| Extensive research has been conducted about the balance limitations in persons who are at heightened risk of falling (those with balance impairments and the elderly). The criteria for when total loss of balance (i.e. falls) occur in normal, healthy adults is unknown. Past research often reports forces or rotation rates while holding other factors, such as time, constant. This makes it impossible to differentiate force from impulse or other parameters. This research had two goals. To determine what mechanical parameter is the best predictor of falling, and to model the chance of falling from rearward perturbations in healthy adults. It was hypothesized that impulse (the integral of force and duration) would be the best predictor.;One phase of the study used 20 volunteers (age 21.8+/-2.53 years) in a within subjects 2x3 factorial design of force (high, low) and duration (83, 250, 750 ms). This design included a 2x2 design of force and impulse (high, low). Trials were conducted in random order for each subject and each condition was repeated twice (12 trials). This experiment showed that the chance of falling increases when impulse is increased, even when force is held constant, and that the chance of falling does not increase for increased force when impulse is held constant. The impulse, relative to body mass, was the best single predictor of falls based on a binary logistic regression analysis. Sex was also an important factor in predicting falls. This model could be improved however if both impulse and the peak force are included. It is suspected that falling requires overcoming multiple balance mechanisms and some may be more or less time dependent than others making both force and impulse important considerations. A separate phase of the study used another cohort of 19 subjects (age 23.1+/-3.29 years) to collect additional data to build a predictive model. Each subject conducted twelve trials in random order with levels in the region of mixed responses. This response data was combined with the phase 2 data to build an improved model using binary logistic regression. Falls can be predicted from a fall factor based on the peak force, impulse, and body mass. |
