They are trying to profile what happens just before a fall--not during it. When a physical therapist observes an elderly person getting "wobbly" they will often say to them "stop, and grab onto something or sit down." By using software analytics to recognize this "wobbly" situation, they hope to use an alarm of even a voice synthesizer to fill-in for the human physical therapist.
The technology for using accelerometers to safely park a hard drive head before it hits the floor is well established. It would seem that initially the technology could be implemented as an alert to caregivers that a fall has happened. The more ambitious goal is to determine how alerting the individual can prevent a fall. If someone is dozing off and starting to topple over, a sharp alarm may enable them to catch themselves. The challenge is likely to be that falls have different root causes - which require different corrective actions.
Many research efforts have attempted to detect falls, but none has been successful enough for widespread commercialization. TI and Texas Tech, if successful, will not only detect falls, but will actually attempt to prevent them, which I believe everyone with an elderly relate will appreciate.
I am eagerly awaiting for such a device to be available. It is almost everybody experience with their elderly people in the family. I myself have seen my father-in-law who is now 86, suffering from such unexpected falls and the resultant injuries. Especially bathrooms where normally such elderly people are susceptible for such falls are places where the caretaker is unlikely to be around . And especially in homes where 24 hours monitoring of such persons is not possible , such kind of device will really be a boon.