Strength Based Body Fat Estimation
Here’s the abstract for the body fat study. The full paper is coming soon.
Objective: To develop a method of body fat estimation specific to the population of experienced weight lifters as a function of age, gender, height, weight, and performance in the classic powerlifting exercises.
Design: A cross-sectional study in which anonymised, self-reported data were collected from a mobile fitness application and then cross-validated with self-reported data collected through a web application.
Subjects: A total of 5278 apparently healthy adults from a mixed non-randomly selected weight lifting population (4766 men and 512 women), varying in age from 18 to 63 y, 1.32-2.03 m height, 47-158 kg weight, 5-60% fat, 23-243 kg bench press 1RM, 25-359 kg squat 1RM, and 25-346 kg deadlift 1RM.
Results: Six variables (age, gender, height, weight, bench press 1RM and squat 1RM) were significantly associated with body fat via univariate analysis. Using these variables, a feed-forward multilayer perceptron (MLP) trained by back-propagation predicted body fat in strong agreement with reported DEXA values (r=0.84).
Conclusions: Body fat estimation requiring no special instrumentation could be of practical value for the population of experienced weight lifters using the classic powerlifting exercises.