Health monitor wearable device
Disclosed are systems and methods for detecting a fall of a user of a wearable device. The described system samples signal data from an accelerometer of the wearable device and performs a preliminary determination of fall detection using pattern recognition. This preliminary detection may be based on whether a first portion of the signal data exceeds a first threshold indicative of a fall of a wearer of the electronic device and whether a second portion of the signal data is within a range indicative of motion stillness following the fall. The described system further applies a machine learning classifier to the signal data associated with the preliminary indication to generate a fall classification indicating the signal data represents a fall of the wearer of the electronic device.
1. An apparatus for detecting a fall of a wearer of the apparatus, comprising:
a wireless communication component;
at least one sensor comprising an accelerometer configured to measure motion of the apparatus;
a processor configured to:
sample signal data received from the accelerometer;
determine that a first portion of the signal data exceeds a first threshold indicative of the fall of the wearer of the apparatus and that a second portion of the signal data is within a range indicative of motion stillness following the fall, wherein the processor is further configured to ignore, based on a tolerance range, solely signal points that are not associated with a falling motion, exclusively between the first portion of the signal data and the second portion of the signal data, while still monitoring for a pattern of an impact followed by inactivity;
generate, in response to the determination, a preliminary indication of fall detection associated with the signal data;
apply, subsequent to generating the preliminary indication, a machine learning classifier to the signal data associated with the preliminary indication to generate a fall classification confirming that the signal data represents the fall of the wearer of the apparatus;
activate an alarm of the apparatus in response to a generated fall classification;
in response to not receiving a button input of the apparatus within a first threshold period of time from the alarm activation:
transmit to a second party associated with the wearer, via the wireless communication component of the apparatus, a notification that the wearer of the apparatus has fallen; and
in response to receiving the button input of the apparatus within the first threshold period of time from the alarm activation:
refrain from transmitting the notification to the second party; and
train the machine learning classifier via feedback indicating that the signal data does not represent the fall of the wearer.
2. The apparatus of claim 1 , wherein the first portion of the signal data comprises a first channel of signal data from the accelerometer during a first time period, and the second portion of the signal data comprises a second channel of signal data from the accelerometer during a second time period subsequent to the first time period.
3. The apparatus of claim 1 , further comprising a heart rate sensor, and wherein the processor is further configured to:
determine a heart rate of the wearer of the apparatus, by:
sampling heart-rate signal data from the heart rate sensor at a first frequency, wherein the heart-rate signal data represents a level of light intensity per signal point;
determining signal points of decreased measurements in the heart-rate signal data;
calculating an average amount of signal points between the signals points of decreased measurements; and
determining the heart rate of the wearer of the apparatus as a ratio of the first frequency of the heart rate sensor to the average amount of signal points.
4. The apparatus of claim 3 , wherein the processor is further configured to:
determine an abnormal heart rate of the wearer of the apparatus based on a comparison of the heart rate to a stored previously-measured heart rate; and
transmit to the second party, via the wireless communication component of the apparatus, a notification of an abnormal heart beat.
5. The apparatus of claim 1 , wherein the wireless communications component comprises a cellular data transceiver and a Bluetooth Low Energy data transceiver.
6. The apparatus of claim 1 , wherein the processor is further configured to generate the preliminary indication in response to determining that a time difference between when the first portion of the signal data ends and when the second portion of the signal data begins is less than a predetermined wait time.
7. A computer-implemented method for fall detection of a wearer of an electronic device, comprising:
sampling signal data received from an accelerometer of the electronic device;
determining that a first portion of the signal data exceeds a first threshold indicative of a fall of the wearer of the electronic device and that a second portion of the signal data is within a range indicative of motion stillness following the fall, wherein the determining further comprises ignoring, based on a tolerance range, solely signal points that are not associated with a falling motion, exclusively between the first portion of the signal data and the second portion of the signal data, while still monitoring for a pattern of an impact followed by inactivity;
generating, in response to the determination, a preliminary indication of fall detection associated with the signal data;
applying, subsequent to generating the preliminary indication, a machine learning classifier to the signal data associated with the preliminary indication to generate a fall classification confirming that the signal data represents the fall of the wearer of the electronic device;
activating an alarm of the electronic device in response to a generated fall classification;
in response to not receiving a button input of the electronic device within a first threshold period of time from the alarm activation:
transmitting to a second party associated with the wearer, via a wireless communication component of the electronic device, a notification that the wearer of the electronic device has fallen; and
in response to receiving the button input of the electronic device within the first threshold period of time from the alarm activation:
refraining from transmitting the notification to the second party; and
training the machine learning classifier via feedback indicating that the signal data does not represent the fall of the wearer.
8. The computer-implemented method of claim 7 , wherein the first portion of the signal data comprises a first channel of signal data from the accelerometer during a first time period, and the second portion of the signal data comprises a second channel of signal data from the accelerometer during a second time period subsequent to the first time period.
9. The computer-implemented method of claim 7 , wherein the electronic device comprises a heart rate sensor, further comprising:
determining a heart rate of the wearer of the electronic device, by:
sampling heart-rate signal data from the heart rate sensor at a first frequency, wherein the heart-rate signal data represents a level of light intensity per signal point;
determining signal points of decreased measurements in the heart-rate signal data;
calculating an average amount of signal points between the signals points of decreased measurements; and
determining the heart rate of the wearer of the electronic device as a ratio of the first frequency of the heart rate sensor to the average amount of signal points.
10. The computer-implemented method of claim 9 , further comprising:
determining an abnormal heart rate of the wearer of the electronic device based on a comparison of the heart rate to a stored previously-measured heart rate; and
transmitting to the second party, via the wireless communication component of the electronic device, a notification of an abnormal heart beat.
11. The computer-implemented method of claim 7 , wherein the wireless communications component comprises a cellular data transceiver and a Bluetooth Low Energy data transceiver.
12. A non-transitory computer readable medium comprising computer executable instructions for fall detection of a wearer of an electronic device, the instructions causing the electronic device to perform steps comprising:
sampling signal data received from an accelerometer of the electronic device;
determining that a first portion of the signal data exceeds a first threshold indicative of a fall of the wearer of the electronic device and that a second portion of the signal data is within a range indicative of motion stillness following the fall, wherein the determining further comprises ignoring, based on a tolerance range, solely signal points that are not associated with a falling motion, exclusively between the first portion of the signal data and the second portion of the signal data, while still monitoring for a pattern of an impact followed by inactivity;
generating, in response to the determination, a preliminary indication of fall detection associated with the signal data;
applying, subsequent to generating the preliminary indication, a machine learning classifier to the signal data associated with the preliminary indication to generate a fall classification confirming that the signal data represents the fall of the wearer of the electronic device;
activating an alarm of the electronic device in response to a generated fall classification;
in response to not receiving a button input of the electronic device within a first threshold period of time from the alarm activation:
transmitting to a second party associated with the wearer, via a wireless communication component of the electronic device, a notification that the wearer of the electronic device has fallen; and
in response to receiving the button input of the electronic device within the first threshold period of time from the alarm activation:
refraining from transmitting the notification to the second party; and
training the machine learning classifier via feedback indicating that the signal data does not represent the fall of the wearer.
13. The non-transitory computer readable medium of claim 12 , wherein the first portion of the signal data comprises a first channel of signal data from the accelerometer during a first time period, and the second portion of the signal data comprises a second channel of signal data from the accelerometer during a second time period subsequent to the first time period.
14. The non-transitory computer readable medium of claim 12 , wherein the electronic device comprises a heart rate sensor, further comprising instructions for:
determining a heart rate of the wearer of the electronic device, by:
sampling heart-rate signal data from the heart rate sensor at a first frequency, wherein the heart-rate signal data represents a level of light intensity per signal point;
determining signal points of decreased measurements in the heart-rate signal data;
calculating an average amount of signal points between the signals points of decreased measurements; and
determining the heart rate of the wearer of the electronic device as a ratio of the first frequency of the heart rate sensor to the average amount of signal points.
15. The non-transitory computer readable medium of claim 14 , further comprising instructions for:
determining an abnormal heart rate of the wearer of the electronic device based on a comparison of the heart rate to a stored previously-measured heart rate; and
transmitting to the second party, via the wireless communication component of the electronic device, a notification of an abnormal heart beat.
16. The non-transitory computer readable medium of claim 12 , wherein the wireless communications component comprises a cellular data transceiver and a Bluetooth Low Energy data transceiver.