The investigators constructed a 12-state mechanistic ordinary differential equation (ODE) model describing calcium partitioning among bone, blood, urine and renal stone formation. The core model (M0) included 11 physiological parameters which were jointly calibrated to eight targets derived from International Space Station (ISS) data using Bayesian identification. An alternative model (M1) adding a direct galactic cosmic radiation (GCR) → bone coupling term was tested for parsimony but found to be unidentifiable at current ISS exposure levels; M0 was therefore retained for primary inference.
Calibration targets and priors were drawn from previously published peer-reviewed literature and NASA Human Research Program public reports cited by the authors. The M0 base model had a 19-dimensional state for the Bayesian identification exercise, and the calibrated posterior distributions were carried forward for scenario analysis.
The M0 posterior predictive distribution was propagated to four mission environments: the ISS, Lunar subsurface, Lunar surface, and Mars. Scenario inputs varied by mission duration, gravity (partial unloading), and cumulative GCR exposure profiles. The authors used a factorial analysis to quantify the relative contributions of duration, gravity, and GCR to outcome variance.
Modeled lumbar lower bone mineral density (BMD) loss increased with mission duration and with greater partial-gravity unloading. Representative outputs reported in the source include lumbar lower BMD change from a 180-day ISS mission (−4.83%) to a 730-day Mars mission (−12.15%). In a 2^3 factorial decomposition the effect of mission duration explained the largest share of variance (~82.9%), gravity explained ~12.5%, and the modeled direct GCR main effect was approximately zero.
These results support the expectation that cumulative unloading and longer exposures drive progressive BMD loss beyond six-month ISS data, motivating explicit extrapolation for deep-space mission planning.
Modeled renal stone incidence showed an inverse gradient relative to BMD loss: the ISS profile produced higher modeled stone rates than the Mars profile (ISS 16.1 vs Mars 13.1 per 1000 person-years as reported in the source). The authors interpret this pattern as a shift in the dominant mechanistic pathway for stone risk across environments. On the ISS the dominant pathway is bone-centric—bone resorption drives urinary chemistry changes that increase stone risk—whereas on Mars the dominant driver is kidney-centric, with residual urinary-chemistry alterations independent of ongoing bone resorption accounting for most of the modeled stone risk.
This environment-dependent pathway shift arises because partial-gravity bone resorption weakens in some deep-space profiles while residual urinary chemistry alterations persist.
The direct GCR → bone coupling term included in the M1 extension was tested for identifiability. At current modeled ISS GCR doses the coupling term was unidentifiable, with a Delta WAIC of +0.0076 ± 0.126 SE versus M0. Given lack of evidence for a stable identifiable effect in the available calibration targets, the authors retained M0 as the main inference model and reported sensitivity checks indicating that forcing GCR coupling terms to zero left composite tier counts unchanged in supplemental analyses.
The model was used to evaluate pharmaceutical countermeasures. Bisphosphonates were predicted to provide substantial protection of bone mineral density (≥84% BMD protection as reported). However, bisphosphonate monotherapy did not remove modeled residual urinary-chemistry changes associated with stone risk; therefore, relying on bisphosphonates alone would likely underestimate stone risk for Mars-duration missions.
Combination therapy including potassium–magnesium–citrate was simulated and reported to reduce the modeled renal stone risk further (reported relative risk reduction metric RRR_RSS 51% in the source). Based on these simulations the authors recommend adding potassium–magnesium–citrate combinations to deep-space countermeasure packages rather than relying solely on bone-protective agents.
Using composite risk tiers defined in the manuscript, a 365-day Lunar surface mission was identified as the earliest NASA-roadmap environment predicted to cross a composite RED threshold. The authors note that the shielded 180-day Lunar profile differs from the 365-day Lunar-surface profile both in cumulative GCR (~69-fold difference reported) and in duration (2× difference), so a shielding-specific effect could not be isolated with the profiles assessed. Forcing the GCR coupling terms to zero reportedly left composite tier assignments unchanged in supplemental analyses.
Independent hold-out validation was reported using a 60-day head-down tilt (HDT) bedrest randomized controlled trial (Culliton 2025) control arm (n=8). That hold-out supported the M0 posterior predictive distribution on the lumbar-BMD sub-scope. All observational calibration data were drawn from previously published literature and NASA Human Research Program public reports cited in the manuscript. The authors state that all data produced by the present study are available upon reasonable request.
Key limitations emphasized in the source include the unidentifiability of the direct GCR → bone coupling at current ISS doses and the inability to isolate a pure shielding effect between the compared Lunar profiles because of concurrent differences in duration and cumulative GCR. The authors explicitly retained the simpler M0 model for main inference due to these identifiability constraints.
The analysis projects an environment-dependent mechanistic shift whereby early or shorter partial-gravity missions (such as ISS-length missions) will remain primarily bone-centric with bone resorption driving urinary stone risk, while longer-duration deep-space missions (example: Mars) may become kidney-centric, driven by persistent urinary-chemistry changes even when bone resorption is relatively weaker. Countermeasure planning should therefore combine bone-protective agents (bisphosphonates) with urinary-chemistry-directed therapies (potassium–magnesium–citrate) for deep-space missions. The model-based predictions and their validation on BMD support these inferences, but the authors note remaining uncertainties tied to GCR effects and extrapolation beyond current ISS durations.