Systems and methods for finger ring device for health monitoring

Abstract

Methods and systems are herein provided for acquiring and analyzing health monitoring data acquired by a wearable finger ring device. In one example, a method comprises acquiring sensor data with one or more motion sensors of a wearable finger ring device; analyzing the sensor data to determine one or more quantities of interest; and outputting the sensor data and the one or more quantities of interest to one or more remote devices.

Description

SYSTEMS AND METHODS FOR FINGER RING DEVICE FOR HEALTH MONITORINGCROSS REFERENCE TO RELATED APPLICATIONS

[0001] The present application claims priority to U.S. Provisional Application. No. 63 / 608,761, entitled “SYSTEMS AND METHODS FOR FINGER RING DEVICE FOR HEALTH MONITORING”, filed on December 11, 2023. The entire contents of the above-listed application(s) are hereby incorporated by reference for all purposes.TECHNICAL FIELD

[0002] Embodiments of the subject matter disclosed herein relate to health monitoring devices, and more particularly, systems and methods for health monitoring in real time through a finger wearable ring device.BACKGROUND

[0003] Regular health monitoring is important for accomplishing long term health goals as well as detecting health changes over time. Vital signs are monitored for patients at various intervals depending on setting. For example, vital signs such as heart rate, blood pressure, and the like are monitored at yearly or other long-term intervals in an outpatient setting for healthy patients, while vital signs are monitored every few hours in inpatient settings for patients demanding acute care. As examples, blood pressure may be determined via a blood pressure cuff and heart rate may be determined via a pulse oximeter device. Other parameters such as gait, sleep habits, and the like, are more difficult to quantify, however.BRIEF DESCRIPTION

[0004] In one example, a method comprises acquiring sensor data with one or more motion sensors of a wearable finger ring device, analyzing the sensor data to determine one or more quantities of interest, and outputting the sensor data and the one or more quantities of interest to one or more remote devices. In some examples, analyzing the sensor data may comprise deploying one or more neural networks to determine the one or more quantities of interest based on a determined activity status of the wearer. In other examples, analyzing the sensor data may usealgorithms to determine the one or more quantities of interest based on a determined activity status of the wearer.

[0005] It should be understood that the brief description above is provided to introduce in simplified form a selection of concepts that are further described in the detailed description. It is not meant to identify key or essential features of the claimed subject matter, the scope of which is defined uniquely by the claims that follow the detailed description. Furthermore, the claimed subject matter is not limited to implementations that solve any disadvantages noted above or in any part of this disclosure.BRIEF DESCRIPTION OF THE DRAWINGS

[0006] Various aspects of this disclosure may be better understood upon reading the following detailed description and upon reference to the drawings in which:

[0007] FIG. 1 shows a block diagram of an exemplary embodiment of a finger ring health monitoring system, in accordance with one or more embodiments of the present disclosure;

[0008] FIG. 2 shows a block diagram of an exemplary embodiment of a neural network training system for training a neural network, in accordance with one or more embodiments of the present disclosure;

[0009] FIG. 3 shows a flowchart illustrating an exemplary method for training a neural network, in accordance with one or more embodiments of the present disclosure;

[0010] FIG. 4 shows a flowchart illustrating an exemplary method analyzing data acquired by a finger ring health monitoring system, in accordance with one or more embodiments of the present disclosure;

[0011] FIG. 5A shows an example finger ring device, in accordance with one or more embodiments of the present disclosure;

[0012] FIG. 5B shows the example finger ring device of with an outer layer; and

[0013] FIG. 6 shows an exploded view of the finger ring health monitoring device of FIG. 5.

[0014] FIG. 7 shows example graphs of acquired and normalized sensor data.

[0015] The drawings illustrate specific aspects of the described systems and methods.Together with the following description, the drawings demonstrate and explain the structures, methods, and principles described herein. In the drawings, the size of components may be exaggerated or otherwise modified for clarity. Well-known structures, materials, or operations arenot shown or described in detail to avoid obscuring aspects of the described components, systems and methods.DETAILED DESCRIPTION

[0016] Methods and systems are provided herein for acquisition and analysis of health monitoring data via a finger ring device. Analysis of acquired health monitoring data may be accomplished with one or more neural networks. Some analysis of deterministic parameters like finger range of motion and reaction times can be conducted using conventional inertial mechanization algorithms and statistical methods. Various health parameters, including vital signs, are monitored for patients both in and out of acute care settings. In some examples, physical activity parameters, such as steps, distance, speed, cadence, as well as various sleep parameters, can be monitored by a variety of personal health monitoring devices. Such personal health monitoring devices may be worn on a wrist (e.g., a watch device), a finger (e.g., a ring device), or other body part. These wearable monitoring devices may be used to detect motion and position and to infer parameters such as physical activity (e.g., steps, distance, speed, etc.), rest vs activity statuses, and sleep patterns.

[0017] Continuous monitoring of various health parameters for detection of abnormalities, such as falls, has been included in some wearable devices, notably those worn on the wrist. However, certain movements of the hands and fingers are more subtle and cannot be analyzed by wrist devices. Further, in some examples, variations in motion and orientation of fingers may provide additional insight into parameters, and changes of those parameters, such as mental status and cognitive function. Humans constantly use their hands to interact and engage both intentionally and spontaneously with the environment around them. Certain natural movements may be associated with various health statuses and health status changes. However, continuous monitoring of finger movement and orientation while analyzing the sensed motion information to detect abnormalities or changes in health status in real-time is challenging.

[0018] Systems and methods are therefore herein provided for a wearable finger ring device health monitoring system that is configured to acquire data, including motion and orientation data, and analyze the acquired data to assess a plurality of quantities of interest. Such quantities of interest may include physical activity status, posture, gait, sleep, mental status, and / or cognitive function. Analysis of such quantities of interest may allow for detection of abnormalities, such asgait changes, losses of balance, falls, mental status changes, and more, which may be indicators of health status. Further, progress of intervention and / or therapies may be determined based on the analyzed data.

[0019] In some examples, analysis of acquired data may include deploying one or more trained neural networks. The one or more trained neural networks may be trained on training pairs for particular quantities of interest. For example, a first neural network may be trained for analysis of walking or other active statuses and a second neural network may be trained for analysis of sleep statuses. In this way, depending on a determined activity status, a corresponding neural network may be deployed to analyze acquired data. The ring device may therefore allow for analysis of acquired data and detection of abnormal states.

[0020] In an embodiment, health monitoring data may be acquired and analyzed by a finger ring health monitoring system, such as the health monitoring system 102 of FIG. 1. The health monitoring system may include one or more neural network models that take as input raw acquired data, and output a corresponding analysis (e.g., a label, a trend, a change, etc.) thereof. Each of the one or more neural network models may be trained by following one or more steps of the method of FIG. 3, as described in relation to the neural network training system of FIG. 2. Analysis of acquired data, in some examples via the one or more neural networks, is described by a flowchart in FIG. 4. An exemplary finger ring device and components thereof are depicted in FIGS. 5-6. Example sensor data is shown in graphs in FIG. 7.

[0021] Referring to FIG. 1, a finger ring health monitoring system 100 is shown. In some examples, the finger ring health monitoring system 100 may include, at least in part, a wearable finger ring device 101. In some examples, the finger ring health monitoring system 100 may comprise a microcomputer 102, an inertial measurement unit (IMU) 136, and a communication unit 138 included within the wearable finger ring device 101. In some examples, one or more additional sensors 140 may also be included in the wearable finger ring device 101. The IMU 136 may comprise a plurality of motion sensors, including orthogonal accelerometers, gyroscopes, and / or magnetometers. The IMU 136 may be configured to acquire data of motion and orientation. In some examples, the IMU 136 may acquire data continuously and in real-time. The communication unit 138 may comprise a radio communication device, a Bluetooth ™ device, or other device capable of transmitting and / or receiving signals, including signals that include sensor data. The one or more additional sensors 140 may comprise temperature sensors, optical sensors,and / or the like. For example, temperature sensors may be configured to detect temperature of a user and optical sensors may be configured to sense reflected light in order to determine parameters such as heart rate and respiratory rate.

[0022] In some examples, the microcomputer 102 may be communicatively coupled to a display device 134 and a user input device 132. For example, the microcomputer 102 may be communicatively coupled to the display device 134 and the user input device 132 via the communication unit 138. In some examples, the microcomputer 102 may be additionally in communication with one or more other health monitoring devices, such as a watch style device worn on a wrist of the user.

[0023] In some embodiments, at least a portion of the microcomputer 102 is communicably coupled to the IMU 136, communication unit 138, and / or one or more additional sensors 140 via wired and / or wireless connections. In some embodiments, the display device 134 and / or the user input device 132 may be disposed at a separate device than the wearable finger ring device 101. For example, the user input device 132 and the display device 134 may be included in a smart phone, tablet, workstation, desktop computer, or virtually any other similar computing device. In other examples, the wearable finger ring device 101 may comprise the display device 134 and / or the user input device 132 (e.g., via a touchscreen display). Further, while the microcomputer 102 is shown and described as being located within the wearable finger ring device 101, however it should be understood that in other examples, the microcomputer 102 may be located remotely and in communication with the IMU 136 and one or more additional sensors 140 so as to analyze the acquired data at the remote location. In this situation, the finger ring may utilize the processor 104 with reduced resources and lower power consumption. The processor’s role may include collection of data from the sensors, including the IMU 136 and the one or more additional sensors 140, execution of calculations, and transmission of this information to a remote device.

[0024] In some embodiments, the finger ring health monitoring system 100 may incorporate various types of haptic feedback in a haptic feedback system 144 that is communicatively coupled to the microcomputer 102. These can include, but are not limited to, commonly used methods such as vibration, electrical stimulation, force feedback, or even thermal changes. The haptic feedback system 144, in its simplest use, may function akin to a cell phone in vibration mode, transmitting coded messages to the wearer. However, the system may take on additional importance in the context of health monitoring. Specifically, the haptic feedback system 144 may be employed toassess reflex timing and reaction speed in individuals. This may be accomplished by identifying whether the wearer’s finger moves subsequent to a triggered haptic action, for example thereby akin a neurological reflex test. The ability of the system to time precisely the reaction time may be used for assessing cognitive function / dysfunction as a measure of mental processing speed as well as onset of neurodegenerative disease such as Parkinson’s and / or Alzheimer’s.

[0025] The haptic feedback system 144 of the finger ring health monitoring system 100 may be implemented through one or more methods. For example, a first strategy may be a proactive strategy, whereby the wearer is prepared for the receipt of the haptic signal. Consequently, the wearer may anticipate a reaction, translating into a movement of their finger in response to the stimulation, either preemptively or as an unanticipated reflex. A second strategy may be a spontaneous strategy, whereby the haptic feedback is unexpectedly triggered when the wearer is in a resting state, allowing for observation of spontaneous reactions. This methods may significantly aid in identifying the wearer’s activity status and could potentially help with early detection of neurological disorders or age-related diseases that are evident in abnormal physical responses, or balance and fall risk.

[0026] Additionally, the haptic feedback system 144 of the finger ring health monitoring system 100 may be utilized to discern sleep stages. By applying haptic impulses during sleep and monitoring responses, different sleep stages, for example from REM sleep characterized by decreased responsiveness due to body paralysis to lighter sleep stages like stage 1 or stage 2 characterized by higher responsiveness to haptic feedback, can be identified.

[0027] In this way, the finger ring device herein disclosed includes specialized hardware components, such as the IMU, the haptic feedback system, and other sensors that are configured specifically for acquiring and processing motion and other sensor data from the user’s finger.

[0028] The microcomputer 102 may include a processor 104 configured to execute machine readable instructions stored in non-transitory memory 106. Processor 104 may be single core or multi-core, and the programs executed thereon may be configured for parallel or distributed processing. In some embodiments, the processor 104 may optionally include individual components that are distributed throughout two or more devices, which may be remotely located and / or configured for coordinated processing. In some embodiments, one or more aspects of the processor 104 may be virtualized and executed by remotely-accessible networked computing devices configured in a cloud computing configuration.

[0029] In certain cases, non-transitory memory 106 of the finger ring health monitoring system 100 may comprise an algorithm module 116 configured to conduct deterministic analysis of data from IMU 136 and the one or more additional sensors 140. The algorithm module 116 may employ well-established techniques to gauge factors such as finger orientation, range of motion, and temporal parameters. Such analysis may aid in evaluating reflex timing and reaction speed, which may be parameters central to health monitoring. Specifically, these metrics provide valuable contributions to comprehending and assessing cognitive functionality, signaling potential emergence of neurodegenerative conditions, determining sleep stages, assess balance and fall risk, and monitoring overall health status.

[0030] Non-transitory memory 106 may store a neural network module 108, a network training module 110, an inference module 112, and sensor data 114. Neural network module 108 may include a deep learning network and instructions for implementing the deep learning network to analyze data acquired by the finger ring health monitoring system 100, including the sensor data 114, as described in greater detail below. Neural network module 108 may include one or more trained and / or untrained neural networks and may further include various data, or metadata, pertaining to the one or more neural networks stored therein.

[0031] As an example, the neural network module 108 may comprise different neural networks trained for different patient statuses. For example, a first neural network may be trained to analyze patient data obtained during a first patient status (e.g., sleep) and a second neural network may be trained to analyze patient data obtained during a second patient status (e.g., walking). Additionally, the neural network 108 may comprise different neural networks trained for different types of acquired data, such as subtle movements vs large / gross movements. In one example, the system may deploy one or more of the neural networks according to the data that is acquired. As a non-limiting example, the system may deploy a first neural network trained to analyze data of subtle movements acquired during sleep in response to determining that acquired data includes subtle movements during a sleep state and a second neural network trained to analyze data of large movements acquired during walking in response to determining that acquired data includes gross limb movement during walking.

[0032] Training module 110 may comprise instructions for training one or more of the neural networks implementing a deep learning model stored in neural network module 108. In particular, training module 1 10 may include instructions that, when executed by the processor 104, causemicroprocessor 102 to conduct one or more of the steps of method 300 fortraining the one or more neural networks in a training stage, discussed in more detail below in reference to FIGS. 2 and 3. In some embodiments, training module 110 includes instructions for implementing one or more gradient descent algorithms, applying one or more loss functions, and / or training routines, for use in adjusting parameters of the one or more neural networks of neural network module 108. Non- transitory memory 106 also stores an inference module 112 that comprises instructions for analyzing acquired data with the trained deep learning model.

[0033] Non-transitory memory 106 further stores sensor data 114. Sensor data 114 may include for example, data acquired with the IMU 136, including data acquired with the plurality of motion sensors thereof. Sensor data 114 may further include temperature data, optically acquired data, and more, as acquired by the one or more additional sensors 140. In this way, the sensor data 114 may store data acquired of the user via the wearable finger ring device 101. In some examples, the stored sensor data 114 may be acquired and stored in real-time (e.g., without an intentional delay). Further, the sensor data 114 may be time stamped and time-aligned so as to allow for analysis of change over time. In some examples, the sensor data 114 may include acquired data over a defined time period, for example over 7 days, where data older than the defined time period is automatically deleted from the memory 106. In other examples, the sensor data 114 may include all acquired sensor data and / or all acquired sensor data that has yet to be downloaded for long term storage on a separate remote device communicatively coupled to the wearable finger ring device 101, for example to a storage device 142. The sensor data 114 may be used as inputs for the one or more neural networks in order for the data to be analyzed to determine a plurality of quantities of interest, including information about sleep, gait patterns, mental status / cognition, and more, including detecting abnormalities.

[0034] In some embodiments, the non-transitory memory 106 may include components disposed at two or more devices, which may be remotely located and / or configured for coordinated processing. In some embodiments, one or more aspects of the non-transitory memory 106 may include remotely-accessible networked storage devices configured in a cloud computing configuration.

[0035] The microcomputer 102 may be operably / communicatively coupled to the user input device 132 and the display device 134. As described previously, user input device 132 may comprise one or more of a touchscreen, a keyboard, a mouse, a trackpad, a motion sensing camera,or other device configured to enable a user to interact with and manipulate data within the microcomputer 102. Display device 134 may include one or more display devices utilizing virtually any type of technology. In some embodiments, display device 134 may comprise a computer monitor, and may display notifications indicating outputted analyses and / or raw data (e g., graphs, timelines, etc ). In examples in which the microcomputer 102 is a remote device separate from the wearable finger ring device 101, display device 134 may be combined with processor 104, non-transitory memory 106, and / or user input device 132 in a shared enclosure. Alternatively, display device 134 may be peripheral display devices and may comprise a monitor, touchscreen, projector, or other display device known in the art, which may enable a user to view data in raw form, in graphs and / or timelines, or in another form, and / or interact with various data stored in or downloaded from non-transitory memory 106.

[0036] The finger ring health monitoring system 100 as herein described may provide for more nuanced data collection and analysis. Fingers have a tendency to demonstrate more spontaneous and / or independent movements compared to the whole hand or wrist, thus the monitoring system herein disclosed may capture data that would otherwise be missed by other devices. Moreover, due to their heightened sensitivity relative to the wrist, fingers provide a more suitable platform to implement the haptic feedback system for the testing of reactive movements as previously described.

[0037] It should be understood that finger ring health monitoring system 100 shown in FIG. 1 is for illustration, not for limitation. Another appropriate monitoring systems may include more, fewer, or different components.

[0038] Referring to FIG. 2, an example of a neural network training system 200 is shown, which may be used to train one or more neural networks, such as neural network 202. The neural network 202 may be any of the one or more neural networks described with respect to FIG. 1. In some examples, the neural network training system 200 may be an example of one or more neural network training systems stored in memory (e.g., memory 106) used to train neural networks. Each of the one or more neural networks may be trained for a particular quantity of interest, parameter, activity status / level, etc. as will be herein described. In some examples, each neural network may be trained by a separate neural network training systems. In other examples, a single neural network training system may be used to train multiple neural networks.

[0039] Each of the one or more neural networks, including neural network 202, may be trained to analyze data acquired by an IMU and / or one or more sensors, including temperature and optical sensors, such as IMU 136 and one or more additional sensors 140 of FIG. 1, in accordance with one or more operations described in greater detail below in reference to method 300 of FIG. 3. The neural network training system 200, and in some examples other neural network training systems, may be implemented by a health monitoring system, such as finger ring health monitoring system 102 of FIG. 1, to train one or more neural networks to analyze data acquired with a finger ring device of the health monitoring system. In some examples, data processing may be executed via a microcomputer within the finger ring device. In other examples, data processing may be executed remotely via a device (e.g., a smart phone, smart watch, etc.) communicatively coupled to the microcomputer within the finger ring device.

[0040] The neural network 202 herein described may be an example of one of the one or more neural networks included in the health monitoring system. While each of the one or more neural networks may be trained on a different training dataset and may be deployed to analyze different acquired data, the training systems therefore may be similar and / or the same. In some embodiments, the neural network 202 may be a deep neural network with a plurality of hidden layers. In one embodiment, neural network 202 is a convolutional neural network (CNN).

[0041] Neural network 202 may be stored within a neural network module 201 of neural network training system. Neural network module 201 may be a non-limiting example of neural network module 108 of microcomputer 102 of FIG. 1. Neural network training system 200 also includes a training module 204, which includes a training dataset comprising a plurality of training pairs of data, such as pairs divided into training pairs 206 and test pairs 208. Training module 204 may be a non-limiting example of training module 110 of microcomputer 102 of FIG. 1.

[0042] A number of training pairs 206 and test pairs 208 may be selected to ensure that sufficient training data is available to prevent overfitting, whereby the neural network 202 learns to map features specific to samples of the training set that are not present in the test set.

[0043] Each pair of the training pairs 206 and the test pairs 208 comprises an input and a corresponding target. In some examples, the neural network 202 may be a supervised network. In such examples, the inputs of the training pairs 206 may be sensor data 212 and the targets of the training pairs 206 may be corresponding labels 216, wherein each sensor datum or group of sensor data has an associated corresponding label. As a non-limiting example, for a first neural networktrained on sleep data, the sensor data 212 may be sensor data acquired during sleep activity and the labels 216 may be labels such as rapid eye movement (REM) sleep, non-REM sleep, deep sleep, light sleep, and the like that corresponds to portions of the sensor data 212. The labels may note that the corresponding sensor data was captured during, for example, REM sleep or non-REM sleep. The labels may be based on, for example for sleep analysis, polysomnography data which allows for identification of various sleep cycles. As an example, when creating the sensor data 212 and the corresponding labels 216 for the first neural network, a subject may undergo polysomnography while wearing an IMU device on a finger. The polysomnography analysis may identify sections of data that were captured during various sleep cycles, which can then be associated with the IMU sensor data that was acquired at the same time as the polysomnography data. Labels may therefore be assigned to sections of data identified as being captured during the various sleep cycles. The IMU sensor data may be portions of the sensor data 212 and the labels assigned to sections of the IMU sensor data may be the corresponding labels 216. In this way, the first neural network may be trained for sleep sensor data with corresponding labels thereof.

[0044] During training, the first neural network may learn to distinguish between various types of sleep activity based on the labels associated with different sensor data. Other neural networks of the one or more neural networks may be trained in a similar fashion, including neural networks trained to analyze gait (e.g., during a walking activity), mental status (e.g., during rest), and cognitive function (e.g., during finger tapping activity). Further, in some examples, one or more neural networks may be specifically trained to distinguish between activity statuses, including walking, resting, sleeping, and the like. In other examples, activity statuses may be determined in a different manner, as will be described below.

[0045] The one or more neural networks herein described may have a variety of artificial intelligence (Al) model architectures including deep neural networks, transformers, random forests, and / or Gaussian processes, depending on setting and application. Architectures may be predetermined or selected in real-time when analyzing data based on accuracy, precision, recall, computational speed, demanded processing power, and / or memory storage demands. As such, while neural networks are herein described, it should be understood that any machine learning architecture may be used without departing from the scope of this disclosure.

[0046] Neural network training system 200 may include a dataset generator 210, which may be used to generate the training pairs 206 and the test pairs 208 of the training module 204. Sensordata 212 (e g., individual data or subsets of data) may be paired with corresponding labels thereof from the labels 216 by dataset generator 210. As an example, the labels 216 may include a predefined set of labels for a given scenario, for example sleep labels. The dataset generator 210 may assign a label to a datum or sets of datum of the sensor data 212. For example, a first chunk of sensor data that corresponds to REM sleep may be given the label “REM” and a second chunk of sensor data that corresponds to non-REM sleep may be given the label “non-REM”.

[0047] In some examples, the sensor data 212 may be generated via transfer learning, wherein data obtained under different settings is combined. For example, laboratory generated data and real-world acquired data may be combined in order to extract useful information from different sources of data and increase usefulness of separate datasets.

[0048] Further, data augmentation techniques may be employed to further enrich the dataset by taking advantage of invariants and symmetries (e g., spatial and / or temporal) in the data. As an example, gathering enough real data of falls of subjects may be difficult, and as such simulated data may be included in the training data to increase robustness thereof. Additionally, in some examples, supervised neural network models may be augmented with unsupervised learning capabilities tasked with identifying anomalous data signatures without using data labels. In this way, the models may be built even in situations where data labels are absent. Methods such as principal component analysis (PCA)-based anomaly scoring, local outlier factor, and / or one-class support vector machine may be employed for detection of anomalous patterns in finger ring sensor data (e.g., motion measurements, orientation measurements, temperature measurements, and the like).

[0049] Once each pair is generated, the pair may be assigned to either the training pairs 206 or the test pairs 208. In an embodiment, the pair may be assigned to either the training pairs 206 or the test pairs 208 randomly in a pre-established proportion. For example, the pair may be assigned to either the training pairs 206 or the test pairs 208 randomly such that 90% of the pairs generated are assigned to the training pairs 206, and 10% of the pairs generated are assigned to the test pairs 208. Alternatively, the pair may be assigned to either the training pairs 206 or the test pairs 208 randomly such that 85% of the pairs generated are assigned to the training pairs 206, and 15% of the pairs generated are assigned to the test pairs 208. It should be appreciated that the examples provided herein are for illustrative purposes, and pairs may be assigned to the trainingpairs 206 dataset or the test pairs 208 dataset via a different procedure and / or in a different proportion without departing from the scope of this disclosure.

[0050] Neural network training system 200 may include a validator 220 that validates the performance of the neural network 202 against the test pairs 208. The validator 220 may take as input a partially trained neural network 202 and a dataset of test pairs 208, and may output an assessment of the performance of the partially trained neural network 202 on the dataset of test pairs 208.

[0051] Once the neural network 202 has been validated, a trained neural network 222 (e.g., the validated neural network 202) may be used to generate a set of outputted labels 234 from a set of acquired (e.g., real) sensor data 232. For example, the acquired sensor data 232 may be acquired by the IMU and other sensors of a finger ring device, which may be a non-limiting example of the IMU 136 and the one or more additional sensors 140 of the wearable finger ring device 101 of FIG. 1. Trained neural network 222 may be stored within an inference module 221 of the finger ring health monitoring system (e.g., inference module 112 of FIG. 1).

[0052] As previously described, a first neural network may be trained on a first set of training pairs, a second neural network may be trained on a second set of training pairs, a third neural network may be trained on a third set of training pairs, and so on, via the neural network training system 200 and, in some examples, additional training systems. As an example, the first set of training pairs may include data and labels specific to sleep, the second set of training pairs may include data and labels specific to a walking activity, including falls and various gait abnormalities, the third set of training pairs may include data and labels specific to mental status, and a fourth set of training pairs may include data and labels specific to cognitive function. The data for each of the training pairs may include motion data acquired by motion sensors (e.g., motion sensors included in an IMU), temperature sensors, optical sensors, and the like, as well as data without labels (e.g., via unsupervised model augmentation, as previously described), in some examples. Furthermore, changes indicative to mental status and cognitive function may be further mapped to other disease states that include but are not limited to various infections, sepsis, stroke, drug overdoses, etc.

[0053] Additionally, the finger ring health monitoring system 100 and the training system 200 herein described may be configured to ascertain the wearer’s baseline patterns, forming a standard that can aid in detecting abnormalities. The baseline patterns may be learned as part of the trainingprocess of the one or more neural networks, as herein described, in some examples. Alternatively, the one or more neural networks may be trained on early analysis data when available. This understanding of the wearer’s normal state is pivotal, especially in scenarios concerning balance and mental status. For instance, individuals prone to falls might exhibit regular balance most days while occasionally experiencing unusual instances of balance loss that may culminate in falls. By establishing this baseline state, the system may be equipped to identify these anomalies amongst the everyday patterns. Consequently, during such deviations from the baseline, alerts may be triggered for the user or their caregivers. This not only ensures immediate attention to possible risks but also assists in ongoing health evaluation and potentially, preemptive interventions.

[0054] Establishing a wearer-specific baseline may be important not only for recognizing health variations, but also for tracking disease advancement or assessing the efficacy of medical interventions. The baseline may be a personalized marker that outlines the normal range for various parameters for the individual wearer. Long-term deviations from this established baseline could potentially indicate transformations in the disease’s path, for example either its progression or its remission. This information may furnish healthcare professionals with instantaneous insights into disease evolution or therapy effectiveness. Consequently, it may facilitate a more personalized, data-based approach to medical decision-making.

[0055] Referring now to FIG. 3, a flowchart illustrating a method 300 for training a neural network is shown. The neural network may be a non-limiting example of the neural network 202 of the neural network training system 200 of FIG. 2, according to an exemplary embodiment. The neural network may be one of a plurality of neural networks trained by the neural network training system 200. Method 300 may be executed by a processor of a finger ring health monitoring system, such as processor 104 of the microcomputer 102 of the finger ring health monitoring system 100 of FIG. 1. In one example, some operations of method 300 may be stored in non-transitory memory of the finger ring health monitoring system, for example in a training module such as the training module 110 of memory 106 of FIG. 1) and executed by the processor of the system. The neural network may be trained on training data comprising one or more sets of training pairs. Each training pair of the one or more sets of training pairs may comprise sensor data and corresponding labels thereof, in some examples.

[0056] At 302, method 300 includes accumulating sensor data and corresponding labels of a plurality of quantities of interest. Data may be laboratory generated data and / or real-world acquireddata. For example, as described above, data may be accumulated from real sources, wherein motion sensors and / or other sensors are employed to acquire data of movements, temperatures, heart rates, etc. of one or more users during various activities, statuses, etc. Variables for the real sources may be known such that labels, such as “fall”, “trip”, “delirium”, “normal sitting”, “normal walking”, “REM sleep”, and the like, may be assigned to corresponding chunks of the data. In some examples, disease states such as “sepsis” and “stroke” may also be included as labels. In other examples, post-processing may map analyzed data to disease potentials. For example, postprocessing may include applying rules to analyze predictions from a neural network to an increased risk for a certain disease.

[0057] In some examples, the real sources may be targeted towards a particular quantity of interest, for example sleep. And labels may be assigned to data or chunks of data based on that quantity of interest, for example REM, non-REM, deep sleep, light sleep, etc. Additionally, data may be accumulated that does not include known labels (e.g., during unsupervised augmentation of training data) and one or more methods may be applied to identify anomalies in and / or assign labels to such data.

[0058] At 304, method 300 includes generating training data with training pairs based on the accumulated sensor data and corresponding labels. As described previously, each training pair may comprise a datum or chunk of data and a corresponding label therefor, wherein the datum or chunk of data is an input and the corresponding label is a target. In some examples, the input and target may be paired by a dataset generator, such as the dataset generator 210 of the neural network training system 200 of FIG. 2. Once the pairs have been created, they may be divided into training pairs and test pairs, as described above with respect to FIG. 2.

[0059] At 306, method 300 includes training the neural network on the training pairs. More specifically, training the neural network on the pairs includes training the neural network to map the sensor data to the corresponding labels. In some embodiments, the neural network may comprise a generative neural network. In some embodiments, the neural network may comprise a generative neural network having a U-net architecture. In some embodiments, the neural network may include one or more convolutional layers, which in turn comprise one or more convolutional filters (e.g., a convoluted neural network architecture). The convolutional filters may comprise a plurality of weights, wherein values of the weights are learned during a training procedure. The convolutional filters may correspond to one or more features / patterns, thereby enabling the neuralnetwork to identify and extract features from the sensor data. In other embodiments, the neural network may not be a convolutional neural network, and may be a different type of neural network.

[0060] Training the neural network on the pairs may include iteratively inputting sensor data of each training pair into an input layer of the neural network. In some embodiments, each data point of the input sensor data may input into a distinct neuron of the input layer of the neural network. The neural network may map the input sensor data to a corresponding target label by propagating the input sensor data from the input layer, through one or more hidden layers, until reaching an output layer of the neural network. In some embodiments, the output of the neural network comprises a label that corresponds to the inputted sensor data.

[0061] The weights and biases of the neural network may be adjusted based on a difference between the outputted label and the target (e.g., ground truth) label of the relevant pair. The difference, as determined by a loss function, in some examples, may be back propagated through the neural learning network to update the weights and biases of the convolutional layers. In some embodiments, back propagation of the loss may occur according to a gradient descent algorithm, wherein a gradient of the loss function (a first derivative, or approximation of the first derivative) is determined for each weight and bias of the deep neural network. Each weight (and bias) of the neural network is then updated by adding the negative of the product of the gradient determined (or approximated) for the weight or bias with a predetermined or adaptive step size. Updating of the weights and biases may be repeated until the weights and biases of the neural network converge, or the rate of change of the weights and / or biases of the deep neural network for each iteration of weight adjustment are under a threshold.

[0062] In order to avoid overfitting, training of the neural network may be periodically interrupted to validate a performance of the neural network on the test pairs. In an embodiment, training of the neural network may end when a performance of the neural network on the test pairs converges (e.g., when an error rate on the test set converges on or to within a threshold of a minimum value). In this way, the neural network may be trained to more generalizable, and be able to more accurately generate labels for newly acquired sensor data that it has not previously seen during training.

[0063] In some embodiments, an assessment of the performance of the neural network may include a combination of a minimum error rate and a quality assessment, or a different function of the minimum error rates achieved on each pair of the test pairs and / or one or more qualityassessments, or another factor for assessing the performance of the neural network. It should be appreciated that the examples provided herein are for illustrative purposes, and other loss functions, error rates, quality assessments, or performance assessments may be included without departing from the scope of this disclosure.

[0064] As is previously described, the method 300 may be applied for training of a plurality of neural networks. For example, a first neural network may be trained based on a first set of accumulated data and corresponding labels and a second neural network may be trained based on a second set of accumulated data. The first neural network may be targeted for a first type of acquired sensor data, for example gait motion sensor data, and the first set of accumulated data and corresponding labels may comprise gait motion data and corresponding gait labels (e.g., normal walking, one or more abnormal gait types, falls, etc.). The second neural network may be targeted for a second type of acquired sensor data, for example sleep sensor data, and the second set of accumulated data and corresponding labels may comprise sleep data (e.g., motion data, temperature data, optically acquired data, etc.) and corresponding sleep labels (e.g., REM, non- REM, light, deep, etc.). In this way, depending on setting, a neural network trained for a particular setting may be deployed to analyze acquired data.

[0065] Finger digits, excluding the thumb, may move in a coordinated fashion, with the index finger demonstrating the most individual control. As such, the finger ring device as herein described may be worn on the index finger, though wearing the finger ring device on the middle, ring, or small finger may be possible as well. Further, which finger the device is worn on may be detectable and the data acquired may be analyzed according to the detected finger, for example using a trained neural network or other method. In some embodiments, more than one wearable finger ring device may be worn by a user. For example, a first ring device may be worn on a first hand and a second ring device may be worn on a second hand. Alternatively, a first ring device may be worn on a first finger (e.g., an index finger) of a first hand and a second ring device may be worn on a second finger (e.g., a ring finger) of the first hand. Data may be acquired by each of the finger ring devices and may be combined together prior to analysis. Additionally or alternatively, the user may also wear a smart watch that comprises an IMU for motion and orientation determination. Data from the one or more finger ring devices may be combined with or compared to data from the smart watch. Combining the data may increase robustness of analysis.Comparing data and analysis between the finger ring device(s) and the smart watch may allow for sanity checks.

[0066] Referring now to FIG. 4, a flowchart illustrating a method 400 for analyzing acquired sensor data, in some examples via deployment of one or more neural networks, such as the neural network 202 of FIG. 2, is shown. Method 400 may be executed by a processor of a finger ring health monitoring system, such as the finger ring health monitoring system 100 of FIG. 1. Some operations of method 400 may be stored in a non-transitory memory of the finger ring health monitoring system (e.g., in inference module 112 of the microcomputer 102 of FIG. 1) and executed by the processor. In various embodiments, the one or more neural networks may be trained as described above in reference to the method 300 of FIG. 3, on accumulated data and, in some examples, corresponding labels thereof.

[0067] At 402, method 400 includes acquiring data of a patient with a plurality of sensors of a finger ring device. In some examples, the finger ring device may be the wearable finger ring device 101 described with respect to FIG. 1, and may comprise an IMU with a plurality of motion sensors, including accelerometers, gyroscopes, and / or magnetometers. In some examples, the finger ring device may further comprise one or more other sensors, including temperature sensors, optical sensors, and / or the like. Acquiring the data may comprise employing the IMU and / or the one or more other sensors to acquire motion data, orientation data, temperature data, and / or optical data (e.g., for detection of heart rate and respiratory rate). The data may include both spontaneous movements of the finger as well as responsive movements, such as movements in response to haptic stimuli (e.g., from haptic feedback system 144 described with respect to FIG. 1). The data may be acquired in real-time and stored in memory of a microcomputer of the finger ring device, in one example. In some examples, the data may be acquired at a sampling frequency not in excess of 100 Hz, which may allow for adequate capture of details of motion and orientation while effectively managing data volume. The acquired data may include within a sliding window, the mean, variance, quantiles of linear and angular velocity and acceleration of the finger, and Fourier- transform of the time series into frequency domain, in some examples.

[0068] In some examples, motion data acquired may include subtle and large movements and orientation changes as well as intentional, unintentional, spontaneous, and deliberate movements and orientation changes of the patient’s finger on which the finger ring device is worn. In some examples, the motion data that is acquired may be compared to data acquired by a wrist-wearabledevice (e g., a health monitoring watch) worn on the same limb as the finger ring device. If the motion data acquired by the finger ring device matches that of the wrist-wearable device, thus indicating that the motion is gross movement or large movement detectable by the wrist-wearable device, the finger ring system may proceed to analyze the data, as described below, according to a neural network trained for gross or large movements. Conversely, in this example, when the acquired data does not match that of the wrist-wearable device, the finger ring system may proceed to analyze the data, as described below, according to a neural network trained for subtle movements. In this way, a more targeted neural network may be deployed for the specific circumstance detected, increasing processing efficiency as compared to deploying a larger / more generic model which may take more processing power and time to produce an output.

[0069] In another example, when the acquired data matches that of the wrist-wearable device mounted on the same limb as the finger ring device, the finger ring device may not proceed to analyze said motion data. Rather, the wrist device may analyze the data it acquires, for example via whatever algorithms or models (e.g., neural networks) the wrist device is configured to execute for analysis purposes. If the motion data acquired by the finger ring device is different from that of the wrist-wearable device, for example in instances in which the detected motion is subtle and not detected by the wrist-wearable device, the finger ring device may proceed to analyze the data as is described below. In this way, overall processing demands of the finger ring computing system may be reduced as data that may be analyzed already by another device (e.g., gross movement data detected by a wrist device) may be ignored, thus mitigating duplicate analysis.

[0070] The finger ring device may include known limits of finger extension, flexion, and lateral movement as based on anatomical limits. Acquired data that is outside of said limits may indicate to the system that drift or other abnormal acquisition has occurred. If detected, a notification may be presented on a display device, for example a remotely connected display device, indicating that re-calibration of the IMU is indicated. In this way, more accurate motion data may be provided by the systems herein described. Further, while a single finger ring device is described herein, it should be understood that more than one finger ring device of the present disclosure may be worn by a wearer. For example, the wearer may wear one finger ring device on one hand and a second finger ring device on the other hand, or the wearer may wear one finger ring device on an index finger of a specified hand and a second finger ring device on a ring fingerof the specified hand. As such, sensor data may be acquired from multiple sources and may be inputted for processing together to provide for more robust analysis.

[0071] In some examples, the acquired data may be offloaded or downloaded to a remote device, such as a workstation, desktop computer, a smart watch, a smart phone, etc., following acquisition of data. The acquired data may then proceed with analysis, including determination of activity status and deployment of neural networks for data analysis, at the remote device. Offloading the acquired data to the remote device may reduce computational demands of the finger ring device and therefore may increase longevity of a battery of the finger ring device. Alternatively, the acquired data may be analyzed be one or more neural networks and / or signal processing methods at the finger ring device. Further, in some examples, a portion of analysis may be employed at the finger ring device, such as timing of events and measuring finger kinematics, while other portions of the analysis, such as deployment of the one or more neural networks may be performed by the remote device.

[0072] At 404, method 400 includes determining a patient status based on the acquired data. The patient status may comprise an activity status or activity level and may be categorized as one of a plurality of predefined categories. In some examples, determination of the patient status may comprise deploying a neural network trained to detect a plurality of patient activity statuses based on acquired data. In other examples, other signal processing methods may be used for determination of patient activity status. In some examples, the patient status may be determined at a first time to inform which neural network to deploy and may not be checked again for a specified duration based on the determined activity status and an associated estimated rate of change. For example a rate of change in activity status may be less for a sleep activity than a walking activity. For example, following detection of sleep activity, repeated determination of activity status may be delayed a duration of an hour, two hours, or other specified timeframe in order to reduce computational demands.

[0073] In some examples, determination of patient activity status may be performed by the microcomputer of the finger ring device. Following determination of the patient activity status, data may be offloaded to one or more remote devices for further processing, including deployment of one or more neural networks based on the determined patient activity status.

[0074] At 406, method 400 includes analyzing the acquired data according to the determined patient activity status. Analyzing the acquired data may comprise assigning labels to individualdata points or to chunks of data in order to determine and / or detect various patterns, actions, etc. For example, for a determined patient status of sleep, labels may be assigned according to sleep patterns seen in the acquired data. As another example, for a determined patient status of walking, labels may be assigned according to walking patterns, such as normal walking, a fall, etc. In some examples, analysis of data may be executed by deployment of one or more neural networks, as noted at 408. Which neural network is deployed may be based on the determined patient status, as described above, whereby a neural network trained on sleep data is deployed during a determined sleep status, a neural network trained on walking data is deployed during a determined walking status, and so forth. Further, in some examples, the type of movement (e.g., subtle movement vs large movement) in the acquired data may also determine which neural network is deployed. In some examples, the window size of data that is entered into the various neural networks may be based on inherent dynamic timescales of the labels and how they change, for example walking may be a short time scale while sleeping may be a longer time scale, what size of neural network and computational

[0075] As a non-limiting example, data may be acquired by the IMU of the finger ring device while the patient is walking. A first section of acquired data may be labeled by a corresponding trained neural network as normal walking. A second section of the acquired data may include a change compared to the first section and may be labeled by the corresponding neural network as a fall. For example, the first section may comprise data consistent with cyclical arm swing and the second section may comprise data consistent with a protective behavior like bracing and orientation changes indicative of a fall.

[0076] As another non-limiting example, data may be acquired by the IMU and other sensors of the finger ring device while the patient is sitting / resting. A first section of acquired data may be labeled by a corresponding trained neural network as normal and a second section of the acquired data may be labeled as abnormal. The abnormal section may be labeled as such due to a change in finger motion or other parameter. In some examples, the abnormal section may be labeled according to a defined abnormality, such as mental status change.

[0077] As previously described, in some examples, as previously discussed, the finger ring device may be communicably coupled to a smart watch that also includes an IMU and processing systems to analyze data from the IMU. Data and analysis from the smart watch may be compared to data and analysis from the finger ring device as a sanity check. Further, data from the smartwatch IMU may be compared to the data from the finger ring device prior to analysis in order to determine whether to deploy a neural network and / or which neural network to deploy. Alternatively, data from the smart watch IMU and the finger ring device IMU may be combined together at a remote device and analyzed together.

[0078] At 410, method 400 includes determining if an abnormality is detected. One or more labels may be predefined as abnormalities and if analysis of the acquired data outputs a label that is predefined as an abnormality (YES at 410), method 400 may proceed to 412. If an outputted label is not predefined as an abnormality (NO at 410), method 400 may proceed to 416.

[0079] At 412, method 400 includes outputting a notification of the abnormality. Type of notification and where the notification is outputted may depend on the abnormality detected. As an example, detection of a fall may output a notification to a connected remote device for display, as noted at 414, such as the user’s smart phone, as well as to one or more other remote devices, including the user’s healthcare provider and emergency services. For example, the user may have one or more settings turned on or off that indicate parties to be contacted in the event of a detected fall event. As another example, detection of a change in mental status may cue output of a notification on a display device of a care provider device remotely connected to the finger ring device, such as during acute care settings where patient status is monitored continuously. In this way, detection of such abnormal events may be more efficient.

[0080] At 416, method 400 includes outputting acquired data to a remote device. In some examples, the finger ring device through which the data is acquired may have an internal memory of a microcomputer (e.g., memory 106 of the microcomputer 102 of FIG. 1). The internal memory of the finger ring device may store the acquired data and the acquired data may be processed and analyzed by the microcomputer (e.g., via one or more stored trained neural networks). In some examples the stored acquired data may be outputted from the finger ring device to a remote device for long-term storage. Outputting the acquired data may occur at regular intervals, upon manual request, or when data storage of the internal memory is full, or some combination thereof. The acquired data may be outputted both in raw form and in analyzed form, in some examples. The acquired data may be accessible via the remote device to which it was outputted for additional analysis. In this way, the processing efficiency of the finger ring device may be maintained and the acquired data may be accessible on a long term basis.

[0081] The method 400 may be executed in a continuous, iterative manner, whereby data is continuously acquired and analyzed in real-time or near real-time. Data may be analyzed as individual data points or sections of data over longer periods of time. Analysis of data over time may allow for determination of response to therapy, including physical / occupational therapy as well as medication therapy, changes in mental status or cognitive function, and more. Notifications may be outputted and displayed for relevant sections of data in real-time as well. It should be appreciated that data may continue to be acquired by the sensors while a notification is outputted and / or displayed.

[0082] In this way, as is herein presented, the method 400 may allow for a plurality of quantities of interest to be determined from analysis of data acquired with the finger ring device. As previously described, the various quantities of interest may include sleep patterns, gait patterns including abnormal gaits and falls, trips, slips, etc., mental statuses including changes, and cognitive functions. Further, various parameters including posture, activity level (e.g., speed, distance, heart rate statistics, etc.), and more may also be analyzed by the methods and systems herein. Additionally, changes over time for one or more of the quantities of interest and / or parameters may be determined, which may allow for monitoring of patient progress for a given intervention or treatment. For example, gait pattern and walking speed may be monitored for changes over time as a patient recovers after a spine surgery or undergoes physical therapy for a chronic knee injury, as non-limiting examples. As another example, sleep patterns may be monitored over time as a patient begins treatment for parasomnia, obstructive sleep apnea, or other condition. This may allow for a more robust analysis of continuously acquired motion data. Further, the data processing pipeline herein disclosed, which involves acquisition of sensor data from the finger ring device, determination of the user's activity status, and then deploying specialized neural networks trained on that activity status to analyze the sensor data allows for more robust health monitoring using the device.

[0083] Turning to FIGS. 5A, 5B, and 6, an exemplary finger ring device 500 is shown. In FIGS. 5A and 5B, the finger ring device 500 is depicted assembled and in FIG. 6, components of the finger ring device 500 are shown unassembled / exploded. The finger ring device 500 may be an electronic ring that is configured to acquire and analyze motion of a user’s finger according to methods as herein presented. The finger ring device 500 is a non-limiting example of wearable finger ring device 101 of FIG. 1.

[0084] The finger ring device 500 may comprise an inner layer 502, as shown specifically in FIG. 5A, a battery 504, a CPU 506, and one or more sensors 508. The one or more sensors 508 may comprise an IMU (e.g., IMU 136 of FIG. 1) and one or more additional sensors such as optical sensors (e.g., the one or more additional sensors 140 of FIG. 1). The CPU 506 may incorporate both a microcomputer and a communication system, such as e the microcomputer 102 and communication unit 138 described with respect to FIG. 1, in some examples. The CPU 506 may be mounted on a flexible printed circuit board (PCB) 510. The one or more sensors 508 may also be mounted on the flexible PCB 510. The flexible PCB 510 and the battery 504 may be coupled to the inner layer 502. As shown in FIG. 5B, an outer layer 520 may be present to cover and protect the components of the finger ring device 500.

[0085] The one or more sensors 508 may transmit data of motion (e.g., from motion sensors of the IMU), temperature, and / or optics (e.g., for determination of heart rate) to the microcomputer of CPU 506. The microcomputer may also store the obtained data in memory, for example in its own memory element or in a separate memory element mounted on the flexible PCB 510.

[0086] The microcomputer may analyze and process the obtained sensor data as herein described. For example, in some examples, one or more trained neural networks may be deployed to analyze the obtained data to assign labels to individual data points or sections of data in order to determine various activity statuses and / or events. The microcomputer 506 may wirelessly communicate the sensor data and / or the analysis of the sensor data to a remote device, e.g., a smart phone, laptop computer, workstation, or the like, via the communication unit (e.g., a radio communicator, a Bluetooth ™ communicator, NFC, or the like).

[0087] The battery 504 may be used to power the one or more sensors 508, the microcomputer, and the communication unit. The battery 504 may be rechargeable, in some examples wirelessly (e.g., via a wireless charging coil, in some examples incorporated as part of the flexible PCB) and in other examples via a wired connector.

[0088] The finger ring device 500 may further comprise a haptic motor 512 of a haptic feedback system such as haptic feedback system 144 described above with respect to FIG. 1. The finger ring device 500 may further comprise a battery charger 514 electrically coupled to the battery 502 and a voltage regulator 516 configured to maintain voltage through the various components. The haptic motor 512, battery charger 514, and voltage regulator 516 may also be mounted to the PCB 510.

[0089] As shown in FIG. 6, the flexible PCB 510 may be positioned opposite the battery 504. When worn on a finger of a user, the finger ring device 500 may be worn in any orientation with respect to a ground and the patient’s finger anatomy (e.g., with the battery at a dorsal surface and the PCB at a ventral / palmar surface, or the battery at the ventral / palmar surface and the PCB at the dorsal surface, or any orientation therebetween). Because IMU comprises one or more sensors to determine orientation, orientation of the ring with respect to ground may be determined no matter the position of the finger ring device on the user’s finger.

[0090] It should be understood that the relative positioning of the various components, including the CPU 506, the one or more sensors 508, the battery charger 514, the voltage regulator 516, and the haptic motor 512, mounted to the PCB 510 as depicted in FIGS. 5A, 5B, and 6, is exemplary in nature and other relative positionings may be possible. The various components mounted to the PCB in any manner that fits within the partially circular design of the PCB. In some examples, the haptic motor 512 may be positioned on a palm side of the device (e.g., a side of the device intended to be oriented towards the wearer’s palm) to increase sensitivity of the haptic feedback system when the wearer is touching or holding objects. Further, positioning the communication unit towards an outer edge of the ring may help to mitigate RF occlusion.

[0091] Turning now to FIG. 7, example graphs of sensor data are shown. A first graph 700 describes angular rates in degrees per second are plotted with respect to time and a second graph 702 describes acceleration in meters per second per second with respect to time. The data may be acquired by an IMU of a finger ring device, such as IMU 136 of the finger ring device 101 of FIG. 1. The data may be acquired during a walking activity, for example. The first graph and second graphs 700 and 702 may demonstrate a cyclic nature of walking as a wearers arm, hand, and fingers swing back and forth during a baseline arm swing. The sensor data as herein shown may be inputs to one or more neural networks, for example a neural network trained for analysis of walking data, in order to determine normal walking, as is seen with regular, cyclic patterns, and abnormal walking events like falls, which may interrupt the regular, cyclic patterns.

[0092] A third graph 704 shows activity level with respect to time during sleep and a fourth graph 706 shows activity level with respect to time during the day. Expected activity levels during sleep are significantly lower than activity levels during the daytime. The data may be acquired by the IMU of the finger ring device, such that activity level corresponds to amount of finger motion. The IMU may be configured to detect motion and activity even during periods of low activity, likesleep. This may allow for the data to be used as inputs for one or more neural networks in order to detect sleep cycles during sleep activity and / or changes in activity during the day time as one may expect during a mental status change like delirium.

[0093] FIGS. 5A-6 are drawn to scale, although other relative dimensions and relative positioning may be used, if desired.

[0094] The technical effect of the methods and systems herein provided is that movements of a person’s finger may be sensed by an IMU and analyzed to determine a plurality of quantities of interest. By utilizing a plurality of trained neural networks and deploying one of the plurality of trained neural networks based on a determined patient activity status, a more accurate and robust analysis of sensor data may be accomplished. Further, the finger ring device may allow for sense of more subtle movements, which may provide for more detailed information to be analyzed.

[0095] The disclosure also provides support for a method, comprising: acquiring sensor data with one or more motion sensors of a wearable finger ring device, analyzing the sensor data to determine one or more quantities of interest, and outputting the sensor data and the one or more quantities of interest to one or more remote devices. In a first example of the method, the one or more motion sensors are included in an inertial measurement unit (IMU). In a second example of the method, optionally including the first example, the one or more motion sensors of the IMU comprise at least one accelerometer and at least one gyroscope. In a third example of the method, optionally including one or both of the first and second examples, analyzing the sensor data comprises detecting normal activity states and abnormal events of the one or more quantities of interest. In a fourth example of the method, optionally including one or more or each of the first through third examples, the one or more quantities of interest comprise one or more of normal gait, abnormal gait, fall event, mental status change, REM sleep, non-REM sleep, and cognitive function assessment scores. In a fifth example of the method, optionally including one or more or each of the first through fourth examples, the method further comprises:, in response to detection of an abnormality during analysis of the sensor data, outputting a notification indicating the abnormality to the one or more remote devices. In a sixth example of the method, optionally including one or more or each of the first through fifth examples, analyzing the sensor data comprises deploying one or more machine learning architectures trained to assign quantity of interest labels to sections of the sensor data. In a seventh example of the method, optionally including one or more or each of the first through sixth examples, the one or more machine learningarchitectures are deployed based on a determined activity status of a wearer of the wearable finger device. In a eighth example of the method, optionally including one or more or each of the first through seventh examples, the method further comprises: employing a haptic feedback system of the wearable finger device for additional analysis of finger motion.

[0096] The disclosure also provides support for a system, comprising: a wearable finger ring device comprising an inertial measurement unit (IMU), a microcomputer coupled to the wearable finger ring device, wherein the microcomputer is communicatively coupled to the IMU and to one or more remote devices and comprises a processor and memory storing instructions executable by the processor that, when executed, cause the processor to: acquire motion data with the IMU, analyze the motion data, and in response to detection of an abnormal event, outputting a notification to the one or more remote devices. In a first example of the system, analyzing the motion data comprises assigning labels to sections of the motion data, wherein the labels correspond to quantities of interest. In a second example of the system, optionally including the first example the microcomputer comprises one or more machine learning architectures stored in memory that when deployed, analyze the motion data to assign labels to corresponding sections of the motion data. In a third example of the system, optionally including one or both of the first and second examples, the one or more machine learning architectures are each configured to analyze motion data corresponding to a particular wearer activity status. In a fourth example of the system, optionally including one or more or each of the first through third examples, the one or more machine learning architectures are supervised neural networks augmented by unsupervised techniques. In a fifth example of the system, optionally including one or more or each of the first through fourth examples, the system further comprises: transmitting data acquired by the IMU to the microcomputer for analysis.

[0097] The disclosure also provides support for a wearable finger ring device, comprising: a microcomputer, one or more motion sensors, wherein the microcomputer and the one or more motion sensors are mounted on a flexible printed circuit board (PCB), a haptic feedback system mounted on the PCB, wherein the haptic feedback system is communicatively coupled to the microcomputer, and a battery, wherein the microcomputer comprises a processor and non- transitory memory configured to analyze data acquired by the one or more motion sensors. In a first example of the system, the system further comprises: a communication unit configured to communicate data from the microcomputer to one or more remote devices. In a second exampleof the system, optionally including the first example, the microcomputer comprises a processor and non-transitory memory configured to analyze data acquired by the one or more motion sensors and the haptic feedback system. In a third example of the system, optionally including one or both of the first and second examples, the non-transitory memory comprises one or more neural networks trained to detect a plurality of quantities of interest in the motion data. In a fourth example of the system, optionally including one or more or each of the first through third examples, the battery is rechargeable.

[0098] When introducing elements of various embodiments of the present disclosure, the articles “a,” “an,” and “the” are intended to mean that there are one or more of the elements. The terms “first,” “second,” and the like, do not denote any order, quantity, or importance, but rather are used to distinguish one element from another. The terms “comprising,” “including,” and “having” are intended to be inclusive and mean that there may be additional elements other than the listed elements. As the terms “connected to,” “coupled to,” etc. are used herein, one object (e.g., a material, element, structure, member, etc.) can be connected to or coupled to another object regardless of whether the one object is directly connected or coupled to the other object or whether there are one or more intervening objects between the one object and the other object. In addition, it should be understood that references to “one embodiment” or “an embodiment” of the present disclosure are not intended to be interpreted as excluding the existence of additional embodiments that also incorporate the recited features.

[0099] In addition to any previously indicated modification, numerous other variations and alternative arrangements may be devised by those skilled in the art without departing from the spirit and scope of this description, and appended claims are intended to cover such modifications and arrangements. Thus, while the information has been described above with particularity and detail in connection with what is presently deemed to be the most practical and preferred aspects, it will be apparent to those of ordinary skill in the art that numerous modifications, including, but not limited to, form, function, manner of operation and use may be made without departing from the principles and concepts set forth herein. Also, as used herein, the examples and embodiments, in all respects, are meant to be illustrative only and should not be construed to be limiting in any manner.

Claims

CLAIMS1. A method, comprising: acquiring sensor data with one or more motion sensors of a wearable finger ring device; analyzing the sensor data to determine one or more quantities of interest; and outputting the sensor data and the one or more quantities of interest to one or more remote devices.

2. The method of claim 1, wherein the one or more motion sensors are included in an inertial measurement unit (IMU).

3. The method of claim 2, wherein the one or more motion sensors of the IMU comprise at least one accelerometer and at least one gyroscope.

4. The method of claim 1, wherein analyzing the sensor data comprises detecting normal activity states and abnormal events of the one or more quantities of interest.

5. The method of claim 1, wherein the one or more quantities of interest comprise one or more of normal gait, abnormal gait, fall event, mental status change, REM sleep, non-REM sleep, and cognitive function assessment scores.

6. The method of claim 1, further comprising, in response to detection of an abnormality during analysis of the sensor data, outputting a notification indicating the abnormality to the one or more remote devices.

7. The method of claim 1, wherein analyzing the sensor data comprises deploying one or more machine learning architectures trained to assign quantity of interest labels to sections of the sensor data.

8. The method of claim 1, wherein the one or more machine learning architectures are deployed based on a determined activity status of a wearer of the wearable finger device.

9. The method of claim 8, further comprising employing a haptic feedback system of the wearable finger device for additional analysis of finger motion.

10. A system, comprising: a wearable finger ring device comprising an inertial measurement unit (IMU), a microcomputer coupled to the wearable finger ring device, wherein the microcomputer is communicatively coupled to the IMU and to one or more remote devices and comprises a processor and memory storing instructions executable by the processor that, when executed, cause the processor to: acquire motion data with the IMU; analyze the motion data; and in response to detection of an abnormal event, outputting a notification to the one or more remote devices.

11. The system of claim 10, wherein analyzing the motion data comprises assigning labels to sections of the motion data, wherein the labels correspond to quantities of interest.

12. The system of claim 10, the microcomputer comprises one or more machine learning architectures stored in memory that when deployed, analyze the motion data to assign labels to corresponding sections of the motion data.

13. The system of claim 12, wherein the one or more machine learning architectures are each configured to analyze motion data corresponding to a particular wearer activity status.

14. The system of claim 12, wherein the one or more machine learning architectures are supervised neural networks augmented by unsupervised techniques.

15. The system of claim 10, further comprising transmitting data acquired by the IMU to the microcomputer for analysis.

16. A wearable finger ring device, comprising: a microcomputer; one or more motion sensors, wherein the microcomputer and the one or more motion sensors are mounted on a flexible printed circuit board (PCB); a haptic feedback system mounted on the PCB, wherein the haptic feedback system is communicatively coupled to the microcomputer; and a battery, wherein the microcomputer comprises a processor and non-transitory memory configured to analyze data acquired by the one or more motion sensors.

17. The wearable finger ring device of claim 16, further comprising a communication unit configured to communicate data from the microcomputer to one or more remote devices.

18. The wearable finger ring device of claim 16, wherein the microcomputer comprises a processor and non-transitory memory configured to analyze data acquired by the one or more motion sensors and the haptic feedback system.

19. The wearable finger ring device of claim 18, wherein the non-transitory memory comprises one or more neural networks trained to detect a plurality of quantities of interest in the motion data.

20. The wearable finger ring device of claim 16, wherein the battery is rechargeable.

Images