DEVICES AND METHODS FOR FALL DETECTION BASED ON PHASE SEGMENTATION

• US 20180070889A1
• Filed: 09/09/2016
• Published: 03/15/2018
• Est. Priority Date: 09/09/2016
• Status: Active Grant

• Current Assignee: Qualcomm, Inc.
• Original Assignee: Qualcomm, Inc.
• Inventors: Lee, Jin Won, Wu, Xinzhou, Attar, Rashid Ahmed Akbar, Han, Feng
• Granted Patent: US 10,506,990 B2
• Time in Patent Office: Days
• CPC Class Codes: A61B 2562/0219 Inertial sensors, e.g. accelerometers, gyroscopes, tilt switches
• CPC Class Codes: A61B 5/1117 Fall detection
• CPC Class Codes: A61B 5/1122 of movement trajectories
• CPC Class Codes: A61B 5/6801 specially adapted to be attached to or worn on the body surface
• CPC Class Codes: A61B 5/7225 Details of analog processing, e.g. isolation amplifier, gain or sensitivity adjustment, filtering, baseline or drift compensation A61B5/30, A61B5/316 take precedence; transformers for specific applications H01F38/00; amplifiers H03F
• CPC Class Codes: A61B 5/7267 involving training the classification device
• CPC Class Codes: A61B 5/7282 Event detection, e.g. detecting unique waveforms indicative of a medical condition A61B5/4094, A61B5/4818, A61B5/349, A61B5/0823 take precedence
• CPC Class Codes: G08B 21/0446 worn on the body to detect changes of posture, e.g. a fall, inclination, acceleration, gait
• CPC Class Codes: G16H 20/30 relating to physical therapies or activities, e.g. physiotherapy, acupressure or exercising
• CPC Class Codes: G16H 40/63 for local operation
• CPC Class Codes: G16H 40/67 for remote operation
• CPC Class Codes: G16H 50/20 for computer-aided diagnosis, e.g. based on medical expert systems
• CPC Class Codes: G16H 50/30 for calculating health indices; for individual health risk assessment

Abstract

Aspects of the subject matter described in this disclosure can be implemented in a fall detection device and method. One or more motion sensors can access a user'"'"'s acceleration data. The acceleration data can be segmented using a segmentation algorithm to identify a potential fall event. The segmentation algorithm can determine a cumulative sum of the acceleration data, where the cumulative sum is based on acceleration values being greater than or less than an acceleration threshold value, and a potential fall event can be identified where the cumulative sum is greater than a cumulative sum threshold value. Statistical features can be extracted from the segmented acceleration data and aggregated, and a determination can be made as to whether the potential fall event is a fall event based at least in part on the statistical features.

Claims

1. A method of identifying a fall event, comprising:

accessing acceleration data of an electronic device from one or more motion sensors;

segmenting the acceleration data to identify a potential fall event, comprising;

determining a cumulative sum for the acceleration data; and

identifying the potential fall event if the cumulative sum is greater than a cumulative sum threshold value;

extracting one or more statistical features from the segmented acceleration data; and

determining whether the potential fall event is a fall event based at least in part on the one or more statistical features.

2. The method of claim 1, wherein segmenting the acceleration data includes segmenting the acceleration data into at least three phases, wherein the at least three phases include (i) a before-fall phase, (ii) a during-fall phase, and (iii) an after-fall phase, and wherein the during-fall phase includes the potential fall event.

3. The method of claim 2, further comprising:

analyzing the duration of time of the during-fall phase, wherein determining whether the potential fall event is a fall event is based at least in part on the duration of time of the during-fall phase.

4. The method of claim 2, wherein the one or more statistical features include one or more of the following:

variance, maximum, minimum, 75^th percentile, 25^th percentile, mean, median, mean crossing rate, absolute area, or any combination thereof.

5. The method of claim 4, wherein the one or more statistical features include:

a mean crossing rate of acceleration extracted from the during-fall phase and variance of acceleration extracted from the after-fall phase.

6. The method of claim 4, further comprising:

accessing angular velocity data and orientation data from the one or more motion sensors; and

segmenting the angular velocity data and the orientation data into the at least three phases, wherein the one or more statistical features include;

a mean crossing rate of acceleration extracted from the during-fall phase, a maximum acceleration extracted from the during-fall phase, an absolute area of acceleration extracted from the after-fall phase, a variance of angular velocity extracted from the during-fall phase, a maximum of angular velocity extracted from the during-fall phase, a minimum of angular velocity extracted from the after-fall phase, a median of orientation extracted from the after-fall phase, or any combination thereof.

7. The method of claim 1, wherein the electronic device is a wearable device.

8. The method of claim 1, wherein the one or more motion sensors consist of one or more accelerometers.

9. The method of claim 1, wherein the cumulative sum is based on acceleration values greater than or less than a first acceleration threshold value.

10. The method of claim 9, wherein the first acceleration threshold value and the cumulative sum threshold value are identified by a machine-learning algorithm.

11. The method of claim 1, wherein the cumulative sum has the formula:

CUSUM[i]=max(0, CUSUM[i−

1]+acceleration[i]−

A_th1), wherein CUSUM[0]=0, and wherein A_th1 is the first acceleration threshold value and acceleration[i] is an acceleration value at time [i], and wherein i>

0.

12. The method of claim 1, wherein determining whether a potential fall event is a fall event comprises:

aggregating the one or more statistical features from the segmented acceleration data;

calculating a metric based on the one or more statistical features; and

classifying the potential fall event as a fall event based at least in part on the calculated metric.

13. The method of claim 12, wherein the metric is calculated as a linear or non-linear combination of the extracted statistical features weighted by coefficients.

14. The method of claim 12, wherein the potential fall event is classified using a classifier trained by a machine-learning algorithm.

15. An electronic device for identifying a fall event, comprising:

one or more motion sensors configured to access acceleration data of the electronic device;

a processor coupled to the one or more motion sensors, the processor configured to;

segment the acceleration data to identify a potential fall event;

extract one or more statistical features from the segmented acceleration data; and

determine whether the potential fall event is a fall event based at least in part on the one or more statistical features, wherein the processor configured to segment the acceleration data is configured to determine a cumulative sum for the acceleration data, and identify a potential fall event if the cumulative sum is greater than a cumulative sum threshold value.

16. The electronic device of claim 15, wherein the processor configured to segment the acceleration data is configured to segment the acceleration data into at least three phases, wherein the at least three phases include (i) a before-fall phase, (ii) a during-fall phase, and (iii) an after-fall phase, and wherein the during-fall phase includes the potential fall event.

17. The electronic device of claim 16, wherein the one or more statistical features include:

a mean crossing rate of acceleration extracted from the during-fall phase and variance of acceleration extracted from the after-fall phase.

18. The electronic device of claim 15, wherein the electronic device is a wearable device and the one or more motion sensors consist of one or more accelerometers.

19. The electronic device of claim 15, wherein the cumulative sum has the formula:

CUSUM[i]=max(0, CUSUM[i−

1]+acceleration[i]−

A_th1), wherein CUSUM[0]=0, and wherein A_th1 is the first acceleration threshold value and acceleration[i] is an acceleration value at time [i], and wherein i>

0.

20. The electronic device of claim 15, further comprising:

a low-pass filter configured to smooth out the acceleration data accessed from the one or more motion sensors.

21. An electronic device for identifying a fall event, comprising:

means for accessing acceleration data; and

means for analyzing the acceleration data coupled to the means for accessing acceleration data, the means for analyzing the acceleration data configured to;

segment the acceleration data to identify a potential fall event;

extract one or more statistical features from the segmented acceleration data; and

determine whether the potential fall event is fall event based at least in part on the one or more statistical features, wherein the means for analyzing the acceleration data configured to segment the acceleration data is configured to determine a cumulative sum for the acceleration data, and identify a potential fall event if the cumulative sum is greater than a cumulative sum threshold value.

22. The electronic device of claim 21, wherein the means for analyzing the acceleration data configured to segment the acceleration data is configured to segment the acceleration data into at least three phases, wherein the at least three phases include (i) a before-fall phase, (ii) a during-fall phase, and (iii) an after-fall phase, and wherein the during-fall phase includes the potential fall event, and wherein the one or more statistical features include:

a mean crossing rate of acceleration extracted from the during-fall phase and variance of acceleration extracted from the after-fall phase.

23. The electronic device of claim 21, wherein the electronic device is a wearable device and the one or more motion sensors consist of one or more accelerometers.