The ISS's ARED replaces free weights (useless in microgravity) with two vacuum cylinders. Each cylinder is evacuated to a low pressure Pv; atmospheric pressure Patm pushes the piston back in, and that push is the resistance you lift against. As the piston extends by x, the sealed volume grows and — by Boyle's Law, PV = const at fixed temperature — the internal pressure drops, so the load rises through the stroke:
P(x) = P_v0 · V0 / (V0 + A·x)
F(x) = n · (P_atm − P(x)) · A
with n = engaged cylinders (1 or 2), A the piston area, and V0/A ≈ 0.5 m the stroke length. This is why ARED's resistance curve rises through a rep rather than staying flat like a barbell, and why it can reach ≈2,700 N (600 lbf) per side — enough to load bone as well as muscle.
The right-hand readouts use a dose-response model built from measured ISS crew outcomes: unexercised astronauts lose roughly 1–1.5%/month of muscle mass and 1–1.5%/month of bone mineral density in microgravity. Define a dose fraction relative to the full protocol (2.5 h/day, 6 days/week ≈ 900 min/week):
dose = (duration × frequency) / 900 (capped)
load = piston peak force / 2,700 N (capped)
muscleTrend = −1.5 · (1 − 0.85 · dose) %/month
boneTrend = −1.3 · (1 − 0.60 · dose · load) %/month
Muscle responds mainly to time-under-tension, so it improves with dose alone; bone remodeling needs mechanical load as well, so a low vacuum setting still leaves bone loss closer to the unexercised baseline even at full duration — matching why real ARED protocols specify near-maximum resistance, not just time on the machine.