Automated Fall Detection For Elderly

For a research project in my school my friend and I are embarking to figure out a way to detect falls for the elderly without the need of a button which will also send a text notification to a care giver and possibly measure heart rate or other vital signs. We have all seen those life alert commercials with the "I've fallen and can't get uppppppppp" lady. Any coding suggestions on how to do this or anything that will help us is appreciated. We would like to know what other materials we will need to make the circuit. Thank you for your time and interest.
Materials:
Photon https://docs.particle.io/datasheets/photon-(wifi)/photon-datasheet/
Adafruit Triple-Axis Accelerometer - ±2/4/8g @ 14-bit - MMA8451 : ID 2019 : $7.95 : Adafruit Industries, Unique & fun DIY electronics and kits

Needed but don't know what to get:
Battery/shield
Board/circuit components

You should be aware that Particle and Arduino are not the same thing. Particle tried to copycat Arduino to cash in on their success but also made some completely unnecessary changes to intentionally cause code incompatibility.

General advice you get here regarding detecting a fall with an accelerometer will likely be just as applicable to Particle as Arduino but once you get to the details you should probably move on to the Particle forum and hope their community is even a fraction as good as Arduino's is.

extvn:
We have all seen those life alert commercials with the "I've fallen and can't get uppppppppp" lady.

Well, as a matter of fact, no idea what you're talking about here.

For your fall detector: first figure out what you want to detect as fall.
Movement? (going down fast - but not when sitting down a bit fast, or taking a big step or jumping down)
Shock? (from the impact of hitting the ground - but not from a jump, or bumping into a table, or walking steps, or sitting down a bit hard, or riding the bus on a bumpy road)
Orientation of the body after a fall? (horizontal - but not when taking a nap - and maybe the person fell in a sitting position)
You probably need a combination of the above, and maybe a few more parameters that I didn't think of.

When you got the fall detected reliably (detect every fall without failing, but no false positives as after 3, 4 false positives in a row the human responder will stop responding), then you can think about the rest. Battery power, GSM module, making it wearable, whatever. Those parts are dead easy compared with the fall detection itself.

You'll have to take 2 steps: first you find out which criteria allow to detect a fall, and in the next step find available sensors for measuring the identified effects.

There exist sensors (MPU6050...) which detect a free fall, typically a notebook falling from a table. Its questionable whether persons fall in just the way that is detected by such a sensor, this is what you have to find out yourself.

A single (knock...) sensor won't be enough for covering all possible parts of a body that may hit the floor in a fall. And I'm not available for a series of fall studies! :wink: