Intelligent closed-loop system for real-time reversal of acute disease

Abstract

A system is provided for detecting and reversing an acute health condition, comprising: at least one wearable sensor configured to sense at least one health indicator of a user; a controller configured to receive signals from the at least one wearable sensor representing the at least one health indicator, the controller including an AI model trained to determine from the received signals whether the user is in an acute health condition; and at least one therapeutic device configured to deliver a treatment to the user in response to a determination that the user is in the acute health condition.

Description

INTELLIGENT CLOSED-LOOP SYSTEM FOR REAL-TIME REVERSAL OF ACUTE DISEASECROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims the benefit of U.S. Provisional Application No. 63 / 558,182, filed on February 27, 2024, entitled “INTELLIGENT CLOSED-LOOP SYSTEM FOR REAL-TIME REVERSAL OF ACUTE DISEASE,” the disclosure of which being expressly incorporated herein by reference.STATEMENT OF GOVERNMENTAL RIGHTS

[0002] This invention was made with governmental support under DA056242 awarded by the National Institutes of Health (NIH). The Government has certain rights in the invention.FIELD

[0003] The present disclosure pertains generally to disease detection and reversal, and in particular to a closed-loop system for real-time reversal of acute disease.BACKGROUND

[0004] Treatments exist for the reversal of acute diseases such as acute anaphylaxis or opioid overdose. However, such treatments have several shortcomings. For example, sudden onset of anaphylaxis induced by food, medications, and other environmental factors usually causes an individual to stop breathing, lose consciousness within minutes, and even die if not treated. Despitethe relative effectiveness of the EPIPEN®, used to inject epinephrine, it still has a failure rate up to 30% for several reasons. First, the patient may not immediately recognize anaphylaxis onset. Second, the patient may miss the brief time window for the self-administration of EPIPEN®. Third, some patients (especially children or the elderly) cannot self-administer.

[0005] Similarly, the various developed treatments for opioid overdose, a condition that caused almost 50,000 deaths in the U.S. in 2019, are very challenging to use effectively. Often the patient or their associates do not immediately recognize the onset of an opioid overdose. Also, the time between the onset of respiratory depression and death is brief. Thus, time window to seek medical help and intervene is very short. Additionally, self-administered or bystander-administered opioid overdose methods or devices are largely unavailable or complicated for lay individuals to use. Accordingly, there is an urgent and unmet need for real-time detection and automated treatment of acute diseases.SUMMARY

[0006] The system of the present disclosure provides a wearable intelligent device that can detect and treat an acute disease such as acute anaphylaxis or opioid overdose. While the system is primarily described with reference to opioid overdose detection and reversal, it should be understood that it may have various applications for various diseases or conditions. The system significantly improves current patient care in a number of ways. As described herein, the system improves reliability and automation by automatically detecting the onset of disease and connecting to emergency care with high reliability, avoiding failure from lack of intervention. The system enhances safety as well. For example, patients may require different epinephrine doses, because of individual characteristics. However, the current EPIPEN® may result in an overdose of epinephrine causing seriousadverse effects (e.g., ventricular arrhythmias, hypertensive crisis, and pulmonary edema). The present system provides individualized doses via a pain-free device, avoiding dosage safety concerns. Additionally, the present system is simple, portable, affordable, and able to automatically detect, manage, and report an individual's onset of acute disease anytime in daily life. This system will enhance patients' confidence to fully enjoy their lives without concerning the sudden onset of the disease.

[0007] According to one embodiment of the present disclosure, a system is provided for detecting and reversing an acute health condition, comprising: at least one wearable sensor configured to sense at least one health indicator of a user; a controller configured to receive signals from the at least one wearable sensor representing the at least one health indicator, the controller including an Al model trained to determine from the received signals whether the user is in an acute health condition; and at least one therapeutic device configured to deliver a treatment to the user in response to a determination that the user is in the acute health condition. In one aspect of this embodiment, the at least one wearable sensor includes a respiratory sensor configured to sense breathing behavior of the user. In another aspect, the at least one health indicator is at least one of a breathing pattern of the user, a blood pressure of the user, a blood oxygen level of the user, or a biochemical molecule. In another aspect, the at least one therapeutic device and the controller are mounted to a patch worn by the user. In yet another aspect, the controller is part of a portable computing device and receives the signals from the at least one wearable sensor over a wireless communications link. In another aspect of this embodiment, the acute health condition is an acute disease condition. In a variant of this aspect, the acute disease condition is an opioid overdose condition. In another aspect, the therapeutic device is an acoustofluidic patch configured to deliver a programmable dosing of an opioid antagonist. In a variant of this aspect, the acoustofluidic patch includes a hollow microneedle. In yet another aspect, thesystem further comprises a communication interface in communication with the controller, the communication interface being configured to transmit a notification requesting healthcare assistance to an emergency personnel device via a wireless network in response to a determination that the user is in an acute health condition. In another aspect, the Al model includes a convolutional neural network. In a variant of this aspect, the convolutional neural network includes three layers of onedimensional convolutions.

[0008] In another embodiment, the present disclosure provides a method of detecting and reversing an acute health condition, comprising: sensing at least one health indicator of a user with at least one wearable sensor; receiving, by a controller, signals from the at least one wearable sensor representing the at least one health indicator; determining, by an Al model of the controller, from the received signals whether the user is in an acute health condition; responding, by the controller, to a determination that the user is in an acute health condition by activating at least one therapeutic device to deliver a treatment to the user. In one aspect of this embodiment, sensing at least one health indictor includes sensing breathing behavior of the user by a respiratory sensor. In another aspect, the at least one health indicator is at least one of a breathing pattern of the user, a blood pressure of the user, a blood oxygen level of the user, or a biochemical molecule. In another aspect, the method further comprises attaching a patch including the at least one therapeutic device and the controller to the user. In another aspect, the controller is part of a portable computing device and receives the signals from the at least one wearable sensor over a wireless communications link. In still another aspect, the acute health condition is an acute disease condition. In a variant of this aspect, the acute disease condition is an opioid overdose condition. In another aspect, the therapeutic device is an acoustofluidic patch configured to deliver a programmable dosing of an opioid antagonist. Another aspect further comprises transmitting, in response to a determination that the user is in an acute healthcondition, a notification requesting healthcare assistance to an emergency personnel device via a wireless network by a communication interface in communication with the controller. In another aspect of this embodiment, the Al model includes a convolutional neural network. In a variant of this aspect, the convolutional neural network includes three layers of one-dimensional convolutions.

[0009] In yet another embodiment, the present disclosure provides a system for detecting and reversing an acute health condition, comprising: at least one wearable sensor configured to sense at least one health indicator of a user; and a controller configured to receive signals from the at least one wearable sensor representing the at least one health indicator, the controller including an Al model trained to determine from the received signals whether the user is in an acute health condition; wherein, in response to a determination of an acute health condition, the controller is configured to provide a notification to the user to operate a therapeutic device to deliver a treatment for the acute health condition.BRIEF DESCRIPTION OF THE DRAWINGS

[0010] The above-mentioned and other advantages and objects of this disclosure, and the manner of attaining them, will become more apparent, and the disclosure itself will be better understood, by reference to the following description of embodiments of the invention taken in conjunction with the accompanying drawings, wherein:

[0011] FIG. 1 is a conceptual diagram of one embodiment of a real-time closed-loop system for disease detection and treatment activation;

[0012] FIG. 2A is a testing timeline for the embodiment of FIG. 1 ;

[0013] FIG. 2B is a box flow graph corresponding to normal breathing by a subject when in a healthy condition before the opioid overdose induction;

[0014] FIG. 2C is a box flow graph corresponding to the respiratory distress experienced by a subject when an overdose is induced by opioids;

[0015] FIG. 2D is a box flow graph corresponding to recovery of the normal breathing pattern by the subject after naloxone treatment using a system according to the present disclosure;

[0016] FIG. 3A provides charts depicting opioid overdose caused reductions in respiratory rate, tidal volume and minute volume;

[0017] FIG. 3B is a graph depicting representative minute volume tracings with and without treatment after overdose induction;

[0018] FIG. 4 is a conceptual diagram depicting a model training pipeline for a deep learning module according to the present disclosure;

[0019] FIG. 5A is a chart depicting accuracy using minute volume of respiratory data calculated and normalized as the input data processed with a sliding window;

[0020] FIG. 5B is a chart depicting a confusion matrix illustrating the accuracy of overdose detection by the system of the present disclosure after training using labelled heathy and overdose data;

[0021] FIG. 6 is a graph depicting minute volume during sleep and wake periods and after overdose induction;

[0022] FIG. 7A is a graph depicting the validation loss of the deep learning module of the present disclosure;

[0023] FIG. 7B is a graph depicting the performance of the deep learning module of the present disclosure as compared to other machine learning models;

[0024] FIG. 8A is a conceptual diagraph of one embodiment of a system according to the present disclosure;

[0025] FIG. 8B is a schematic illustration of a test procedure timeline;

[0026] FIG. 8C is a chart depicting comparisons between the intelligent patch system according to the present disclosure and manual treatment by threshold;

[0027] FIG. 9A is a graphic depiction of representative minute volume tracings of the system according to the present disclosure;

[0028] FIG. 9B is a graphic depiction of representative minute volume tracing of manual patch tests; and

[0029] FIGS. 10A-E are graphs illustrating the performance of a system according to the present disclosure.

[0030] Corresponding reference characters indicate corresponding parts throughout the several views. Although the drawings represent embodiments of the present disclosure, the drawings are not necessarily to scale, and certain features may be exaggerated or omitted in some of the drawings in order to better illustrate and explain the present disclosure.DETAILED DESCRIPTION

[0031] In general, the present disclosure provides an intelligent theranostic system that detects and manages acute onset disease in real-time. The system links three innovative components, incorporating automated feedback: (a) wearable sensors that track multiple physiological parameters and disease symptoms in real time; (b) wearable therapeutic devices for on-demand delivery of therapeutics; and (c) a closed-loop controller, using machine learning, for real-time analysis of sensing signals, triggering therapeutic devices, and connecting with local emergency services. The system provides a wearable, portable, and user-friendly system for daily use that can provide effective personalized therapy and medical care for an individual with acute onset disease such as opioid overdose and acute food allergy.

[0032] In certain embodiments, acoustofluidics are integrated with a hollow microneedle patch to achieve programmable and precise dosing of an opioid antagonist. As an additional feature, an artificial intelligence (“Al”) model tailored for overdose detection based on respiratory readouts is trained and optimized toprovide rapid detection of an overdose. Experimentation with an animal model of opioid overdose demonstrated that a real-time closed-loop system according to the present disclosure outperforms the conventional methods and presents a promising platform for precise and personalized treatment against therapeutic opioid overdose. Moreover, the principles of the present disclosure may be used in other applications to address challenges of other pathological conditions that require a rapid response.

[0033] Referring now to FIG. 1 , a conceptual diagram of a real-time closed- loop system for disease detection and treatment activation is provided. The system 10 generally includes at least one sensor 12 and an intelligent patch 14. In certain embodiments, the intelligent patch 14 includes an acoustofluidic therapeutic patch 16 and controller 18 that uses an Al model 19 for guiding overdose classification as is further described herein. In other embodiments, the controller 18 is separate from the intelligent patch 14 and located, for example, as part of a smart phone or other portable computing device. In such embodiments, the controller 18 may be in communication with the therapeutic device 16 over a wireless communication link such as Bluetooth®. In still other embodiments, the therapeutic device 16 is separate from the intelligent patch 14 and, for example, carried by the user such as an EPIPEN®. In such embodiments, in response to a determination that the user is in an acute health condition, the controller 18 may provide a notification (e.g., audible and / or visual) to the user via an interface of, for example, a smart phone or other portable computing device. The notification may instruct the user to operate the therapeutic device 16 to deliver a treatment for the acute health condition.

[0034] In one example, the at least one sensor 12 is a respiratory sensor which, at a high level, senses the breathing behavior of the subject 20 and outputs signals representing such breathing to the controller 18. It should be understood that other physical characteristics may be monitored such as, but not limited to, blood pressure and / or blood oxygen level. In any case, the at least one sensor 12 generally senses at least one physical characteristic of the subject 20 indicative of ahealth condition which is hereinafter referred to as “a health indicator.” In the present example, the controller 18 determines whether the breathing pattern of the subject 20 represents a healthy condition 22 or an unhealthy condition 24. In the example described below, the unhealthy condition 24 corresponds to an overdose condition. If the breathing pattern of the subject 20 is determined by the controller 18 to correspond to an unhealthy condition 24, then the therapeutic device 16 is activated to provide a precise dosing of an opioid antagonist such as Naloxone to the subject 20. Additionally, the system 10 may further include a communication interface 15 which is activated upon detection of an unhealthy condition to provide a notification to an emergency personnel device 17 requesting healthcare assistance. The communication interface 15 may be any suitable transmitter for communicating wirelessly over a network to the emergency personnel device 17. If a healthy condition is detected, the system 10 continues to monitor the subject’s 20 breathing using the at least one sensor 12 as indicated by the dynamic feedback control line 26 in FIG. 1.

[0035] By using an animal model of opioid overdose, it was shown that the system could effectively and robustly detect and reverse overdose symptoms in realtime, showing better performance than the conventional threshold-based method, thereby demonstrating the potential of the system 10 for personal and precision medicine to combat pharmaceutical opioid overdose.In Vivo Test for Opioid Overdose Induction and Reversal

[0036] In one experiment, the system 10 was used to test opioid overdose detection and reversal using conscious, spontaneously breathing CD1 wildtype mice with whole-body plethysmography (WBP). More specifically, the patch 14 was attached to subject 20 (i.e. , the mouse) and the subject 20 was placed inside a WBP chamber. The respiration readouts from the at least one sensor 12 werecontinuously monitored for up to 90 minutes allowing real-time assessment of the breathing patterns of the subject 20. Referring to FIG. 2A, the test was conducted in three stages: the healthy baseline (starting from time Tx); the induction of opioid overdose (starting from time To); and the recovery after naloxone patch treatment (starting from time Tyand ending at full recovery time Tz). The time point for full recovery Tz was defined as when the respiratory readouts returned and stabilized at the range of a healthy baseline.

[0037] FIGS. 2B-D display volume-time WBP readouts following the testing sequence. FIG. 2B is a box flow graph corresponding to normal breathing by the subject 20 when in a healthy condition before the opioid overdose induction. FIG. 2C shows the box flow changes corresponding to the respiratory distress experienced by the subject 20 when an overdose is induced by opioids (i.e. , between the times To and Ty). FIG. 2D shows the box flow changes corresponding to recovery of the normal breathing pattern by the subject 20 after naloxone treatment by the intelligent patch 14 (i.e., after time Tz).

[0038] In the experiment, the box flow was measured every millisecond and the output parameters of the WBP (i.e., respiratory rate (bpm), tidal volume (mL) and minute volume (mL / min) were recorded every two seconds. FIG. 3A shows the healthy readouts were the minute average value of 10 min before the overdose induction and the overdosed readouts 10 min after the overdose induction. The data is shown as mean ± s.d., n = 5 independent experiments. As shown in FIG. 3A, opioid overdose caused reductions in all three parameters with the most significant drop in minute volume. This is because, as is known in the art, minute volume is the product of frequency and tidal volume and the physiological quantity of airflow in / out of the lungs. The time series value of minute volume was thus used as the input for assessing overdose symptoms using the Al model 19 of the intelligent patch 14 described above.

[0039] FIG. 3B shows the representative minute volume tracings with or without patch treatment after overdose induction. Representative minute volume (shown as minute average) exhibited a decrease approximately 10 - 20 minutes after fentanyl patch treatment.

[0040] As shown by the data series 28, without patch treatment, the ventilation volume remained below 50% baseline throughout the whole course of the test. In other words, there was no recovery. Upon administering the naloxone patch treatment, respiration recovered within about 10 - 15 minutes (represented by the data series 30) was able to reverse the overdose effects within 20 min. These results demonstrated the successful application of the patch system 10 in small rodent models for opioid overdose study.Deep Learning Guided Detection of Opioid Overdose

[0041] To detect overdose conditions in real-time, the Al model 19, which in certain embodiments uses a convolutional neural network (“CNN”), was developed for the classification of the overdose status. Referring to FIG. 4, a schematic of the model training pipeline is shown. A binary classification task was developed based on the CNN featuring three layers 32, 34, 36 of one-dimensional convolutions to differentiate between overdose respiration samples 38 and healthy baseline respiration samples 40.

[0042] Referring now to FIG. 5A, the minute volume of respiratory data was calculated and normalized as the input data that was processed with a sliding window (2 seconds per step). The model accuracy was compared at different sliding window lengths (1 , 3, 5, 7, and 10 minutes). Five minutes was identified as the most optimized setting for training (78 mice in total). During the training process, 50 percent of the data from the training set was set aside for model validation. For model optimization, hyperparameter tuning was combined with 5-fold cross-validation in every iteration. An early stopping strategy was used to avoid overfitting, i.e. , the training would be stopped when the performance on the validation set was not improved for ten consecutive rounds. The test dataset was generated from the opioid induced overdose tests of independent 12 mice and were not used for training and validating.

[0043] The result of the testing data (six independent experiments) was aggregated in the confusion matrix depicted in FIG. 5B. The testing data was collected with or without opioid overdose induction up to 40 minutes. As shown, there was an overall binary classification accuracy of 95% and an F1 score of 94.8%.

[0044] FIG. 6 shows representative minute volume tracings showing variations in respiratory patterns during wake periods, sleep periods and overdose induction periods captured with WBP systems. The time points below 50% of the baseline value were considered “unsafe” in the study. It is noted that the Al model 19 of the controller 18 was able to distinguish well between the breathing pattern of sleep and overdose. For any inputs longer than the 5-minute fixed window length, the Al model 19 processed the time series stream in a sliding window fashion and output the classification sequentially. As shown, the Al model 10 classified positive overdose data points just over one minute after overdose induction as indicated by data point 44.

[0045] Referring now to FIG. 7B, the performance of the Al model 19 of the present disclosure was compared with two other machine learning models: recurrent neural network (“RNN”) and artificial neural network (“ANN”). A threshold gated method (i.e., below 50% baseline) and random chance were also included in the comparison. As shown in FIG. 7B, the receiver operating characteristic (“ROC”) curves of all three machine learning models had a higher area under the curve (i.e., an “AUC” of greater than 0.90) value than the conventional method. The AUC of the CNN model is shown by data series 50. The AUC of the RNN model is shown bythe data series 52. The AUC of the ANN model is shown by the data series 54. This verified the feasibility of the present framework. The CNN model of the Al model 19 had the best performance with an AUC of 0.969.Training and Validating Deep Neural Network

[0046] Animal WBP data were generated from 78 individual mice. All data were sectioned into fixed window lengths of 5 minutes and further split 50 / 50 between training and validation. For training, data were manually labelled as “healthy” or “post-overdose induction.” The CNN architecture has a total of three alternations between convolution layers and BN-ReLu activation functions (combined batch normalization and rectified linear unit). After the global pooling, SoftMax was used to generate the final output. To enable the automatic patch treatment, a Python script was implemented to handle the I / O of deep learning: it was listening to the classification outputs with a 30-second refreshing rate and automatically activated the acoustofluidic patch based on the overdose probability.

[0047] As shown in FIG. 7A, the validation loss 58 of the Al model 19 of the present disclosure decreases to a stable level after approximately 200 to 300 iterations of training loss 60. As shown, the validation loss is slightly higher than the training loss, which indicates an optimal case.Intelligent Closed-Loop for Real-Time Treatment

[0048] A flowchart depicting one embodiment of an intelligent closed-loop system according to the present disclosure is shown in FIG. 8A. The real-time data corresponding to one or more disease markers 62 was collected by one or more sensors 64 from animal models and further processed to obtain sensor signals 66, which were used by the Al model 19 to classify conditions 70. With a refreshing rateof 30 seconds, the Al model 19 of the controller 18 would activate the therapeutic device 16 for treatment delivery 68 when the output classification indicated an unhealthy condition 70.

[0049] To validate its reversal efficacy, the intelligent patch test was compared with a conventional manual-checked threshold-based method. In this study, a 50% decrease in baseline was chosen as the trigger of the manual treatment. This threshold value (as the “unsafe” level of respiration) is more conservative than other reported studies (e.g., 10%-40%). The schematic illustration of the test procedure is shown in FIG. 8B. After collecting baseline data during the period before To, an opioid overdose was induced in healthy mice (i.e. , at To). When an overdose was detected by either method, a patch treatment was applied to deliver the same amount of naloxone (1 .5 mg) at Ty(see FIG. 2A). The respiratory parameters were continuously monitored by the WBP system until full recovery at Tz(see FIG. 2A).

[0050] FIG. 8C shows representative comparisons between the intelligent patch 14 of the present disclosure and manual treatment by threshold. The intelligent patch 14 achieved quicker overdose detection when compared to the manual threshold method. The average overdose detection time for the intelligent patch 14 was 12.6 minutes ± 1 .5 minutes post-induction, which was around 11 minutes faster than the conventional threshold method. Furthermore, these faster rapid reversal efforts contributed to significantly less overall overdose duration (FIG. 9A) as well as less duration of an “unsafe” respiration period. The overall “unsafe” respiration was quantified as the area under the 50% baseline. The data series 72 in FIG. 9A depicts representative minute volume tracings of the system according to the present disclosure. The data series 74 in FIG. 9B depicts representative minute volume tracing of manual patch tests. The time points 76 in FIG. 9A represent overdose detection by the controller 18 and the time points 78 represent more than a50% decrease from baseline (“unsafe” overdose symptoms). The duration of patch treatment is depicted by the shadowed area 80.

[0051] As illustrated by the comparison between FIG. 9A and 9B, the controller 18 achieved faster overdose detection than the manual threshold group (comparing point 82 to point 84). Additionally, the use of the present system demonstrated better reversal effects with significantly less overdose duration, as well as a shorter “unsafe” duration of overdose symptoms (comparing points 78 to points 86).

[0052] Referring now to FIGS. 10A-E, faster overdose detection guided by the Al model 19 of the controller 18 is shown as compared to the conventional threshold method. The timepoints 90 represent overdose detection by the Al model 19 and the timepoints 92 represent over 50% decrease in ventilation volume. The Al model 19 was able to detect overdose 11 ± 5.2 minutes earlier than the <50% baseline threshold.

[0053] Any directional references used with respect to any of the figures, such as right or left, up or down, or top or bottom, are intended for convenience of description, and do not limit the present disclosure or any of its components to any particular positional or spatial orientation. Additionally, any reference to rotation in a clockwise direction or a counter-clockwise direction is simply illustrative. Any such rotation may be implemented in the reverse direction as that described herein.

[0054] Although the foregoing text sets forth a detailed description of embodiments of the disclosure, it should be understood that the legal scope of the invention is defined by the words of the claims set forth at the end of this patent and equivalents. The detailed description is to be construed as exemplary only and does not describe every possible embodiment. Numerous alternative embodiments may be implemented, using either current technology or technology developed after the filing date of this patent, which would still fall within the scope of the claims.

[0055] The following additional considerations apply to the foregoing description. Throughout this specification, plural instances may implement components, operations, or structures described as a single instance. Although individual operations of one or more methods are illustrated and described as separate operations, one or more of the individual operations may be performed concurrently, and nothing requires that the operations be performed in the order illustrated. Structures and functionality presented as separate components in example configurations may be implemented as a combined structure or component. Similarly, structures and functionality presented as a single component may be implemented as separate components. These and other variations, modifications, additions, and improvements fall within the scope of the subject matter herein.

[0056] In various embodiments, a hardware module may be implemented mechanically or electronically. For example, a hardware module may comprise dedicated circuitry or logic that is permanently configured (e.g., as a special-purpose processor, such as a field programmable gate array (FPGA) or an applicationspecific integrated circuit (ASIC)) to perform certain operations. A hardware module may also comprise programmable logic or circuitry (e.g., as encompassed within a general-purpose processor or other programmable processor) that is temporarily configured by software to perform certain operations. It will be appreciated that the decision to implement a hardware module mechanically, in dedicated and permanently configured circuitry, or in temporarily configured circuitry (e.g., configured by software) may be driven by cost and time considerations.

[0057] Accordingly, the term “hardware module” should be understood to encompass a tangible entity, be that an entity that is physically constructed, permanently configured (e.g., hardwired), or temporarily configured (e.g., programmed) to operate in a certain manner or to perform certain operations described herein. Considering embodiments in which hardware modules are temporarily configured (e.g., programmed), each of the hardware modules need notbe configured or instantiated at any one instance in time. For example, where the hardware modules comprise a general-purpose processor configured using software, the general-purpose processor may be configured as respective different hardware modules at different times. Software may accordingly configure a processor, for example, to constitute a particular hardware module at one instance of time and to constitute a different hardware module at a different instance of time.

[0058] Hardware modules may provide information to, and receive information from, other hardware modules. Accordingly, the described hardware modules may be regarded as being communicatively coupled. Where multiple of such hardware modules exist contemporaneously, communications may be achieved through signal transmission (e.g., over appropriate circuits and buses) that connect the hardware modules. In embodiments in which multiple hardware modules are configured or instantiated at various times, communications between such hardware modules may be achieved, for example, through the storage and retrieval of information in memory structures to which the multiple hardware modules have access. For example, one hardware module may perform an operation and store the output of that operation in a memory device to which it is communicatively coupled. A further hardware module may then, at a later time, access the memory device to retrieve and process the stored output. Hardware modules may also initiate communications with input or output devices, and may operate on a resource (e.g., a collection of information).

[0059] The various operations of example methods described herein may be performed, at least partially, by one or more processors that are temporarily configured (e.g., by software) or permanently configured to perform the relevant operations. Whether temporarily or permanently configured, such processors may constitute processor-implemented modules that operate to perform one or more operations or functions. The modules referred to herein may, in some example embodiments, comprise processor-implemented modules.

[0060] Similarly, the methods or routines described herein may be at least partially processor-implemented. For example, at least some of the operations of a method may be performed by one or more processors or processor-implemented hardware modules. The performance of certain of the operations may be distributed among the one or more processors, not only residing within a single machine, but deployed across a number of machines. In some example embodiments, the processor or processors may be located in a single location (e.g., within a home environment, an office environment or as a server farm), while in other embodiments the processors may be distributed across a number of locations.

[0061] The performance of certain of the operations may be distributed among the one or more processors, not only residing within a single machine, but deployed across a number of machines. In some example embodiments, the one or more processors or processor-implemented modules may be located in a single device or geographic location (e.g., within a home environment, an office environment, or a server farm). In other example embodiments, the one or more processors or processor-implemented modules may be distributed across a number of devices or geographic locations.

[0062] In addition, the aspects and functionalities described herein may operate over distributed systems (e.g., cloud-based computing systems and / or network-based computing systems), where application functionality, memory, data storage and retrieval and various processing functions may be operated remotely from each other over a distributed computing network, such as the Internet or an intranet. User interfaces and information of various types may be displayed via onboard computing device displays or via remote display units associated with one or more computing devices. For example, user interfaces and information of various types may be displayed and interacted with on a wall surface onto which user interfaces and information of various types are projected. Interaction with the multitude of computing systems with which aspects of the invention may bepracticed include, keystroke entry, touch screen entry, voice or other audio entry, gesture entry where an associated computing device is equipped with detection (e.g., camera) functionality for capturing and interpreting user gestures for controlling the functionality of the computing device, and the like.

[0063] Unless specifically stated otherwise, use herein of words such as “processing,” “computing,” “calculating,” “determining,” “presenting,” “displaying,” or the like may refer to actions or processes of a machine (e.g., a computer) that manipulates or transforms data represented as physical (e.g., electronic, magnetic, or optical) quantities within one or more memories (e.g., volatile memory, nonvolatile memory, or a combination thereof), registers, or other machine components that receive, store, transmit, or display information.

[0064] As used herein any reference to “one embodiment” or “an embodiment” means that a particular element, feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment. The appearances of the phrase “in one embodiment” in various places in the specification are not necessarily all referring to the same embodiment.

[0065] Some embodiments may be described using the expression “coupled” and “connected” along with their derivatives. For example, some embodiments may be described using the term “coupled” to indicate that two or more elements are in direct physical or electrical contact. The term “coupled,” however, may also mean that two or more elements are not in direct contact with each other, but yet still cooperate or interact with each other. The embodiments are not limited in this context.

[0066] Additionally, some embodiments may be described using the expression “communicatively coupled," which may mean (a) integrated into a single housing, (b) coupled using wires, or (c) coupled wirelessly (i.e. , passing data I commands back and forth wirelessly) in various embodiments.

[0067] As used herein, the terms “comprises,” “comprising,” “includes,” “including,” “has,” “having” or any other variation thereof, are intended to cover anon-exclusive inclusion. For example, a process, method, article, or apparatus that comprises a list of elements is not necessarily limited to only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus.

[0068] In addition, use of the “a” or “an” are employed to describe elements and components of the embodiments herein. This is done merely for convenience and to give a general sense of the description. This description, and the claims that follow, should be read to include one or at least one and the singular also includes the plural unless it is obvious that it is meant otherwise.

[0069] The patent claims at the end of this patent application are not intended to be construed under 35 U.S.C. § 112(f) unless traditional means-plus-function language is expressly recited, such as “means for” or “step for” language being explicitly recited in the claim(s).

Claims

WHAT IS CLAIMED IS:1 . A system for detecting and reversing an acute health condition, comprising: at least one wearable sensor configured to sense at least one health indicator of a user; a controller configured to receive signals from the at least one wearable sensor representing the at least one health indicator, the controller including an Al model trained to determine from the received signals whether the user is in an acute health condition; and at least one therapeutic device configured to deliver a treatment to the user in response to a determination that the user is in the acute health condition.

2. The system of claim 1 , wherein the at least one wearable sensor includes a respiratory sensor configured to sense breathing behavior of the user.

3. The system of claim 1 , wherein the at least one health indicator is at least one of a breathing pattern of the user, a blood pressure of the user, a blood oxygen level of the user, or a biochemical molecule.

4. The system of claim 1 , wherein the at least one therapeutic device and the controller are mounted to a patch worn by the user.

5. The system of claim 1 , wherein the controller is part of a portable computing device and receives the signals from the at least one wearable sensor over a wireless communications link.

6. The system of claim 1 , wherein the acute health condition is an acute disease condition.

7. The system of claim 6, wherein the acute disease condition is an opioid overdose condition.

8. The system of claim 1 , wherein the therapeutic device is an acoustofluidic patch configured to deliver a programmable dosing of an opioid antagonist.

9. The system of claim 8, wherein the acoustofluidic patch includes a hollow microneedle.

10. The system of claim 1 , further comprising a communication interface in communication with the controller, the communication interface being configured to transmit a notification requesting healthcare assistance to an emergency personnel device via a wireless network in response to a determination that the user is in an acute health condition.11 . The system of claim 1 , wherein the Al model includes a convolutional neural network.

12. The system of claim 11 , wherein the convolutional neural network includes three layers of one-dimensional convolutions.

13. A method of detecting and reversing an acute health condition, comprising: sensing at least one health indicator of a user with at least one wearable sensor; receiving, by a controller, signals from the at least one wearable sensor representing the at least one health indicator;determining, by an Al model of the controller, from the received signals whether the user is in an acute health condition; responding, by the controller, to a determination that the user is in an acute health condition by activating at least one therapeutic device to deliver a treatment to the user.

14. The method of claim 13, wherein sensing at least one health indictor includes sensing breathing behavior of the user by a respiratory sensor.

15. The method of claim 13, wherein the at least one health indicator is at least one of a breathing pattern of the user, a blood pressure of the user, a blood oxygen level of the user, or a biochemical molecule.

16. The method of claim 13, further comprising attaching a patch including the at least one therapeutic device and the controller to the user.

17. The method of claim 13, wherein the controller is part of a portable computing device and receives the signals from the at least one wearable sensor over a wireless communications link.The method of claim 13, wherein the acute health condition is an acute disease condition.

19. The method of claim 18, wherein the acute disease condition is an opioid overdose condition.

20. The method of claim 13, wherein the therapeutic device is an acoustofluidic patch configured to deliver a programmable dosing of an opioid antagonist.21 . The method of claim 13, further comprising transmitting, in response to a determination that the user is in an acute health condition, a notification requesting healthcare assistance to an emergency personnel device via a wireless network by a communication interface in communication with the controller.

22. The method of claim 13, wherein the Al model includes a convolutional neural network.

23. The method of claim 22, wherein the convolutional neural network includes three layers of one-dimensional convolutions.

24. A system for detecting and reversing an acute health condition, comprising: at least one wearable sensor configured to sense at least one health indicator of a user; and a controller configured to receive signals from the at least one wearable sensor representing the at least one health indicator, the controller including an Al model trained to determine from the received signals whether the user is in an acute health condition; wherein, in response to a determination of an acute health condition, the controller is configured to provide a notification to the user to operate a therapeutic device to deliver a treatment for the acute health condition.

Images