System and method for fall detection

• US 8,909,497 B1
• Filed: 04/05/2011
• Issued: 12/09/2014
• Est. Priority Date: 04/06/2010
• Status: Expired due to Fees

• Current Assignee: KEYnetik, Inc.
• Original Assignee: KEYnetik, Inc.
• Inventors: Shkolnikov, Mark
• Primary Examiner(s): Charioui, Mohamed
• Assistant Examiner(s): Kuan, John
• Application Number:
  US13/080,580
• Time in Patent Office: 1,344 Days
• Field of Search: 702/141, 702/150, 340/573.1, 340/669, 600/595
• US Class Current: 702/141
• CPC Class Codes: A61B 2560/0257 using atmospheric pressure
• CPC Class Codes: A61B 2562/06 Arrangements of multiple sensors of different types
• CPC Class Codes: A61B 5/1116 Determining posture transitions
• CPC Class Codes: A61B 5/1117 Fall detection
• CPC Class Codes: A61B 5/1121 Determining geometric values, e.g. centre of rotation or angular range of movement
• CPC Class Codes: A61B 5/1123 Discriminating type of movement, e.g. walking or running A61B5/1116, A61B5/112 take precedence
• CPC Class Codes: A61B 5/725 using specific filters therefor, e.g. Kalman or adaptive filters A61B5/316 take precedence
• 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: G01P 15/00 Measuring acceleration; Measuring deceleration; Measuring shock, i.e. sudden change of acceleration
• CPC Class Codes: G01P 15/0891 with indication of predetermined acceleration values G01P15/135 takes precedence
• CPC Class Codes: G08B 21/0438 Sensor means for detecting
• CPC Class Codes: G08B 21/0446 worn on the body to detect changes of posture, e.g. a fall, inclination, acceleration, gait

Abstract

Embodiments of the invention relate to spectrally and spatially dissecting accelerometer data from a signal of a motion sensor. Data received from the dissected signal is reduced to statistical averages for selected frequency bands and spatial dimensions. A time segment subject over which the signal was acquired is segmented so that the dissected signal may be evaluated per the segmentation of the time. Change of orientation towards gravity of the sensor is analyzed with respect to a change of orientation from a before impact segment to an after impact segment. The analyzed change in orientation value with respect to a threshold is a factor in determination of a fall of a body to which the sensor is attached.

Claims

1. A system comprising:

a sensor attached to a body;

the sensor to acquire sensory data, including acceleration data, and

a tool to detect fall of the body in accordance with the acquired sensory data, the tool comprising;

an orientation manager to extract orientation data from the acquired sensory data;

an impact manager to quantitatively compute cumulative high frequency energy of impact data from the acquired sensory data; and

an analyzer to analyze the orientation data and the impact data and to detect a fall, including the analyzer to differentiate the detected fall from an intentional change of posture, including weighting acceleration data extracted from the sensor, the extracted data including impact, orientation, and energy.

2. The system of claim 1, further comprising a vertical manager in communication with the fall detection tool, the vertical manager to learn a true vertical value of the sensor before impact, and to replace a prior vertical value of a before impact time segment with the true vertical value.

3. The system of claim 1, further comprising a compression manager in communication with the analyzer, the compression manager to compute and compress statistical data over time received from the acquired sensory data, the tool including a technique selected from the group consisting of:

a stepping window and a sliding window.

4. The system of claim 3, wherein the stepping window technique extracts statistical properties of data within a stepping window and includes a processing of adjacent time segments to sample sensory data.

5. The system of claim 3, wherein the sliding window technique extracts statistical properties of data within a sliding window and includes an overlap of adjacent time segments to average features of the data over time, wherein each average comprises features of the data from impact segments in the adjacent time segments.

6. The system of claim 1, further comprising the impact manager to monitor an impact of the body within an impact time segment, including monitoring for a parameter reflective of cumulative energy of a sensory signal within the segment to exceed a threshold.

7. The system of claim 1, wherein before and after the orientation data extractions are conducted from the sensory data before and after impact time segments, and the cumulative high frequency energy is computed during an impact time segment.

8. The system of claim 1, further comprising the impact manager to provide resolution between a fall and a device drop by detection of attachment of the sensor to the body.

9. The system of claim 1, further comprising a secondary sensor attached to the body, the secondary sensor being a second accelerometer, the secondary sensor in communication with the tool, wherein the tool extracts rotation data associated with an impact time segment.

10. The system of claim 1, further comprising a secondary sensor attached to the body, the secondary sensor being a compass, the secondary sensor in communication with the impact manager, the impact manager to extract magnetic field data associated with the detected fall.

11. The system of claim 1, further comprising the impact manager to extract rotation data using a differential signal from two tri-axis accelerometers, and to determine excessive rotation within an impact time segment.

12. The system of claim 11, further comprising a secondary sensor in the form of a pressure sensor, and the impact manager to extract altitude data from the secondary sensor, including a change in altitude from a before impact time segment and an after impact time segment.

13. The system of claim 1, further comprising a secondary sensor in the form of a microphone, the impact manager to acquire sound data from the microphone during an impact time segment.

14. The system of claim 1, further comprising a conversion manager to convert and weight physical signal properties, including assignment of truth values to objective properties, and assignment of a weight to each objective property.

15. The system of claim 1, further comprising a secondary sensor attached to the body, the secondary sensor in the form of a proximity sensor, and the impact manager to detect if the secondary sensor was on a user during an impact time segment to differentiate a device drop from a human fall.

16. A method comprising:

providing a processing unit in communication with memory and a data structure to store sensory data, the processing unit to process the sensory data;

receiving the sensory data from a sensor attached to a body in communication with the processing unit, the sensory data including acceleration data;

detecting fall of the body in accordance with the acquired sensory data, including;

extracting orientation data from the acquired sensory data; and

quantitatively computing cumulative high frequency energy of impact data from the acquired sensory data; and

analyzing orientation data and impact data and detecting a fall, including differentiating the detected fall from an intentional change of posture, including weighting acceleration data extracted from the sensor, the extracted data including;

impact, orientation, and energy.

17. A computer program product for use with a sensor attached to a body, the sensor to acquire sensory data, the computer program product comprising a non-transitory computer readable storage medium having computer readable program code embodied thereon, which when executed causes a computer to implement the method comprising:

receiving sensory data from a sensor attached to a body, the sensory data including acceleration data;

detecting fall of the body in accordance with the acquired sensory data, including;

extracting orientation data from the acquired sensory data; and

quantitatively computing cumulative high frequency energy of impact data from the acquired sensory data; and

analyzing orientation data and impact data and detecting a fall, including differentiating the detected fall from an intentional change of posture, including weighting acceleration data extracted from the sensor, the extracted data including impact, orientation, and energy.

18. The computer program product of claim 17, further comprising learning a true vertical value of the sensor before impact, and to replace a prior vertical value of a before impact time segment with the true vertical value.

19. The computer program product of claim 17, further comprising computing and compressing statistical data over time received from the acquired sensory data, including a technique selected from the group consisting of:

a stepping window and a sliding window, wherein the stepping window technique extracts statistical properties of data within a stepping window and includes a processing of adjacent time segments to sample sensory data and the sliding window technique extracts statistical properties of data within a sliding window and includes an overlap of adjacent time segments to average features of the data over time.

20. The computer program product of claim 17, further comprising monitoring an impact of the body within an impact time segment, including monitoring for a parameter reflective of cumulative energy of a sensory signal within the time segment to exceed a threshold.

21. The computer program product of claim 17, wherein before and after the orientation data extractions are conducted from the sensory data before and after impact time segments, and the cumulative high frequency energy is computed during an impact time segment.