Health and wellness management technology

Patent metadata

Field	Value
Publication number	US-12046366-B1
Application number	US-202016891403-A
Country	US
Kind code	B1
Filing date	Jun 3, 2020
Priority date	Apr 24, 2012
Publication date	Jul 23, 2024
Grant date	Jul 23, 2024

Abstract

Health and wellness management technology, in which events that relate to activity within the monitored property are sensed based on output from sensors located at a monitored property. Behaviors are detected based on the sensed events that relate to activity within the monitored property and, in accordance with the detected behaviors, one or more models are created based on a likelihood of similarly expressed events happening at similar times with similar characteristics. Additional behaviors detected after creation of the one or more models are evaluated against the one or more models to determine whether the additional behaviors are consistent with the one or more models. At least one action is performed based on the determination of whether the additional behaviors are consistent with the one or more models.

First claim

What is claimed is:

1. A computer-implemented method comprising:
determining, using first sensor data for a property, an activity classification that indicates a first behavior i) performed by an individual at the property and ii) that has an activity type;
generating a synthesized data stream from a plurality of streams of sensor data obtained from a plurality of sensors at the property using stream logic that enables arithmetic and logical operations, wherein:
i) a stream logic expression for the stream logic represents a method by which each stream of sensor data is defined programmatically; and
ii) the stream logic expression operates on data objects that represent respective streams of sensor data from the plurality of streams, the operating including performing arithmetic and logical operations via metadata expressions;

detecting, using a portion of the synthesized data stream that represents the first behavior performed by the individual as indicated by the activity classification as input to a predictive model that was generated using behavioral event data comprising a) an event of the activity type that was detected in sensor data captured by a sensor at the property and b) an annotation for the event that was defined by user input, a behavioral deviation between the first behavior that has the activity type and corresponding behavior of the activity type modeled by the predictive model in an activity trend of the individual; and
triggering, using an output of the predictive model, an automated action when the behavioral deviation is indicative of an event that compromises health or safety of the individual.

2. The method of claim 1 , wherein detecting the behavioral deviation comprises:
detecting the behavioral deviation using the portion of the synthesized data stream that does not include image data of the individual.

3. The method of claim 1 , wherein detecting the behavioral deviation comprises:
determining, using the predictive model, that a data value representing a measurement of behavior for the individual exceeds a threshold value; and
detecting the behavioral deviation in response to determining that the data value exceeds the threshold value.

4. The method of claim 1 , comprising:
translating statistical measurements of parameter values of the synthesized data stream into respective approximations of different detectable activities, each detectable activity being associated with the activity trend modeled by the predictive model.

5. The method of claim 4 , wherein translating the statistical measurements of parameter values comprises:
translating the statistical measurements of parameter values through a heuristic classification using a multiplicity of specified variables comprising one or more of: current sensor state, change in sensor state, duration in sensor state, time of day, type of sensor, location of sensor, identity of individual triggering sensor, and stream states.

6. The method of claim 1 , comprising:
generating a confidence score that represents a measure of confidence in the activity classification using multivariate analysis performed using the predictive model; and
using, as the portion of the synthesized data stream used as the input to the predictive model, measurements of the first behavior that correspond to the activity classification for which the confidence score was generated.

7. The method of claim 1 , wherein generating the synthesized data stream comprises:
generating the synthesized data stream at a physical point of data aggregation that corresponds to a gateway device or a remote server of a property monitoring system.

8. A system comprising one or more processing devices and one or more non-transitory machine-readable storage devices storing instructions that are executable by the one or more processing devices to cause performance of operations comprising:
obtaining, from each of a plurality of sensors at a property, respective streams of sensor data;
detecting, in at least one stream of the plurality of streams of sensor data, an event that has an activity type;
accessing, for the event, user input defining an annotation for the event;
generating behavioral event data using data for the event detected in the at least one stream from the plurality of the streams of sensor data and the annotation of the event;
generating a predictive model that models activity trends of an individual at the property using the behavioral event data comprising the event and the annotation;
determining, using first sensor data for the property, an activity classification that indicates a first behavior i) performed by the individual and ii) that has the activity type;
generating, using a data synthesizer configured to define a stream aggregation behavior using a stream logic expression, a synthesized data stream from the plurality of streams of sensor data obtained from the plurality of sensors, wherein:
i) the stream logic expression represents a method by which each stream of sensor data is defined programmatically; and
ii) the stream logic expression operates on data objects that represent respective streams of sensor data from the plurality of streams, the operating including performing arithmetic and logical operations via metadata expressions;

detecting, using a portion of the synthesized data stream that represents the first behavior performed by the individual as indicated by the activity classification as input to the predictive model that was generated using the behavioral event data comprising a) the event of the activity type that was detected in sensor data captured by a sensor at the property and b) the annotation for the event that was defined by the user input, a behavioral deviation between the first behavior that has the activity type and corresponding behavior of the activity type modeled by the predictive model in an activity trend of the individual without direct visual observation of the individual; and
triggering, using an output of the predictive model, an automated action when the behavioral deviation is indicative of an event that compromises health or safety of the individual.

9. The system of claim 8 , wherein detecting the behavioral deviation comprises:
determining, using the predictive model, that a data value representing a measurement of behavior for the individual exceeds a threshold value; and
detecting the behavioral deviation in response to determining that the data value exceeds the threshold value.

10. The system of claim 8 , wherein the operations comprise:
translating statistical measurements of parameter values of the synthesized data stream into respective approximations of different detectable activities, each detectable activity being associated with the activity trend modeled by the predictive model.

11. The system of claim 10 , wherein translating the statistical measurements of parameter values comprises:
translating the statistical measurements of parameter values through a heuristic classification using a multiplicity of specified variables comprising one or more of: current sensor state, change in sensor state, duration in sensor state, time of day, type of sensor, location of sensor, identity of individual triggering sensor, and stream states.

12. The system of claim 8 , wherein the operations comprise:
generating a confidence score that represents a measure of confidence in the activity classification using multivariate analysis performed using the predictive model; and
using, as the portion of the synthesized data stream used as the input to the predictive model, measurements of the first behavior that correspond to the activity classification for wh

Assignees

• Alarm Com Inc

Inventors

• Hanson Mark Andrew
• Silverman Christopher
• Barth Adam T

Classifications

• G16H50/20 Primary

  for computer-aided diagnosis, e.g. based on medical expert systems · CPC title

• G16Z99/00

  Subject matter not provided for in other main groups of this subclass · CPC title

• G06F17/18

  for evaluating statistical data {, e.g. average values, frequency distributions, probability functions, regression analysis (forecasting specially adapted for a specific administrative, business or logistic context G06Q10/04)} · CPC title

• G06N20/00

  Machine learning · CPC title

• G16H50/50

  for simulation or modelling of medical disorders · CPC title