG sensors, they are placed appropriately for capturing signals from muscles used in writing.

[0057] Fig 4 shows a simplified block diagram of a neural network 400. The neural network can be configured as a wearable accessory so it is close to the skin. For example, the wearable accessory may be a pair of eyeglasses, a cap, clothing, a sleeve or a patch. Other types of wearable accessories may also be useful.

[0058] In one embodiment, the neural network includes an Al module 412. The Al module is a chip, such as an FPGA, configured with the Al engine, such as RNN and CNN. Other types of Al modules may also be useful. The engine is capable of performing real-time analysis onAttorney Docket No.: SYNP2024PCT07WO0 the internal environment as well as parallel analysis of external environment inputs and outputs in real-time. The parallel processing pathways may be switched on and off or configured to run simultaneously for various data collection, training, assessment or treatment scenarios.

[0059] The Al module may include storage for storing the results of the analysis by the Al engine. The neural network includes a communication module 416 configured for high-speed internet connectivity for transmitting and receiving data. A power module provides power for the modules of the neural network.

[0060] The neural network also EEG and EMG sensor modules 442 and 446. The EEG sensor module includes EEG sensors and the EMG sensor module include EMG sensors. The sensor modules may communicate with the Al module for providing brain and muscle signals. The communication may be wired or wireless communication.

[0061] The wearable neural network is configured to be easily removable and chargeable. The wearable neural network may also be worn as an assistive device, apart from being an intelligence augmentation tool. The neural network can be configured to be cloud-enabled. Furthermore, treatment and training for augmenting intelligence can be on via a mobile display or projected on any surface. As discussed, the neural network can be configured to be subcutaneously implanted. In such cases, the communication module may be a wireless communication module, such as Bluetooth which can connect to an App on a mobile device

[0062] The Al can be employed to assess and diagnose various disorders or diseases, including: i) neuro-development disorders, including low IQ; ii) neuro-adaptive disorders, e.g., emotional and mental trauma; iii) neuro-degenerative disorders and diseases; iv) neuro-event disorders, e.g., injury, physical trauma; v) movement disorders;Atorney Docket No.: SYNP2024PCT07WO0 vi) mobility disorders; vii) cognitive disorders; viii) emotional and mental disorders, ix) addiction disorders; x) performance disorders and dysfunction; xi) chronic diseases such as hypertension, diabetes, heart / kidney disease, autoimmune disease, chronic pain, chronic headache; xii) genetically predisposed disorders and disease - downregulating / upregulating gene expression; and xiii) endocrine disorders and dysfunction.

[0063] The basic assessment modality is to identify maladaptive instantaneous reactions in brain and muscle using Al trained by the datasets, as described, for various demographics, types of individuals and for specific performance scenarios.

[0064] Once the assessment is made, treatment is then prescribed. The basic treatment modality includes the real-time feedback of the existing brain-muscle reactions and an Al driven gap analysis to understand the variance from optimum reaction frequency and amplitude distributions in individual signal streams and combination signal streams.

[0065] The Al-backed analysis of existing behavior / reaction patterns can be used to predict the onset of disorders and disease A recommendation or prescription of activities to alleviate symptoms of the disorders and diseases, and the ultimate reversal leveraging neuroplasticity and the self-healing nature of the human brain and neuro-musculo-skeletal-organ system.

[0066] The Al with a human in the loop can also be employed to enhance performance and interaction. For example, human emotional expression can be converted into bio-potential signatures and expressed through wearables worn on the fingertips as a touch sequence. Micro-Atorney Docket No.: SYNP2024PCT07WO0 failure signatures in performance scenarios or situations can help identify risks of potentially larger failures or injuries

[0067] Creative blocks can be characterized as bio-signal signatures, identified in individuals. However, the creative blocks can be overcome through instantaneous selfregulation supported by the Al engine in real-time. In the case of injury of illness, the reestablishment of performance benchmarks can be tracked using bio-signal signature benchmarks across various muscle and brain channels. A heatmap of high / low / medium risk channels can be generated, indicating which require specific attention and priority.

[0068] The present disclosure may be embodied in other specific forms without departing from the spirit or essential characteristics thereof. The foregoing embodiments, therefore, are to be considered in all respects illustrative rather than limiting the invention described herein. The scope of the invention is thus indicated by the appended claims, rather than by the foregoing description, and all changes that come within the meaning and range of equivalency of the claims are intended to be embraced therein.

Claims

Attorney Docket No.: SYNP2024PCT07WO0CLAIMS1. An artificial intelligence for augmenting human performance comprising: a neural network configured with a human in the loop, wherein the neural network is trained with brain-muscle datasets of different activities and from subjects of different demographics; biosensors for sensing brain and muscle signals from a subject; inputs of the neural network from the biosensors, wherein the inputs comprise external and internal environmental inputs to a central nervous system(CN S) and a peripheral nervous system (PN S) of a subj ect, and real-time audio- visual feedback inputs of the reactions to the CNS and PNS; outputs of the neural network, wherein the outputs comprise external environment outputs through motor and cognitive processes(MCPs) and through a musculo-skeletal system (MSK), internal environmental outputs from the CNS and PNS, and modulated outputs from the real-time audio-visual feedback of the CNS and PNS, wherein the modulated output serves as the real-time audiovisual feedback inputs of the reactions to the CNS and PNS to form the loop; wherein the neural network is configured to perform a gap analysis of the musclebrain signal combinations of the subject when performing an activity with ideal muscle-brain signal combinations to establish thresholds for the subject to be achieved for rehabilitation.

2. A method of improving performance of a subject comprising: providing a neural network trained with brain-muscle data sets of different activities from subjects of different demographics;Atorney Docket No.: SYNP2024PCT07WO0 fitting the subject with biosensors to collect brain and muscle signals when performing a prescribed activity; and collecting brain and muscle signals as inputs to the neural network; and performing a gap analysis of the muscle-brain signal combinations of the subject with ideal muscle-brain signal combinations to establish thresholds for the subject to achieve to improve performance.

Images