Skin BioDynaMo (SkiBiDy)

SkiBiDy is a hybrid agent-continuum simulation of skin tissue biology built on BioDynaMo. It models wound healing, immune response, fibroblast and collagen dynamics, scar formation, vascular perfusion, hemostasis, cellular senescence, cutaneous innervation, tumor growth, diabetic impairment, and therapeutic interventions, all from mechanistic first principles backed by DOI-linked source papers across 29 modules.

How it works

Healthy skin runs as a composite field coupling 41 diffusion grids with no agents. When an event occurs (wound, infection, tumor), cells spawn from local field state, interact with the fields, and dissolve back once stable. This is the UWYN (Use What You Need) paradigm: the simulation only creates agents where the biology demands cellular resolution.

Corneum    [continuum]  barrier, desquamation
Granulosum [continuum]  keratohyalin, tight junctions
Spinosum   [continuum]  desmosomes; agents on event
Basale     [continuum]  stem/TA cells on event
---------- basement membrane ----------
Dermis     [continuum]  vasculature, O2/KGF, collagen, nerves; agents on event

Quick start

source <path_to_biodynamo>/bin/thisbdm.sh
./run.sh                                     # interactive menu
./run.sh --study=wound                       # punch biopsy wound healing
./run.sh --study=diabetic-wound              # chronic diabetic ulcer
./run.sh --study=tumor                       # basal cell carcinoma growth
./run.sh --compare                           # normal vs diabetic side-by-side
./tests/test.sh                              # 130 unit tests

Modules

29 modules, each self-contained in modules/ with its own config, source, data, and README. See docs/ for the full module index with biology, parameters, coupling, and validation.

Module	Default	Module	Default
tissue	on	dermis	on
wound	on	elastin	off
immune	on	hyaluronan	off
inflammation	on	glucose	on
fibroblast	on	temperature	on
scar	on	lactate	on
mmp	on	nitric_oxide	on
fibronectin	on	ph	off
angiogenesis	on	hemostasis	off
perfusion	on	biofilm	off
diabetic	off	tumor	off
senescence	on	neuropathy	on
ros	on	bioelectric	on
lymphatic	on	mechanotransduction	on
body_site	off

Configuration

Config is layered TOML, merged at runtime:

bdm.core.toml              core tissue params
  + modules/*/config.toml   29 module configs (auto-merged)
  + profiles/*.toml          skin phenotype overlay
  + studies/*/preset.toml    study experiment overlay
  + studies/*/treatments/*.toml  therapeutic intervention overlay
  = bdm.toml                 runtime config (gitignored)

Skin profiles: normal (default), aged, diabetic, aged_diabetic, rheumatoid. Studies: wound, diabetic-wound, tumor, tumor-wound, rheumatoid, baseline.

Treatment study

The diabetic treatment study compares 8 therapeutic interventions against an untreated diabetic baseline:

# Single treatment schedules (NPWT/HBO/MSC at different start days)
python3 scripts/study/gen_treatment_schedule.py --treatments npwt,hbo,msc --days 0,7,14,21

# Pairwise combinatorial scheduling
python3 scripts/study/gen_treatment_schedule.py --treatments npwt,hbo,msc --days 0,7,14,21 --combo

# Run all generated schedules
python3 studies/run_experiments.py studies/diabetic-wound/experiments/schedule_*.toml

# Full comparison study (all 8 treatments + baseline)
python3 studies/diabetic-wound/treatment.py --combos all

Available treatments: anti-inflammatory, HBO, NPWT, doxycycline, growth factor (PDGF-BB), MSC therapy, moisture dressings, senolytic (dasatinib+quercetin), and a rational combination targeting all four dysfunction axes. See docs/treatments.md for mechanisms and references.

Rheumatoid arthritis study

The RA study models synovial inflammation and cartilage erosion with a study-scoped TNF-alpha/IL-6 dual cytokine module:

# RA baseline (untreated, 10-run consensus)
python3 batch/batch.py -n 10 --skin rheumatoid --study rheumatoid

# Treatment comparison (baseline + 4 biologics)
python3 studies/rheumatoid/study.py

# Biologic comparison experiment
python3 studies/run_experiments.py studies/rheumatoid/experiments/biologic_comparison.toml

Available RA treatments: anti-TNF (infliximab/adalimumab), tocilizumab (anti-IL-6R), methotrexate (DMARD), and triple combination. See studies/rheumatoid/ for details.

Validation

Validated against published literature across 12 observables with per-module SOURCES.yaml files providing DOI-linked citations. Four optional mechanistic toggles (gradient-driven recruitment, efferocytosis M1 to M2, constitutive + TGF-beta collagen, HIF-1alpha VEGF) can replace parametric models for biophysical fidelity testing.

python3 batch/batch.py -n 10 --study wound --validate          # parametric (default)
python3 batch/batch.py -n 10 --study full-model --validate      # mechanistic toggles

Documentation

Document	Description
Docs index	Module index, metrics columns, validation overview
Guide	Configuration, skin profiles, studies, running, visualization
Architecture	UWYN hybrid agent-continuum design, CompositeField, field coupling
Parameters	Parameter index with module links and config layering
Treatments	8 therapeutic interventions for diabetic wounds
Studies	Packaged studies, experiments, example outputs
Batch	Multi-run consensus, parameter sweeps
Literature	Validation framework, RMSE dashboard, reference data

Third-party

Library	Version	License
toml++	3.4.0	MIT
BioDynaMo	1.04+ (master)	Apache 2.0 (linked, not bundled)

License

Apache 2.0