Blood pressure and arterial compliance
Compare a cited lumped-parameter Physiome model in JSim with the browser pressure model, HumMod scenarios, and empirical reference ranges.
Integrating evidence, models, and clinical reasoning
Generative-Mechanism Research
Physiolog integrates findings from empirical physiological research into explicit, mechanism-based models. These models are tested across appropriate simulation environments and compared back against empirical evidence, then translated into transparent teaching tools and reproducible research models. The ultimate goal is to improve physiology learning and bring stronger mechanistic models into clinical reasoning through models4PT and the Clinical Inference Engine.
Research Pipeline
The workflow connects existing physiological evidence to explicit mechanisms, tests those mechanisms without privileging one engine, and returns their predictions to the evidence.
- 01
Integrate the evidence
Synthesize findings, theories, and measurements from existing physiological research.
- 02
State the mechanism
Express the proposed explanation as equations, assumptions, units, state variables, and observable predictions.
- 03
Implement the model
Use the model scale and simulation environments best suited to the mechanism and research question.
- 04
Compare simulations
Test behavior across versioned models, engines, scenarios, parameters, outputs, and tolerances.
- 05
Return to the evidence
Compare predictions back to empirical findings and investigate disagreement, uncertainty, and missing mechanisms.
- 06
Advance learning and reasoning
Translate supported mechanisms into teaching tools, models4PT research models, and the Clinical Inference Engine.
Model Ecosystem
These resources are complementary. Each contributes a different scale, capability, or source of evidence to the same research process.
| Resource | Role | Contribution |
|---|---|---|
| Physiolog | Learning and question formation | Small browser models make focused causal relationships visible and independently explainable. |
| Physiome Model Repository | Public model structures | Cited CellML models provide established mechanisms, model families, and parameter references. |
| JSim | Equation-based execution and analysis | Unit checking, parameter estimation, mixed equation systems, and CellML/SBML interchange support formal model testing. |
| HumMod | Whole-body comparison | Offline experiments examine integrative feedback, longer time scales, and cross-system responses. |
| Empirical evidence | Independent constraint | Broad PubMed searches will support evidence discovery and import, while experimental and clinical findings test whether modeled mechanisms remain plausible. |
| models4PT | Research destination | Reproducible physiological mechanisms and evidence support models for physical therapy reasoning. |
Current Status
In progress
JSim modernization
Reproducible builds and numerical-compatibility baselines must be completed before modernized JSim outputs are accepted as research baselines.
Available offline
HumMod 1.6.1 reference
The intact standalone model is available for whole-body experiments, with protocols, analysis, results, and provenance kept in a separate research repository.
Planned
Common experiment record
A shared comparison format will record model and engine versions, units, initial conditions, interventions, outputs, tolerances, and acceptance criteria.
First Validation Pilots
The initial studies test both the physiology and the reproducibility of the comparison workflow.
Orthostatic response
Compare focused circulation and autonomic models with intact HumMod, then identify the minimum mechanisms needed for a transparent teaching model.
Muscle or oxygen transport
Test the workflow outside lumped cardiovascular models and determine whether the comparison method generalizes across model families.
Research Standards
- 1No model or engine is treated as ground truth.
- 2Model and engine versions, architecture, units, assumptions, scenarios, and initial conditions are recorded.
- 3Solver settings, tolerances, outputs, and acceptance criteria remain part of the evidence.
- 4Disagreements are investigated and reported rather than averaged away or hidden.
- 5Model output is hypothesis-generating evidence, not a patient-specific prediction or clinical calculator.
- 6Only the minimum mechanism needed for a learning objective moves into a public teaching simulation.
Connected Programs
Physiolog Teaching Models
Transparent browser models that translate supported physiological mechanisms into focused learning experiences.
models4PT
Reproducible physiological and computational models intended to support physical therapy reasoning and research.
Clinical Inference Engine
A downstream framework for incorporating physiological mechanisms into diagnosis and treatment reasoning.
Collaborate
Collaboration is welcome on clinical physiology, model comparison, reproducible simulation, clinical inference, and educational model development. Please use the contact information provided on this site or review related work in Sean Collins's professional portfolio.