Proposed hallsim framework for composing aging models and testing interventions in silico
The authors propose hallsim, a software framework meant to let humans and AI agents combine small mechanistic aging models, recalibrate them against data and test perturbations in simulation; they demonstrate it on three published kinetic models. [1]
Automatic summary · not editorially reviewed
The key findings
The authors propose hallsim as a compositional simulation framework intended as a substrate for agent-orchestrated aging research. [1]
Three independently published kinetic models covering genomic instability, nutrient sensing and proteostasis were co-simulated and perturbed in silico. [1]
A Neural ODE surrogate was trained and composed with the mechanistic modules in a hybrid composite. [1]
The source at a glance
- Study type / format
- Abstract describing a proposed computational framework with demonstration
- Population / context
- Three published kinetic models of aging-related subsystems and a trained Neural ODE surrogate, in simulation
- Summary based on
- Study abstract
- Publication status
- Preprint · peer-review status not confirmed
- Original publication
- September 23, 2026
Proposed substrate for agent-built aging models
The abstract describes aging as a network-level phenomenon whose hallmarks interact through feedback loops across very different timescales. Decades of reductionist work yielded thousands of models of narrow subsystems, but no simple way to merge them, so whole-cell and multi-hallmark aging models remain rare, manually built and relatively small. The authors propose hallsim as a compositional simulation framework intended as the mechanistic substrate for research orchestrated by autonomous agents, with built-in validation checks and modeling guidance.
Supporting source [1]
Three kinetic models co-simulated and perturbed
The authors report co-simulation and in-silico perturbation of three independently published kinetic models covering genomic instability, nutrient sensing and proteostasis. These models are connected by four edges and calibrated against a public dataset. The framework is described as JAX-native, composable, end-to-end differentiable, GPU-executable and equipped with automated analysis and model-selection tools. A Neural ODE surrogate was also trained and combined with the mechanistic modules in a hybrid composite.
Supporting source [1]
Abstract-only scope of this proposal
This is an abstract of a proposed framework, not a validated product, and the full text was not reviewed. The demonstration covers three kinetic models and one surrogate; the abstract reports no lifespan, clinical or human outcome, and simulation results do not show that any intervention extends life. Whether the framework performs as described in wider use cannot be judged from the abstract alone.
Supporting source [1]
How to interpret this
Abstract only; the full text was not reviewed. The work is a proposed framework demonstrated on three kinetic models and one surrogate, with no reported lifespan, clinical or human outcome.
Based on the study abstract; the full paper, tables and supplementary data were not reviewed.
Source & supporting passages
Belic M, Furman D. · europepmc.org · September 23, 2026