# US 10,319,209 B2 · Method and system for motion analysis and fall prevention

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 10,319,209 B2
- Document ID: US10319209B2
- Status: Granted
- Kind: B2
- Application number: 15/613,623
- Filed: 2017-06-05
- Published: 2017-12-07
- Granted: 2019-06-11
- Earliest claimed: 2016-06-03
- Applicants: John Carlton-Foss
- Inventors: John Carlton-Foss
- Web page: https://patentgrep.com/patent/US10319209B2

## CPC

- G08B 21/0446: {worn on the body to detect changes of posture, e.g. a fall, inclination, acceleration, gait}
- A61B 5/11: Measuring movement of the entire body or parts thereof, e.g. head or hand tremor or mobility of a limb {(A61B5/1038 takes precedence; motion detection to correct for motion artifacts in physiological signals A61B5/721)}
- A61B 5/1116: {Determining posture transitions}
- A61B 5/1117: {Fall detection}
- G06N 7/01: Probabilistic graphical models, e.g. probabilistic networks
- G06N 20/00: Machine learning
- G06N 20/10: using kernel methods, e.g. support vector machines [SVM]
- G08B 5/36: using visible light sources
- G08B 6/00: Tactile signalling systems, e.g. tactile personal calling systems
- G08B 25/08: using communication transmission lines {(G08B13/19658, G08B21/0286, G08B25/016 take precedence)}
- G08B: SIGNALLING SYSTEMS, e.g. PERSONAL CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- A61B: DIAGNOSIS; SURGERY; IDENTIFICATION (usefulness limited to only animals A61D)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

A system and method of motion analysis, fall detection, and fall prediction using machine learning and classifiers. A wearable motion sensor for collecting and transmitting motion data for use in a fall prediction model using features and parameters to classify the motion data and notify when a fall is emergent. Using machine learning, the fall prediction model can be created, implemented, evaluated, and it can evolve over time with additional data. The system and method can use individual data or pool data from various individuals for use in fall prediction.

## Claims

### Claim 1

A system for detecting an emergent fall comprising:
one or more sensors wearable by one or more users, at least one of the one or more sensors being configured to collect and transmit sensor data, including motion data;
a hub for receiving and labeling the sensor data, wherein labeling the sensor data includes date/time and whether a subsequent fall actually occurred;
a processor on the hub configured to:
classify the sensor data as a data classification including at least whether a fall is emerging or not according to a fall prediction model;
store the data classification, wherein the labeled sensor data is used to create one or more parameters, strings, features, data models, and classes to be used in subsequent data classification via supervised or unsupervised machine learning;
process one or more parameters, where the one or more parameters include time duration and time placement for the one or more strings;
process one or more strings, where the one or more strings comprise one or more streams of sensor data including N-dimensional event phase space information which can be matched to behavioral motion, such that the one or more strings are classified into one or more classes and the one or more classes are used to determine individual and group thresholds used by an alert system; and
match output from the data classification with the one or more parameters, strings, features, and data models from the fall prediction model; and
a transmitter for transmitting information to the alert system and to a repository of data for use by the fall prediction model;
the alert system being configured to send a notification that a risk of a fall is above a threshold based, in part, on a confidence level for a match to the one or more parameters, strings, features, and data models from the fall prediction model to the data classification.

### Claim 2 (depends on 1)

The system for detecting an emergent fall of claim 1, wherein the at least one sensor is an accelerometer.

### Claim 3 (depends on 1)

The system for detecting an emergent fall of claim 1, wherein the notification is in the form of an audio signal.

### Claim 4 (depends on 1)

The system for detecting an emergent fall of claim 1, wherein the notification is in the form of a visual signal.

### Claim 5 (depends on 1)

The system for detecting an emergent fall of claim 1, wherein the notification is in the form of tactual/vibratory signal.

### Claim 6 (depends on 1)

The system for detecting an emergent fall of claim 1, further comprising a receiver for receiving data from the repository of data for further use by the fall prediction model.

### Claim 7

A method of detecting an emergent fall comprising:
providing at least one wearable sensor on at least one user;
measuring at least motion data with the at least one sensor;
creating a fall prediction model, wherein an input for the fall prediction model comprises at least the motion data and time data;
implementing the fall prediction model;
comparing at least one stream of sensor data including N-dimensional event phase space information of the motion data with at least one stream of sensor data including N-dimensional event phase space information of the fall prediction model, wherein the at least one stream of sensor data can be matched to behavioral motion, such that the at least one stream of sensor data is classified into one or more classes and the one or more classes are used to determine individual and group thresholds used by an alert system;
creating a data classification including at least whether a fall is emerging or not using the comparison of at least the motion data and optionally the time data to the fall prediction model;
calculating a probability that, and a time frame within which, a fall is emergent;
calculating a confidence level for the probability that a fall is emergent;
indicating whether the fall is emergent based, in part, on the data classification;
evaluating the fall prediction model in real-time;
evolving the fall prediction model with additional data from the at least one sensor on at least one user;
determining if the fall is emergent based, in part, on whether a risk of a fall is above a threshold; and
communicating a notification with the alert system, if the fall is emergent, that the fall is emergent so that the fall can be prevented.

### Claim 8 (depends on 7)

The method for detecting an emergent fall of claim 7, wherein the at least one sensor is an accelerometer.

### Claim 9 (depends on 7)

The method for detecting an emergent fall of claim 7, wherein the notification is in the form of an audio signal.

### Claim 10 (depends on 7)

The method for detecting an emergent fall of claim 7, wherein the notification is in the form of a visual signal.

### Claim 11 (depends on 7)

The method for detecting an emergent fall of claim 7, wherein the notification is in the form of a tactual/vibratory signal.

### Claim 12 (depends on 7)

The method for detecting an emergent fall of claim 7, further comprising a receiver for receiving data from a repository of classified data for further use by the fall prediction model.

### Claim 13 (depends on 7)

The method for detecting an emergent fall of claim 7, further comprising adjusting threshold parameters to adapt the fall prediction model to modify a sensitivity of the fall prediction model for the at least one user.

### Claim 14

A wearable for detecting emerging falls and an actual fall comprising:
a power source;
one or more sensors wearable by at least one user, at least one of the one or more sensors being configured to collect and transmit at least motion data;
at least one processor for classifying at least the motion data and time data received from the one or more sensors according to a fall prediction model, such that data classification includes at least whether a fall is emerging or not;
one or more hubs configured to:
receive raw sensor data and the classified data as well as information about the data classification,
process one or more parameters including time duration and time placement for one or more strings, where the one or more strings comprise one or more streams of sensor data including N-dimensional event phase space information which can be matched to behavioral motion, such that the one or more strings are classified into one or more classes and the one or more classes are used to determine individual and group thresholds used by an alert system: and
transmit the information about the classification to the alert system;
a communications system capable of sending data and commands, as well as information contributing to sending alerts between a remote server, a local server and the one or more hubs; and
the alert system for sending a notification when an emergent fall has been identified based in part on whether a risk of a fall is above a threshold based on the data classification by the fall prediction model.

### Claim 15 (depends on 14)

The wearable for detecting a fall of claim 14, wherein the at least one sensor is an accelerometer.

### Claim 16 (depends on 14)

The wearable for detecting a fall of claim 14, wherein the notification is in the form of an audio signal.

### Claim 17 (depends on 14)

The wearable for detecting a fall of claim 14, further comprising a receiver for receiving data from a repository of classified data for further use by the fall prediction model.

### Claim 18 (depends on 14)

The wearable for detecting a fall of claim 14, further comprising a machine learning server to offload demand from the one or more hubs and to function as a data repository to optionally implement a Big Data-based approach to creation and evolution of data models, features, and parameters for use by the fall prediction model.

### Claim 19 (depends on 18)

The wearable for detecting a fall of claim 18, wherein the machine learning server runs in a communitarian mutual self-help mode with different views of individual and group data:
to enable use of one or more models, thresholds, and/or parameters by at least one user to identify actionable likelihood of falls;
to enable mutual monitoring and help by the one or more participating users;
to enable privacy by the one or more users;
to enable monitoring and data acquisition by professionals serving patients or participating users;
to enable record keeping of the one or more users; and
to enable communications among the professionals and the one or more participating users using the communitarian system.

### Claim 20 (depends on 1)

The system for detecting an emergent fall according to claim 1, wherein a false positive and a false negative may appear as a pause in motion between 0.5 second and 5 seconds, with false negatives being closer to 0.5 seconds and false positives being about 5 or more seconds.