Verily Life Sciences LLC

Mark Murphy

Russell N Mirov

David Andrew Gibson

Nikhil Bikhchandani

Christopher Towles Lengerich

Ali Shoeb

Sarel Kobus Jooste

Raphael Cendrillon

Erin Soderberg

Anupam J Pathak

Emre Demiralp

Tushar Anil Parlikar

Scott Brian Kennedy

Michael Jastrzebski

Philip John Stephens

Hongsheng Wang

Systems and methods are provided for continuously monitoring a user to determine when cardiovascular events are likely occurring and to responsively provide a prompt to a user to engage in additional physiological assessment of the putative cardiovascular event. Additional assessment can include the user engaging an additional sensor to provide signals that are more accurate, lower noise, or otherwise improved relative to a continuously-monitoring sensor used to initially detect the cardiovascular event. Detection of cardiovascular events includes using a cardiovascular classifier to determine, based on the output of such a continuously-monitoring sensor, whether the event is likely occurring. Such a classifier can be received from a cloud computing service or other remote system based on sensor outputs sent to such a system. Use of such a classifier can facilitate reduced false-positive detection of cardiovascular events based on the continuously-monitoring sensor, and thus reduce extraneous prompts to the user.

Continuous detection and monitoring of heart arrhythmia using both wearable sensors and cloud-resident analyses

A system is provided to monitor, over time, one or more physical variables related to a severity or progression of a movement disorder and/or of symptoms thereof. The monitored physical variables can include speech sounds; keyboard outputs; or accelerations, rotations, or other properties of the motion of one or more body parts. The system operates to detect, based on the monitored one or more physical variables, potential changes in the degree and/or character of the movement disorder symptoms. In response to detecting such a potential change, the system provides the user with one or more tasks that the user can perform. The system detects one or more properties of the user's performance and, based on that detection, determines the severity or progression of the movement disorder. The tasks can include stepping, turning, standing, sitting, reaching, typing, pointing, manipulating an object, speaking into a microphone, or other tasks.

System for continuous detection and monitoring of symptoms of Parkinson's disease

Systems and methods are provided for determining the frequency of a cardiovascular pulse based on a first physiological signal that is continuously available and a second physiological signal that is less available and that is more accurate or otherwise improved relative to the first signal with respect to pulse rate estimation. When the second signal is available it controls the determination of the pulse rate. When the second signal is unavailable, the first signal is used to determine the pulse rate. This can include using the first signal to estimate the pulse rate until the second signal is available, at which point the pulse rate is estimated based on the second physiological signal. Alternatively, the first signal could be used to determine a number of candidate pulse rates, and the second signal could be used to select a pulse rate from the set of candidate pulse rates.

Method for improving heart rate estimates by combining multiple measurement modalities

A method for tagging and organizing data is provided. In one example, physiological data detected from a wearer of a wearable device is received and associated with a tag based, at least in art, on an input by the wearer. The input may be a state of the wearer, such as physical or mental state, or a rule. The collected physiological data may be organized based on the tag and, in some examples, on other types of received data, such as a wearer's personal data. In other example methods, data may be stored in a database based on one or more tags associated with the data.

Data tagging

System for ubiquitous detection and monitoring of symptoms of Parkinson's disease

A system is configured to discriminate amongst different environments based in part on characteristics of ambient light. Ambient light intensity is measured using a light-sensitive element configured to generate an output signal indicative of an intensity of light incident on the light-sensitive element. A controller is configured to obtain a set of ambient light measurements using the light-sensitive element, and determine that the measurements correspond to a particular ambient light profile. The particular ambient light profile can be one of multiple ambient light profiles that each correspond to a different environment and/or context.

Context discrimination using ambient light signal

A system including two or more wearable devices includes at least one sensor in each device to detect signals related to the body of a wearer. User inputs, sensor outputs, or other information are used to determine an activity being performed or otherwise engaged in by the wearer and footsteps, heartbeats, discrete turns of the torso, or other events can be detected, from one or more of the sensor outputs, during a period of time that the user is engaged in the activity. A duration, amplitude, timing, mean value, or other characteristic of each such detected event can be determined, based on the sensor outputs, and used to generate a sample of clinically relevant information about the wearer. This information can then be used to determine the presence, stage, degree, progression, or other information about a disease state of the wearer or some other information about the wearer's body.

System and methods for integrating feedback from multiple wearable sensors

A wearable device described herein includes a housing and a mount configured to mount the housing to an external surface of a wearer. The wearable device further includes one or more sensors configured to measure at least one physiological parameter of the wearer. The wearable device may obtain an activity metric that is based on at least one physiological parameter of the wearer measured by the one or more sensors and demographical data specific to the wearer. In some examples, the wearable may be configured to calculate the activity metric or a preliminary activity metric and to indicate the activity metric and/or preliminary activity metric to the wearer. In some examples, the wearable device may transmit the physiological parameter measurement to an external device and receive an indication of the activity metric from the external device.

Activity metric calculation for wearable devices

Systems and methods are provided for determining the frequency of a cardiovascular pulse based on a photoplethysmographic measurement of blood in a portion of subsurface vasculature. A plurality of samples of the photoplethysmographic signal are obtained and a frequency spectrum determined based on a first set of the plurality of samples. A frequency is determined based on a maximum frequency component of the frequency spectrum. Further, a confidence level of the determined frequency of the maximum frequency component is determined based on the magnitude of the maximum frequency component and the overall energy in the frequency spectrum. A pulse rate is then determined by updating a predicted pulse rate of a second, earlier set of the plurality of samples based on the frequency of the maximum frequency component and the determined confidence level.

Method for improving accuracy of pulse rate estimation

The present disclosure is directed to waveform synchronization in multi-modal sensor networks. An example method includes providing a reference signal to a translation circuit. The method also includes generating, by the translation circuit, (i) a first synchronization signal capable of exciting a first emitter to produce a first wave in a first modality and (ii) a second synchronization signal capable of exciting a second emitter to produce a second wave in a second modality, wherein a modality is a domain within a form of energy. The method further includes producing, by the first emitter, first wave in the first modality and, by the second emitter, the second wave in the second modality, wherein the first wave is substantially directed toward a first sensor capable of interacting with the first wave, and wherein the second wave substantially directed toward a second sensor capable of interacting with the second wave.

Time synchronization of multi-modality measurements

Example embodiments assess cardiovascular health by receiving, from at least one monitoring device, pulse rate data from a subject. The pulse rate data includes pulse rate measurements and corresponding measurement timestamps. The embodiments select, with at least one processor, a plurality of recovery periods in the pulse rate data. Each recovery period corresponds to a respective decrease in pulse rate from a respective upper pulse rate to a respective lower pulse rate. The respective upper pulse rate is less than a peak pulse rate of the subject or the respective lower pulse rate is greater than a resting pulse rate of the subject. The embodiments determine, with the at least one processor, an exponential decay from the plurality of selected recovery periods. The exponential decay characterizes a pulse rate recovery for the subject. The embodiments provide, via a user interface, the pulse rate recovery based on the exponential decay.

Growing community of inventors

Palo Alto, CA, United States of America

Mark Murphy